ML20133C782

Impact of TMI Nuclear Accident Upon Pregnancy Outcome, Congential Hypothyroidism & Infant Mortality. Reprints of Medical Journal Article Excerpts Encl

ML20133C782
Person / Time
Site:	Crane
Issue date:	12/31/1981
From:	Tokuhata G PENNSYLVANIA, COMMONWEALTH OF
To:
Shared Package
ML20132B167	List: ML20132B163 ML20133C709 ML20133C724 ML20133C733 ML20133C747 ML20133C751 ML20133C767 ML20133C782 ML20133C790 ML20133D391 ML20133D591 ML20133D652 ML20133E380
References
FOIA-85-285 NUDOCS 8507200566
Download: ML20133C782 (100)

Text

{{#Wiki_filter:';~3s jy= + Ics-sel =a= Q' r-n,.,... a g e.. 3

9.,_

ic Cf, s n. in - .1, IMPACT OF TMl NUCLEAR ACCIDENT UPON PREGNANCY OUTCOME, k CONGENITAL HYPOTHYROIDISM AND INFANT MORTALITY * ~ George K. Tokuhata, Dr.P.H., Ph.D. Director Division of Epidemiological Research Bureau of Epidemiology and Disease Prevention Pennsylvania Department of Health and Professor of Epidemiology and Biostatistics (Adjunct) Graduate School of Public Health University of Pittsburgh 1 l.

• . f-Chapter prepared for " Energy, Enr/ronment and the Economy"publiehod by the Pennsylvenle Acceemy of Science.

Y \\ R 68 \\ l-8507200566 850524 f PDR FOIA pop LODGEOS-PRS c1 x._ b g ^ ._ 1..;.

O IMPACT OF TMl NUCLEAR ACCIDENT UPON PREGNANCY OUTCOME, CONGENITAL HYPOTHYROIDISM AND INFANT MORTALITY The Three Mile Island nuclear accident of March 28,1979 has resulted in marked social unrest locally, nat'ionally and world wide, particularly with respect to the health and safety aspects of nuclear energy. Sub-sequent to the accident the Pennsylvania Department of IIealth initiated a comprehensive evaluation of possible health effects of the accident upon local population. During the 10-day period of crisis, it was not possible to ascertain accurate information regarding radioactive emissions from the damaged nuclear reactor into the environment. However, the presence of rather diffuse and growing psychological disturbance in the area was apparent. Within a short period of days following the accident, we were able to conceptualize and develop a multidisciplinary plan for a variety of different research studies specifically designed to assess the health im-pact of the TMI accident. Studies conceived during this critical period mostly reflected the existing epidemi-ological knowledge regarding biological effects of low level ionizing radiation and severe emotional stress. I was designated by the Governor of Pennsylvania to coordinate and manage all health related research activities relative to TMI. At the same time, a special Advisory Panel was commissioned by the Secretary of Health to oversee and guide all TMI related studies administered by the Department of IIealth. A. PREGNANCY OUTCOME AROUND THREE MILE ISLAND One of the most important studies developed shortly after the accident was to determine if the TMI nuclear accident has had any measurable impacts upon pregnancy outcome and infant health in the vicinity of the damaged nuclear reactor. We knew that both ionizing radiation and emotional stress can affect human reproductive process and pregnancy outcome. We also recognized that the embryo and the fetus are highly sensitive to such environmental insults, depending upon their severity, the mode of exposure, and the gestational age when exposed. Before describing the methodology and study design in detail, let me review briefly the current state of epidemiology of pregnancy outcome, particularly in relation to radiation and stress. I

Radiation and Pregnancy Outeoase: Much information is now available regarding the effects of ionizing radiation on the embryo and fetus. Most of the more reliable data are derived from animal experiment; however, certain experimental findings may be applicable to the humans, at least in a qualitative sense, while recognising inherent limitations or difficulties in such cross species inferences. The most significant damage from exposure to ionizing radiation results from the direct inter. action of the stream of ions produced by radiation with the nucleus of the irradiated cells. The cell may be killed, the radiation may produce no damage, or such damage may be repaired. There is another type of damage which is probably the most significant one,i.e., the damaged cell sur-vives and reproduces a clone of abnormal cells which may result in malignancies or congenital anomalies. Possible effects of radiation on pregnancy outcome are (a) intrauterine and extrauterine growth retardation, (b) embryonic, fetal or neonatal death, and (c) grou congenital malformations. The tissue (organ) most readily and consistently affected by radiation is the centrol nervous system. Laboratory and clinical studies by and large support the contention that doses of radiation less than 10 rada do not contribute to intrauterine or extrauterine growth retardation or to gross congenital malformations (1). Distribution of the absorbed dose from X rays or gamma rays externally exposed is considered to be rather uniform in the developing embryo or fetus; thus, a child with multiple radiation induced malformations is also likely to have intrauterine growth retardation and some CNS abnormalities. To determine the effect of radiation upon pregnancy outcome, one must consider (a) the ab-sorbed dose, (b) the dose rate (acute or chronic; continuous or intermittent), (c) the stage of, gestation at which the exposure occurred, (d) the age of the mother when conceived, and (e) the health condition of the mother, in general. If the dose rate is reduced significantly, the damaged cell may recover from it in time. The pre implanted stage of the embryo is the most sensitive to lethal effects of radiation. Embryos destroyed at this stage of pregnancy may never be recognized or recorded. However, preplantation irradiation has no apparent relationship to teratogenesis. Radiation has its greatest effectiveness in producing con 7enital malformations during the organo-l genesis period. In humans, this corresponds to the 14th 19th day of gestation (2). The peak incidence of grcss malformations occurs when. the fetus is irradiated during the early organogenesis period, although cellular, tissue and organ hypoplasia, including growth retarda. j tion can be produced by radiation throughout organogenesis, and fetal and neonatal periods, if ( the dose is high enough. These are usually limited to CNS abnormalities and other organs, which l continue to differentiate throughout gestation. Thus, cerebral hypoplasia, microcephaly, cerebel. I tar hypoplasia, and testicular atrophy can be produced by "high" doses at specific stages of gesta-l tion. 2

A number of studies'suggest that " low" levels of radiation with less than 10 rads of acute or chronic exposure may produce some pathologic effects in the embryo (3),(4), (5),(6), but these minor effects may be subtle and thus difficult to detect. For this reason, the National Council on Radiation Protection and Measurements has established the maximum permissible dose (MPD) to the fetus from occupational exposure of the expectant mother well below the known teratogenic dose. The neonatal death rate is highest in the surviving embryos irradiated during the early organogenesis period. Radioactive isotopes administered internrJ1y to the pregnant woman have a variable distribution in the embryo and fetus depending upon (a) the stage of gestation, (b) whether the radioactive material crosses the placenta, and (c) the biochemical affinities of the type of radiation emitted (alpha, beta, or gamma). Thus, the evaluation of the relative risk is much more complex and dif-ficult for radiations absorbed from internally administered radioactive materials than for radia-tion delivered from external X rays or gamma ray sources. It is generally assumed that embryonic germ cells are susceptible to the mutagenic effects of radiation throughout gestation. However, there is some uncertainty as to whether " low" doses below 10 P or low dose rates can produce significant cytogenetic defects. There has been no con. vincing evidence that cytogenetic (chromosome) abnormalities as such caused by radiation in utero have caused any significant increase in the incidence of clinical diseases. There is no doubt that "high" doses of radiation can be carcinogenic, flowever, whether or not " low" doses, such as below 2 reds, can induce leukemia and/or other malignant tumors in the humans has been debated by some epidemiologists and radiation biologists. It appears improbable that radioactive fallout as reported in the past or natural background radiation as such, signifi-cantly affects the incidence of congenital malformations, growth retardation or fetal death. The exact nature and extent of damages caused by " low" doses of radiation upon humans are still unknown. Ilowever, if the cell nucleus is damaged by radiation and some genetic materials (DNA) are lost or impaired, one may not conclude that the risk is zero. It is logical to assume then, that there is no threshold in radiation effect which may increase more or less with the in-crease or accumulation of exposure, liowever, the problem we are facing today is that anomalies caused by such " low" doses of radiation,if any, may not be detectable with the existing method of epidemiologic inquiry. It is also important to recognize that not all persons run the same risk of developing a malignancy or other abnormalities from a given radiation exposure. These variations depend upon individual genetic. constitutional makeups, as well as different individual experiences and environmental exposures. 3

i Stress and Pregnancy Outcome: Stress or psychoemotional disturbance is considered by many researchers as a precursor to disease. There are a number of studies in humans which have found an association between prenatal anxiety / stress and gestational, perinatal and developmental pathology. While some of these stud-les seem to have methodological flaws, several have found a significant relationship to either complications of pregnancy (7), (8), or to infant growth and development (9). Nuckolls, in particular, studied the effect of " social support" upon pregnancy outcome (10). Women with a high number of " psychological assets" had one third the pregnancy and perinatal complication rate of women whose " psychological assets" were low. Newton (11)in a retrospective study of postpartum women showed that pregnancies terminating in premature labor were more likely to have been stressful. In terms of the average number of life events per pregnancy, it was clear that the more premature the onset of labor, the higher the level of psychological stress was likely to be. The groups were matched for age, gravidity, and parity. The results of the study were independent of socioeconomic levels. 1 The findings of the studies cited above suggest a number of practical and scientific questions to I be addressed within the context of TMI Health Effect Research Program. The first, and most l obvious question, is whether or not the local population, including pregnant women, as a whole experienced any detectable stress effects. Previous studies of stress and pregnancy complications have found relationships which are either relatively weak (12) or restricted to subgroup of the f overall study population. A arcond question concerns factors which render individual women, particularly vulnerable to stress effects. As reviewed earlier, stress may be associated with morbidity only in the absence of supportive interpersonal relations. This obaervation is in accord with other studies of stress and 111nese(13), as well as with Burchfield's(14) argument that a maladaptive response to stress is i atypical, and likely to occur only when adequate coping resources are unavailable. An assessment of the role of social support, as well as other possible mediating factors, would contribute to the study of stress as a scientific concept and provide information as to which seg. ments of the pregnant population might be at risk for stress. induced morbidity. While specific mechanism of stress induced morbidity is not yet fully understood, there may be several different explanations with respect to pregnancy outcome; e.g., stress anxiety induced changes (a) in maternal behavior, such as increased smokirg, drinking or medication during preg. nancy (b) in obstetric practice, such as increased prescription of analgesics and psychotropic drugs or use of special procedures, (c) in maternal. infant bonding and child. rearing practices, and (d) in the hypothalamic. adrenocortical mechanisms (15). 4

l: A carefully designed retrospective cohort study of Pregnancy Outcome was initiated in August,1979 following four months of preparation. This study covered all pregnant women residing within a 20 mile radius of the TMI, who gave births during a one year period from March 28,1979 through March 27,1980. This study cohort consisting of approximately 4,000 deliveries will be com-pared with a control cohort of another 4,000 deliveries during a one year period immediately follow-ing the study cohort in the same geographic area. The study cohort will also be compared with similar data collected in the same general area during the immediately preceding four year period. This design will make it possible to compare pregnancy outcome measures among three cohorts, one study group and before and after control groups. Measures of adverse pregnancy outcome being investigated are: fetal deaths (stillbirths with and without abortions of 16 week or more gestation) as expressed per 1,000 deliveries, neonatal deaths (deaths within 28 days postpartum) as expressed per 1,000 live births, hebdomadal deaths (deaths within seven days postpartum) as expressed per 1,000 live births, perinatal deaths (combined measure of fetal and neonatal deaths) as expressed per 1,000 deliveries, prematurity (gestation less than 37 weeks) as expressed in percent, immaturity (birth weight less than 2,500 grams) as expressed in percent, congenital malformations (one or more defects observed at birth) as expressed in percent, and low Apgar score (less than seven at one minute of delivery) as expressed in percent. As indicated earlier, the main objective of the present investigation is to determine if the TMI nuclear accident has had a measurable effect on pregnancy outcome. liowever, there are numer-ous factors other than radiation and stress that are known or suspected to influence the course of preg-nancy and fetal outcome. In order to delineate the effect of the TMI accident, other known influences must be taken into account. This necessitated ascertainment of the appropriate data on a large number of variables pertaining to the pregnant women themselves and the surrounding complex environment to which they have been exposed. The maternalfactors considered in this study include: soclodemographic characteristics, such as race, age, education, occupation, employment, marital status, religion and residence; behavioralat-tributes such'as smoking, drinking, and birth control practice; and medical-obstetric histories, such as-diabetes, hypertension, thyroid disease, obesity, previous abortions miscarriages, previous fetal deaths, prematurity / immaturity, congenital malformations, and gravidity /index birth order. The provider factors that were taken into account are: medical specialty of the attending physician (obstetrician; general family practitioner; osteopath; etc.); type of practice (solo vs. group); and prenatal care (initiation of medical care, frequency of visits, special procedures or tests done,in-structions given, medications administered, X ray exposures, etc.). Afaternal stress during the index pregnancy is being measured by overt personal statements of " anxiety fear" as experienced by individual women during the crisis, as well as by actual stress-coping patterns, such as taking tranquilizers and sleeping pills. 3faternalradiation exposure during the 10-day crisis following the nuclear accident is being estimated by the Department of Radiation llealth of the University of Pittsburgh Graduate School of Public licalth. For this purpose, all available and reliable radiation source data compiled by various agencies are being reviewed carefully, consolidated and computer analyzed on the digitized electronic maps with respect to distance and direction of each I pregnant woman from the Three Mile Island. Also, added to this body of data is detailed data relative to individual whereabouts during the 10-day period so that more accurate radiation exposure can be

estimated. Eventually, two series of dose estimates on an individual basis will be established, namely, maximum possible done and most likely dose. These radiation dose estimates, together with the measures of psychological stress will be re-lated to each of the eight pregnancy outcome measures, while holding constant influences of all other factors considered in the study. The impact of the TMI nuclear accident will be assessed in terms of both radiation and stress combined, as well as each factor considered independently. Since the level of radiation exposure is considered to be very low and thus no major radia-tion effect upon pregnancy outcome is expected. On the other hand, it is possible that some measur-able radiation effects might be detected if the originally " reported" radiation dose data were signifi. cantly underestimated. Our ability to detect relatively small differences, if any,in the incidence of ad-verse consequences attributable to the nuclear accident will heavily depend upon how well we can con-trol (take into account) the influences of all the other factors, some of which are known to be much more important than low level radiation and/or psychological stress being investigated. B. CONGENITAL HYPOTHYROIDISM One of the important radioactive releases during the TMI nuclear accident was that of It 31. Since it is 131 known that radioactive iodine can cause hypothyroidism and that 1 can be taken up by pregnant women in the vicinity of TMI which, in turn, absorbed by the fetal thyroid gland through placenta after the 10th week of gestation, we decided to examine the incidence of congenitalhypothyroidism among newborn infants. The fetal thyroid gland is much more unsitive to radioactive iodine than is the mother's thyroid gland (fetal thyroid affinity for iodine is greater than maternal thyroid affinity); i.e., a relatively small dose to the mother can be a relatively large dose to the fetus. Beginning in July 1978, all children born alive in Pennsylvania are required to be screened for hypo-thyroidism, the condition characterized by lack of or insufficient level of thyroid hormone in the in-fant's blood. The purpose of this screening program is to find newborns with metabolic defects early, enough after birth to prevent mental retardation. The thyroid screening procedure involves testing for " low" thyroxine (T4) and "high" pituitary thy. roid stimulating hormone (TSil). Confirmation of diagnosis is done through thyroid scan, which can help determine various types of the abnormality. During the initial six. month period, testing proce. dures were not fully standardized and the results were not considered to be complete. There are several different diagnostic classes in congenital hypothyroidism: (a) agenesis (absence of the thyroid gland), (b) dysgenesis or ectopic type (incomplete maturation and/or displacement of the thy. roid gland from the normal position), (c) dyshormonogenesis or genetic type (lack of enzyme neces-sary to syr.thesize thyroxine and/or difficulty in the release mechanism of thyroxine; the condition usually inherited from the parents es an autosomal recessive trait) and (d)other types (abnormalities caused by environmental agents). In a normal population, the incidence of congenital neonatal hypo-thyroidism is in a range of one in 4,500 to 5,000 live births. 6

During the March 28,1979 March 27,1980 period only one case of congenital hypothyroidism was identified within a ten mile radius of TMI among approximately 4,000 newborn infants. This inci-dence rate is well within a normal range of expectation. The Statewide incidence of congenital hypothyroidism for 1979 (12 month period) was one per 4,600 live births, which is also within a normal range of expectation. The rate for 1978 (only the lat-ter six month period) was considerably lower; this was expected because of the fact that the thyroid screening program in Pennsylvania was started in July 1978 and that during this start up period the screening procedures and standards were not yet fully established, making data unsuitable for compari-son. The Statewide incidence for 1980 was one per 4,427 live births, again indicating that the level of congenital hypothyroidism for Pennsylvania as a whole remained within a normal range. An apparent clustering of seven cases of congenital hypothyroidism in Lancaster County during 1979 was subjected to a specialin depth analysis and investigation because of physical proximity and timing of the Three Mile Island nuclear accident. The following diagnostic and epidemiological features are of interest: (a) One of the seven cases identified was reported in January of 1979, prior to the TMI ac-cident, thus has no connection with radioactive iodine released from the damaged nuclear reactor. (b) One with severe multiple central nervous system anomalies was born three months after the accident; this case is unlikely to be associated with TMI accident because of the late gestation period of the fetus when the nuclear accident occurred (most, if not all, of these defects would have come about prior to the TMI accident) and also of coexisting developmental anomalies which are unlikely to be as-sociated with radiation. (c) One case was of dysgenesis, representing one of discordant Amish twins, thus, non. supportive of the etiology secondary to radiation exposure. (d) Another case of dysgenesis in whom the thyroid glands were displaced from the normal position. (e) One case of dyshormono-genesis from an Amish family where the condition (lack of enzyme to synthesize thyroxine) was in-herited from the parents. (f) For the remaining two cases no thyroid scan was conducted. ( Having completed detailed diagnostic analysis and epidemiological assessment of all the cases reported ) in Lancaster during 1979, we concluded that cases of congenital hypothyroidism were not related to the TMI nuclear accident. Except for the two cases for which diagnostic scan was not performed (un-known type), these types of anomalies are not expected to result from direct or indirect exposure of the fetus to radiolodine. This conclusion was also supported by an independent !!ypothyroidism Investigative Committee organized by the State llealth Department, which included expertise in the fields of epidemiology, pediatric endocrinology, obstetrics, medical genetics, blostatistics, and radia-tion physics. Apart from the incidence analysis presented above, there is also an important consideration with re-spect to radiation in relation to congenital hypothyroidism. Mrst, after March 28 through December 31,1979, no single case of congenital hypthyroidism was re-ported in Dauphin, Cumberland, Perry, Northumberland, Juniata, Snyder, Mifflin, and Union Counties, 7

the areas downwind (N, NW, NNW) from the Three Mile Island during the first 48 hours of the ac. cident, when probably the largest amount of radioactive releases took place, thus the largest amount of contamination including I*. Second, the maximum combined (inhalation and ingestion) human thyroid dose of radioactive 10 dine in the vicinity of the TMI following the March 28,1979 accident through April 1979 is estimated to be 7.5 mrad (Editorial: Annals of Internal Medicine, Vol. 91, No. 3, September 1979). At least 1,000 times greater thyroid doses (i.e., 7.5 rads) would be required to have significant acute damages to the thyroid gland; however, even at this dose level, many of the damaged cells may be repaired. Based on the experiences of the Marshallese exposed to fresh radioactive fallout and atomic bomb vic-tims, it is considered likely that as much as 50 to 100 rads fetal thyroid doses would be necessary to cause irreversible tissue damages, such as congenital hypothyroidism and/or thyroid cancer. Acknowl-edging the fact that the fetal thyroid is much more sensitive to radio iodine than is the maternal thyroid (a conservative upper bound estimate is that the thyroid dose to a fetus may be as high as ten times the maternal thyroid dose), the maximum likely fetal thyroid dose (approximately 75 mrad) and the maximum possible thyroid dose of 190 to 200 mrad in the vicinity of the damaged nuclear plant are still far too small to have caused congenital hypothyroidism. In an epidemiological investigation of possible " clustering" of a disease or morbid condition, it is important to recognize the technical difficulty and methodologicallimitations associated with such investigation. It is the overall consistent pattern of observation that provides useful basis for conclu-sion, rather than a single isolated change or difference, which in most cases occurs without substantive epidemiologic significance. This is particularly true when relatively small populations are being stud-led. One may or may not find a " statistically significant change, difference, or clustering in morbid rates in an area depending upon how such population is delineated geographically and/or temporally. It is equally important that investigators carefully m.unine the observed relationships and determine if such relationships are consistent with the known biological theory or orientation, which is based on the previous studies and experiences. Our conclusions regarding congenital hypothyroidism around the Three Mile Island nuclear plant have been based on both the overall pattern of epidemiologic observa-tions and in reference to the existing scientific knowledge. C. INFANT MORTALITY lonizing radiation is often related to infant morbidity and mortality in a general context of biological effects because of the greater sensitivity of the newborns to radiation, as compared with the adult population. The infant mortality is defined as the risk of infants dying within the first year of life and is expressed per 1,000 live births. Subsequent to the March 1979 nuclear accident, we initiated a comprehensive evaluation of the exist-ing vital statistics data in order to determine if the TMI accident has had any measurable influence upon infant mortality in the vicinity of the damaged plant. 8

For the purpose of the present study, we considered a 10 mile radius of the Three Mile Island, wherein approximately 4,000 infants are born annually (Table 1). Both levels of radiation exposure and psychological distress within the 10 mile radius communities were higher than those beyond the 10 mile radius communities. The available mortality data were analyzed by calendar quarters, as well as annually, for each of the three consecutive years,1977,1978, and 1979, for the entire 10-mile area, including Harrisburg, the 10 mile area excluding Harrisburg, and Harrisburg separately. For cross. sectional comparison, corresponding mortality data for the State of Pennsylvania as a whole, were evaluated for the same historical time frames. As indicated in Table 2, the infant mortality rate was not significantly different between the 10 mile area with or without Harrisburg and the State of Pennsylvania for any of the three years under con-sideration. The higher infant death rate indicated for Harnsburg separately is a reflection of the fact that approximately one half of the infants born in the city were nonwhite. The infant mortality rate within the 10 mile radius, including Harrisburg, was already considerably high (19.3 per 1,000 live births) during the first que'rter of 1979 prior to the TMI accident. The rate remained at the same level during the second quarter of 1979 immediately following the accident, but declined substantially during the third 12.7) and fourth (13.4) quarters. This temporal pattern of change in the rate is consistent with the view that the TM1 accident has had no measurable impact upon infant mortality. Otherwise, the infant mortality rate would have increased steadily (or, at least, would have remained high as a result of interaction between seasonal downward trend and TMI re. lated upward trend), particularly during the third and early fourth quarters. Fetal sensitivity to radia-tion and maternal distress is much greater in the earlier period of gestation or organogenesis when ex. posed and this would have been reflected on the gradually rising mortality trend following the acci. dent for a period of nine to ten months. However, the actual observation was contrary to this hypo-thesis. Within the 10 mile radius of TMI, the 1979 infant mortality rate (16.1) was not significantly differ. ent from the 1977 rate (12.5). The 1978 infant mortality rate (10.8) in the same area was somewhat atypical and unusually low, particularly within the immediately surrounding communities outside of Harrisburg (8.4). This is largely because of the small population, wherein marked statistical variations from year to year are not at all uncommon with no particular epidemiologic significance. For this ren-son, the 1978 infant mortality rate should not be used as a normal base for comparison. Having considered both cross sectional and temporal analyses of the available vital statistics data com. I piled by the State Health Department, we found no evidence that the TMI nuclear accident has had any significant impact upon infant mortality. Statistical variations or differences, as observed in the i 10 mile radius, are considered to be a typical random phenomenon in a relatively small population with no particular epidemiologic significance. Theoretically, too, the low levels of radiation exposure, as reported offsite, cannot be directly related to such massive destruction or impairment of cells that cause infant deathr. 9

The pattern of fetal mortality rate or the risk of the fetus being born dead in the vicinity of the Three Mile Island was also analyzed by the same method as applied to infant mortality. We found that there is no indication that the TMI nuclear accident was related to its quarterly or annual variations in fetal mortality (Table 3). The level of fetal mortality within the 10 mile communities was, in fact, con-siderably lower than that for the State as a whole.

SUMMARY

AND CONCLUSIONS The Three Mile Island nuclear accident has caused an extensive social and political unrest world wide. At the same time, it has presented social scientists and biomedical investigators a unique opportunity to evaluate its impact upon local population. Probably the most important concern is that of safety and health effects of this unprecedented event. From the currently available epidemiological knowledge, no significant physical health effects are ex. pected from the low level radiation reported to have been released from the damaged TMI facility, llow. ever, in the absence of absolute certainty as to the exact amount of radioactive contamination of the local environment and population, particularly during the early period following the accident, carefully designed epidemiological studies, such as those described in this report, are justified. Furthermore, some substantial psychological impacts upon local populace have been documented. It is not known at this time what signifi. cant physical manifestations, if any, may actually ensue from psychological distress over an extended per. iod of years. In addition, because of high sensitivity of the fetus to ionizing radiation and severe matemal stress, timely evaluation of pregnancy outcome should be pursued. Based on the already established TMI Population Registry, there should be a continuous and long. term epidemiologic surveillance of the exposed general population, which includes approximately 37,000 individuals. The TMI Population Registry is now updated annually so that annual mortality (rate and cause)' and periodic health survey, can be conducted. Such endeavor will make it possible to determine and docu. ment if there is any measurable health impact in humans from the low level ofionizing radiation that has not yet been fully studied. It is the responsibility of government agencies and academic communities to properly inform the gen. eral public with the objective results of carefully-designed scientific studies of possible health effects of the TMI nuclear accident. It is also critical that safety of nuclear energy is properly addressed in relation to that of other available means of energy production, as well as many other potential risks in human life. Equally important is the public understanding of various alternatives, so that the society as a whole, rather than selected few, can make a rational choice for its constituents. It is hoped that TMIlicalth Effect Re. search Program will serve as a means to achieve such goals. 10

TABLE 1: RESIDENT LIVE BIRTHS f QUARTER: PENNSYLVANIA AND TEN MILE TMl AREA COMMUNITIES, 1977-1979 TMI Ten Mile Aree Year /Ouarter Harrisburg Excluding Pennsylvania Total City Hbg. City 1977 (153,415) (3,750) (1,001) (2,749) Jan. - March 36,911 886 242 644 April-June 38,414 937 248 689 July - Sept. 40,181 977 274 703 Oct. - Dec. 37,909 950 237 713 1978 (151,438) (3,803) (1,057) (2,746) Jan. - March 37,084 926 261 665 April-June 36,339 922 262 660 July - Sept. 39,932 1,029 302 727 Oct. - Dec. 38,083 926 232 694 1979 (157,533) (3,905) (1,185) (2,720) Jan. - March 38,326 932 296 636 April-June 38,351 983 303 680 July - Sept. 41,933 1,023 302 721 Oct. - Dec. 38,923 967 284 683 a

TABLE 2: RESIDENT INFANT DEATHS, NUMBER AND RATE, BY OUARTER: PENNSYLVANIA AND TEN MILE TMI AREA COMMUNITIES, 1977-1979 Number of Deaths Death Rate Per 1,000 Live Births Ten Mile TMI Area Ten Mile TMI Aree Year /Ouarter Harrisburg Excluding Harrisbu.g Excluding Pa. Total City Hbg. City Pa. Total City Hbg. City infant Deaths 1977 (2.137) (47) (15) (32) (13.9) (12.5) (15.0) (11.6) Jan. - March 544 13 6 7 14.7 14.7 24.8 10.9 April-June 554 11 2 9 14.4 11.7 8.1 13.1 July - Sept. 520 9 3 6 12.9 9.2 10.9 8.5 Oct. - Dec. 519 14 4 10 13.7 14.7 16.9 14.0 1978 (2.031) (41) (18) (23) (13.4) (10.8) (17.0) (8.4) Jan. - March 533 13 8 5 14.3 14.0 30.7 7.5 April-June 509 9 3 6 14.0 9.8 11.5 9.1 July - Sept. 473 5 1 4 11.8 4.9 3.3 5.5 Oct. - Dec. 519 14 6 8 13.6 15.1 25.9 11.5 1979 (2.118)* (63) (31) (32 (13.4) (16.1) (26.2) (11.8) Jart. - March 511 18 10 8 13.3 19.3 33.8 12.6 April-June 537 19 9 10 14.0 19.3 29.7 14.7 July - Sept. 507 13 3 10 12.1 12.7 9.9 13.9 Oct. - Dec. 562 13 9 4 14.4 13.4 31.7 5.9

• Includes one death, month af occurrence, unknonn.

s

TABLE 3: RES40ENT FETAL DEATHS (TOTAL), NUMBER AND RATE, BY OUARTER: PENNSYLVANIA AND TEN MILE TMI AREA COMMUNITIES, 1977-1979 Number of Deaths Death Rote Per 1,000 Deliverses' Ten Mile TMI Area Ten Mile TMI Aree Year /Ouarter Harrisburg Excludes Herrisburg Excluding Pa. Total City Hbg. City Pa. Total City Hbg. City Fetel Deaths (Total) 1977 (4.068) (80) (45) (25) (25.8) (20.9) (43.0) (12.6) Jan. - March 1,062 16 10 6 28.0 17.7 39.7 9.2 April-June 992 18 9 9 25.2 18.0 35.0 12.9 July - Sept. 1,026 23 12 11 24.9 23.0 49.0 15.4 Oct. - Dec. 978 23 14 9 25.1 23.6 55.8 12.5 1978 (4.034) (77) (M) (39) (25.9) (19.8) (34.7) (14.0) Jan. - March 1,003 15 5 10 26.3 15.9 18.8 14.8 April-June 1,047 21 12 9 28.0 22.3 43.8 13.5 July - Sept. 1,001 20 10 10 24.5 19.1 32.1 13.6 Oct. - Dec. 983 21 11 10 25.2 22.2 45.3 14.2 1979 (3,608) (67) (37) (30) (22.4) (16.9) (30.3) (10.9) Jan. - March 938 24 13 11 23.9 25.1 42.1 17.0 April-June 916 12 6 6 23.3 12.1 19.4 8.7 July - Sept. 937 16 10 6 21.9 15.4 32.1 8.3 Oct. - Dec. 817 15 8 7 20.6 15.3 27.4 10.1

• Definries: Live births and fetal deaths (including abortions).

r REFERENCES

1. Brent, R.L. and Gorson, R.O.: Radiation Exposure in Pregnancy, Current Problems in Radiology, Vol.

II, No. 5,1972. e

2. Brent, R.L.: Effects of Ionizing Radiation on Growth and Development. Contributions to Epidemi.

ology and Biostatistics, Vol. I,1979. N~ ^

3. Meyer, M., Diamond, E. and Meta, T.: Sex Ratio of Children Born to Mothers Who Had Been Exposed

. to X spys in Utem. Johns Hopkins M.J. 123: 123,1968. 4.- Segall, k., MacMahon, B., and Hannigan, M.: CongenitalMalformations and Background Radiation in Northern New England. J. Chron. Dis.17:915,1964.

5. Tabuch!, A.:fetalDisorders vue' to Tordsing Radiation, Hiroshima J. M.Sc.13:125,1964.

i

6. 'Kinlon, L.J. and Acheson, E.D.: Diagnostic Irradiation, Congenital Malformations and Spontaneous

) Abortioru, Brit. J. Radiol. 41:648,1968. ~

7. Nuckolla, K.B.: Ptychological Assets, Life Crisis and the Prognosis of Pregnancy. American J. Epid.

i' 95:431,'1972.

8. Morishima, H.O.: The influence of Maternal Psychological Stress on the Fetus, Amer. J. Obs. and Gyn.

131:286, 1978.

9. Barlow, 8.M.: Delay of Postnatal Growth and Development of Offspring Produced by Maternal Re-straint Stress During Pregnancy in the Rat. Teratology 18:211,1978.

10. Nuckolls, K.B.: op. cit.

i

11. Newton, R.W.: Psychosocial Stress in Pregnancy and its Relation to the Onset of Premature Labor.

Brit. Med. J. 2:411, 1979.

12. Gorsuch, R.L. and Kay, M.K.: Abnormalities in Pregnancy as a Function of Anxiety and Life Stress.

Psychosomatic Med. 36:352,1974.

13. Berkman, L.F. and Syme, S.L.: Social Networks, Host Resistance, and Mortality. American J. Epid.

109:186, 1979.

14. Burchfield,8.R.: The Stress Responar: A New Perspective. Psychosomatic Med. 41:661,1979.

I

15. Smith, D.J.: Modification of Prenatal Stress Lffects in Rats by Adrenalectomy, Dexamethasone, and Chlorproma:Inc. Physiology and Behavior 15.461,1975.

16. Houts, P.S., Miller, R.W., Tokuhats, G.K., and Ham, K.S.: Health.Related Behavioral Impact of the Three Mile Island NucIsarIncidence. Report submitted to the TMI Advisory Panel on Health Research Studies, Pennsylvania Department of Health. Part I. April,1980.

1 l 14

C r. I. e ChapterFourteen 'y*?' Psychological and Social Effects on the Population Surrounding Three Mile Island After the Nuclear Accident on March 28,1979 i ! Peter S. Houts, Ph.D. Marilyn K. Goldhaber, M.P.H. The Milton S. Hershey PENNSYLVAN!A DEPARTMENT Medical Center OF HEALTH PENNSYLVANIA STATE UNIVERSITY College of Medicine Hershey, Pa.17033 Dr. liouts received a B.A. degree from Antioch College and a Ph.D. in Social Psychology from ihe University of Michigan. lie has taught at Goucher College and was a post-doctoral research fellow in the Psychiatry Department of the Stanford University Medical School before joining the faculty of the Depart-ment of Behavioral Science of The Pennsylvania State University College of Medicine in 1%7. lie is now associate professor in that department. Ile has di-rected studies on the Ilealth Related Behavioral Impact of the Three Mile Island t Nuclear incident under contracts with the Pennsylvania Department of 11ealth. I Ms. Goldhaber graduated from ihe University of California, Berkeley, with a B.A. in Mathematics and an M.P.11. in Biostatistics. She has worked in cancer research at the Resource for Cancer Epidemiology in Berkeley, California and is currently employed by Ihe Division of Epidemiological Research at the Pennsyl- ' G-l vania Department of IIcalth. The Division of lipidemiological Research is con- ?} ducting a comprehensive Three Mile Island Research Program where Ms. j t g Goldhaber is in charge of the Three Mile Island Population Registry. a fi ~

q Pochologirst med Sorisf F,ffirers oss the Mym&rtion Sumnuedper TMI... 153 152 Energy, Envirsnment, and the Economy. o The crisis at Three Mile Island (TM I) nuclear power plant began on March 28, ficial order other than the Governor's " advisory" the choice to evacuate was an 1979. It lasted a littie' more han a wcek and caused the temporary exodus of a individual or family decision. Only one community, Middletown, prepared and substantial portion of ihe population in the vicinity of1he plant. During the crisis distributed hand bills on the fifth day after the accident depicting plans for a many area residents were fearful of their safety and many who evacuated were possible evacuation. A Red Cross sponsored shelter for women and small chil-not sure that they would ever be able to return to their homes. After the im-dren was set up in the town of Hershey (approximately 10 miles from the plant) mediate crisis passed, there was a continuing cont ros ersy about how to deal with to accommodate persons from the five mile radius who came under the Goser-the damaged reactor. As a result, the situation at Three Mile Island received nor's advisory but had no family or friends with which to stay. it was iater found almost continuous attention in national and kical news media for over a year that 85 % of all evacuees stayed with family or friends and only 15% were obliged ^ after the original accident. What began as a short term crisis became, for many to find hotel or motel accommodations (3). persons in the area, a chronically disturbing situation. Evacuation behavior, ascessed through surveys of the population after the This chapter is concerned with social, psychological and economic impacts of crisis, showed that the older the head of the household, the less likely that the the TMt crisis on persons and institutions in the vicinity of the reactor. Studies household evacuated. Families with small children, higher than average educa-dealing with the decision to evacuate, t he cconomic impact of the crisis, the effect tional attainment, and higher than average income w ere more likely to evacuate. of the crisis on mobiliry of persons living near the plant, attitudes toward TMI The presence of a pregnant woman also had a significant affect on the decision io following the crisis and psychological stress experienced by peopic in the vicinity leave the area (3). of the facility will be reviewed. Studies in each of these areas were carried out In general women were more prone to evacuate than men. Younger persons either directly by the Penns3tvania Department of Health or under its auspices were more likely to evacuate than older persons. Young parents (approximately and these studies will be the primary focus of this chapter. Other studies carried in their 20's) and their small children were more likely Io stay away for longer out under auspices of the Presidential Commission on Three Mile Island, the periods than were other age groups. Evacuation patterns aiso differed slightly by Nuclear Regulatory Commission and the National Institute of Mental Health geographic location. This was especially apparent in several small communities will also be reviewed. where the perception of distance to the plant differed substantially from actual distance. Evacuation A survey in July 1979 (7) asked respondents w ho had left the area during the Voluntary evacuation in the vicinity of Three Mile Island began almost im-crisis their reasons for evacuating. The most common reasons were that the situa. mediately after the discovery of the nuclear accident was broadcast to the public. tion seemed dangerous (82%) and that the information available was confusing During the first two days after the accident the evacuation was minimal with only (78%). Those who had stayed during ihe crisis were also asked their reasons for 5 or 6 percent of the population within the 5 mile radius Icaving for ostensibly staying. The most frequently given reasons were: "whatever happens is in God's safer ground. On the third day, however, nearly 50 percent of the population hands" (70%) and " waiting for evacuation order" (62%). Interestingly only within a 5 mile radius had evacuated (7). This was largely precipitated by the 30% of the stayers said they saw no danger. Governor of Pennsylvania's advisory on Friday, March 30,1979, for voluntary evacuation of pregnant women and small children residing within 5 miles of the facility. This, compounded by the increasingly confusing and sometimes alarm-ECONOMIC IMPACT ing stories being reported by news media and the fact that the weekend was ap-proaching, led io the exodus of an estimated 39% of the total residential popula-The accident at TMI had an immediate, detrimental impact on the economy of tion within 15 miles of the plant by the weekend. (This involved 60% of the the area surrounding the plant according to a report of the Governor's Office on population in a 0-5 mile ring,44% of the population in the 5-10 mile ring and Policy Planning (6). However the impact was short-lived and the economy of 32% of the population in a 10-15 mile ring.) The total number of evacuees was South Cent ral Pennsylvania returned to nearly normal levels w ithin 2 weeks after estimated to be almost 150,000 persons. The median distance traveled by evacu-the accident. Some specific aspects of Ihe economy, notably food processing ces was 100 miles and Ihe average evacuation period was 5 days (3). (dairy), real estate and tourism took longer io return to normal. Ilowever, wit hin At no time did the evacuation take on the tenor of mass panic. That is, most 3 months the economy, other than the nuclear industry, seemed imperceptably families left in a relatively orderly fashion in their family cars at a genera) flow changed if at all. (Effects of the crisis on the nuclear industry will not be ad-w hich did not noticeably disrupt normal traffic patterns. Since there was no of-dressed here.)

t$4 Energy. Environment, and the Economy Psychologinst ond Social Effat.s on the Popukusson Surraumhnx TMI.., t$$ MOBILITY The costs of evacuation was the greatest source of economic loss for both in- .lividuals and communities. With 39% of the population residing within 15 miles of the plant temporarily away, the routine of economic life was drastically A telephone survey of residents within 5 miles of TMI in July,1979, three altered. Retail sales were reduced, industrial production was interrupted and months after the crisis, had found that 17% of respondents living within 5 miles normal business patterns wcre disrupted. Schools wcre closed, conventions had of TMI said that someone in their household had " considered" moving because 1o be cancelled, and, in some of ahe communities near ihe plant, curfews were en-of the situation at Three Mile Island and 6% reported ihat someone had definite-ly decided" to move because of that situation. It therefore seemed likely that a forced. Short term losses to business were estimated, in the report of the Governor's substantial number of peopic might subsequently move away from the area Task Force on Three Mile Island, at 7.7 million in the value of production and because of the Three Mile Island crisis. Ilowever, interview data of this sort are manufacturing,74.2 million in business sales and non-manufacturing and.25 to not definitive since they only deal with intentions to move rather than actual .50 million in agriculture (13). However, both tourism and housing were able to behavior. Furthermore, it is not known whether these people might have moved rebound to normal levels and even make up for their temporary losses (adjusting in any case with or without the Three Mile Island crisis. Because of this, a study for fuel price increases, high interest rates and shortage of mortgage funds). was undertaken by the Pennsylvania Department of itealth in collaboration with Thus, the total economic loss to businesses in the area was estimated to be about Pennsylvania State University to examine mobility rates for the area within 5 miles of Three Mile Island during ihe one yur period immediately following the 82 million dollars. Two other studies assessed personal costs (direct costs to citizens) in terms of Three Mile Island crisis in March,1979 (5). Three questions were addressed in evacuation expenses, wages lost, additional drug-alcohol, cigarette consumption this study: expenditures (due to stress) and increased health facility usage. These studies

1. Was the percent of persons within 5 miles of Three Mile Island who were sponsored by the Nuclear Regulatory Commission and the Pennsylvania moved during the year following the crisis different from percent of Department of Heahh (3,11). They found that costs of evacuation were sub-movers in other comparable groups during this time period?

stantial Io many households, averaging about 200 dollars per evacuated

2. For persons within 5 miles of TM L at the time ofIhe accident, how great a household. Subtracting out insurance reimbursements, the total net cost to the role did the situation at theThree Mile Island play in the decision to move residents within a 15 mile radius around the plant was estimated at 610 million or stay, or the choice of how far to move?

dollars. These estimates include actual evacuation expenses, pay loss and other

3. Were the people who moved out of the area within 5 miles of TMI in the incidental costs to residential families.

year following the crisis different from those who moved into the area? Ilu et al(11) carried the investigation further by exploring possible health-It was possible to answer these questions by using the Three Mile Island related economic costs resulting from changes in physical or mental health status population registry, a 95% complete registry established by Ihe Pennsylvania and/or a change in the use of health care services. They analyzed the situation Department of IIcalth of all persons residing within 5 miles of the nuclear plant from both the individual perspective (population surveys) and provider perspec-shortly after the March 28,1979 accident. The registry was des eloped as part of a tive (physician survey and health insurance reimbursement data). There was long-term study of possible health effects of the nuclear accident and thus re-some indicalion from ihe population survey Ihat increased stress may have quired yearly updating the current addresses of all persons included iherein. As a brought about a small increase in health care utilization (physician visits). Flow-bi-product of the yearly update, residential mobility could be systematically ever, surveys of local physicians and studies of Illue Cross / Blue Shield and monitored. In addition, samples of movers were identified after the first year and Medicare / Medicaid data indicated little, if any, increase in heahh care utilization interviews were conducted with random samples of in-migrants, out-migrants, attributable to the crisis, at Three Mile Island. Ilouts et al(10) in a study of and non. migrants and a control group from communities 10 to 55 miles away. . utilization patterns among patients in a primary care family practice near Three The first question studied was whether the mobility rate in the vicinity of TMI Mile Island, came to similar conclusions. They found no increase in practice was different from that of other comparable groups. The post-TMI annual utilization that could be attributed to the Three Mile Island crisis. Interestingly, mobility rate (percent of persons who moved to another dwelling) for the 5 mile though, Ihey did find that people who wcre high practice utilizers prior to the community was computed io be i 1.6%. T his was low er ihan the 13.6% mobility crisis tended to be more upset during ihe crisis han were low utilizers. Ilowever, rate for Ihe same population during ihe one year prior to Ihe Three Mile Island this cone rn apparently did not cause the high utiliiers to increase utilitation crisis and was the reverse of conjectures that the TMI crisis precipitated mass beyond it ir baseline rates following the crisis. migration out of the area. This drop of almost 2% was, however, difficult to in-terpret because of a sharp escalation ofinterest rates and a shortage of mortgage

s. evergy, unwonnwns, ams ahe Leonomy ) p,yckegag;cor,,g Socs,IKffeers oos the Popularoon Surroundma TMt... IST loan funds during the year studied. Therefore, a second comparison was made with the mobility rate for ehe cont rol group living from 48 to S5 miles from Three ATTITUDES TOWARD THREE MILE ISLAND Mile Island. The rate for ihat group was 12.5%, only 1% greater than for the TM I area. The difference bet w een ihis rate and Ihe TMI mobility rate was evalu-Following is a summary of attitudes from surveys carried out by the Pennsyl-ared statistically using analysis of covariance controlling for variables which are vania State University and the Pennsylvania Department of Health in July,1979, frequently associated with mobility: age, education, how many times the respon-January,1980, and October, 1980 (7,8,9). All three surveys included persons dent had moved in his/her lifetime, w het her he respondent owned or rented and living within 5 miles of TMI as well as a control group from 41-55 miles away. maritalstatus. The difference between the adjusted means was less than 1% and Each survey addressed slightly different issues related to attitudes toward Three was far from statistical significance. The conclusion drawn from these analyses Mile Island. Some questions were repeated in more than one survey making it was that there was no conclusise esidence that mobility rates within the 5 mile possible to examine changes over time. Responses to these questions will be radius of Three Mile Island w cre significantly effected by the crisis in March of discussed first. 1979. The first question concerned restarting Reactor #1, the undamaged reactor, The second question addressed was how great a role Ihe situation at Three Mile at Three Mile Island, in January,1980, when the question was first asked,60% Island played in Ihe decision to mos e or in how far to move. This question was of respondents w ithin 5 miles of Three Mile Island and 30% of respondents 41 55 addressed utilizing multiple regression analysis. The conclusion from these miles away opposed restarting the facility. Ilowever, in October,1980, nine analyses was that, after traditional predictors of mobility (age, cducation. own-months later, opposition to re-opening had decreased in the area close to the rent, etc.) were controlled, people's feelings about Three Mile Island had only a plant with only 46% of respondents within 5 miles opposing restarting compared small effect of mobility, f low ever, there w as some evidence that, for very mobile to 42% of respondents living from 41-55 miles away opposing the restart. The people (young and educated) attitudes ioward Three Mile Island did relate to the decrease from 60% to 46% among the close group was statistically significant, distance moved. The more concerned these people were about the situation at but the increase from 30% 1o 42'*e among the controf group w as net. The differ-TMI, the further they moved from the facility. ence between the close and far groups was statistically significant in January. The Ihird question addressed in ihe mobility study was whether Ihe type of per-1980, but was not so in October,1980. son moving into the arca differed from those moving out and, therefore, whether The second question repeated from earlier surveys concerned political activity population characteristics of the area were changing. Once again, resuks were and asked respondents whether they had, personally, been active in any largely negative. Only Iw o differences in demographic characteristics were statis-organization or had gone to any meeting to influence what happened at Three tically significant bet w een the two groups. T he age of t he oldest household mem-Mile Island. Results showed that in January, 1980,13% of respondents gave af-ber for in-migrants was slightly younger than for our-migrants and there was a firmative answers to this question and in October, 1980,15% did so indicating significantly larger percentage of TMI employees among in-migrants compared little change over the 9 month interval. It should be noted that this participation to out-migrants. There wcre, however, consistent differences between in-and rate near Three Mile Island was high by Ihe usual standards of political actis ity in out-mi rants in their attitudes oward the situation at Three Mile Island. Five at-this country. For example, a survey by Ihe National Opinion Research Center in F titude measures dealing with perceived safety, dangerousness of krypton vent-1973 found that only 5% of Iheir respondents had ever participated in any kind ing, being upset about Three Mile hiand, support for re-starting TMI Unit One, of anti-war or pro-war demonstration and a har only 9.5 % had ever been invoh ed and whether there should be morc or few er nuclear plants in the future all showed in picketing in the course of a labor dispute. The figures for Three Mile Island are more positive TMI attitudes among in-migrants than among out-migrants. even more impressive when one considers that the National Opinion Research The general conclusion from Ihis st udy on mobility was that the TMI crisis had Center questions referred to activities oser an entire lifetime, w here political ac-scry small, if any, effect on mobility in the immediate area of the facility. This livity related Io Ihe TMI accident could have occurred only wit hin a period ofless conclusion is essentially Ihe same as Ihat of Ihree other studies, one conducted by than 18 months. the Penmylvania Governor's Office (6), one conducted by a local board of In October,1980, respondents were asked how much influence they felt dif-realtors (14) and another under sponsorship of ahe Nuclear Regulatory Com-ferent groups should have in the decision io clean up Three Mile Island Strong-mission (4). All of these studies showed a marked drop in real estate sales in the est support was given to ehe Nuclear Regulatory Commission and energy esperts sicinity of T hree Mile bland immediately following ihe crisis but a return to nor-working for the Pennsylvania Department of Emironmental Resources. It ap-l mal leveh within a mont h, and for the year as a whole, very little change irg real pears, from this, that the pubhe prefers technical esperts working under gm ern-ment auspices to have the most influence in cleaning up the facility. estate tramactions.

i55 knergy, Ennronment, and the Econom) Psychological and Social Effwrs on the Populatwn Surroundmx TMI... l54 Respondents, in October,1980, wcre also asked how strongly they believed or STRESS DURING AND FOI. LOWING Tile Tall CRISIS disbeliesed a number of common rumors concerning the effects of the Three Mile Island accident. Fis e rumors u ere read to respondents including alleged in-One of the most widely studied aspects of the Three Mile Island crisis was the creased numbers of miscarriages, increased birth defects, expectation of higher cancer rates, increased health problems in farm animals, increase in general psychological stress experienced by people in the vicinity of TMI. Results of these health problems, and increase in mental health problems because of the Three studies, w hich indicate some degrec orcontinuing dist ress for up lo a ycar follow-Mile Island crisis. Results showed approsimatcly equal numbers of believers and ing the crisis, have attracted considerable public attention. In this section we will nonbeliesers. The most widely accepted rumor was that there had been an in-discuss primarily results of studies carried out by ihe Pennsylvania Stare Univer-crease in mental health problems because of Three Mile Island (58%), the least sity in collaboration with the Pennsylvania Department of licalth (7,8,9). Rela-accepted was that there had been an increase in birth defects since the crisis tionships between the Iindings of the Penn State-Pennsylvania Department of Health studies and those of other investigators will also be discussed. (31%). For two rumors, increase in miscarriages and increases in birth defects, there was greater acceptance ofIhe rumor among the group far from Three Mile llouts et alin a 1981 report summarized the results of three surveys conducted Island than among the group within 5 miles of the facility. over an 18 month period (9). These surveys wcre of representative samples of Attitudes toward media coserage of the Three Mile Island situation were also households with phones within 5 niles of Three Mile Island as well as in com-assessed in October,1980. Results showed that almost half the respondents felt munities from 41 to S5 miles from the facility. Five distress indices were included as follows: that the media had blow n esents out of proportion and that 1/5 thought that in-formation was withheld or cosered up. l. css than I/3 felt that the events were

1. Ilow upset the respondent was about the situation at Three Mile Island.

reported accurately.

2. Ilow serious a threat the respondent felt 1MI was to safety.

Finally, three questions were included in the October survey which dealt with

3. Frequency of " behavioral symptoms" during a two week period (i.e.,

atfit udes toward Ihe venting of k rypton gas at the Three Mile Island facility in Ju-lack of appetite, overeating, sleeplessness, feeling shaky, trouble think-ly,1980. Although the authorities proclaimed it as safe, the venting of this gas ing, irritability and anger). had been the subject of considerable public debate in he months prior io its oe-

4. Frequency of " somatic symptoms" (i.e., stomachaches, headaches, currence. Partly as a result of this debate, the public was wc!! informed in ad-diarrhea, frequent urination, rash, abdominal pain, and sweating spells) during a 2 week period.

vance of the venting. This gave persons living in the area an opportunity to evacuate during venting periods ifIhey so wished. Fifteen percent of the popula-

5. For those persons who reported either behavioral or somatic symptoms, iion living within 5 miles of Three Mile Island in July,1980, reported Ihat venting whether they attributed those symptoms to the situation at Three Mile Island.

was "an important reason" for heir leaving ihe area during the venting period. Results, shown in Table 1, are as follows. The average length of absence due to the venting was 10 days. Respondents were also asked if they felt the venting was Ihe right way Io get rid of the krypton gas

1. Upset ratings were significantly higher close to TMI compared to farther and also how dangerous they thought the krypton venting was. These questions away from April,1979, through January,1980, but were no longer sigmficantly were asked of respondents within 5 miles of T hree Mile Island as well as the con-different in October,1980. In examination of the mean ratings at each time trol group from 41-55 miles away.1 he results showed that people living close to period shows Ihat levels of upset came dow n over time for bot h Ihe close and far Three Mile Island were significantly less negative about the venting than were groups, but that the drop was sharper for the groups close to TMI.

people further away. This is consistent with Ihe finding mentioned above where

2. Both behavioral and somatic symptoms were reported more frequently some rumors tended Io be accepted Io a greater degree furIher from the plant close oTMIcomparedIo41 o55 milesawayin Apriland July,1979,as wcilas than among persons close to the facility.

January,1980. However, by October,1980 this difference was sharply reduced T hesc attitude st udies indicate that a substantial percent of t he population ricar and the difference between the two groups w as no longer statistically sigmiscant. Three Mile Island had serious concerns about the nuclear facility as long as 18 11 should be noted in interpreting these findings that the general lesels of s)mp-months after the crisis, llowever, there was also evidence that these concerns are tom reporting fluctuated considerably over ihis time period. T his could hase decreasing. This is consistent with the studies of stress discussed in the next sec. been due to many factors including seasonal variations. Therefore, dillerences tion. between close and far groups are better indicators of the effects of ihree Mile Island than are general levels of symptom reporting at each time period.

3. Perceptions of TMI asa serious ihreat oone's family's safety camedow n

. uuss. us...vimeum..nu use tsonomy Psychologscaland SocsalEffn ts on the Popularwn Snorronendung TAff... I6l TABt l' t both groups during the 15 months studied, but, as with perception of thrc '.J-Afcon oestress scoresfor />cesons t.asine other wsrAin 5 Af,fes of TAff or scri,cen tribmion was still significantly higher close to Th11 in October,1980. 4/ and 55.tfs/cs Ai,-arfrrim Dflar t'our Tame Periods The overall pattern of these findings shows reductions in distress for h>th the DIST RESS htli ASURI!S " P W of reductton for the group close to Th11 than for far away. Dntance from Upsei about se u Behavioral somatic y t, While the data shown in Table I indicate long term distress, other studies of TMt (miles) TNtt-i hicat' ssmptome S> mptoms* to Tati, mental distress showed only short term strew following the crisis. Studies by April 1979... Dohrenwend et al for the Presidential Commission on Three Afile Nan Within 5 3.36 3.10 .30 .18 .69 utilized the " demoralization" scale of psychological distress in sescral studies 4I 55 2.21 2.13 .07 .05 .29 beginning immediately a fler the crisis and extending into hlay, Iwo months later. Difference i.15 * .97** .23** .t3** 40** July 1979 They found demoralitation scores markedly raised immediately after the acci. deng, but found that they dropped sharply in Apriland Afay. In the htay sample wiihin 5 + 3.16 .32 .37 + 41 55 + 2.46 .17 .21 + responses of mothers close to l hree hlile Nand were compared wit h responsn of hi d p apptoximately 100 miles away and it was found Ihat D ference .70** .15" . 3 6" they were not significantly different from each other. Eight months later a Penn within s 2.59 2.39 .40 .51 .25 State / Department of IIcalth survey (7) found similar results utilizing the 4t.55 t.95 1.6n .t8 .35 .04 Langner scale of psychological distress which correlates very highly (over.90) Difference .54" .79** .22" .16" .21 a with the demoralization scale. They found, in January,1980, that there was no October 1980 significant difference in scores on the I.angner scale for persons close to Three Within 5 2.29 2.46 .40 .40 .27 h1ile Island compared to persons in a control group from 41 to 55 mdes away. 41 55 2.13 1.82 .39 .38 .I2 On the surface it appears that studies utilizing the Demoralitation and thfrerence .16 . 64 " .nl .02 .15* i Irecis of age, sci, education. manial staius and income hase been controlled. Langner scales contradict results discuwed alxne using symptom checklists and Ilow upset are/ucre Sou about 1518? Scale t.5; 5 - Very upset, t - Not at att upset. attitude questions. Ilowever, a careful examination of Ihe Iwa sets of data indi-1. 2. Ilow senous a ihreat svaas 1sti to famity's safety? scate 14: 4 - very serious, cate two reasons for their differences. First, Ihe I.angner and Demoralitation /t d\\ u scales include many symptoms not included in ihe Somatic or llehavioral symp-3. case of Arnt. in the 2 mecks of the crisns: lack of appetite, overcating, sleeplessness, tom btGee symptom Charactesmere mentapm.cg.,ydenW or more of the following symptoms in the past 2 weeks (or,in the a n shakes, trouble thinking. irritabihty, or anger? I - Yes. O - No. feelings of sadness or depression," "wonderm, g if anyth, g is worthwhile m 4. Do/Did you hase any one or more of the following symptoms in the past 2 weeks tor,in anymore," " feared going crazy," " felt completely helplew," etc. the case of April. in the 2 wecks of the crkis): stomachaches. headaches, diarrhea, fre-A second reason for the different results is in Ihe sescrity of symptoms as-quent unnation, rash, abdominal pain, sucating spells? l - Yes. 0 - No. sessed. For example, the " somatic" symptom scale asked if Ihe respondent had 5. Do you think your 9 mptoms are due to T M t? I - Yes. 0 - No. IOnly for persons experienced a headachein the past two weeks u hile he Langner scaleasked ifIhe reponmg symptoms) Indicates that the difference between groups n significant at the p <.05 level. respondent had experienced headaches "often." Since the Langner and the

• • Indicaies ihat she difference beimeen groups is significant at the p <.01 level-somatic symptom scales were asked in the same sursey (January,1980) it was July 1979, data is not avaitable for upset and attribution of symptoms.

possible to compare these Iwo ways of asking about the same symptoms. When +

• " Data for April.1979 were collected in the July.1979 sursey and. therefore are respondents were asked if they had had a headache in the past Iwo weeks, there was a significant difference hetween persons within 5 miles of Thti and persons over time for both the close and far groups shough, as with ratings of upset, the 41 to $$ miles away. Ilowever, w hen the same respondents werc asked if they es-drop was shatper for respondents within 5 miles of Th11. The differences be-perienced headaches often, there was no difference hetween ihe two groups, Iween close and far groups uere still statistically significant in October,1980, These comparisons of the two sets of scales suggests that one set (the behasioral eighteen months after the original crisis.

and somatic scales) ast.essed responses Io nor mal day Io day st rewes w hile i he see-

4. The attribution of symptoms to l hree Alile Nand, which involved only ond set (Langner and Demoralitation scales) were more concerned with respondent s w ho reported behas ioral or somatic symptoms, dropped sharply for debilitating symptoms characteristic of mental patients. This suggests that the long term stress that resulted from the Tht! crisis was in the " normal" rather

~} IM Energy. Emironment, and ahe Economy Psychologicaland SocialEffects on the l'otmlarwn Surroundutg TSil... Itd than " pathological" range. This conclusion is consistent with findings by out-patient facilities in the vicinity of Thil compared to similar patients near a Ilromet discussed below. nucicar facility 200 miles away. The lindings were generally negatise with little difference in symptom levels between the two groups. Long term strew cl/ iris on 3pwial grrmps In conclusion, a number of studies dealing with stress among persons in the 1heytudies cited above wcre primarily concerned with Ihe total population in vicinity of Three Mile Island showed ihat stress levels w cre high during ihe crisis, t he s iemit y of I hr ee Mile Island. So cral ot her st udies focussed on special groups but dropped shortly afterwards to lescls that v.cre greater than for comparable thought to be opecially sensitis e to slie crisis. I hese groups included mothers of control groups, but, nonetheless, were wit hin a " normal," i.e., nonpathological young children, workers at the ~I MI facility, and mental patients. range. Finally, as time has passed the number of distrewed persons has con-Ilromet (1) earried out an estemise study of 312 mothers in the TMI area and sistently decreased to w here, in October, 1980,18 months after the crisis, several compared their strew loels uith those of 124 mothers in llcaser County, Penn-of the distrew measures which had been greater close to 'I Mi than for a control ghania, approsimately 200 miles away, but also in the sicinity of another group, no longer showed statistically significant differences. Nuclear Power I acehty. Ihe purIwe of this doign was to isolate ihe effects of

SUMMARY

the 1hree Mile Island accident f rom the cifects of lising near a nuclear facility. llromet co!!ceted data approximately 9 months and 12 months after the crisis. A There is a remarkable degree of consistency among different studies of the number of standardized instruments for assessing psychological distress were social. psychological, and economic ef fects of t he T h ree Mile Island nuclear acci-dent in March,1979. These studies generally agree that the major impact of the utilized by llromet, as w cil as a comptes measure of social support. Results of her analyses wcre as follow s. I) T Mi mot hers reported significantly more mental dis. crisis was felt during the few weeks immediately following the accident and ihat, tress than did the control group mothers at bot h interviews (December,1979 and while some cffects persisted over the following year, lone term cffects were of a April,1980). 2) Mothers with a history of psychiatric disorder reported higher low magnitude, frequency rates t han did comparable mothers in the cont ral group, but ont in t he Temporary evacuation, which lasted less than two weeks for most evacuees. April,1980, surs cy. 3) While ihe symptom leveh for TMI mot hers at bot h 9 and caused short term economicloss of approximately $90 mi!! ion to individuals and 1 12 months were higher than for the control group, they were in the " normal" localcommunities. It was during the two week es acuation period that psycholog-rather than " pathological" range. ~ his finding is consistent uith findings of the ical stress was the greatest, whether one evacuated or not. After the immediate Penn State-Department of Ilealth studies discuwed earlier. crisis was over, concern about the situation at Three Mile bland persisted, main-Kast, Chisholm and Eskenari conducted a study of workers at Ihe TMI facility tained, in part, by continuous attention in the new s media. This concern was es i. and compared their responses to workers at another nuclear facility approx. dent in attit udes as well as reports of stress-related somatic and behas iora19 mp-imately 40 miles away (12). Ihey fomid that, compared to the control group, toms. Ilowever, these were within the " normal" rather than " pathological" TMI workers perceised more danger to their health, had lower job satisfaction, range. espressed more uncertainty about their job future and were less likely to want to Concern about TMI was sufficiently high shorIly after ihe crisis to af fect peo-see their son or daughter work for their company. Iloweser, interestingly, they ple's stated intentions to move permanently out of the area, but not sufficiently found that TMI workers were less willing to accept criticism of their company high to significantly effect residential mobility behas ior in the year following the than w ere control w orkers. The authors concluded that TM I workers were show-crisis. Similarly, while symptom reporting was elevated among persons lising ing ambivalence about their place of employment. Ilromet (1) carried out a near TMI, little evidence of increased demand for medical sersiees could be similar study with the same TMI workers se eral months later and compared found. Distress levels gradually fell mer an 18 month period following the acci-their responses to comparable workers at another nuclear power plant 200 miles dent to the point where symptom reporting was no longer higher close to I Mi as from 1MI. Ihomet reported higher lesch of ansiety and depression among compared to further away. Ilowever, widespread concern about what has and workers at the facility compared to control group workers. Ilowever, lhomet may happen at Three Mile Island still esists among persom lising close to the points out that ihese differences should be interpreted cautiomly since rates for facility, as indicated by attitudes ton ard TMI and belief in rumors alvut alleged the 'I MI workers were higher ihan for the control group even before the crisis in negative health effects. Marth of 1979. An interesting additional tinding in llromet's study was Ihat TMI REFERENCES workers leh more rewarded by their jobs than did control group workers, pmsibly because of the challenge of the special problems at the TM) facility. I. Ilromet E. 7hrec Mi/c Naml; Mentalucahh Findm.o. Western Psychianie A tinrd population studied intemively by llromet was clients of mental health Institute and Clinic,1980.

~ ' ~ -~- ~ ~ ~ ~ ( k i, IM Energy Environment, and the Economy .iy L ) 'I 4 i

2. Dohrenwend, B., Dohrenwend, II., Kasi. S. and Warheit, G. Retx>rt of the j

Task Force on RehavioralEffects ofIhe President \\ Commiuinn on ihe Ac-I cident at Three A1de bland, I979. {

3. I'lynn, C. Three Afile bland Tehphone Survey, Preliminary Report on Pro-i I

ceduret and Findmes, U.S. Nucicar Regulatory Commission, Washington, D.C.,1979. E

4. Gamble, Ii. and Das ning, R. Effects of the Accidentat Thrw Afile bland

{ On Residential Property I'ahrs and Sales, U.S. Nuclear Regulatory Com-mission, Washington, D.C.,1981. j

5. Goldhaber, hl., llouts, P., and DiSabella, R. Afobility of the Population if'ithin 5 Afiles of Three Afile bland During the Peri <xlfram August,1979 I

e Throughjuly,1980. Repori submitied IoIhe1 All Adsisory Panelon licalth 9, Research Studies of tc Pennsylvania Department of Ilealth,1981. j ", '? ^ i

6. Governor's Office of Pe,iicy Planning, TheSocio-Econonticimpacts of the i

\\ I l 2',."'. Three Afile Island Accident: Hnal Report, I f arrisbury, PA: 1980.

7. Ilouts, P., Afiller, R., Tokuhata, G., llam, K. Ilealth-Related Behavioral 3

I ,(,, impact ofthe Three Afile IslandNuclear incident Parts 1and11. Report sub-t mitred to the Tht! Advisory Panel on llealth Research Studies of he Penn-l 'i sylvania Department of Ilealth,1980. h'y.f

8. Ilouts, P., Afiller, R., llam, K., and Tokuhata, G. Estent and duration of

( :. - psychological distress of persons in the sicinity of Three N1ile Island Pro-l l credings of the Pennsylvania Academy of Science $4:22 28, l480 t

9. liouts, P., DiSabella, R., and Goldhaber, N1. lleidth.Related Rehavioral

tf impact ofthe Three AfileIslandNuclearIncident Part Ill. Report submitted g y.;,, j to the Th11 Advisory Panelon IIcalth Research Studies of the Penns>hania ea'*C, Department of Ilealth,1981. l

10. flouts, P., tienderson, R., and 51 iller, R. Family Practice Useand Revumse d}

l to the Three Afde Aland Crisis. Paper presented at the North American V, Primary Care Research Group, Lake Tahoe, CA,1981. i V i1. Ilu, T., Slaysman, K. Ilam, K., and Yoder, N1. Heahh-Related Economic Costs of the Three Afile /sland Accident. Report submitted to the T N11 Ad-visory Panelon llealth Research Studies of he Pennsylvania Department of J i IIcalth,1981.

12. Kast, S., Chisholm, R., and Eskenari,11. The impact of the Accident at the Three h1ile Island on the llehavior and Well-being of Nuclear Workers:

Report of the Task Force on Behaviond Effects of the President '.s Conunis. sion on the Awident at Ihree Afile bland, I979 I3. Seranton, W, Report to the Governor's Conuniuion on Three A1de bland. Ilarrisburg, PA,1980. I4. Shearer, D. Three Afile blandNuclear Accident Cennnwnity impact Study on Real Estate. liarrisburg, PA: Greater liarrisburg floard of Reahors, 1980 i 6' g Q-

.i e go. t 'st ~ ( W ( y \\ JOURNAL Or PREVENTTVE PsVCHIATRY ' binase I. Nomiker 3. lW2 beefe Ana tjebert. lee Fehlishers (.'m . nm, r,t, Mt Mental Health of Residents Near the ';m Three Mile Island Reactor: A L 1-Comparative Study of Selected Groups EVELYN J. BROMET,' DAVID K. PARKINSON,8 HERBERT C. /- SCHULBERG,8 LESLIE O. DUN'I.' and PAUL C. GONDEK' s s f ABSTRACT The purpose of the present study was to assess the mental health effects of the accident at the Three Mile Island (TMI) nacient plant. Three groups t thought to have been particularly affected by the accident were studied, namely, mothers of preschool children living within 10 miles of the plant: i workers at the power plant; and psychiatric patienas. Their peers in and around another nuclear plant in Pennsylvania served as controls. Interviews were conducted nir. months and twelve months after the accident. Both clinical diagnostic information (current sad lifetinee) and subclinical symp-tomatology were assessed. The results showed that mothers had an excess risk of experiencing clinical episodes of anxiety and depression during the year after the accident. TMI mothers also reported usere sympteens of sam-lety and depression at subc!!alcal levels at both laterviews _,_:.t. with controls. TMI mothers who were most symptosistic were those who had s [ psychiatric history before the accident, those who lived withis five adies of the plant, and those with less adequate social support. No diffesences were I observed between the TMI nuclear v'orkers sad controls, although TMI workers reported feeling more rewarded by theirjobs. In both asens, work-ers with grester social support were less synoptomatic. With respect to the [ I psychiatric patients, there were no differences between exposed and somen. ., ~ f This research was support by FGMH Contract #27s.3404s (SM) and unining grant .g.

h.,,.,4 j )

MH 15169 04 t >b j { ' Associate Professor of Psychiatry and Epderniology: 'A*sse Profener of Meheme: L Trainee. Psychiatric Epidemiology. Departrnent of Psychiatry. Western 7.AL. s=enne 8'}.'I'.['. 3 ..I ' Professor of Psychiatry and 7.i.c,:crey; *Princmal Research Assocase; 'Postdocsonal d { and Clinic University of Pittsbursh. 3811 OHara Sasset. Pimtmarsh. PA 13213 6 p I 225 2-- L / i..-. ,,a 4,..-> ..be -==T m.

l,, e !. j i \\ l BROMET ET AL. l posed patients. Although for mothers and workers perceptions of TMI as dangerous or living near a nuclear plant as unsafe were unrelated to sympto-matology, these perceptions bore an important relationship to symptom level among the patients. Mt 1 l bt ,i

r. 4 h g

.f= i INTRODUCTION 'd N MAaCH 28,1979, a serious acddent occurred at the Three Mile Island (TMI) 3 nudear plant. Unlike previously studied disasters, this accident unfolded over several days and threatened to destroy an entire region. The plant and the surround-t ing area continued to be threatened long afterward by technical problems and in-creased risks assemrM with the decontamination process and occasional radiation leaks to relieve the pressure inside the damaged system. Moreover, the controversy involving Metropolitan Edison's efforts to reopen the remammg undamaged re- ,i - l actor and constant attention by the media to TMI's persirtent danger ensured con-tinued public awareness of the incident and its aftermath. In light of these unusual - l stresses, the psychological well-being of residents of the TMI area became a major f public health concern. ~ r f The National Institute of Mental Health responded to this concern by funding an / epidemiologic invarigation of the mental heahh of three groups thought to have been parucularly affected by the stress of TMI: mothers of preschool children living l within 10 miles of the plant; mental-health system clients living within 10 miles of the plant; and workers at the nuclear power plant. Specifically, the study had three goals:(1) to eya== the psychological consequences of exposure to prolonged stress by comparms the snental health of the TMI residents with that of umdar residents living near a nonproblem nuclear power facility 8-9 months after the incident; (2) to determme dianges in mental health resulting from the additional stress of the anni-versary of the event; and (3) to assess the role of social support networks in modify-ing the impact of the stress. " Die long-term mental. health consequences of disasters have rarely been studied from an epidemiologic Ww. The majority of published reports are case W i sendies describing psydtiatric sequels of disasters such as floods (Bennet,1970; I.ogue et al.,1979; Titchener and Kapp,1976), tornadoes (Moore,1958; Moore and Friedsam,1959), cydones (Parker,1977), and fires or explosions (Adler,1943; Cobb andIi-L = = 1943; Leopold and Dillon,1%3). Each of these disasters occurred i suddenly and caused many deaths, extensive property damage, evacuation, and at F tirnes resulential relocaucu. In spite of the diversity of type of disaster, timing of oc-currence, and cuhural characteristics of the affected regions, most reports consis-1 tently show an increase of at least minorlevels of m;J hic disturbance (Edwards, ~ 1976; Kinston and Rosser,1974; 'tierny and Batsden,1979). Some investigators, however, have reported a high rate of psychiatric sequelae. ./ - } i q Moore (1958), for example, Interviewed community residents who had experienced 'r ] two tornadoes 13 months apart. More than 50% reported emotional problems in a --(,, i j family meneer one year later. Thsrty-six persons survived a marine explosion on the .Y.s s j 8,f Y zu at ~. . em N ~. t l u -m e =eg e o W(,e-d f M MW a

... - ~. - .~. ,y. e ~ j i s MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS l s I Delaware River in 1957; in the succeedmg four years,72% sought help for psychi-I atric problems (Leopold and Dillon,196 ). "Ihe most dramatic evidence for residual j emotional effects of a disaster was reported by Titchener and Kapp (1976) who eval-a usted survivors of the Buffalo Creek disaster. Two years after the event,90% of the

• . : g1r j

survihrs suffered symptoms such as anxiety, despair, severe sleep disturbances, and {$7 i*f nightmares. In fact, the descriptions of Buffalo Orek survivors were remarkably !,,I similar in nature to charactenstics of concentration camp survivors (Dor.Shav, I. } 1978). Finally, in the months following the Hurncane Agnes floods, Zusman (1976) ' d pgJ j found a 20% increase in intake interviews at the local community mental health cen-j 'ter, a sharp increase in state mental hospital admitsions, an incre2se in deaths due to 4 heart disease, and an increased number of suicides in the affected locales. ~ 1 The findings from disaster studies, although suggestive, have limited applicability I because of their methodological weaknesses. The mWority of the published data are ,1 based upon clinical impressions, retrospecdve judgments, and/or responses to un-I validated questionnaires obtamed from unrepresentative samples. Parker's (1977) i the only study that employed a standardtad instrument. Parker's study was also '/ ~ assessment of cyclone victims using the General Health Questionnaire appeared to be i unique in that he attempted to collect data longitudinally (immediately post. disaster, i two months later, and 12 months later). However, the representativeness of his small sample (67 out of 30,000 evac' aces) was somewhat questionable, and his attrition rate reached 73% by 12 months. The other major methodological weakness of disaster j' d research is the unavailability of baseline data (Zusman,1976). Disasters are unex. j pected, and this limitation may not be surmountable. A comparison group design , '/ provides a reasonable alternative, yet it has rarely been implemented in disaster re-search. Agamst this plethora of problems, it is not clear whether observations about the psychiarne sequelae of disasters are methadala91 artifacts or genuine effects. 2 Nevertheless, the consistency of previous results is intriguing. The ultimate goal of the research described above has been to demonstrate a causa! association between the stress caused by disaster and mentalillness. Since this asso- ] ciation could not be ascertained from existing disaster studies, we turned our atten-j tion to a related, albeit equally problematic set of studies, on the impact of stressful life events. Although the relative impact of environmental stressors and personal ,J, stressors is uncertain, the conceptual similarity is apparent (Dohrenwend and j Dohrenwend,1978). It is surprising to note that with the exception of Logue, Han- 'l sen, and Struening's (1979) five. year follow-up study of flood victims in central Ii Pennsylvania, the mental-health disaster literature has failed to consider relevant re- ,l search on major stressful events. Similarly, with a few excepnons, researchers exam-l. Ining the impact of stressfullife events on mental hes'th have not drawn'upon studies of disaster victims. Life events st'; dies were sekstively reviewed with three goals in mind. First, is there convincing evidence that psychological problems ensue from stressful life evems? Second, what factors are thought to moderate this association? l Third, can we extrapolate from these findings to the Three Mile Island situation? 4 A considerable body of research suggests tha: stress is related to numerous health . g

k l!

problems, both somatic and psychiatric. There are two kinds of life. event studies . q, g, L! which have investigated this association, one focused on effects of specific life crises .S. ll (e.g., rape, death of spouse, unemployment, ), and one utilizing life-y, Ij f 'f 1.. u 5 j M .'I i l s 5- ~ c ? l

t m

I y m"1B R F W" m* $**"f

i. >.,,y I BROMET ET AL s } events checklists. A variety of studies have shown that major life crises, particularly death of an immediate family member and rape, often have adverse psychological ramifications, at least in the short run (e.g., Moos,1976). With regard to life-events checklists, several case-control studies have reported an excess of adverse life events .l among psychiatric patients compared with normal controls (Paykel,1974; Brown et {lW ff q ty N] al.,1973a,b; 1978; Mue!!er et al.,1978). Several community studies have also re. ported a relationship between stressful life events and self-report psychological and fl t '*k ej )ff '($, 3.g physiological symptoms (Brown et al.,1975; Myers et al.,1971,1972; Mueller et al., j 1977). Because of many serious statistical and methodological problems (Brown, } 1974; Dohrenwend and Dohrenwend,1978. Rabkin and Struening,1976). It is diffi-a cult to assess the nature and strength of the observed relationship. However, the

i most parsimonious conclusion seems to be that vulnerable individuals (i.e., psychi-L

'g f.l atric patients) and persons suffering sustained loss may develop an illness episode j upon exposure to substantial stress (Zubin,1978), whereas less-vulnerabic individ-uals (untreated community respondents) respond with minor, transient synptoms or I show no effects at all(Dohrenwend and Dohrenwend,1978). Because life events appear to have their strongest effects on vulnerable individuals, 1 this study focused on three groups thought to be particularly affected by TMI. Psy. '/~ j chiatric patients were selected because the risk of subsequent episodes or increased i symptomatology fo!!owing a stressful occurrence appears to be heightened (Andrews j and Tennant,1978). Mothers of preschool children were considered to be at greater /- e risk than the general population because (a) there was some evidence, albeit con-at troversial, that mothers with young children have a high rate of untreated minor de. j 'f. ,3 pression (Brown et al.,1975; Richman,1976); (b) those living within five miles of l TMI incurred an additional stress of evacuation; (c) mothers had concerns about i radiation damage to their children which increased the stress of TMI; (d) surveys im-mediately fo!!owing the accident showed increased anxiety among such mathers ji (Dohrenwend et al.,1979). Nuclear plant workers were selected because they had the greatest exposure to radiation as well as the uncertainty about future employment. It j should be noted that the mental health of nuclear plant workers has never been 4 studied, apart from a descriptive report appearing in a French journal (Sivaden and Fernandez,1968) and a telephone survey conducted by the Behavioral Effects Task Force of the President's Commission on the Accident at Three Mile Island, showing relatively high demoralization scores among TMI workers 6-7 months after the inci- 'j dent (Dohrenwend et al,1979). Given the size of the nuclear industry, the data from this part of the study will contribute substantially to the field of occupational mental health. In addition to an individual's predisposition to psychiatric disorder,it has been ' suggested that the noxious effects of stress may be modified or reduced by social sup- ? 5 port (Antonovsky,1979; Cassel,1976; Cobb,1976; Dean and Lin,1977; Henderson, 1977; Nuckolls et al.,1972). According to Caplan (1974), the most important factor affecting outcome of people exposed to stress is "the quality of the emotionalsup. J' port and task. oriented assistance provided by the social network within which that 'I individual grapples with the crisis event"(page 4). A number of studies attempted to .9 ,j'h ./ investigate the protective role of social support in psychiatric disorders. The support provided by confiding relationships appears to be salient in a number of studies a.p, 3 (Brown et al.,1975; 1.awenthat and Haven,1968; Miller and Ingham,1976). More. Q L i 228 D'd. o 2 /R ~ " * ' * =. 8-6 t 4 p m. m- -==e.-- - * = = -. ~ -~

I. e ) MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS l s ) over, Myers et al. (1975) found that individuals experiencing many stressful events I but showing few symptoms were more socially integrated than individuals with few life events and many symptoms. Thus social support as a =c L-Gig.vanable of-fered a possible explananon for the consderable diversity observed in responses to m ,1 A;ll ~ stress.. N l Relatively fewrv44t s studies have critically evahtated the role of social sup. , drp 'i ek j port networks in relation to emotional disturbance. Existing studies typicaDy use a ( case. control design in examming differences in networks between psychiatric pa-4 tients and normal controls (Henderson et al.,1977; Pattison et al.,1975; Sokolovsky N j et al.,1978; Tolsdorf,1976). These studies have demaa<trated the the networks of 1 j psychiatric patients tend to be smaller and more asymmetric and to contain a higher proportion of relatives. Two promment prospective studies should be noted. One L 1 was the analysis of Human Population Laboratory data by Berkman and Syme g (1979) showing that less socia!!y integrated individuals subsequently had higher l mortality rates during a nine year follow.up period. The other study which is more j relevant in this context is the longitudinal investigation by Gore (1978) of blue collar / j men whose jobs were ternunated.by plant closure. She showed that suppornve. / ness of wives modified " severity of psychological and heahh.related responses to g f unemployment." a Crisis theorists such as Lindemann (1944) and Caplan (1964) haveasserted that life J / I crises provide significant learning and growth experiences. This position has beenad. vanced by Dynes and Quarantelli(1976) based on the world-wide work of the Ohio ./ Disaster Research Center. They indicated that their long-term, post-disaseer observa- / tions pointed to increased cohesiveness in the social networks of anaffected popula. / l tion following a disaster. It is quite possible that the situation at TMI -- --p existing networks. Although research directed at the role of social support systems is in its infancy, it provides a prarnning frarvework from whids to view the impact af 'I major stressful occurrences. The nuclear incident at TMI was sufficiently different from othermanual chsaseers 4 i that our ability to make pr=&tians based on previous findings was hmand as much by TMI's uniqueness as by the rnsehndological weakneses noted above. Specifically, i no lives were lost, no property was damaged, and although 60% of the pntadatina in l the ares evacuated, no one was suinequently relocated. Moreover,exacerbances of the " initial stress" (Tyhurst,1951) continued, 44 upon new revelations about the incident (e.g., it came within 30-40 minutes of a meltdown), redtaren leakaps ^ into the air and water, znalfunctions==w,=ted with the clean-up process, and the Intensity of discussons about reopening the undamaged reactor. Thus, this study i provided longitudinal data on a unique simation. Since there are 72 nuclear power s facilities in the United States, the rmdings could be of critical importance to the development of maximally efficient and effective nuclear =mdent or dsaster plans s l. involving millions of people. METHODS t Table I presents the six groups who were interviewed for the study. Initialisme. , ' P,4 3 . :h..TI views (Time 1) were condacsed in December,19796 and January,1900 (i.e eightso c l ~ 8* I, 9-m -===m. 0

• Mn 4
• /

,.no, a. . ~ w--o.e.e

p ) 1 BROMET ET AL Tants 1. STtror PortunoNs + Three Mile Island yime i Time 2 Attrition % .,8 1.h"M! Mothers 328 312 4.9 Workers 189 161 14.8 . f Mental Health Chents 177 151 14.7 q' j comparison site Mothers 133 124 6.8 ' Workers 143 124 13.3 i Mental Health Clients 70 64 8.6 3 nine months after theTMIincident). Follow-up interviews (Time 2) began on March 28,1980, the first anniversary of TMI, and were completed by the first week of May. De followns docussion delineates the rationale and method for choosing each of / _. 4 the groups and a justification for the comparison site. ]

1. Rationale and Methods of Sample Selection: Mothers ofpreschool.childres, j

particul.riy those living within five miles of TMI, were' selected because they were /~ thought to have experienced additional stress as a result of the evacuation order. j Furthermore, they may have had concems about potential radiation effects on their children. Data ran,cred shortly after the amdent Indicated that recent mothers ex-j perienced a great deal of distress and anxiety (Dohrenwend et al.,1979). , j-The sample of mothers was selected in the following manner. All birth announce-1 ments appearing between January 1,1978, and March 28,1979, in the Dauphin. York, and I.ancaster County newspapers (TMI sample) and the county newspaper e for the companson site were examined. I.ists of mothers livings within five miles and 5-10 miles of the nuclear plants were compiled for each of the two sites. Distance from the plant was deternuned by checking addresses against detailed maps of the areas. Where ponnbie, addresses were verified using 1979 telephone directories. Ran. g (: dom samples were drawn from these lists and randomly assigned to interviewers. C The target sample of mothers in the TMI area was 432, ofnehom 328 were beer-F viewed,69 refused to be W. J., and 35 were never reached. At follow-up,312 mothers were resterviewed, nine refused, and seven could not be reached. In the pl compenson area,197 mothers comprised the target sample, c,f whom 133 were inter-i viewed,56 infused, and eight were not reached. At follow-up,124 mothers were l reinterviewed, five refissed, and four couki not be reached. We were concerned about thepossibihty of systematic differences between the 35 TMI mothers whom the interviewers had not been able to reach initially and the 328 g who were inserviewed in the-twr. Therefore, brief telephone int:rviews were con-l, ducted in May,1980, to gather information on sociodemographic characteristics, significant life events, and perceptions of danger concernmg the TMI accident. From the group of 35 women,27 telephone interviews were completed. Of the remaining

p.,;'.f l eight wornen, five refused to be interviewed, and three could not be reached in spite 1 -Q k

e l of numerous attempts. An examination of the informat2on obtained from the tele-tap. i phone interviews indicated that the 27 nonrespondents were not different in any re- ', Y,[. spect from the 328 women who compnsed the study sample. Telephone interviews v ? 230 .R l ..e ll l - ~ - 1-, 1. 4 y 4 l, ~ l' ,/ . ~,.. _. ,,--...._wn -w--.

J MENTAL HEALTH OF THREE Mll.E ISI AND RESIDENTS were also completed with five of the eight comparison site mothers whom the inter-viewers had been unable to contact. No unusual characteristics were found in this group. Workers at the Three Mile Island nuclear plant faced the greatest exposure to radiation as well as potential job loss Since the planned study groups were relatively -f '[t small, the decision was made to icterview a homogeneous group of workers rather t than attempt to select a random sample of all people employed at the plants. We , la,-(g , NN

• (d [

therefore restricted the respondents to persons selected randomly from union mem. bership lists provided by the International Brotherhood of Electrical Workers .(IBEW) in the two study areas. Members of some of the smaller Building Trades Council unions (construction unions) at TMI also participated. These workers were then randomly assigned to interviewers. j e At the time when both sets of interviews were conducted, workers in the two areas were working 12 hours / day, 6-7 days / week. We recognize that their participation was a sacrifice of their scarce free time. There were 347 TMI workers who comprised the target sample, of whom 189 were interviewed, 82 refused, and 76 were never reached by the interviewers. At follow-up,161 workers were reinterviewed,17 re- / fused, and !! could not be reached. The 17 refusals included three operstmg room technicians who were counseled by their lawyer against participating because of a N pending lawsuit. In the comparison site,199 workers comprised the known target r samp!c, of whom 143 were interviewed,34 refused, and 22 were not contacted. At follow.up,124 were reinterviewed,9 refused, and 10 could not be reached. i Additional attempts were made to contact workers whom the field interviewers had been unable to reach initially. Of the 76 TMI workers and 22 comparison site j workers,48 and 13, respectively, were reached by telephone. Again, no differences f were found between them and the study group with respect to their demographic characteristics or to their responses to questions about the dangers associated with I the accident. Mental health system clients were theoretically the most vulnerable group in the i study. The possibility existed that the stress of Three Mile Island may haveincreased j their symptom levels. Furthermore, some of the clients were believed to be chron-ically disabled and lacked the financial and mechanical means to leave the area at the height of the accident. A report carried out under the auspices of the Nuclear Regu- 'l !atory Commission indicated a socioeconomic bias with respect to evacuation (Flynn,1979). i Anonymous lists of adult mental health system clients treated durms the six months prior to the TMI accident (i.e., active cases), excluding those with retarda. tion and organic brain syndrome, were obtained from three community mental health center facilities near TM1 and the one mental health center near the comparison l, site. A list comprising all addresses within a 10-mile radius of the nuclear facilities was constructed for each site. These lists contained only iderttification rtumbers, not names, to preserve the confidential nature of treatment. Random samples were l-drawn from these lists and randomly assigned to eligible interviewers. The lists were [$ then returned to mental health center staff who reattached names to the addresses. g.$ ) '2 Professional staff of the mental health centers conducted the interviews with the f g]b pi i

clients, L

The c!!ents were the most residentially unstabigand hence the most difficult group

• p*'

. n.h., l y 't h 23i 9 1 i,

' --~

e se. m= b

• %e.

o* e l ./ ... ~ A - -~ w e ~

........ ~.. -.. .,e BROMET ET AL s to locate. The final sample represented extensive work in the field by the inteniewers in attempting to locate many of the clients. In the TMt area, the known target sam. ple was 244, of whom 177 were interviewed,41 refused to be interviewed, and 26 were never located. At the follow-up interview,151 were interviewed,16 refused a second interview, and 10 could not be contacted. In the comparison area,118 clients comprised the target sample, of whorn '10 were interviewed,20 refused to be inter- - 3 .t' g,(((( viewed, and 28 were never located. At foDow-up,64 were interviewed, one refused. j4 one was deceased, and four emdd not be contacted. q

2. Comparison Site: Beaver County (Western Pennsylvania) was selected as the h,[ry j

comparison site for several reasons. The adjacent Shippingport and Beaver Valley . '; r nuclear reactors are located there. According to 1970 census information, the popu.

• '(

i. lation characteristics and area density levels were reasonably comparable to those of TMI. This comparison site was selected to rule out the possibility that differences in 2

I mental health between respondents In the two areas could be attributed to livirg near

~~ L '.i a nuclear facility per se. Ideally, a second set of control groups from an area)vith a fossil-fuel power plant should also have been studied, but financial constraints pre.

i c!uded this possibility. Our follow-up research on TMI includes this additional com-parison site.

\\ j - .I s B. Study procedures

1;;

A letter was sent to each potential pamcipant explaining the nature of the study and noting that an interviewer from the project would be calling shortly to arrange an appointment at a mutually convenient place and time. Most of theinterviews were j 1 conducted in respondents' homes and lasted 1M hours on the aversge. Participants who agreed to the study signed a consent form indicating that they understood its nature, and that their agreement was voluntary and could be withdrawn at any time. ~' i Furthermore they understood that there were no risks or benefits associated with the interview. They were also issued a U.S. Cme.

Certificate of Confidentiality to L

further protect the privacy of their rapnnene i C. Data collectiort l Table 2 provides an overview of the informanon obtained at each imerview and reported on in the present paper. "Ihe following anninn brieDy describes each of the instruments used to obtain this information. ~ TAstz 2. AnaAs of DaTk Col.taCTEN Nm* e Month Interview Follow up Interview Background Information Current Symptomatology Social Support Life. Time Mental Health Current Symptomatology Current Job Stress Current Job Stress Mental Health in Past Year r ~ Measures of Stress Y.j, 3 k or l

• t'y f *
• k.p M

,Q ." ;c.d b f 0 s I %I g. i ~ s 'l ~ / -v,e .w-

... ~ MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS s ' Background h formation: A sociodemographic questionnaire was designed to' t collect standard information on age, sex, race, edu* cation, occupation, income, fam-ily size, age of children, and length of time at present residence. Respondents were also asked to estimate how far they lived from the power plant. Since demographic i (" g ;,.Q characteristics are sometimes associated with mental health, adjustments for demo-r t. f, s graphic differences were employed when companns the mental health of respon-dents from the two areas. ,'. e f.* Social Support: One possible determinant of the subjective impact of a major T Mel stressful event is the support offered by an individual's primary social network. A

• M
• b social support interview schedule developed by Mueller (1978) was modified and
• 'h 3

,4 used in the present study of TMI and Beaver Valley respondents. This interview pro-vides detailed, systematic information about an individual's social support. J The social support interview schedule, administered during the first session, con-LA 1 tained four sections. Respondents were first asked to list a maximum of 15 persons (family, neighbors, work assoczates, club members, friends) with whom they had an i ongoing personal relationship during the past year. Five questions were then asked . ~ about each of these 15 individuals. These questions dealt with geographic proxunity, / ,{ frequency of contact, feelings toward the person, and chances in the relationship or i in the amount of contact. Respondents were then asked to select from the original 15 j perso.ts no more than five persons to whom they felt closest (i.e., the inner circle). ~ *, Six questions were asked about each of the inner circle members. These questions

• i dealt with practical and emotional suppon from these people and similar support i

given to these people by the respondent. Respondents were then asked to select the one person to whom they felt closest (i.e., their confidant), and four additional ques- / i tions dealing with expressiveness in the relationship were asked. Finally, a fourth seo. tion contained questions about general availability and use of support from the _ = network. g Factor analyses were performed on the social network interview schedule for the combined mother group (N = 449), the combined worker sample (N = 327) and the combined client group (N = 207). The analyses used a principal components procedure and varimax rotation. The results of all three sets of factor analyses were ^' comparable in deriving the same six factors for each of the groups (i.e., mothers, workers and clients). These factor analyses provided the basis for the development of six subscales composed of the items with the highest factor loadings. Table 3 de-i scribes the content of each of the subscales. Three of the subscales deal with struo. tural characteristics of the network (size, accessibility, stability), and three deal with quality cf support (general network support, inner circle support, and confidant support). The average intercorrelations between the subscales were 0.17 for the mothers. O.14 for the workers, and 0.19 for the clients. Current Symptomatology: The Symptom Checklist, a 90-item self.admmistered scale, was given at both interviews. This checklist inquires about symptoms during the two weeks preceding the interview. The items are rated ort a five-point distress scale (not at all, a little bit, moderately, quite a bit, extremely). The psychometric i ' j,.3 y properties of the Symptom Checklist have been established (Derogatis et al.,1977). +"PQ i The scale has been factor analyzed, and nine symptom dimensions have been identi-l A . %,, g' - + 233

1. y, g t'. >}

b.g "I'- - * =. _ r g w.. w 'I ---3:se -y e"Te-gM' gp-g-' S

,e + BROMET ET AL TAat2 3. SUasCAtis of soc 1Ar. SUFFoaT Structural Characteristics 1. Network Size - Total number of members (up to a maximum of 15) F' ' ccessibility. f,;[r, 2. A b. - Gwgic proxim'ty - Frequency of contact if,9'd I

• ,,{ %

3. Stability

- Change in relationslup during past year l - Change in frequency of contact during past year Quality of Support 4. General Network Support. j t - Generally someone to turn to - Generallevel of satisfacuan with network - Assessment of thoughts and feelings toward all members 5. Inner Circle Support - Reliance on inner circ!c members for emotional and practical support f - Reliance of members on respondent for emotional and practical support

6. Confidant Support

~ s - Expressiveness with confidant f. - Confidant's expressiveness with respondent / / fled. In this report, we will focus on thee of the symptom dimensions, namely De- ./ pression, Anaiety, and Anger, and an overall distress measure (Global Severity 8 ' Index). / Table 4 shows the items whida comprse the Depressen, Anxiety, and Anger sub-scales of the Sympte.n Clwektar I 4fe. Time Menef Healds: A modification of the life-time version of a standard-Ized psychiarne interview schedule (Schedule for Affective Disorders and Schizo. phrenia or SADS.L) was used for makmg mental health diagnostic evaluations. The SADS is a structured interview that provides systematic and reliable data for classi-fying psychiatric disorders according to the Research Diagnostic Criteria (RDC) of Spitzer, Endicott, and Robins (1978). The SADS was originally developed in con-junction with the RDC"... In an effort to reduce information variance in both the descriptive and diagnonne evaluation of a subject "(Endicott and Spitzer,1978, page 837). Because the SADS has been widely used in efineal investigations, information exists regardmg its reliability and validity. The SADS-L was selected for the present investigation because it off.:rs a cxxnpiehensive assessment of a person's mental health history. l-The two kinds of clinical disorder with the highest prevalence in the general popu-lation are depression and anxiety. The criteria for depression are as follows: (1) hav-ing a period of at least two weeks duration charactenzed by symptoms of depression, sadness, hopelessness, irritability, or the like; (2) seeking help from someone for those feelings, taking medication for the symptoms, or acting differently with people ,i I during these thnes; and (3) havmg three or more of the following symptoms (four or l snore if the epssode was current): %p ( t' !.M 7' *p 234 ~- ~ !~ l \\ M l ,o l ~ - - - .-,ne

e-MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS TABt.a 4. SYMFroMs CoWralstNo THz ANxnmr, DsratssoN, AND ANCza SUBsCAtas Anxiety. Depression 3 p, Nervousness or. shakiness inside Loss of sexual interest or pleasure [Mt Trembling Feeling low in energy or slowed down Suddenly scared for no reason Thoughts of ending your life ,py'A' ! = l Feeling fearful Crying easily , I PJ Heart poundin:g or racing Feeling of being trapped or caught ~ Feeling tense or keyed up Blaming yourself for things Spells of terror or panic Feeling lonely 3 Feeling so resdess you couldn't Feeling blue L sit still Worrying too much about things Feeling that familiar things are Feeling no interest in things strange or unreal Feeling hopeless about the future Feeling pushed to get things Feeling everything is an effort i c-done Feelings of worthlessness / Anger ~ Feeling easily annoyed or irritated Temper outbursts that you could not comrol - "I Having urges to beat, injure, or harm someone Having urges to break or smash things / Getting into frequent arguments . /- Shouting or throwing things I 1. poor appetite or weight loss, or increased appetite or weight gain. 2. trouble sleeping or sleeping too much. 3. loss of energy, easily fatigued, or feeling tired. 4. loss of interest or pleasure in your usual activities or sex. 5. feeling guilty, worthless, or down on yourself. 6. trouble concentrating, thinking, or making decisions. 7. thinking about death or suicide. 8. being unable to sit still and having to keep moving or the opposite-feeling slowed down and having trouble moving. "Ilie criteria for anxiety are: (1) having a period of at least two weeks duration char. acterized by feeling anxious, tense, jittery, " uptight" or the like most of the time: (2). seeking help from someone, taking medication, or behaving differently with other people because of these symptoms; and (3) having at least one of the following symptoms: I. difficulty fa!!ing asleep. 2. sweating, blushing, dizziness, palpitations or shortness of breath.

r. /,f Gey

! g.l,*tr[*g.~ 3. muscular tension or tremors. P= 4. worrying much of the time about things that might happen. o 'f tj g 5. fidgeting or being unable to sit still.

1. ,g,,

235 I ene o e... m e. f

• ep..

'I ,____s ~ ~ ~ ~ ~ ~ ~ ~ ,vf g y e4 ee-- e-g-*

BROMET ET AL In each category, respondents had to meet all three criteria to qualify as having h an episode of either anxiety or depression. Other major disorders covered in the SADS.L include mania, schizophrenia, alcohol abuse, and drug abuse. More tran-s sient disorders, such'as minor depression, hypomania, phobia, panic, and obsessive- .' ' i compulsive disorder, are also assessed. The last four disorders about which the Q!t SADS inquires are recorded as hfe tirne personality disorders. These include labile. cyclothymic and anti social personality, and Briquet's disorder. 4;0 Wo,-k stress is also an importut aspect of overall emotional well-being. A ques.

• 7+' f~'h

.tionnaire developed at the 11niversity of Michigan (House et al.,1979) was adminis. tered to all gainfully employed respondents at both sessions.This questionnatre con. tains information on the following: responsibility pressure, quality concern, role i 3 conflict, job vs. non. job conflict, work load, Type A behavior pattern, job satisfac-tion, occupational self-esteem, intrinsic newards, extrinsic rewards, importance rewards, and control rewards. These 12 subscales were factor-analyzed, using a prin-ciple components procedure and vartmax rotation in order to group them into mean-ingful categories. Two factors emerged. The first factor encompasses the first six

• / -~

subscales and the second factor includes five of the six remairung subscales. As a result of these analyses, two summary scales were derived by summing the in-dividuti subscale scores. The first scale is indicative of Overall Job Pressure, and the ~' second summary scale assesses Overall Job Rewards. Stress Variables We were concerned with the possibility ofinfluencing responses 3 on the mental health questionnaire by foassing on perceptions of TMI's danger. Herefore, only two items were included and analyzed in this report. One item in. quired about the respondent's judgment as to whether the situation at Bl! was cut-rently dangerous (yes or no). De second item assessed the respondent's judgment ' about the safety of living near(within a lO. mile radius) a nuclear power facility. Two other variables considered to be indicative of stress were dastance of residence from r the plant and the decision to evacuate. D. Interviewers The quality of data obtained in mental health interviews is clearly a function of the experience and competence of the interviewer. Dus we were i.u,- J not onlywith selecting the most valid and rehable mental-health measures possible but in making sure that our interviewers were highly quah5ed to administer the mental health por-tions of the interview. Twenty-six TMI asesinterviewers and 11 Beaver County inter. viewers were hired. neir qualifications incinded a minimum of four years of clinical experience in mental heahh settings and a Masters or Doctoral degree hs a related field. We also asked about the interviewerf attitudes toward nuclear power in order to eliminate candadates with strong biases in either direction. Respondents within each of the three study groups were randomly assigned to the interviewers. E. Data analysis yf In addition to Nye analyses that provided means and standard deviations or - P'ti % counts for the many variables measured in this study,several different analyses were (p,' P,Q designed to investigate specific questions. . ', L,, .) ,6 '.s s -. ' *--me. .L. l .e.., w,. i eem. e ~~ '

MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS Diagnorist Differences between sites in rates of diagnosable mental illness were investigated using risk ratios. Risk ratios were calculated by dividing the rate for TM1 iespondeats by the rate for Beaver County respondents. If the lower limit of the con. fidence interval for the risk ratio is greater than I, then the risk ratio is considered to be significant. For these analyses, diagnosable illness was defined in one case as any dy<mble illness and in a second case as anxiety and/or depression. These analyses ,.7 !u. were not ccaducted for clients since, by definition, all of the clients had a diagnos-l E, S able condition. r,i Contributions to 'any differences in the rate were examined separately for each site I using log-linear analysis. The design of these analyses was usually three dimensional,

h. f 6

h crossing diagnosis (presence or absence) by stress variable (TMI dangerous versus not; nuclear plant safe versus not; evacuated versus not; living within five miles of TMI versus not) by trichotomized social support variable (high, medium or low inner g circle support; generaly network support; network size; confidant support; accessi-bility; or stability) in all combinations. The two sites were analyzed separately be-cause at times the Beaver County sample did not fill the cells for the three dimen-sional design and a simpler design had to be used. Further information about the tog linear procedure can be found in Dixon and Brown (1979). /- Symptom Levels; Differences between sites on Anger, Depression Anxiety and the Global Severity ladex were examined in analyses that also considered respon-dents' scores at Time I and Time 2 and presence or absence of psythiatric history (this last dimension was omitted for clients). This analysis was a 2-between, l.within (repeated measures) analysis of variance (ANOVA). BMDP.79 routine 2V was used [ for this analysis so that reasonably unbiased estimates could be obtained despite un-equal cell sizes. It was possibic that sociodemographic and social support variables /,.- could have affected the means and thus influenced the observed differences between 3 ' sites. Thus, analyses described above were extended by including covariates. Spe-cifically, the differences between the means for TM1 and Beaver County respondents were reexamined adjusting for known differences in variables that either predated I the accident or also correlated with mental health. In addition, these analyses were replicated with the data transformed by the ex. perssion In(x + I) because of concerns that the data might not fit the assumption of the ANOVA (i.e., some heterogeneity of variance and a loose, but possibly syste. matic, relationship between cell means and standard deviations was observed; the natural lognrithm transformation is the standard adjustment for these problems). Since the results for the transformed and untransformed analyses were nearly iden-tical, only the untransformed results are f.esented in this report. Further investigations of the effects of psychiatric history and social support on symptom levels were carried out using ANOVA that crossed those variables with each other and with the four stress variables (three for Beaver County) in all possible combinations. These analyses were performed separately for each site because of ce!! size problems with the Beaver County site. FINDINGS: MOTHERS ~ Jackgrowrd Characterirrics-The 312 mothers in the TMI area and the 124 moth. ...# *I ersla the Beaver County area were quite similar (Table 5). He median age was 27-28 l: A (- 237 1i !.. fr i ~ '., _.t .e OGb

q..

-me m== l \\, \\ 4 = ..,,.agepen

? BROMET ET AI_ TAatz 5. BAcxcmouxo CHAaAcTsatsTics of MoTurns Tbil Beaver County ~ N N Total - 312 100.0 124 100.0 'l.[ii Age s 25 92 29.5 35 28.2 fW2 [ M,. ;I# 26-29 114 36.5 40 32.2 30 + 106 34.0 49 39.5 f Education s 12 years' 158 $0.6 55 46.3 Some college 77 24.7 47 37.9 Co!!ege graduate 77 24.7 22 17.7 or more Religion Catholic - 86 27.6 54 43.5 Protestant 205 65.7 59 47.6 Other or None 21 6.7 11 r9 Employment Employed outside 115 37.0 32 26.0 / Status the home Housewife 1% 63.0 91 74.0 (Missing) (1) (1) Total Family s 14,999 60 22.1 7 5.8 Income 15-19,999 98 36.0 25 20.7 20-24,999 63 23.2 61 50.4 25,000 + SI 18.8 28 23.1 (Missing) (40) (3) ,/ Where Grew Up Same Area 188 60.3 86 69.4 Elsewhere 124 39.7 38 30.6 i Number of I 118 37.8 52 41.9 Children 2 13 6 43.6 47 37.9 3+ 58 18.6 25 20.2 Distance from 0-5 miles 134 42.9 56 45.2 Plant 6+ miles 178 57.1 68 54.8 years, and two-thirds of the mothers in both areas were under 30 years of age. Most of the mothers had one or two children. About half of the mothers in each area had some formal education beyond high school. In both sites, most had grown up in the same area in which they were currently living, and slightly fewer than half lived l within five miles of the nuclear plants. Although not depicted in Table 5,98% of the i mothers in both areas were married, and 97% were Cair=6m In addition,19 TMI mothers reported that they were pregnant at the time of the accident. There were some demographic characteristics on which the mothers differed. TMI i mothers were predominantly Protestant, whereas in Beaver County, Catholics and Protestants were equally represented. Relatively more mothers in theTMt area were emplopd outside the home, although the total family income appeared to be lower

.*f,, )

238 .c '. O. u l .. l.' i .s., l ~. m m em.. i ./ e e e= 3se.-- . =.. + + -ee upsm-- -we-= + - ~,

e MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS among TMI mothers on the average. (However,40 TMI mothers did not answer the question about income, and this may have influenced the findings on this variable). It should be noter! that there were no demographic differences between mothers =- : ' who were Interviewed at both the 9-and 12 month sessions and those who were not . -7

• V reinterviewed at the anniverstry.

[. h i Percept /ons of Strusar As Table 6 shows, mothers in the TMI area were more likely to view the plant as dangerous than comparison site mothers. Two. thirds of

• 'l' '.I the TMI mothers perceived the situation at TMI as currently dangerous, whereas half of the Beaver County mothers viewed the situation as dangerous. Similarly, two

{ - thirds of the TMI mothers felt that !!ving near a nuclear facility was not safe, whereas half of the Beaver County mothers felt that way. Consistent with the perception of k g TMI as dangerous, most TMI mothers had participated in the evacuation (273 out of 312) even though the Governor's order was aimed at mothers with small children liv-Ing in the five. mile radius. In fact, all of the mothers in our sample from the five-mile radius did evacuate, and 78% of the remaining mothers in the 6-to 10. mile radius / - also evacuated. The 39 mothers who stayed in the area at the height of the accident were similar demographically to the rest of the mother sample. Errent of Cfinical Disorderr The key issue addressed in the present study was whether the stress =~4='M with the accident increased the rate of psychopathology '/. In mothers of small children. Hgure I depicts the monthly prevalence rates of affec-tive disorders (anxiety and depression combined) from November,1978 through De- ' *: e ',. cember 1979, and the rate in April,1980. The data for the November,1978-De-amber,1979 period were derived from retrospective accounts obtained during the first interview. In the first interview, a mental health evaluation modeled after the / SADS.L was used, and mothers were probed about the timing of any affective epi-l Tant.:6. DisTaratrrioN or Resrowsas to QtmsnoNs Anotrr Tus Accmerr: Montras TM1 BC N N 312 100 124 100 Do you currently feel the situation at TMIis dangerous? Yes 203 67.7 65 53.2 No 97 32.3 $7 46.7 (Missing) (12) (2) 1 Do you feel that it is safe to live near (within 10 I miles) a nuclest power ~ l facility? ',r[ l' Yes 97 32.1 63 S t.6 No 205 67.9 59 48.4 4 (Missing) (10) (2) ..' s.. y k4.Q: .t . 6 'pJ+g, l 239 '^ i eo e.-* l [ t l l.

e-I BROMET ET AL. 2s e--* TMI. Ms312

• *j.'.*W 3,

20' . ~ e s.0. Nst24

• l;l

& R t;>t e c . ', g

• s w.

,3.. G z T 10 1 t 5 ll. .L x. .....y.,___, / rt NOV JAN aPR. JUL OCT DEC APR s 52 78 73 ) FIGURE 1 Point Prevalence of Anxiety /D,4 by Month: Mothers. / sodes occumns in the past year.The April,1980 rate is based on the formal SADS-L ,/ administered during the second interview. A peak period of episodes of affective disorder can be seen in TM1 mothers at the time of the accident. Of the TMI mothers reporting at ! cast one episode of anxiety and/or depression during the 13-month period frota November,1978-December, 1979, more than half had reported an episode at the time of the accident. By con-trast. there were no outstanding peak times for the Beaver County mothers with a disorder. Among the TMI mothers with a disorder half spontaneously attributed their state to the TMI situation. By April 1980, the lines seem to have converged. und the rue for TMI mothas was similar to that of Beaver County mothers at that time. However, point prevalence rates based on a relatively small sample are un-stable. The more judicious w. , Mus are those which namine differences in pe-riod prevalence rates for the year following the accident. Tables 7a and 7b show the risk istics for all psychiatric disorders and for anxiety / depression in particular before and after the TMI accident. These data are based on information from the formal SADS.L nommitrered daring the second interview and the percentage of cases during the year afterthe accident is somewhat lower than the rate estimated by the mental heahh instranent used in the first interview. In this case the risk ratios were calculated by dividing the rate for TMI mothers by the rate for Beaver County mothers. 'Ibese tables show that (1) there were no differences be-gl tween sites in the proportions of mothers reporting inness episodes before the acci- - j dent; (2) TM1 mothers reported significantly more disorder during the year after the ) :,.'

• .4 accident than Beaver County mothers.This was particularly true for new cases devel.

.,( > ; oping post 'IMI. More specifically, theriskintio for new cases of anxiety / depression ,, ',*'p o *.., . 'u.. l}', 2e ',',Q .- :: -~ d I. I ob -* " * '~* -~-

• • +

g e. MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS TAaLI 7a. I rrrrnicon or ANxzrTY/DErnzssrow IN RrunoN j.'. To Tus TMI ACCIDENT: Moturns ,I Risk (Confidence .. c Hhtory Dil BC Ratio Interva0 g p .p' 1.M . Cases pre.TMI 84 31 1.08 (0.756, 1.534) 0.411 0.341 !] Cases post.TMI 45 10 1.788 (0.950, 3.J66) 1.802 0.036 New cases post.TMI 25 3 3.399 (1.157, 9.984) 2.226 0.013 Recurrent cases ( g post.TMI 20 7 1.054 (0.495, 2.246) 0.137 0.445 T4at:7b. Lzxxunoon or ANY DIsotDEn IN REr.AMoM To Tus TMI ACCIDENT: MoTHEas Risk (Confidence l History Dil BC Ratio Intervay z p Cases pre.TMI 93 44 1.232 (0.870, 1.746) 1.174 0.1:0 / Cases post-TMI 78 17 1.824 (1.154, 2.882) 2.573 0.005 New cases post.TMI 26 3 3.720 (1.286, 10.761) 2.424 0.008 /.. Recurrent cases i/ . post.nll $2 14 1.198 (0.801, 1.793) 0.880 0.190 __ _ /.. post.TMI was 3.4 compared with 1.8 for all cases occurring during that one. year

• period.

What effects did perceptions of danger, distance of residence from the plant, evao-nation, and perceptions of the safety of living near a nuclear facility have on re-por:ad episodes of disorder after the TMI accident? How did social support affect illness reports after the accident? Since differences in rates post.TM1 were found, separate log. linear analyses for each area were used to examine factors that might have affected those rates. These analyses were conducted for all possible combina-tions of diagnosis (anxiety / depression; any diagnosis), stress (danger; safety; dis-tance from plant; evacuation), and social support (general network support: inner cirde support; confidant support; network size; accessibility; and stability). De analyses revealed that neither stress nor social support, nor their interactions, af-fected the presence or absence of diagnosable illness. These results were true for both the 'Bl! and Ber.ver County sites (see Table 8). Sepa. ate analyses of possible effects of demographic variables also failed to iden-tify specific correlates of diagnosable disorder. Thus, the only variables that affected ..,[ b the likelihood of disorder occurring after the accident was site, with TMI mothers Y',7 i being significantly more likely to report a diagnosable disorder. .M Cwrent Symptomatology: Current degree of symptomatic impairment was as-

• • ?

sessexi at each interview using the Symptom Chectlist.90. Anxiety, Depression. An- ,, (.*}g ' ser, and Overall Distress (Global Severity Index) are described in this report. This section is divided into three parts: (1) comparisons between TMI and Beaver County ..T. i...z ..:s ,n 241 /

s'. BROMET ET AL TAat: 8. Post TMI ANXIETY /DEFaZuloN IN RELAT1oM To STaEss ann Soc AL Sc7 Pott: Moruxas ,'; j TMI BC ,ea Disorder Disorder Duorder Dhorder Present Absent Praent Absent r as g'. Distance from s 5 mi. 18 115 4 51 i plant > $ mi. 26 151 5 62 TMI dangerous yes 31 171 6 57 L 2 no 13 83 3 54 Living near a unsafe 34 170 4 54 nuclear plant safe 9 87 4 58 no 4 35 ')~ Evacuation yes 40 231 General Network Low 18 89 1 26 Support Medium 9 86 6 45 High 18 80 3 41 r' Inner-Circle Low 20 86 1 38 + Support Medium 12 87 1 39 s High 12 81 7 34 Confidant Low 6 61 4 27 f Support Mediuta 26 118 5 49 High 12 67 1 33 Accessibility Low 19 100 3 25 Medium 12 90 0 32 High 14 65 7 54 Stability Low 11 87 5 43 Medium 16 88 2 27 High 18 80 3 39 Size Small 12 68 5 50 Medium 19 95 3 41 ~, Large 13 91 2 20 mothers; (2) effects of percepdons of danger and social support on symptom levels at the anniversary; (3) comparisons of mothers who were snore symptomatic at the an-niversary with mothers who symptoms stayed the same or decreased. (1) Comparisons between TMI and Beaver Cbemty Mothers. Figures 2a-24 .- y ,., b ? (' N show the mean symptom scores at both interview sesmons (Thne 1 ard Time 2). S, t, Four groups are delineated: TMI mothers with and without ahistory of pA '

• e,.. g,L.qQ disorder prior to the TMI accident; and Beaver County mothers with and withom a prior history. It should be pointed out that the mean symptom scores fer moth-

,e ers from both ptes at both times were within the normal range (as defined by ,y, 242 " \\f* O 'u-.~.. s " es% . = - - -. ' G@ e e 'l / ~ -

MEhTAL HEALTH OF THREE MILE ISLAND RESIDENTS

• 1.00 5---e TM I - HISTORY,

N=118 O oTMI - NO HISTORY, N=193 + - $ BC - NO HISTORY, N:83 ', i.' ~ ' )

• --- BC - HISTORY, N:41 0.90 z,j SITE, HISTORY & TIME OF=1, 431

. '.* ',. l., 0.80 '. g, I ' r, ,1 t 0.70 \\ ./ ~ O.60 's 's 's 0.50 o' s \\ N / ys a, s., 'g 0.40 -s, N. 1 '-s -s, 'g O.30 0.20 0.10 0.00 T1 T2 srns r-e. acre.coas, nine r= ta.4eir<o. con, me r n.esce.o. con

• 2 a4 FIGURE 2

..,k,( Mothers. (a) SCL-90 Depression Scores. (b) SCL.90 Anxiety Scores. (c) SCL-90 Anger Scoses. (d) SCL 90 Giobal Severity Index Scores. ..;p-e' . '4 4,4 243 , c'lr .d r. =. ~ ....p. f g he, ~ s/ b e6 e.Me D hh - e e m .e --weeg.m.-e.i

4 q BROMET ET AL. 1.00 C e TH1 - HISTORY, N=118 -j' o,' I o oTMI - NO HISTORY, N 193 '>l. 4 W 0.90 A--4 BC - HISTORY, N:41 .. ".e s t - $ 8C - NO HISTORY, H:83 .' dd SITE, HISTORY & TIME CF=1, 431 ' O.'. " I rQ 0.80 L ~~ 0.70 s' / 0.60 i i l0.50 ,/ i O. a J;, 0.40 I .., ? 4 g 0.30 o 4 0.20 5 0.10 =t-o,co

• 1 T2

~- ' 2.s 7:e 51Ts F.10.utP=0.00. f. MISTS Fe 11.14(P4.0013. ?!sta Fett.35tPec.001) g p f .l. D) i ,I.

• I v.

1. [.5 '6I 18.

• l

( , * ( ' ?' - ( e.<N-.g.

• • ~=a o

me... m. e e .e. se e e...s , +. - 4 .mo e-e - ~~- -~~ ~ s

. -. -. - -. ~ -~ MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS ,,[j.) 1,00 e--e TMI - HISTORY, N=118 ... ;}., f .vj ; o oTHI - NO HISTORY, N=193 A---A BC - HISTORY, N=41 ,j 0.90 t - + BC - NO HISTORY, N=83

• .,'y SITE, HISTORY & TIME OF=1, 431 L 3

~ 0.80 l-0.70 /.. r 0.60 l l0.50 e,,7'~~ ~ 8 ' ~~s e o,40 ~~.,_s-o r, 0.30 's o l 0.20 0.10 N 0.00 " ~- 12,5g) T1 T2 ? ' sn+ 1 s:Tc r s.tstr<o.cas, ntsr r s.sscr<o.coas, r14 : r=ta.s44re.oot .y F. i e . '4.. s !, l' s = g) l l'

• k ',8"c 245 i

= f g l I *.

s. e BROhET ET AL. -,}l' f, ' 6, ~/ 1.00 ~. e--e TM I - H I STORY, N=118 ..".,.S ~ 1 ...h o eTMI - NO HISTORY, N=193 '. i.'. ? {

• ---* BC - HI STORY, N:41 0.90 e - 4 BC - NO HISTORY, N=B3 SITE, HISTORY & TIME OF=1, 431 g

0.80 - L \\ ~ / \\ 0.70 s 0.60 4 ,e

./

I 0.50 w I E A C.).' N 0.40 .. ~, ,. ~. .... * ~,.% 's, 3 0.30 0.20 0.10

41. s. v.*

' i'

• 0.00

'$ \\ Y, Q b .fR J2 a 4 SITE: F=4.74tPec.0033. MIETs Fa 14.38tPes.es13. T!*a Fue.844P40.8013 e. ) . (, ,t, 94

• s

.<,...e, \\ 't l

MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS berogans,1977). Repeated-measuret analyses of variance was used to determine. whether the differences between the groups were significant. As indicated by Figures 2a-2d, TMI mothers reported significantly higher levels of Anxaety. Depression and Anger as well as greater Overall Distress (Global Sev-crity Index) Knup isd to Beaver County mothers. In addition, on the average the mothers were less ~ symptomatic at Time 2 than at Time 1. This probably reflects j i seasonal vananons in mental health. Finally, in both sites, mothers with a history D.*

• L of psychiatne disorder reported more symptoms than mothers without a prior A

~5 history. [ P 'd Analyses of covariance were also computed with spouses' occupational status, motherf education, and network support (general and inner circle) entered as co. g,. 'f

• s r

vanates. Spouse occupational status and general network support were highly sig-nificant covariates. That is, mothers whose husbands' jobs had less status and who [ } viewed their social networks as less supportive reported greater symptomatology. The Tesults of the overs!! analyses of variance described above were unchanged. Table 9 presents the results of special comparative analyses (planned simple sanin effects tests) in which TMI mothers with and without a psychiatric history were compared to their peers in Beaver County at each interview time. TMI moth- / ers with a history of disorder were significantly more symptomatic than similar Beaver County mothers at the anniversary, although at nine months, the differ-ences were not significant. Comparisons of TMI and Beaver County mothers with ~ no history of disorder showed that they had similar symptom levels at both inter-view times. 'Ihus, the most symptomatic group were the TMI mothers with a his. / / tory of psyclusme disorder. f In summary, TMI mothers were significantly more symptomatic than Beaver / County mothers at both interview times. However, more specific analyses (made possible by v.rs.;AL-ic history data collected at Time 2) showed that TMI moth-ers with a history of disorder predaung the TMI accident were relatively more symptomatic at the anmversary than their Beaver County counterparts. (2) FBacts ofStress VariaNes and Social Support on Symptomatology at the

f. For TMI rnothers, the four symptom subscales were examined in reisuon to each of the four stress variables (perceptions of danger, the safety of TAata 9. Sumtmcauca Lavats som BaTwuru Stra CoMFAalsoNs or SYMFroM ScoRIs von Marzuus Wrnt AND WrrHotrr A HaToar or PsycmATa!C DUotDEa lime 1 7ime 2 Katory No History Ristory No HLitory F

p F p F p F p Anxiety 3.64 ns 3.60 ns 6.59 <0.05 2.67 as Depressen 4.03 <0.05 1.26 ns 7.99 <C.01 1.66 ns Anger 2.13 ns 1.61 ns 2.76 ns 1.75 ns GlobalIndex 3.11 ns 2.15 as 6.37 <0.05 2.68 ns Seventy

• q*l.. j' e.,

9*

247

• ...)

c.,, t ;

• 'hh e,.

D m. .m. o O. e / t, ..__e w e eg,,w,4 1pe--i-

8 BROMET ET AL living near a nuclear facility, distance of residence from the plant and evacuation), the six support measures, and history of disorder using analyses of variance. Since the findings for the four symptom dimensions were similar, only the specific re-sults for Overall Distress are described below. History of disorder, as shown in the previous analyses, was strongly related to symptom level, with mothers reporting a I history of psychiatric disorder being significantly more symptomatic. With respect to stress, perceptions of danger, evaluations of the safety of living near a nuclear '. r,'h facility, and evacuation were unrelated to Overall Distress. On the other hand, ?l [,? f TMI mothers who lived within five miles of the plant had significantly higher

• +

e. M.d{

symptom scores than TMI mothers living outside the five-mile radms. Thus, the only stress variable that correlated with symptom level was distance of insidence rh from plant. Two of the social support measures were also related to n u,wo-tology. namely, inner circle support and general network support. TMI mothers who per-i 3 ceived relatively more support from their network reported less symptomatology. It is important to note that there was no interaction between social support and distance of residence from TMI. Table 10 summartzes these results for general net-work support and distance of residence from the plant. Among Beaver County mothers, network support, perceptions of stfety ofliv-ing near a nuclear facility and living inside or outside a five. mile ra.lius of Ship-pingport were all unrelated to current symptomatology. These results were striking because Shippingport has received a lot of attention in the last decade concermns . Jf possible low-level radiation leaks and heahh effects on the populmr= living j nearby. 1-' ~ (3) Esamination ofMothers Who Were More Symptomaticat the An=xiversey. Although there was a general trend for me:hers to report fewer symptoms at Tiene s / 2,40% of the sample had the same or higher symptom scores at follow.up. On the depression and anxiety scales, the mothers wer: divided into three groups-the i 16% group whose symptoms increased the most, the 16% group whose symptoms P decreased the most, and the remaming 68% whose scores were relanvely stable. These divisions.were carried out separately by site. One-way analyses of uraance - TAats 10. GzwunAr. DIsTasss Scoazs or TMI Morazas Acconocco w DtsTANCE oF Rzsmaxes ruou TMI, Lryst, or GewsmAr. NsTwoar Sorroar. AND PsycmATalc Hisroav History of Disorder No History'ofDisorder Level of Support s3 mi. >$mL s3 mL > 3 mL High 0.424 0.318 0.220 0.157 Medium 0.417 0.385 0.327 0.243 Low 0.695 0.511 0.406 0.333 F (History) = 16.67; d.f. = 1,300; p < 0.001. F (Radius) = 3.96; d.f. - 1,300t p <0.05.

• ~

F (Support) = 12.66; d.f. = 2,300; p < 0.001. '. y

• ?'4 All other effects were nonsignificant.

,'.,p) . r,.y 148 .*i.o-kf..l(

• \\ Q.

. L..A, e-- eg

e. e.

S M. *e 1 /

u.,.. MENTAL HEALTH OF THREE MILE ISI.AND RESIDENTS were computed to look at effects of age, education, family income, distance of residence from the plant, and the six social support subscales. In both sites, changes in scores were not significantly related to the demographic variables or distance of residence from the plant. General network support was related to changes in deorssion and anxiety in TMI and changes in depression in Beaver ,,, / p" County. As depicted in Table I1, mothers who became more depressed had some-c; ~^ ' what lower levels of support compared to the other two groups. However, al-

5 ;.,,,* f though significant, the differences were very small. The most parsimonious con-

,1,,7 c' d clusion we can reach at this point is that there do not appear to be any major variables that differentante between mothers who became more symptomatic and

• V,
• M

. those who did not. Further analyses are planned to look at this issue in the future. ~ '. t Other Analyses Three additional sets of analyses were conducted. First, we ex-amined the mental health of the 122 TMI mothers who were interviewed closer to the '4 L anniversary (prior to April 15) with that of the 190 TMI mothers interviewed subse-quent to that date. On the Symptom Checklist, TMI mothers interviewed gfter April [ 15 had higher scores than mothers interviewed closer to the anniversary. On the anx. lety subscale, the difference between the means was highly significant (t = 2.75; ~ p < 0.01).This may have occurred beca:Le of the publicity associated with the im- '/ pending release of krypton gas. In Beaver County, the 51 mothers interviewed on or e before April 15 had scores almost identical to those of the 73 mothers interviewed a later. There were no differences in either area in responses to questions about TMI's danger or the safety of living near a nuclear facility. Because Middletown has received so much attention by the media, we also com-pa ed the mental heakh of Midd!ctown mothers (N = 69) with that of the other TMI mothers (N = 79) living in the 5-mile radius. There were no differences on the Symptom Checklist subscales. There were also no differences in the proportions who viewed TMI as dangerous (66% in Middletown; 68% of the other mothers living ~' nearby) cnd who felt thatliving near a nuclear facility was unsafe (63% and 67%, I respectively). Finally, we compared the memal heakh of the 19 TMI mothers who were pregnant at the time of the accident with three control groups drawn from a pool of TMI

• mothers matched on age, number of chudren, race, marital status, and distance of residence from the plant. The control groups were randomly selected with replace-ment. A larger proportion of the pregnant mothers reported clinical episodes of anx-lety/ depression after the TMI accident (37% wmi-ed with 29%,14%, and 30% in the three control groups). None of the reported episodes were considered to be post-TAat.E II. LaTEIS oF GassaAr. Nrrwaar SurroaT*

nr RnArtow To CHANGES IN DIrazssmx: McTaras More Less Depressed Stable Depressrd F df p 1 TMI 12.26 12.79 12.55 5.63 2,308 <0 004 B.C. 12.28 13.02 13.03 3.84 2,121 <0.02 ., h e, %wer scoresindicate less suppon.

i. E a

.;N 249 \\. .kd 'x t N t-

• e d e?,

l

• • ma.

l m. mo

• es e a b

-a ~ ~ - ~

s BROMET ET A1 partum depressions. The higher rate of clinical disorder after the accident was**m sharp contrast to the similar rates found for pregnant and nonpregnant mothers prior to the TM1 accident. Pregnant mothers also reported more distress on the Symptom Checklist than mothers in the three control groups, particularly on the depression and anger subscales. Summaryr 'llius. TM1 mothers had an excess risk of experiencing clinical episodes of anxiety and depression during the year after the accident. These clinical episodes ..S were not associated with other stress and support factors. TMI mothers also reported .;[ p more symptoms of anxiety and depression at subclinical levels at both interviews i.' a [ '?.' compared to Besier County mothers. TM1 mothers who were most symptomatic were those who had a prior psychiatric history before the accident, those who lived .. l ' t' - ' 'J. I. within five miles of the plant, and those with less adequate social support from their network of friends and relatives. Being pregnant at the time of the accident also was 'r an important risk factor. 11 FINDINGS: WORKERS Background Characteristicsr Most of the workers from the TMI and Beaver l County nuclear plants were married, male, and White. As can be seen in Table 12 TM1 workers were older than their counterparts in Beaver County, and there were /- TAaLa 12. BACKomoUND CHARACmusrICst WomKZas g TMI Beaver County N N Total 161 100.0 124 100.0 ./ [ Age s 29 37 23.0 59 47.6 30-39 71 44.1 44 35.5 e' 40 + 53 32.9 21 16.9 i Sex Male 146 90.7 !!0 88.7 Female 15 9.3 14 11.3 Marital Married 137 85.1 107 86.3 Status Unmarried 24 14.9 17 13.7 Education s 12 years 94 58.4 50 40.3 > 12 years 67 41.6 74 59.7 Religion Catholic 30 18.6 56 45.2 Protestant !!8 73.3 54 43.5 Other or None 13 8.1 14 11.3 Income s 25,000 61 42.7 60 49.2 i i 25,000 + 82 57.3 62 50.8 + i (Missing) 18 2 Distance of home s5 miles 49 30.4 17 13.7 from plant 6-10 miles 34 21.1 25 20.2 ~ 11+ miles 78 48.4 82 66.1 c,, l.7

p.,;.{,,

250 ,. rj a

• I k,

. u,. y ,.4 I . n, 'A +..s.. l l ~ / =

p MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS fewer TMI workers with formal education beyond high school. Consistent with the different age distribution of the groups TMI workers had a slightly higher total fam-ily income. TMI workers were predominantly Protestan:, whereas in Beaver County. Catholics and Protestants were almost equally represented. At the anniversary, most of the workers (N = 120) from the TMI area were still employed at TMit 27 were employed chahe, and 14 were unemployed. Most of those no longer employed at -;j.,' % i r* ' TMI belonged to the Building Trades Council unions. (It should be noted that there a.*- 9.. were no demographic differences between the workers who were interviewed at both sessions and those not availabic for the second session.) ,3,.;(' Perceprions o/Stresst Table 13 presents the responses to questions about the cur-p'.'. 7 t-rent situatson at TMI and perception of the safety of living near a nuclear facility. r Most workers in both sites felt it was safe,to live within 10 miles of a nuclear power facility. However, somewhat fewer TMI workers perceived the situation as danger- [ ~ ous m y W with Beaver County workers (25% vs. 33%). Very few workers left the TMI area at the time of the accident. However,50 work-1 ers had family members who left. Most of these families lived within five miles of the plant. i Errait of C2uucal Diso.-der: Mwaments of affective disorder (depression and / anxiety) for each month from November,1978-January,1980 (when most of the worker, Time I laterviews took place) were made during the first interview. The rate I for April,1980 was obtamed from the SADS.L which was administered at the fol-low.up session. Figure 3 presents a graph depicting the monthly point prevalence / rates of depresson/ anxiety from November,1978 to January,1980, and the rates for April,1980 for the two worker groups. As is evident from the figures, the rates were , f." consistently higher for TMI workers (Mann-Whitney U Testt p < 0.001), although TAat 13. Danuatmow or Rzsrowsss To Quxsnows Anotrr Tax ACCIDENU WosEIns TMI Beant County N N 161 100.0 124 100.0 Do you currently feel the si:uation at TMIis dangerous? o Yes 39 24.8 41 33.3 No I18 75.2 82 66.7 (Missms) (4) (1) Do you feel that it is safe to live near (within 10 miles) a nuclear power facility? Yes 129 81.6 105 86.8 l ' 1.,'; ~ '~ No 29 18.4 16 13.2 (Missing) (3) (3) ,[. y'

• ,4 Q

.v . i., a{- \\, 25: '.'h 1 j -~ '"** m

== 7 r-l. e ? 'I i w ./ .aL=.. sj.,_ m..J a-. ----

a u ^ =- 4 BROMET ET AL. to +---e TN (. Na te t e-e 8.C.. Nst24 a .a i, Af f e e eg

• .;, g e

p,. 2 4 4 a g e e _e...e.e_ g ~ i ' gb-( ~ ~ ~ ,.... 4 '4 6 ...,....... i.. - - 1 NOV JAN APR JUL. OCT A APR 78 73 i HGURE3 Workers. Point Prevalence of Anxiety / Depression by Month: s the actual percentage differences were not very large. Moreover, the rates are con- ~ ; siderably lower than those reported for the mothers. Further analyses were conducted to examine the relative risks of disorder before c', and after the accident using the SADS-L data from the follow-up interview (Table j t ' 14). Unlike the TMI and Beaver County mothers whose lifetime rates were umune ~ before the TMI accident, a larger proportion of TMI workers roet eriterta for an disorders and for anxiety / depression in parucular before the accident compared to Beaver County workers. (Although not shown on the table, the resuhs were umilar i when only nonconstruction union TMI workers were compared with Beaver County workers.) ne relative risks post-TMI were also higher for TMI workers, although the risk of new cases occurring after TMI was significant only for annety/depres-sion. Thus, these results support the observanen from the graph that the rates of clinical disorder after TMI were higher for TMI compared with Beaver County workers. However, these differences must be interpreted cautiously since the rates before the accident were already higher in the TM1 cohort. What factors affected the imam of workers' reportmg annety/ depression? NM-agraphic characteristics, social support measures, and stress variables were exanuned. With respect to sociodemographic variables. TMI workers reporting a } disorder either before or after the accident were somewhat younger than workers without a clinical disorder. This result is consistent with findings from an epidemio-J logic study using the same interview schedule (Weizman and Myers,1978).There were no demographic variables that distinguished Beaver County workers with and without anxiety / depression. Log-linear analyses were used to examme the offacts of social support and stress on the distribution of anxiety / depression. No sigr*==f ( ,[ e, ll findings emerged. (See Table 15.) Thus, consistent with the pattern of results for the 4 I.. . 6 '.,".. \\, *" h.f 2,$2 )' I$ 4 1 '. ' 'u(n. 8 j ~---- ~ -- amise e u-mm,.e-. ii.e-,

,- e MENTAL HEALTH OFTHREE MILE ISLAND RESIDENTS TAat 14a. I.nztmoon or ANXfFTY/DEFaEssioN IN RatAnoN To Tus TMI Accrot. r: Woaxxas v RLsk (Confidence History TMs BC Rario interva0 t p Cases pre.TMI 41 18 1.754 (1.077, 2.858) 2.258 0.012 Cases post TMI 25 7 2.751 (1.289, 5.872) 2.615 0.004 New cases post-TMI 11 3 3.239 (1.005, 10.442) 1.968 0.025 Recurrent cases

1. ,,g(

i

'l post-TMI 14 4 1.537 (0.608, 3.883) 0.908 0.182 ',[.' ;[ l ,. ~. ' i , / '.. TAata 14b. RISE oF ANY DIsoaDEa IN Rztanow To Tur TMI AccIntxt: Woazzas .y, y4 + Risk (Confidence t. filstory TMI BC. Ratio Interva0 g p Cases pre.TMI 61 34 1.382 (.982, 1.945) 1.855 0.032 ~ Cases post TMI 37 11 2.591 (1.433, 4.684) 3.150 0.001 New cases post-TM1 7 3 2.100 (0.578, 7.630) 1.127 0.130 Recurrent cases post-TMI 30 8 2.090 (1.154, 3.785) 2.434 0.007 / mothers, the major vanable that was==== red with the distribution of clinical dis-order was site, with TMI workers having somewhat higher rates. Current Syrnprontarology: Symptoms of ansucty, depression, anger, and overall / distress were assessed at each interviewand are described in this section. The section is divided into three parts:(1) compansons between TMI and Beaver County work- / ers; (2) effects of stress and social support on symptom levels at the anniversary; (3) ,T comparisons of workers who were more syr.apcomatic at the anniversary than at the first interview with workers whose scores stayed the same or improved.

1. Cor spernons between 7Mf and Asever Comary Forters: Figures 4a-4d show the mean symptom scores at both interview sessions (Tisne I and Time 2). Four

,j groups are depicned-TM1 worhers with and without a prior history of disorder; and Beaver County workers with and wnhout a prist history. Repeated-measures anal-yses of variance was used so determee whether the differences between sites were und*Wify significant. As indicated in the figures, TMI workers did not differ on .l any of these measures from Beaver County workets. He analyses were repeated ex - cluding members of the construction umsons from the TMI cohort, and the same re. sults were obtamed. Corniment with the dasa frota the mother cohorts, workers in both sites with a priorpsycinatre history had signi5c:amfy higher scores on the three subscales (Anxiety, Depresson, Anger)and the global nwemres than workers with-out a prior history. In addition, lower syrnptoun levels were observed in both groups at the anniversary than at Time 1. Analyses of covariance were aho ccmputed with age, edni:ation, and network sup-port entered as covarines. Inner circle support was the only significant covariate, in-dicatmg that workers with greater levels of mapport were less symptomatic. s f J. g,* '253 [?. d ~ UY

> $.i.9 (

1 ,.h.:l ,6g y 4.d., I ~

• ea e.

e-.- g, I ,/ N

~ ' ~ BROMET ET AL. TAaLe 15. Post-TMI RELATIoNsWP Bstwsrx AxxrsTv/Derazssrow Ann STassi AND SoCtAL SUPPonTt WoaKaas TMI BC Disorder Discrder Disorder Disorder Present Absent Present Absent Evacumuon yes 6 44 ' ).,; _ no 19 88 Danger yes 8 30 4 37

'i,y

) no 17 98 3 79 ' Living near a unsafe 6 23 2 14 -',', ' P nudcar plant safe 17 108 4 101

g. :

General Network - Low 8

• 43 4

36 Support Medium 11 45 2 34 L High 6 44 1 46 Inner Circle Low 9 38 4 37 Support Medium 11 41 1 42 Highi 5 49 2 37 Confidant Low 8 40 2 23 / \\ Support Medium 8 ' 60 4 58 High 9 30 1 34 . Stability Low 7 15 1 27 Medium 8 73 2 48 / High 10 44 4 40 Accessibility Low 8 40 3 45 ~ 8 ' Mdm 6 4 1 M High 11 44 3 37 Size Small 8 52 1 41 8 Medium 6 35 1 30 Large 11 45 5 45 Table 16 presents the results of planned comparisons in which TMI workers with and without a history were compared to their peers in Beaver Coynty at each inter- ~ view time. The results indicate that the overall site similarities depicted in Figures da-d were unaffected by psychiatric history. That is, TMI workers with a history of chmcal disorder were not more symptomatic nine months or 12 months after the scrittent than Beaver County workers with a history of distress. "Ihere were also no differences between sites when workers without a history of disturbance were s compared.

2. Effects ofSocialSupport on Symptomatology at the Anniversary Each of the four symptom variables were examined in relation to each of the six social support subscales and psychiatric history. The stress variables were not included because too "N-few workers responded in a negative direction. The two variables that were signifi-

, a.4: .ke 1st t e. S** i ..pi ;- h y ma i h. amo p. e.. e m. O.d + e

~ -

i.,

~. MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS 1.00 s s---e TM I - H ISTORY, N:61 O oTMI - NO HISTORY, N=100 ,4 A---* BC - HISTORY, N=33 0.90 ,%f p t - $ 8C - NO HISTORY, N=86 '. ' '*{ SITE, HISTORY & TIME DF=1, 276 f.'.. l 5 .. c '. 0.80 9,'. 74 -t d - ~ 0.70 t ~ 0.60 ./ s $ 0.50 W j A, o,40 8 ii 4 g I 0.30 3, s 's +...,, ' ' ' * ~ '... 0.20 Q ~. i 0.10 0.00 T1 T2 SITE: FANS, HIST: Fa 15.75(P<0.0013. TDC: F4 7.28tP=0.001) g[.

ro FIGURE 4 Workers. (a) SCL-90 Depres

ion Scores. (b) SCL-90 Anxiety Scores. (c) SCL 90

\\ d'.' L Anger Scores. (d) SCL-90 Global Severity Index Scores. ,.i) a' e,-) 255 ?. '%

• I ',
• I S

g , 'e s.U. t I g, a g e eWD. 4 1- .f I-s

.L: T -- ~ ~. l BROMET ET AL. I 1.00 i i ~ . e-e---e TM I - HISTORY, N=61 o oTMi - NO HISTORY, N=100

3. ' E h

. #' U A---* BC - HISTORY, N=33 . ;G' s i- ) 0.90 ~ t - + BC - NO HISTORY, N=86 r 8 ;;

. (* j SITE, HISTORY & TIME Dr=1, 276

. s'. " ? i t, l 0.80 3 1 O.70 1 / 0.60 j .= / 0.50 o' a .J t / 0.40 A,*,, I ','s 0.30 o.... ..***a...,,** i. 4.,*%., 4 i .i 0.20 ,e 7 4 0.10 \\ 0.00 T1 T2

c. -ce -2.

$1TE: FsNS HIST: Fa 8.46(P40.004), TINE: F=14.01tP<0.001) [g ,..P'. , 2 l g y ', a? i i }

• Qt 1

8 256 ',5 Ai6 e ~. < *.NO.3_&. _ \\ ! ".g-I,- '~ ~. b .eee me R S l. m.e w s s a g e i L

s

=

e MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS 1.00 i s---e TMI - HISTORY, N=61 ,, e o o TMI - NO HISTORY, N=100 .j t,y 6.90 a---a BC - HISTORY, N=33 t - + BC - NO HISTORY, N=86 i + e-SITE, HISTORY & TIME OF=1, 276 ' '. i ' 'N ..,,.', p 0.80 - t } t I 0.70 i 0.60 s 9 0.50 /' A', s.- s 0.40 ,a . I O.30 U. t e..~..' * '.. t 0.20 ,~.* e ,'s i 0.10 i t 0.00 T1 T2 stTE: rans. Hist: r= ss.sstr<o. cots, itME: rate.7steso. cot) 'S glp W.' s s ,,' - 1) 25, .b..,..Ig , a(#y ..s. 4 am.., e .l'. p M.* -e o I m, / g. Ww ,4,,, ---_a. w.,e9-s---+-n -egy--- g,,, e 4

t BROMET ET AL 1.00 e--e TMI - HISTORY, N=61 t

~

o eTMI - NO HISTORY, N=100 ' y',, - 0.90 - A---4 BC - HISTORY, N 33 4 .'..I,Lff-6 -

• BC - NO HISTORY, N=86 SITE, HISTORY & TIME OF=1, 278 4

' i'. M o.80 E 7 d i L 0.70 1 l '\\ ~j ~ 0.60 = I 0.50 a r 4 l /* ' i O.40 t ,'%,N 0.30 o...'.., ~s .I s 96.,'~ ~., . c ' i.' 0,2n , ~ - ~.., i 0.10 0.00 T1 T2 stTE: FANS. HIST: Fa 13.31(P=0.001). TINC: F=34.09tP<0.0013 }.s *.*J j i, C.i sa ., t,,' f 258 . k r.,t - 'h[^b_. g pg h.

e..

e es a m. + Me 'i i I l ~ a /: .-r

<e. MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS TAala 16. SloNIFICANc2 LEvita Poa BrrwarN $rTI COMPAalsoNs or $rMFroadScoaEs .-4 rom Woaxxas WrTH AND WrrNottr A H2svoar or PsycutAtsuc Daoansa - 4*

• I

.-[s' i { l.g

9. Months Anniversary History No History History No History 5

.6 F

p F p F p F p ~ \\g. }~ Aruuety 0.55 ns 0.53 ns OAl na 0.71 ns Depression 0.98 ns 1.14 ns 2.27 as 0.07 ns 1 Anger 0.47 as 1.93 ns 0.11 as 0.33 ns t Global Severity index 0.63 ns 2.57 ns 0.78 as 0.66 ns cantly related to symptom scores of TM1 workers were level ofinner circle support .I and psychiatric history. Table 17 summarizes these results for the TMI workers. Thus, TMI workers with less support from their inner cirde of friends and relatives / and those having a history of clinical disorder were experiencing tuore symptons dis-tress at the anniversary. / / Interestingly, araong Beaver County workers, level of support sessned to affect de-pressive symptomatology but was unrelated to the other symptora dimensions.

3. Examination of Workers Who uwre More Syrnpronantic at she Annuversvry:

Although there was a general trend for workers to report less sytmptom detras at Time 2, some workers had the same or greater levels of syunpromatologyat the sec-g ond interview. On the depression and arudsty suhe=Aa the woriers were divided into three groups-the 16% group whose symptoms increased the most, the 16% group whose symptoms decreased the most, and the romaanns48% who were rela-tively stable. These divisions were carried out zwW, by site. One.way analyses of variance were computed to look at effects of age, h'M incarne, rhar==rw of residence from the plant, and the six social support subscales. In TMI, the only vart-able which had an effect was the measure of general network support. In Beaver County, none of the variables had an effect on the change scores.Thus, there do not appear to be any major variables that difflerentiaeed worisers who became snore symptomatic. Work Stress Figures 5a and $b present the means on the work sesuss ed=rmaa for TMI and Beaver County workers from both interview sessions. Members of the con-struction unions were exduded from these analyses to permit between.sise compari-sons of workers with similar occupations. It should be nosed that higher scores on the Overall Job Pressure subscale indicate more negathe perceptions 0.e., feehnss of greater pressure), while higher scores on the Overall Job Rewards subscale indicate more positive perceptions (i.e., fechass of greater rewards). With regard to Job Re- ,,. 3. ^*/ wards (Figure Sa), there was a significant effect for site indicating that TMI worisers j,i','Q fdt more rewarded by their jobs than Beaver County workers. Job Rewards per- .;g ceptions were unrelated to prior psychiatric histoIy and did not chasse significantly

e ov" time

W. W.,

, y= I l _ i-t 259 M.. e .l s. / w

~ q i l ;. I j-

• . ~

a t-I l if. TABL 17. MsAN SruProu Scomes of TMt Waarans AccomotNo to PsYCHIAtac HatfomY AND INNEm ClactB SUPPOmf J History No History to

o liigh Medium Low fllsh Medium Low F(History)

F(Support) 9 Support Support Support Support Support Support 4f1,131 df 2.131 G l Depression 0.302 0.523 0.386 0.130 0.119 0.433 7.15 (p < 0.01) 2.78 (p < 0.07) Q Anxiety 0.162 0.430 0.315 0.131 0.173 0.409 1.26 (ns) 4.61 (p < 0.01) Anger 0.245 0.300 0.428 0.116 0.137 0.403 3.40 (p < 0.06) 6.59 (p < 0.01) r-Global Severity i Index 0.248 0.430 0.322 0.146 0.138 0.405 4.34 (p < 0.05) 4.06 (p < 0.05) 6 j I 3 i l i s I f l ~ lf y, t sl c Y .G- [,.. .'f' "h i. j ' ~ ~.

v....,,,,. ~.

M S h,N * .m - h (, s I

MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS With regard to Job Pressure, there was a significant effect for time, indicating that workers at both phts felt less pressured at the second interview. These changes in perception were particularly true for the TMI workers (Figure $b). Perceptions of

• *i,.'f ; p Job Pressure were not related to site or to psychiatric history.

r Analyses of covanance were computed with age, education, and support serving as covariates. ' Age was significantly correlated with Job Pressure, indicating that [6,* v g .= younger workers experienced more pressure from theirjobs. General network sup. i[ '; > port was associated with Job Rewards, indicating that workers who had more sup. 4 -4 port felt more rewarded by theirjobs. The overall main efrects described above were ' t.; unchanged. Summaryr Thus, it appears that there was a small rnental health difference be-2 tween TMI and Beaver County workers. However, differences in clinical disorder ~ were seen before as well as after the accident. Subclinical symptom leveis were simi. lar in the two groups. Finally, TMI workers felt more rewarded by theirjobs. In both 100 / M TMt

• HISTORY, Ns35 e.

e TMI = NO Hl$7oRY, Ns62 / ~ " ' 90 ~ t - 9 8C = No HISTORY, Na89 /. SITE & HISTORY Orat, 207 TIPC Orat, att g j 70 So e.. 50 s...........s a so o Tt T2 'S J, g,' SITC: Fa4.23(P*0.05), HIST: Fa Ng, Tgge FmNs g-s

9. W

.4 FIGURE S /. ii Nonconstruction IBEW Workers. (a) Job Rewards. (b) Job Pressure. . \\.'/ 1: >l 6 261 ',',Q w t' ,,_,,...,.m no -m'

r ? BROMET ET AL. 80 i i

i. g ts---e TM I - HI STORY,

N=35 "4 , i,- j V-a N .. '.,'. g. o CTMI - NO HISTORY, N:62

• ---* BC - HISTORY, N=33 l '

. s,','

3.g 70 t - t BC - NO HISTORY, N=89 r.i#;

SITE & HISTORY OF=1, 207 g-L TIME OF=1. 211 60 \\ \\ / s 50 r , i-

1. 9 I

'e' W 40 9 O'"*.,, % g 3 n. 30 i 20 l* T T 0 T1 T2 s 1 .3,..e,. g) 'P' s% SITE: Fs2.41(P<0.12). HISTt Fa NE, TilEt Fs13.32(P40.001) ., l. 8, 3 3 [ ,'. i 3 i s .. \\. t ' (.. .Ia.., 262 ,4 :., ...u,, ' .. ~m ~ ~ ~ 9e e f a I 9 3 -..c

? -_m a MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS atess, workers with greater social support were less symptomatic and felt rnore re. warded by their jobs. FINDINGS: MENTAL HEALTH CLIENTS Background Characteristics-The 151 clients in the TMI area and 64 clients in the I Beaver County area were demographically similar in several respects (Table 18). The '.I${r average age in both areas was 38, approximately two. thirds of the clients in both areas were female, and one third had never been married. Almost all of the clients d were white and born in the United States. However, the groups differed somewhat in ' '5'.'--f' terms of education and income, with relatively fewer TMI clients having college " 5-training and earning more than 510,000 per year. TMI clients were less residentiany 1 ) ~ e TAata 18. BACKGaOttND CkAaAC'tRa!! TICS: CLDDrTs 1 TM1 Beaver County i N N / Total 151 100.0 64 100.0 Age s 29 49 32.5 22 34.4 30-39 44 29.1 17 26.6 40 + $8 38.4 25 39.0 Sex Male 49 32.5 24 37.5 Female 102 67.5 40 62.5 (' t l Marital Married 58 38.4 31 48.4 / Status Single 51 33.8 20 31J Divorced / Separated ' 37 24.5 11 17.2 Widowed 5 3.3 2 3.1 Education s12 years 110 73.3 39 60.9 > 12 years 40 26.7 25 39.1 (Missing) (1) Religion Catholic 36 23.8 27 42.2 Protestant 90 59.6 30 46.9 Other or None 25 16.6 7 10.9 Income 5<5,000 60 41.9 11 17.2 5-9,999 26 18.2 11 17.2 10-14,999 22 15.4 16 25.0 15 + 35 24.5 26 40.6 (missing) (8) Employment Employed 66 43.7 23 35.9 Unemployed 58 38.4 18 28.1 Housewife 27 17.9 23 35.9

• ..,p Distance of home s3 miles 52 34.4 24 37.5 I.

from plant 6+ miles 99 65.6 40 62.5 ..,pj .,y ~ ~ s s...,,

g..q 263

. <y "e &g e 'I n o. .-n-

f e. + i i 1 BROMET ET AL i stable than Beaver County clients, reporting an average of 8.4 years at their currerit l address compared to 11.7 years in the comparison group. As was true for the moth-ers and workers, there were more Protestants among the TMI clients (60%) com-J pared with equal r. umbers of Catholics and Protestants in Beaver County. ,,,.. (y,, . c.- Clients who were not reinterviewed at the anniversary were compared with those d',. j who stayed in the study with respect to demographic composition. Relatively more of the TMI dients who were missed at follow.up were black, less educated, and sep- .t '., ", q d arated or divorced. In Beaver County, four of the six clients who did not continue in e', ' I} the study were under 30 years of age, and all six were Caucasian and married. i d Perceptions ofstress-With respect to fears about TMI and nuclear power in gen. eral, more than half of the chants in both areas felt that TMI was dangerous and that d living near a nuclear facility was unsafe (Table 19). The analyses which follow will L show that the effects of these fears on symptomatology were expressed quite differ-ently in the two areas. Extent of C2siksI Daorderr The treatment histories of the two groups of clients 'I. were remarkably similar. Most clients in both areas had been in long-term treatment. / and approxunately 60% had one or more prior psychiatric hospitalizations. Table 20 presents the life. time prevalence rates of the psychiatric disorders determined on the j basis of the SADS interview. (Individuals can have more than one diagnosis.) Figure -U j 6 presents a graph depictma point prevalence rates of depression from November. i 1978 to April,1980. Twenty five percent of TMI clients and 20% of Beaver County /. I clients had suffered from schizophrenia (according to RDC criteria), and more than .I half of the patiems had had an episode of major depression in their lifetime. I / Chrrent Symptomatologyr Current symptoms were assessed at each interview / j using the Symptosi n=r+he-90. Figures 7a-4 show the mean differences between 4 ~. f TAna.s 19. Deramtmos or RasroNsts To Qurstross ~ Anottr'rus AccmsNT: Cuf.NTs TMf Beaver County N N 151 100.0 64 100.0 i Do you currently feel the situation at TMI is j dangerous? Yes 90 62.1 34 57.6 i No 55 37.9 25 42.4 (Missing) (6) (5) Do you feelthat it is safe to live near a (within 10 toiles) m4rar power facility? . [ * +. p 'd Yes 62 42.5 29 47.5 %.C ' p,. No 84 57.5 32 52.5 (Missi.g> (n (n **.g J. l *6 '.'%.b. g -N. % e.>

== -..m.. "We h = t

i MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS TABLE 20. MarrrAt HEALTH CUI!fTs* LaETrus PREVALENCE OF PsTCHIATRIC DISORDERS TMI Beaver Cosasty , t'. ', f N = 151 N = 64 i . r, y n y g Defnite Disorder i '.,,' Major Depression 84 55.6 45 70.3 ." I # Generalized Anxiety 19 12.6 5 7.8 . e. - { ,I Minor Depression 4 2.6 4 6.3 r-1 Mania 17 11.3 9 14.0 1 Hypomania 1 0.7 1 1.6 g i Schizophrenia 37 24.5 13 20.3 ~

)l Alcoholism 19 12.6 7

10.9 Drug Abuse 29 19.2 4 6.3 Panic Disorder 13 8.6 5 7.8 Phobic Disorder 9 6.0 5 1.6 Obsessive Compulsive 3 2.0 I / D* a der I Lifetime (Personality) Dhorders' { Briquet's Disorder 1 0.7 0 Cyclothymic Personality 3 2.0 2 3.1 ,f l Labile Personality 6 4.0 3 4.7 / j Antisocial Personality 3 2.0 0 / j / 1 l so i

{

~ 25 r..iss .t

• ...e e.c.. n=es to i

4 ?. .. ' l.. { his e to g 3 e s I NOV JAN APR jut. CCT JAN APR k J 78 75 80 ' ' '2[ ' 'H'. M FIGURE 6 Point Prevalence of Depression by Month: Cliegls. V 't 4 s...'61 -) , 265 i ',','g f..L .. L .p o %e. I

• -e g

y. .a 4 W.- ,W' i ~ e. PAL

~ BROMET ET AL 1.50 i Is---e TM I, N=148 a 1.40 - a---* BC, N=61 i SITE, TIME OF=1, 207 -l4,' 1 i 's. v.~{fh ' f :rr j 1.30 - .c ' *; I j .s 1.20 L .1 1.10 i* .s j-a, ' 1.00 m z j m t i 2 0.90 ? 8 ' 0.80 I 0.70 1 0.60 O.50 T T ~ 0.00 T1 T2 SITE: F=NS, TIE: F= 4.76(Pj0.04) i Clients. (a) SCL-90 Depression Scores. (b) SCL-90 Anxiety Scores. (c) SCL.90 Anger o/9' e, FIGURE 7 f Scores. (d) SCL-90 Global Sevedty Indes Scores. p, pj .f 9 I* g g . s. *s)

1. .,\\

• I'
• a.: J }..p D

4 s . 6%

• he

~., -e. I* 1 e'p.e e p. h

1. g m.-

I e e/ 4 a.4- -e ..... - =. .w ...._m w.. 7 WT- * -. * - < - g g,, 9 gg wa ep-e ,== b a

e. y MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS

,' i '

1.50 ;eN, C

3 TM I,

N=148 .. ' I., s + A---* BC, N:61 1.40 e, .fi 3 SITE, TIME OF=1, 207 t _n l, 1.30 .1 t ' e. g t 1.20 1 / 1.10 i 1.00 en a-. t [ .f k

i ta A

~ 2 0.90 's 's e \\,a 0.80 4 4 e 0.70 4 O.60 ' I O.50 i T 0.00 ~*J T1 T2

• M',,

~ !!. ) 'I SITE: F=NS, TIME,: F= 3.77(P<0.05)

.e+

s.,). ' 'k ,,,s '. 9 *g* 267 . eb %m a.=. k; s ./ --_...-....,w- ---.---~---;-~-----*

.o o e .I BROMET ET AL

4

~ 1.50 ~ 3 (B--lO TM I, N=148 ",h, 4 .y. 1.40 ~ '. ', ' ' ' a---A BC, N=61 .3

j SITE. TIME OF=1, 207 J

s....

~' .4 1.30 a 1 1' t 1.20 e-j + / 1.10 i 1.00 .i <n 1 z w ) 0.90 0.80 b 0.70 ,,'s s, s, 's, 0.60 ~ 0.50 C 0.00 .l T1 T2 J. e. N'.,' ui SITE: F=NS, TIME: F 4.41(P<0.04) j ..,) t te ' ' **

• n )

8. s l' [

i M8 y ' k g' J., e.* i C -- - t eI /

. - - -. ~... MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS i, 1.50 e i-llf,., 'g e----e TM I, N=148 ~ .'J Cy A---A BC, N:61 1.40 $} 4 i~ SITE, TIME OF=1, 207 0 -{ L. 1.30 L _1 I, ~ 1.20 1.10 ,e ./ I 1.00 m r t z w a /- j 0.90 F A, '. r O.80 ~~ 9 'A J 0.70 0.60 0.50 T T O.00 T1 T2 3 $.- ?' 'M';. t ii i SITE: F:NS, TIME: F= 8.07(P<0.005 ) ..}), e*.

• • • .,..)'

p.n ..yn. 269 h.g g ~. % em -- e m.mo. ep _ 88m-m em m ene see. Omee Me,e em i gl ./ e- =4* e e a +=-e W 4

• w a

,7w w v --M-g (==gp my*f g P r -- W

i L BROMETET AL. clients at each interview on Depression, Aruusty. Anger, and Overall Distress (Global Severity Index) and the results, of repeated-measures analyses of vanance tests. Although TMI clients had higher symptom scores, the differences between the sites were not statistically significant. There was a significant effect, however, for

• J{

time. Clients in both areas, like the mothers and workers, were less symptomatic at .O.4 k Time 2 than at Time 1. s 'I These analyses were repeated with the schizophrenic clients excluded to see if their - [ $l l contribution to the overall vanance might in part explain the failure to attain statisti. ca! significance. The sarne resuhs were obtained, with the TMI means being con- , !g, h sistently but not significantly higher than those of Beaver County. e Differences between sites were also examined using analyses of covariance In I which years at present address, age, educettan, and inner circle and general network L. - ] ~ { support were entered as covariases. The overall mean differences between the sites remained nonsignificant. Education and general network support were significant 1 covariates, Indicating that clients in both areas with more education and more sup. port reported fewer syrupcoms. EffsetsofSocialSupportandStresronS>wsplomatologyattheAnniversary: Site. 'f -- specific analyses were carried out to summine the effects of social support and per. s ceptions of TMI's danger, radms, evacuation, and perceptions of the safety ofliving near a nucicar facility on symptomatology. These analyses showed that in both areas, clients who had greater general network support reported fewer symptoms of anxiety, depression, and anger and less overall chstress. Although evacuation and distance of residence from the plant were umulated to symptomatology, perceptions of TMI as dangerous and of living near a nuclear facility as unsafe were associated with heightened symptoms. For TMI chents, responses to these two questions were related to anxiety, but not to depressaos or anger. Tables 212 and 21b illustrate these results for the TMI clients. In Beaver County, positive responses to the question

p about TMI's danger were==aci===d with increased angrr, whereas responses to the question about safety ofliving near a nuclear facility wre assocanted with increased depression. There were no signinrartt lateractions when each of the support scales was examined in retadon to each of the danger ruessures. 'Dms, perceptions of the plant and nuclear power were related to symptorns among clients ahhough these variables had not been assodated with sympsomatology in the mother and worker cohort.

Emamination of Curnas Who Were MoreSynpromeric at the Anniversary: Al-though there was a general trend for clients no report less distress at Time 2, some a clients had the same or higher symptom levels at followwp. On the Depression and Anxiety subscales, clients were divided into three groups-the 16% group whose i symptoms increased the most in a nesstive derection, the 16% group whose symp. toms increased the most in a positive direction, and the remaining 68% whose scores were relatively stable. These divisions were carried out arparately by site. One.way analyses of variance were computed to====== the effects of age, education, family income, distance of residence from plant, and the six social support subscales. None ,,., j of these variables differentiated clients who became more symptomatic from those whose scores were stable or decreased over time. M. m, Additional Analysis-One final analysis abandd be reported. In each ares, clients ,,, y ,k. N, 270 ,.M..y Id5%

• Mum h,**

... e. as.

• =eene -

as - amulumme e. _ -em m em - me 9.' W.. tt ~

MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS TAals 21a. ANxrETY AT THz ANNrvEasAaY IN RELAUoN To LavrL or GaNzaAL SUPPOnT AND PsactrnONs or TMI As DANGEmoUs: TMI CuRNTs Current Danger _ ;,, f ,-ll_ ) ] Yes No Soport 'x .sd N x sd N '. *.. l1 .1

c.

si Low I.546 1.019 34 0.716 0.678 21 t4 .I Medium 0.986 0.852 30 0.720 0.962 ' 20 . i j High 0.856 1.032 25 0.367 0.735 13 i F(danger) = lo.g8; df = 1,137; p < 0.001 2 t F(support) = 3.44; dr = 2,137; p < 0.03 F(latenactica) = 1.21; df = 2,137; p = ns TAaLE 21b. ANXIETY AT THs ANNrvtasAaY IN RztaTroN To Lrvrt or GaNEaAL SUPPOnT AND PracznzoNs or THE SArriv / 1 or try No NEAa A NUCLEAa FAczLrTY: TMI CumwTs 3 l, Unspfe Safe soport x sd N x sd N i Low 1.435 1.003 29 0.822 0.663 25 /,- i hd-imm 1.117 0.902 34 0.478 0.69) 18 l High 0.647 0.929 19 0.733 2.043 18 F(safety) = 7.03; df = 1,137: p < 0.01 F(support) = 3.53; dr = 2.137; p < 0.03 F(ksesection) = 134; df = 2,137; p = ns l leserviewed on or trefore April 15 ("Ih!!, N = 611 Beaver County, N = 25) were J whh clients interviewed after April 15 (TMI, N = 88; Beaver County. N = 39).There were no differences in symptomatology in the TMI group; among Beaver County cisents, those interviewed later had somewhat lowr symptom scores then those inta riewed before April 15. There were no differences in the means of the scel.1 support sid==les. Summary: TMI clients who percetved the accident as dangerous and felt that liv-ins near a nuclear facility was unsafe had consistently higher Anxiety scores. On the other hand TMI clients did not show significantly higher symptom levels than Beaver County clients. COMMENT FNer==lan q/the Results Previous reports in the literature about the mental-health effects of natural and manmade disasters have often described high rates of ?*# i duturbance. The present study, using more rigorous sampling and assessment pro-h.' t.i cedures, found elevated rates of psychological distress in mothers of preschool chil- ,1. 1 ( i ei..,n 271 ' s 2: ~~ - f 4 g e s. l m f E 4

m. . ? o BROMET ET AL dren which appear to be attributab!c to the Three Mile Is!and accident.There were no effects on nuclear plant workers or mental health system clients which could be attributed to the accident. The following discursion explores some of the possible cz. planations for ot r results. f 4,( ~ The' data on the cohort of mothers showed that the risk of new episodes of affeo. V f., tive disorder during the year after the accident was more than three times greater aP'. ^ among TM1 mothers than among their counterparts in Beaver County.These resulta

y'.' '.

• f were striking because the rates of disorder predating the accident were virtually iden-tical in the two areas. Most of the episodes of affective diserder occurred shortly

-.g - after the accident took place. On the other hand, the level of subchnical symptoma. tology remained higher for TMI mothers one year after the accident. g The fact that TM1 mothers experienced significantly more distress thantheir coun-terparts in Beaver County is even more icspressive when we consider the characteris-tics of the comparison site. Beaver County's nuclear plants have been subjected to a barrage of criticism and alarms. In 1974, the Commor. wealth of Pennsylvania ~ formally investigated the allegations about health effects resulting from low-level '/ radiation leaking from the Beaver Valley-Shippingpost plants. The Governor's Fact Finding Committee produced a report showing that there were no unusual health problems in the area. Thus, the sde.::fon of Deaver County as a cornpanson site probably produced a conservative estin ste of the extent to which TMI mothers were experiencing increased psychological stress. / The TMI worker cohort had higher rates of clinical disorder before and after the accident. During feedbsck sessions with the workers from both plams, we learned / that the differences in job mobility at the two sites probably explain this result. Spe. I cifically, Beaver County workers v.ho no longer care to work the long hours required l by nuclear plants and to undergo the stzesses that are part of the job can transfer to one of several coal-fired plants in western Pennsylvania. The cansfer can be done with no loss of pay or union seniori y. The TMI workers have no such option. Leav-las TMI would mean losing their seniority and potentially their pay level.*Thus, through this process of selecdon, the two worker samples were probably not com-parable with respect to mental health. 'Ihe Bearer County workers were, therefore, somewhat younger and hearthier than the workers at "IN!. For both TM1 mothers and workers, reporting greater social support from sheir network of friends and relatives was associated with lowered symptevnsenangy at the annivecary. The conceptual literature on social support suggests that it serves to 1 cushion the effects of stress. To some extent this theoretical construct wassupported in the present study. Conversely, social support Icvel was not related to symptom-l atology level in the comparison mothers and workers among whom no similar life. threatening event had taken place. l The few earlier reports In the literature about the reaction of psychiatric paiems l to disurer had concluded that this population adjusts well when faced with the stress of earthquakes, flooding, etc. While these conclusions were derived from poorly designed investigations, they continue to be supported by the present research which 4 2, ,,'.J sought to minimize methodological deficiencies by collecting data through stan- ,...e ,j-dardized instruments at several points in time 'md by including a sn,--issn group. ,,'" !4 l ',*9 We found no differences bety een these twc patient groups on theSCIA0 Global l1 Severity Indea aine and 12 months after the accident., . \\ n.. a ..U . g, _ m ..m h- ) l s / l i .,..,. - ~ ~.

,e j .? f MESTAL HEALTH OF THREE MILE ISLAND RESIDENTS We thus are led to suggest that psychiatric patients in the TMI area were at no ~ . I (,l greater risk for clinical episodes of anxiety and/or depression, and global symptoma. tologyafter the nuclear accident than ps:thiatric patients residing in an area with a

7. c' 4'g nuclear facility which had not experienced a potentially disastrous accident. We recognize that if the comparison group had been psychiatric patients residing in an

...N area lacking a nuclear facility, significant differences might have been found between . '.+ them and our TMI patients. However, increased anxiety and/or depression in the s,'.*Ik TMI group could not then be ascribed to the accident per set it might simply be asso.

• !i i

ciated with living near a nuclear facility. _d I Relation to OtherStudier: Among the surveys which have taken place in the TMI L ares since the arrufent, two reports have received considerable attention. One report was based on a telephone survey of a random sample of the area around TMI(Houts,1979). A major finding from the Houts report was that fewer people outisde a 40 mile radius (a quasi-comparison group) had one or more symp. toms when compared with residents living in the vicinity of TMI. In general, our comparisons between TMI and Beaver County mothers (our comparisort group) show the same difference. However, the differences between areas for the client pop-j ulat2ons and for the worker populations, while in the same direction, are less re- ,/ markable. Another report was a document prepared by the Task Force on Behavioral Effects for the President's Commission on the Accident at Three Mile Island (Dohrenwend /, 6 et al.,1979). Although many complex issues were addressed in the document pre. a / j pared by the Task Force, the President's Commission report included only a one / i paragraph summary indicating that TMI produced "... Immediate, short-lived -l I mes.tal distress"(Kemeny,1979, p. 35). The results of the present study suggest that altt.ough this assumption may be true of some residents living near TMI, adverse mertal heakh effects were seen in mothers as long as one year after the accident. REFERENCES ] ADLER, A. (1943). Neuropsychiatric complications in victims of Boston's Coconut Grove disaster. /AMA 123, 1098-1101. AND REWS, G. and TENNANT, C. (1978). Ufe event stress and psychiatric illness. Prp. Med. g,545-549. ANTONOVSKY, A. (1979). Hve#A, Strusr. and Coping. San Francisco: Jossey-Bass Pub. fishers. BENNET, G. (1970). Bristol floods 1968. Controlled survey of effacts on health of local com. munity disaster. Brir. Med. /. 3,454-458. BERKMAN, L and SYME, S.L. (1979). Social networks, host resistance. and mortality: a nine. year follow.op study of Alameda County residents. Art. /. Epidemiol. 109, 186-204. BROWN. G. (1974). Meaning. measurement and stress oflife events. In B.S. Dohrenwend and B.P. Dohrenwend (eds.).Strust/sd14fe Everris: TheirNature and Effects. New York:Wiley. pp. 217-243. ' 2*y.d

• f -@4 BROWN, G., BHROLCHA!N, M., and HARRIS. T. (1975). Social class and psychiatric dis.

j turbance among women in an urban populadon. Sociology 9, 225-254. ,,',,;.'f ,I BROWN, G.W., HARRIS, T.D., and PETO, J. (1973a). Ufe events and psychiatric disor. ~ ders. Part 2: Nature of causallink. Pry. Med. 3,159-176. ,46,b BROWN. G.W., SKLAIR, F., HARRIS, T.D., and BIRLEY, J.LT. (1973b). Life events and .{

• 6 psychiatrie disorders. Part 1: Some methodological ksues, Pry. Med. 3,74-87.

273 s --man .m.. -i f I

f..

3 g. ./ ~

e. e - y-en e p.* 4 -

f

• r' b

s e BROMET ET AL CAPLAN G. (i964). hinciples e/ Preventive Psychiatry. New York: Basic Books. CAPLAN. G. (6974). Support systens end Community Mental Nas/A New York: Behavioral Publications. . E.t* 6 CASSEL J. (1976). The contribution of the sodat environment to bort resistance. Am. J. Epidemiol. 104.'107-123. ',', y COBB 5. (1776). Social support as a mediator of life stress. Psychosore. Med. 38,300 314. COBB. 5. sad LINDEMANN. E. (1943). NeuiwJd.hk Observatbns. Ann. Surg.117 , 6, 814 824. p*. n b DEAN. A. and LIN, T. (1977). De stress-buffering role of social. support. J. Nervous and . r 'b! Menrel Dsease 166. 433-417. DEROCATis. LR. (1977). De SCf. 90 Manual 1: Scoring, a6ninistration and procedures g for the SCL.90. Baltunore: The Johns Hopkins University School of Medicine. Nimt Psychometrics Unit. DIXON. W.J. and BROWN. M.B. (1979). AMDP 79: Alomediad Computer hierunst A5eries.1979. Berkeley: University of California P ess. DOHRENWEND. B.S. and DOHRENWEND. B.P. (if1). Some issues in research oc stress. fut life events. J. Nervoue sad Mental Dis. 166.7-13. / DOHRENWEND. B.P., DOHRENWEND. !!.5 KASL. S., et al (1979). Technical staff analysis report on behavioral effects. Presented to the President's Commassen on the Acci. dent 6:Three Mile Island. DOR.5HAV N. (1978). On the long-range effects of concentration camp internment on Nad victims: 23 years later. J. Consult. Cia. Psy. 44,1-11. g DYNES, R.R. and QUARANTELLI. E.L. ne fa mily and comnn:nity react. ens to disaster. !a H. Parad. H. Resnik, and L. Parad. Emerferry and Disaster Management. Bowie. MD: 8 The Charles Press,1976 (Chapter 20). EDWARDS. J.G. (1976). Psychiatric aspects of civilias disasters. Dir. Med. J. 1.944-947 1 1976. ENDICOTT. J. and SPITZER R. (1978). A diagnostic interview:The Schedule for Affective i Disorders and Schisophrenia. Arch. Gen. Psy. 35,837-844 FLYNN. C. (1979). Three Mile Island Telephone Survey: preliminary Report en Procedures and Findings. Mountain West Rascarch. Inc September. Unpubished manuscript. GORE S. (1978). The effect of social support in moderating the henkh comwr= aces of anem. 1 ployment. J. Neet A Soc. Beh. 19. 157 163. HENDERSON 5. (1977). The social network, support, and neurosic the fension of attaca. ment in adult life. Brit. J. Psy. 131. 185-191. HOUSE J., McMICHAEL. A WELI.S. J. et al (1979). Oaupational sesess and hashh among factory workers. J. Neal:A Soc. Seh. 29,139-160. HOUT3. P., et aL (1990). Heahh.Related Behavioral Impact of the Three Mile Island Naclear Incident. Submitted to the TM1 Advisory Panel on Henkh Research Studies of the Pennsyl. vania Department of Health. Unpublisheel manuscript. KEMENY. J.G. (1979). De President's Com=uian on the Amident at Three Mile Idand. Washington, DC. KINSTON W. and ROSSER. R. (1974). Disaster: effects on snental and physical state. i ' /. Pspehosom. Res. 13, 437-456. LEOPOLD, R.L. and DILLON. H. (IM3). Pr/cho eastomy of a disester:Along.eenn study of post.tr==n* neuroses in survivors of a marine exphaa Ant. J. Psycht.119,913-921. .. y 1 LINDEMANN. E. (1944). Symptosastology and managenset of acuse grief. Am. J. PocAL 9, ? 101.141. t ?. N R LOGUE, J., HANSEN. H., and STRUENING E. (1979), Fnumiam! and physical distress .}} following Harncane Agnes la Wyoming Valley of - Au publ. NrettA Ag. 94 l[! e. l - , t, {,

1. 5-502.

9. o.y 274

? ' - - ~ ._._...._.l ee e t qNr we F. p -..g9g' g D a

V l MENTAL HEALTH OF THREE MILE ISLAND RESIDENTS V 'o.'i LOWENTHAL, M.F. and HAVEN, C. (1968). Interaction and adaptation: intimacy as a criti. i cal variable. Am. Soc. Rev. 3J. 20 30. .. !)j* ' MILLER, P. and INGHAM, J. (1976). Friends, confidants, and symptoms. Soc. PsycAL 11 SI-58. MOORE, H. (8958). Some emotional concomitants of disaster. Mental #priene 42,45-50. O' i I MOORE, H.E. and FRIEDSAM, J.H. (1959). Reported emotional stress following a disaster. Socist forcer 38,135-139. d MOOS, R. (ed.). (1976). Numan Adaptation! Copint with L(ft Criser. Lexington. MA: DC L H ath. ~ MUELLER, D., EDWARDS. D., and YARYlS, R. (1977). Stressful life emts and psychi. atric symptornaratasy: Change or undecirability. J. Neelsh and Soc. Seh. 18, 307-317. MUELLER. D.. EDWARDS. D. and YARVIS. R. (1978). Stressfullife events and community MUELLER, D. (1978). Social networks and depression: an exploratory study. Unpublished

• /~

mental health center patients. J. Nervous and Mental Dis. 166, 16-24. manuscript. s. MYERS, J.K., LINDENTHAL, J.J., and PEPPER. M.P. (1971). Ufe events and psychiatric impamnent. /. Nervous sad Mental Dis. 152, 149-157. c' MYERS J.K., LINDENTHAL, JJ.. PEPPER. M.P., and OSTRANDER. D.R. (1972). Ufe events and mental status: a longitudinal study.1. Res/th and Soc. Beh. 13, 398-406. MYERS J.K., LINDENTHAL, JJ., and PEPPER. M.P. (1975). Ure events, social integra. ./ tion and psychiatric symptorsatology. J. Health sad Soc. Beh. 16,421-427. / NUCKOLLS. K.B., CASSEL, J., and KAPLAN, B.H. (1972). Psychosocial assets, life crisis j l, . and the prognosis of prignancy. Am. /. Epedemiol 95, 431-441. ~ PARKER. O. (1977). Cyclone Tracy and Darwin evacuees: on the restoration of the species. l7 Brit. J. PrycAL 130, 548-555. PATTISON. E.M., DEFRANCISCO, D., WOOD, P., et al. (1975). A psychosocial kinship 1 model for family therapy. Am. /. PsycAL 132, 1246-1231. PAYKEL E (1974). Ufe stress and psychiatric disorder: Applications of the clinical ap. i preach. la BS. Dohrenwend and B.P. Dohrenwend (eds.). Strest/kl L4fe Events. New York: Waey. 1 RABKIN, J. and STRUENING, E. (1976). Ufe events, stress, and illness. Saence 194,1013-1020. RICHMAN, N,(1976). Depression in mothers of preschool children. J. Child Psychol and 1 PsycAL 11, 73-78. $1VADON, T. and FERNANDEZ, A. (1968). Psychopathological disturbances in a nuclear l . work milies. Annaier Madco.Psychologiguer 2,201-213. l; SPITZER. R., ENDiCDTT, J. and ROBINS. E. (1978). Research Diagnostic Criteria: Ra. tionale and reliability. Arch. Gen. PsycAL 35,773-782. SOKOLOVSKY, J., COHEN, C., BERGER. D.. and CEIGER, J. (1978). Personal networks of en. mental patients in a Manhattan SRO hotel Human Organi:arion 37,5-15. TIERNEY, K. and BAISDEN, B. (1979). Crisis intervention programs for disaster victims. Washington DC, DHEW Publications No. (ADM) 79 675. TITCHNER, J.L. and KAPP. P.T. (1976). Disaster at Buffslo Creek: family and character

• 2.. e, 3

range at Bufyalo Creek. Am. J. PsycAL 133, 295-299. s'e. 4 l' TOLSDORF, C. (1976). Social networks, support, and coping: an exploratory study. remity 'i. p Ascost 15,407-417. /.*fI l8 TYHURST J.S. (1951). Individual remerinm to community disaster. Am.J. PryeAL 197,764

• \\.' ' '

769. .ks'..} lf WEISSMAN, M., and MYERS, J. (1978). Affectives disorders la a U.S. urban community. Arch. Cer. PsycAL 35,1304-1311. % b,,, ZUBIN. J. (1978). How to break the loslam in schizophrenia. Prm presented at American w l 275 D l: i '.t y

.r ~ ~ s>. l... BROMET ET AL. Psychological Anociation meeting. Toronto. Assast.1978. ZUSMAN J. (1976). Meeting mental health needs in a disaster: A public hashh view. Is i Parad. Resnik, and Parad. Einergency and Assiser Maassenser. Bowie. MD: The Qiaries Press. (Chapter 21). p g Address nprint requests to 47,.S Dr. Ewlyn J. Bromet

• • (l.

Dn:partment Q,fPsychiatry ,.~, C, Western Packnatric Instkute and Clinie . University ofPittsburgh O.*

• • \\

3811 O'Hara Street

• • h ri e n w n L

_1 B ./ e>*. 1 i 1 / + f g 1 1 e % J... .4 ,u. p . i.. i.'

• • *..,)

4,f. 1. i 276 ..\\ e z.

==

• \\

o / l. y .y -+-e% w-m

1 (, O, ,) a L Proceedings cf the Prinsylvazia Academy cf Science 57; 99-102. I981 RADIATION DOSE ASSIGNMENT TO INDIVIDUALS RESIDING NEAR THE THREE MILE ISLAND NUCLEAR STATION' DAVID GUR8, WALTER F. GOOD', GEORGE K. TOKUHATA', MARILYN K. GOLDHABER', JERRY C. ROSEN', GOVIND R. RAO', JOHN M. HERRON', DAVID M. MILLER' and ROBERT S. HOLLIS' O. ABSTRACT estimates from the 1972 Metropolitan Edison Company Final A method to assign likely and maximum possible whole-body Safety Analysis Report (FSAR)(ME72). These projections were gamma doses to individuals who resided near the Three Mile based on the 1970 census data in conjunction with information Island Nuclear Station (TMI) during the 1979 accident has been from state and county planning boards. In addition, most of developed. The location of residence in relation to TMI (distance these estimates provide the likely and maximum possible doses to and direction) and information on movements into and out of the one individaal who was expected to have received the largest ex-area during the 10 days following the accident were used in con-posure, but the main emphasis was on the " population" rather junction with estimated time dependent dose rate distributions than on the " individual", to assign likely and maximum possible doses to the 34,000 For educational, public relations and defensive epidemiology members of the 13,000 households residing in the 5 mile radius, purposes a methodology was deseloped to assign and document Estimates of average likely and average maximum possible gam-likely and maximum possible whole body gamma doses to in-ma doses are approximately 9 and 25 millirems, respectively, dividuals in the immediate surrounding population (5 mile These results compare well with doses estimated by other in-radius). Estimated time dependent dose rate distributions were vestigators, superimposed on selected information from the special census of that population. The location of residence in relation to TMI (distance and direction) and information on movements into and INTRODUCTION out of the area during the 10 days following the accident were used in conjunction with estimated time dependent dose rate During the weeks following the accident at the Three Mile distributions to assign likely and maximum possible doses to the Island (TMI) Nuclear Station a major effort was mounted to 34,000 members of 13,000 households located within the 5 mile assess the potential impact on the surrounding population. The radius. This report describes the methodology which was used to primary and immediate concern was the possible health impact assign gamma doses to individuals residing in the five mile radius. from the radiation dose delivered to the population (RE80; RE79; The limitations associated with such estimates are discussed. Ad79). A great deal of public uncertainty arose over the severity Some of the adsantages of establishing such individualized of the accident and whether the situation was controllable. documentation of dosimetry and other information (i.e., evacua. Though no official order was given, a substantial portion of the tion) are also pointed out. nearby population temporarily evacuated their homes. As management of the accident progressed and radiation measurements were refined, it became evident that the popu. METHODS AND RESULTS lation-integrated radiation exposure was very low (US80; Wo79). Because of this knowledge, public anxiety in the vicinity of TMI A. Population Data Base declined somew hat, yet was higher than that in comparison com-In order to provide improved baseline data for future evalua-munities for up to one year after the accident (Br80; Ho81). The tions of possible health effects, the Pennsylvania Department of early uncertainties concerning the consequences of the accident, Health, in cooperation with the Center for Disease Control and and the evacuation of the area by a significant segment of the the U.S. Bureau of the Census, conducted a specialcensus of the local population were important stresses which could well result TMl population in June of 1979. Those included in the census in detrimental mental health changes (RE79; Br80). resided approximately within a five mile zone (FMZ) ofIhe plant Early dosimetric studies in the TMI area used population at the time of the survey (PA8I). Information collected on each individual included such basic demographie data as age, sex, ' Received for publication March 27, 1983; reviewed; accepted marital status and occupation, a brief health history, residential May 18,1983. address and specific information on movements into and from ' Department of Radiation Health, University of Pittsburgh, the five mile area during the ten days following the accident. In an Pittsburgh, PA 15261 effort to document all essential information related to the TMI ' Pennsylvania Department of Health, P.O. Box 90, Harrisburg, PA 17120 \\ population, a computer-based, digitized map of the area was developed and the census data superimposed onto the map. A

m 3 .c 100 PROCEEDINGS OF THE PENNSYLVANIA ACADEMY OF SCIENCE Vol. 57, Issue I,1983 TABLEI. Representative estimated integral doses (10 days) as afunction of direction and distancefrom TMI. Doses are given in miems. Angle (degrees from Distance from TMI (meters) north) 300 600 1000 1500 2000 4000 6000 00 230 100 55 42 40 16 9 20 400 150 81 74 71 29 17 40 91 41 27 25 24 11 6 60 110 46 28 32 17 10 6 90 99 41 32 19 13 5 2 130 79 39 27 31 28 14 7 180 23 11 6 4 3 3 2 240 40 19 12 8 12 5 2 300 380 180 110 80 93 37 20 TABLE 2. Representative estimated and measured in tegralexpo.rures (10 days) in specific locations around TMI. Approximate Estimated Measured Distances Exposures Exposures Location Direction (Meters) (mR) (mR) References 1 NNW 300 980 900-1100 Ad79;US80;Wo79;Au79 2 NNE 1200 $2 51-61 US80;Wo79;Au79 3 SE 3700 14 11 15 US80;Au79 4 WSW 2l00 18 12-17 Ad79;US80,Wo79;Au79 5 North 4200 16 10 Wo79;Au79 6 NE 5500 6' s 5' Sh81 ' Indoor dose estimates. m regional map covering approximately 400 square miles centered developed by Woodard (Wo79). The necessary steps in develop- , on TMI was digitized into a 512 x 512 array. This map can be ing such a distribution have been described in detail elsewhere , shown on a high resolution computer graphic display system and (Wo79). In brief, an assumption is made that other than several ' consists of such general features as paved and unpaved roads and transient peaks, Ihe emission rate during the first 10 days declines highways, rivers and streams, and population centers (Ra82). exponentially. This general trend has been demonstrated in Table The TM1 census was conducted in late June,1979, approx-4-3 of Woodard's report (Wo79). Hourly gamma doses can be imately three months following the accident. One-hundred and ' estimated by sector averaging the plume concentrations around twenty-nine enumeration districts (E.D.'s), comparable to those the wind direction. The inert gases which were considered in the used in the 1970 U.S. census were thoroughly canvassed. The derivation of the time dependent dose rate distributions were TMI Population Registry resulting from the census effort has Xenon-133,133m,135,135m and krypton-88. These dose esti-been e,timated to be 95% complete (PA81). The geographical mates can be adjusted by comparison with TLD readings which boundaries followed during the TMI census extended occasional-have been corrected for background and energy dependent sen-ly beyond fise miles to conform to township lines or natural con-sitivity (Wo79; Au79). The TLD readings are assumed io repre-tours of the land. The estimated location of residence of each of sent the integral outdoor dose at the measurement locations. An the 13,000 households participating in the special FMZ census hour-by-hour time dependent dose rate distribution to fit those was superimposed onto the digitized map. Each household can be integral doses can be derived by an iterative process (Wo79). identified by a household number within one of the 129 E.DJs. Linear or weighted least square fits with or without wake correc-Individuals within a given household can be further distinguished tions can be performed. The main differences between the dose by a person number in thc household. Wit h the assistance of these rate distribution which was used here and the one derised by identification numbers, the residential location (distance and Woodard are the TLD sensitivity corrections (Au79) and the use direc. ion coordinates) and other demographic attributes of any of a rotating sector around the actual wind direction ( 11.25') individual within the FMZ can be easily retrieved. Additional rather than a fixed 16 sector grid. Integral dose estimates in details concerning this comprehensive data base and the pro-various representative locations from the estimated dose rate cedures for its validation are described elsewhere (Ra82). distributions and the TLD readings are gisen in Tables I and 2, respectisely. The limitations of such a procedure have been B. Estimates of Time Dependent Gamma Dose Rate Distribu-discussed (Wo79). tions Several time dependent dose rate distribution estimates for the C. Dose Assignment to Individuals first ten days after the accident are available (Au79; Ad79; US80; Once the time dependent dose rate distributions were esti-Wo79, Kn80). The estimated time dependent dose rate distribu-mated, masimum possible and "likely" gamma doses were tion used here for indisidual dose auignment is similar to that assigned to each indisidual.

r r3 )* RADIATION HEALTH: GUR, ET. AL 101 TABLE 3. TABLE 4 Representative estimated maxsmum posstble gamma doses (mrem) in Representative estimated "hkety" gamma doses Lmrem) in dsfferent sectors as afunction ofdostancefrom TMI. The highest dofferent sectors as afunction ofdistancefrom TMI. The highest assinned (MAX). the average of maxsmum posssble da:es assigned assigned "hkely " dose (MA X), the average of "likely " doses assiened (A VGl end the standarddersation tSD) m each subgroup arepresented. (A VG) and the standarddeviation (SDj in each subgroup arepresented. Combined Combined Population Population Distance Sector Size Distance Sector Site (km) I(N) 2(NNE) 3(NE) (3 sectors) (km) 1(N) 2(NNE) 3(NE) (3 sectors) 0-1.6 MAX a 50 57 60 0-1.6 MAX a 20 29 60 AVG 45.1 46.5 AVG 14.8 11.7 SD 3.4 6.6 SD 7.3 9.1 1.63.2 MAX 98 80 55 348 1.6-3.2 MAX $0 42 26 348 AVG 76.7 47.6 44.8 AVG 35.8 19.2 13.0 SD 19.3 14.9 6.3 SD 13.5 11.8 8.7 3.2-4.8 MAX 64 48 29 5756 3.2-4.8 MAX 33 27 14 5756 AVG 30.2 19.1 19.0 AVG 12.6 6.2 5.9 SD 8.6 5.8 3.9 SD 5.9 5.0 4.3 4.8-6.4 MAX 36 36 19 5611 4.8-6.4 MAX 19 19 9 5611 AVG 21.1 17.7 12.2 AVG 9.0 7.0 3.9 SD 6.1 8.9 2.7 SD 4.1 5.6 2.8 6.4-8.0 MAX 22 19 11 623 6.4-8.0 MAX 12 10 5 623 AVG 13.0 16.7 8.4 AVG 5.5 8.0 2.2 SD 3.7 2.4 1.1 SD 2.1 2.5 1.8

• Population size = 0.

~ Population size = 0. The maximum possible dose assigned to individuals is the 10 may not be greatly overestimated, the population dose is over-day integrated outdoor estimated dose in the location of residence estimated by approximately 25 percent. with no shielding or evacuation corrections and multiplied by a

d. The dose assigned using this procedure is, for the most part, safety factor (S) which is derived from the deviations of the ratios skin dose (entrance dose) rather than "whole body" dose w hich is between estimated and measured doses in the region. For the dose lower.

rate distribution used to generate Table 3, S = 1.5. Due to the above considerations, we believe that the likely dose Each indisidual was assigned a "likely" dose based on the assignments are overestimated by approximately 40 percent for following procedures: most people in the investigated population.

a. A person is exposed if his residence location lies within the Selected dose estimates are presented in Tables 3 and 4. Based

" exposed" sector for a given hour, on the assumptions stated abose, the highest maximum possihle

b. A person is assumed to remain outdoors four hvurs a day dose assigned to an individual was 165 mrem and the highest and remains indoors otherwise. An average protection factor of "likely" dose assigned to an individual was 80 mrem. These 1.35 was used based on recommendations made in other reports results agree well with other published estimates (Ad79; Wo79; (Au79).

US80; Sh81). The average dose within the entire five mile zone c A person is assigned a dose of zero on a given day if he was was estimated to be approximately 9 mrem and temporary evacu-not in the area, ation resulted in an estimated 14% redoction in dose. This is due

d. A person receises a total daily dose if he was in the area for to the fact that a large portion of the dose from inert gases was any part of the da).

receised during the Drst two days prior to the time most indi. Note that the "likely dose" assignment is not based on any viduals had evacuated the FMZ. Average dose to pregnant fe-statistical optimization in a specific region. It is a rather conser. males within the fise mile ione was approximately 8 millirem with vative estimate for the following reasons. a someu hat larger estimated dose reduction ( ~ 30%) due to an

a. Each time an individualindicated that he was within the five earlier esacuation. It should be noted that this report deals only mile zone,it was assumed that he remained there the entire day with "whole body gamma dose" from inert gases and not with (24 hours).

thyroid dose due to inhalation and/or ingestion of radioiodine.

b. Outdoor actisities were awumed to consuine four hours a This dose has been also estimated and found to be estremely low day, which is a fairly high estimate. Other reports use between (Wo79).

two and four hours per day. This estimate is probably escewise for the season in question.

c. The deris ed dose rate distribution ( for each hour) is based on DISCUSSION the integral dose for ten days as measured by the dodmeters.

Because of the higher precision awociated with the doses Maximum possible and "likely" radiation doses were assigned measured near the plant, t he esiirnated dose rate dist ributions are to indisiduals w ho resided near TMI during ihe accident (March. more heasily weighted by the measurements obtained on the April,1979). The procedures far assigning these doses utihres island. This procedure resulted m the oserestimation of dose as estimated time dependent dose rate distributions in conjunction distance from the plant mereases. While the dose to some areas with the location of residence and mosements into and out of the i e

d a 102 PitOCEEDINGS OF THE PENNSYLVANIA ACADEMY OF SCIENCE Vol. 57, Issue I,1983 $ mile area during the 10 days following the accident. Despite the Ho81 Houts. P.,1981, " Health Behasioral Impact of the Three large errors associated with these estimates (approximately a fac. Mile Island Nuclear Incident: Parts I,11,111", Pennsylvania tor of three), and the fact that the "likely" dose is a conservative Department of Health Publication. ' estimate for most persons in the population, the results obtained Kn80 Knox, J.B., Dickerson, M.H., Greenly, G.D., Gudiksen, using this methodology are in good agreement with other P.H., and Sullivan, T.J.,1980, " Utilization of the Atmos-reported collective doses. Because of the unique interest in this pheric Release Advisory Capability (ARAC) Services During population, such individualized dose assignments may prove to and After the Three Mile Island Accident", Lawrence Liver-be an important tool for future epidemiological studies, more Laboratory. ME72 Metropolitan Edison Company,1972, Final Safety Analy-sis Report, Three Mile Island Nuclear Station-Unit 2 (Volume 2). ACKNOWLEDGEMENTS Pa81 Pasciak, W., Branagan, E.F., Congel, F.J., and Fairobent, J.E.,1981, "A Method for Calculating Doses to the Popula-The authors thank Keith Woodard for his major contributions tion from Xe 133 Releases During the Three Mile Island Acci-in the derivation of estimated time dependent dose rate distribu. dent". Health Physics 40,457 465. tions. This work was supported in part by Contract Number PA81 Pennsylvania Department of Health,1981, "The Three 632482 from the Pennsylvania Health Department. Mile Island Population Registry: A General Description". Ra82 Rao. G.R., Good, W.F., Gur, D., Herron, J.M., Toku-hata, G.K., and Goldhaber, M.K.,1982, "The TMI Popula-tion: A Closer Look" Proceedings of the Pennsylvania REFERENCES Academy of Science 56,44-49. l RE80 Report of the Governor's Commission on Three Mile Ad79 Ad Hoc Interagency Dose Assessment Group,1979," Pop-Island, February 26,1980, pp.ll2-Il4. ulation Dose and Health Impact of the Accident at the Three RE79 Report of the President's Commission on the Accident at Mile Island Nuclear Station". Report NUREG-0558, USNRC. Three Mile Island, October 1979, pp 34-35. Au79 Auxier, J.A., Berger, C.D., Eisenhauer, C.M., Gesell, Sh81 Shuping, R.,1981,"Use of Photographic Film to Estimate T.F., Jones, A.R. and Masterson, M.E.,1979, " Report of the Exposure Near TMI", Health Physics 41,195-199. Task Group on Health Physics and Dosimetry" Presented to US80 U.S. Nuclear Regulatory Commission,1980, " Report to the " President's Commission on the Accident at Three Mile the Nuclear Regulatory Commission from the Staff Panel on Island", the Commission's Determination of an Extraordinary Nuclear Br80 Bromet, E.,1980, Psychological, Behavioral, and Social Occurrence (ENO)", Report NUREG--0637 USNRC. Aspects of the Three Mile lsland Incident ("The Mental Health Wo79 Woodard, K.,1979, " Assessment of Offsite Radiation of Residents Near the Three Mile Island Reactor: A Com-Doses from the Three Mile Island Unit 2 Accident" Report prehensive Study of Selected Groups"). Interim Report. TDR-TMI-Il6, Pickard, Lowe, and Garrick, Consultants. l i ( l l ^ i r i L_

\\cg.(,ay W1 . Haaps K. Scand j work environment health 9 (1983) 341-345 / (4 /z'i'd / c: tor of the g ovia r deu.,t of M, NN%g, g,,jfg i ilum. Sct.nd 192-202. 1E. Determi-

al mrterials Correlates of mental health in nuclear and W

au eidung von coal-f. ired power plant workers

1. s'

-itsmed ar.

1971) 122.

by David K Parkinson, BM, MS,' Evelyn J Bromet, PhD2 $I PARKINSON DK, BROMET EJ. Correlates o'f mental health in nuclear and coal-fired 19P3 power plant workers. Scand j tmrk environ health 9 (1983) 341-345. The mental health of 104 nuclear workers at the Three Mile Island plant was compared with that of 122 workers Rom another nuclear plant and 151 workers from two coal fired generating plants.*Ihe coal. fired plant workers were somewhat more symptomatic than the nuclear plant workers. Assessments of work environments Showed that the coal. fired plant workers perceived less stress but more problems with workplace exposures than the nuclear plant workers. Negative perceptions of work and mantal stress were both strongly and independently related to mental distress. Overall. the results suggest that the Three 8 Mile Island accident did not engender long. term psychological difficulties in workers evaluated 2.5 years after the accident. Key terms: job satisfaction. occupational exposure, work stress. Prior to the accident at the Three Mile after the accident). Significant correlations 2 Island nuclear plant in March 1979 the between job tension and sympomatology mental health of nuclear power plant were also found. workers had not been studied from an epi-In our own work, sponsored by the Na-demiologic perspective. The accident at tional Institute of Mental Health, face-to-Three Mile Island provided an opportunity face interviews assessing diagnostic and i' to examine the workers' mental health in subclinical levels of psychiatric disorder relation to the accident. Two studies were were conducted twice during the year after subsequently conducted. Kasl and his the accident, in January and in April of 1980 colleagues (5) carried out a telephone (1). Unionized workers at the Three Mile survey in August-September 1979 under Island plant were compared with members the auspices of the Behavioral Effects Task of the same union at the Beaver Valley-Group of the President's Commission on Shippingport nuclear plant in western the Accident at Three Mile Island. Super-Pennsylvania, approximately 250 miles visory and nonsupervisory workers at away. Minor differences in rates of clini-Three Mile Island were compared with cal psychiatric disorder were found, al-nuclear workers at a similar plant 40 miles though the current sympm levels away. With respect to mental health, non- [ based on scores from the Symptom supervisory 'Ihree Mile Island workers Checklist.90 (2)] of the two groups of work. retrospectively reported more hostility, ers were identical at each interview. Thus psychophysiological symptoms, and de-the symptom differences which Kast moralization than the referents at the time et al(5) noted to have decreased in mag-of the accident. Significant, albeit smaller, nitude over a six-month interval seemed to differences were found in symptoms at have disappeared ten months after the the time of the interview. (six months accident. The problems at the Three Mile Island burMn"by "[SSty of Pitts-plant have continued.The utility itself has n become concerned about the pressures s Department of Psychiatry. University of Pittsburgh, Pennsylvania, United States. to which the workers and their families have been exposed as a result of the acci- 'pa nt dent and contracted with a counseling s ch aIy, 11 t. ttsbur service to provide free psychotherapy to o 15213. USA. 341 3 h.c ,g

f..*W L '-

b Mc - "h l._ ::t f. ). s g mag

v workers and their families. One question tal stress, some of the analyr.es presented addressed in this paper is whether have been based on the married workers Three Mile Island workers have indeed only. been experiencing psychological strain that can be attributed to working at that Methods plant. Data used in this analysis were col-Face-to-face interviews were conducted lected 2.5 years after the accident. The /' second question raised is whether Three in the home and lasted approximately 1.5 Mile Island workers experienced more

h. The subjects were randomly assigned stress at work and at home than other to interviewers, who were mental health TiiVF power plant workers. Union employees at professionals working with the project the Beaver Valley-Shippingport plant and on a part-time basis. They had at least a erk'?

i ~ at two coal. fired generating plants in west-master's degree in a psychiatric f.ield and ern Pennsylvania served as the reference an average of eight years of clinical experi-h.s groups. Finally, the relationship between ence I symptomatology and a variety of psycho-l social factors is examined in relation to Mental health. Current symptom levels ti, the Three Mile Island workers and com, were determined from responses to the 5 pared to the patterns observed for workers Symptom Checklist-90 (2), in which J in the other power plants. Percep-symptoms occurring during the past two tions of stress in the work and home weeks are rated on a five-point distress d environments are the central focus of scale. The average distress level for the 90 V this analysis. items, ie, the Global Severity Index, is -UM., presented in our report. P.3

.dM Subjects and methods Work environment. Three work, environ-ment indicators were selected

work stres s, Subjects job satisfaction and occupational expo-t 7,p a sure. Work stress was assessed by re-l Y~f @

The analyses presented in this paper were sponses to 12 items developed and valid-c f based on information from 377 male ated at the University of Michigan and union workers interviewed in the fall of widely used in studies of occupational i 1981. Included were 104 Three-Mile Island stress (3). These items are rated on a five-workers,122 Beaver Valley-Shippingport point scale and reflect pressures about job workers, and 151 coal-fired generation responsibilities, quality concern, conflicts plant workers. All the employees on the about satisfying other people's expecta-union lists (the entire nonmanagerial work tions, feelings that the job interferes with force, excluding the contractor personnel) nonwork life, and feelings of being trapped were asked to participate. The refusal rate by the job. Scores can range from 0 (no for the three groups averaged 20 %. The stress) to 48 (maximum stress). Job satis-mean age of the workers was 36 years at faction was determined from a five-item the nuclear plants and 37 years at the coal-scale concerning whether the worker fired sites. Relatively more Beaver Valley-was satisfied and happy with his job Shippingport workers had some educa- (and the level of satisfaction and happi. l tion beyond high school (65 % vs 50 % of ness), whether he would make the same the Three Mile Island workers and 46 % decision now to take the job and/or rec-of the coal-fired plant workers). Family ommend it to a good friend, and whether income levels were also higher for the the job measures up to his prior expecta. Beaver Valley-Shippingport workers, tions. Scores can range from 0 (high 60 % reporting annual incomes of USD satisfaction) to 20 (low satisfaction). a 30,000 compared to 44 % of the Three Occupational exposure was assessed by Mile Island and 23 % of the coal-fired having workers rate the extent to which plant workers. Most workers in each plant 11 sets of exposures (eg, toxic chemicals, were married (92 % at Three Mile Island, air pollution, dangerous storage of equip-87 % at the Beaver Valley-Shippingport ment, noise) were problematic; the ratings plants and 80 % at the coal-fired plants). were made on a five-point scale from no Since one focus of this paper is on mari-exposure to exposure presentmg a great 342 M' w'v-v'-ww

.m 7 resented problem. The measure used in this re-hol consumption in the year prior to the , workers port was the average problem rating across interview (ll-point scale from never to the 11 sets of exposures trange = 0-4). three or more times a day), and current smoking habits (5-point scale from non-smoker to smoking more than one pack per day). ' nducted Home environment. The marital relation-l ately 1.5 ship was the focus of the home environ- > issigned ment analysis.pfarital conflict and marital Results I health satisfaction subscales were derived from project a questionnaire which incorporated items Table 1 shows the mean difference in ' t least a from the marital adjustment scales of current symptom levels (Global Severity !ield and Spanier (8) and Pearlin & Schooler (6). Index) among the groups. The Three Mile The marital conflict subscale contained Island and Beaver Valley-Shippingport l enperi. five items that dealt with experiencing workers had similar distress scores, but marital problems, seeking marital advice the workers from the coal-fired plants were from relatives or friends, considering di-somewhat more symptomatic than the 1 levels . ; to the vorce or separation, leaving home because nuclear cohorts. When a cut.off was used which of a fight, seeking professional counseling of one standard deviation above the for the marriage [ range: 5 (no conflict) - mean of the normative male sample past two l distress 25 (high conflict)]. Marital satisfaction in-from Derogatis' validation study,25 cc of ,,r the 90 cluded reports about affectionate behavior the coal-fired platit workers had scores idex. is and positive evaluations of the marri-reflecting high distress in comparison to age [ range: 6 (low satisfaction) - 30 (high 21 "r of the Three Mile Island and Beaver satisfaction >J. Valley-Shippingport workers. mviron. With the one-way analysis of variance Table 1 also gives the results for the ' k stress, the delineated variablas were compared work and marital stress perceptian There il expo. for differences among the plants. Mul-were no significant differences among by re. tiple regression analyses were computed the plants with respect to perceptions of d valid. to evaluate the amount of variance in the marriage. Moreover the mean scores tan and current symptom levels accounted for by were at the positive end of the scales. pational the work and marital environment mea-With respect to the work environment, 1 a five, sures. In these analyses. other independent the workers from the coal-fired plants re- . >out job variables shown to affect current symp-ported less work stress and more problems onflicts tomatology in psychiatric epidemiologic with exposure than the nuclear workers. research were also ' entered. These vari-Table 2 presents the correlations be-expecta. es with ables included number of stressful life tween the Global Severity Index and work trapped events occurring during the year prior and marital stress scales. For the Three a 0(no to the interview, perception of friendship Mile Island workers work stress correlated .b satis. support. age, education. frequency of alco-the strongest with the Global Severity .ve-item worker Table 1. Differences in current symptoms and work and mantal stress among the Three Mile Island his job (TMI), Beaver Valley-Shippmgport (BV-S). and coal-fired plant (CF) workers - One-way analysis of vanance. I happi-TMi BV-S CF . lor ree-y,jges, ,,y,,,, .vhether Mean SD N Mean SD N Mean SD N xpecta. monat Sewnty 1 (h31 h 0 index 0.27 029 104 028 0.21 122 0.34 0.30 151 0.10

action).

Work stress 18.39 7.68 104 19.63 6.98 122 17.42 7.40 151 0.05 ised by Job satis- > which faction 6.96 2.50 104 6.70 2.43 122 6.36 2.58 151 NS [ ~micals' CCCSP"" "*' a exposure 124 0.78 104 1.59 0.79 122 2.40 0.68 151 0.001 f equip-uantai

• I11 tings conflict 7 59 2.43 96 7.34 2.63 106 7.68 2.80 121 NS

' rom no Mantai satisfaction 24 19 4 05 94 23.85 4.19 105 23.51 4.19 121 NS a great 343

~. Index. Job satisfaction and exposure were life events contributed the most strongly unrelated to the Index. Higher symp-to the variance. By contrast the same tom levels were also associated with array of variables explained considerably i greater conciet and less satisfaction in less of the variance among the Beaver the marriage. In the Beaver Valley-Ship-Valley-Shippingport workers (27 "c ). pingport and coal-fired plant groups, all work stress, marital conDict and friend-W, of the work and marital environment ship support accounting for almost all of .J scales correlated with the Global Severity the variance. Among the coal-fired plant Index at significant levels. The coal-workers 39 c of the variance was ex-t 3I fired plant workers showed a more striking plained chiefly by marital conflict and - %"f. relationship between occupational ex-satisfaction work stress and exposure. posure and mental health than the nuclear life events, and education. Thus work 4 workers. and marital stress made significant inde. Table 3 presents the results of stepwise pendent contributions to the workers' I regression analyses which consider the ef-mental health in all three groups. N l fect on symptom levels of the occupational O i 'I. and home environments, along with other independent variables associated with Discussion ?7 mental health in epidemiologic research. i ~ Y Among the Three Mile Island workers, The results indicated that the coal-fired / 40 '*c of the variance was explained by plant workers reported somewhat greater M the array ofindependent variables entered. psychological impairment than the work. 'A Work stress, job satisfaction, marital ers in nuclear power facilities. 'Ihis pattern J satisfaction, friendship support, and was not accompanied by differences in l Q psychosocial stress at home or at work. M-In fact the coal-fired plant workers report- "F Table 2. Relationships between current symptoms ed lower levels of job stress than the n"g,% and work and marital stress by site - Pearson nuclear worirers. The major area in which 7 correlation couffecients. (TMI = Three Mile island. 9 9et - BV-S = Beaver Valley-Shippingport. CF = coal-the workers differed was their exposure fired plants) to potentially harmful agents in the work-place, coal-fired plant workers reporting 4p, Global Seventy indu significantly more problems with ex-s . M' % vanab'* posures than either of the nuclear groups. Tui Bv-S CF Although the majority of exposures work strus 0.48 ~ 0.2e"- 0.38 " studied in the present investigation have Job satisfaction - 0.14 -020- -026 " no known neurotoxic effects (ie, dust. Occupational u posure 0.13 0.17-0.28 "- smoke, fibers. equipment stored danger-Mantai conthet 0.27" 0 37*" 0 44*" ously, etc). the correlation between ex-uantai satisfaction - 0.32 ~ - 0.31 "* -041" posure and the Global Severity Index

• p < 0.05. " p < 0 01. " p < 0.001.

was 0.28 for the coal-fired plant workers. ' ", - ~' Table 3. Regressiott of current symptoms on stress and other variables affectrng rnental health. (TMI = N Tnree Mile Island workers. BV-S = Beaver Valley-Shippingport workers. CF = coal-fired plant workers) TMI (N = 93) BV-S (N = 105) CF (N = 117) vanable Cumula-Cumula-Cumula-Beta tive R8 Beta teve R8 Beta tive R8 i I work stress 0.52"* 0.23 0.23"* 021 024*" 0 32 Job satisfaction 0.16 " 0.35 - 0.10 0.39 Occupational exposure - 0.07 0.38 0.13 0.26 024*" 0 26 Mantal confhet 0 19" 0 14 0 21*" 0 18 Mantal satisfaction - 0.26"' O 28 - 0.13 0 26 - 0.19*" 0.30 Fnendship support 0.23*" 0.33 - 0.15' O 24 - 0 05 0 39 Lste events 0.16 " 0 36 - 0.02 0.27 0.14" 0.37 Age 0.10 0.39 0.02 0 27 0.05 0.39 Education 0.07 0.40 0.08 0.27 - 0.12* 0.36 Onnkmg frequency 0 03 0 40 - 0.02 0.39 Smokmg habits - 0 05 0 38 - 0.02 0.27 -0 04 0 39

• p < 0 05. " p < 0.01. "* p < 0 001.

344 t -y,.----,.3 ,,-,...w -g pm.,--_- _m .g.g.

-.-..~ ~.... -. - V ~, f compared to 0.13 and 0.17 for the Three 1979. there do not appear to be any resi. t strongly Mile Island and Beaver Valley-Shipping-dual effects beyond that period. the same , tsiderably port workers. respectively. Thus work-place exposure problems (perceived or te Beaver real) may partially explain the elevated Acknowledgments 127 9) id friend. symptom levels found among the coal- ' iost all of fired plant workers. This research was funded in part bv The relationship between work. stress National Institute of Mental Health gran't ired plant and mental health has been well docu. was ex. %* 1H35425* mented for white. collar workers (4). Al. g,,;3h to thank JR Bush and Ms M iflict and though work stress and job satisfaction Connell for their help with the data anal. exposure' correlated with each other (-0.38 for Three tus work ysis and members of the International ' ant inde. Mile Island. -0.54 for Beaver Valley-Brotherhood of Electrical Workers and workers-Shippingport, and -0.39 for coal-fired Utility Workers of America for their help plant workers), they were differentially in implementing the study. 3' related to mental health. particularly among the Three Mile Island workers. l Thus it was important to include both variables in the regression analyses. References j ' coal. fired One explanation for the similarity of the correlation coefficients of work and

1. Bromet EJ. Parkinson DK. Schulberg HC.

Dunn LO. Gondek PC. Mental health of I at greater the work. marital stress in relation to mental health residents near the Three Mile Island reac. I is pattern is consistehey in response style in that tor: A comparative study of selected groups. rences in the workers tended to perceive the vari-J prev psychiatt 1 11982) 225-275. at work. ous domains of their lives consistently as

2. Derogatis LR. The SCL.90 Manual 1:

Scoring, administration and procedures more or less stressful. The fact that both than the the work and marital variables entered he'hoo$ of sheo eths rs report. edicine in cal in which the regression equation indicates. how-Unit. Baltimore. MD 1977. ever, that they were not so interdependent

3. House JS. McMichael AJ. Wells JA. Kaplan exposure BH Landerman LR. Occupational stress the' work.

as to cancel one another out. It also sug. reporting gests the need to reevaluate inferences $alth s beha io 20 1 79 139 160 with ex. about the work stress-mental health

4. Kas! SV. Epidemiological contnbutions to ir groups, relationship since perceptions of the the study of work stress. In: Cooper CL.

work environment may reflect more about Payne R. ed. Stress at work. John Wiley & Sons. New York NY 1978. pp 3-48.

• xposures tion have the individual worker than the setting

5. Kas! SV. Chisholm RF. Eskenazi B. The sie, dust.

in which he is employed. In this vein. it impact of the accident at Three Mile Island 3 danger. is instructive to note that objective mea. on the behavior and well.being of nuclear w rkers. Amj publ health 71(1981)472-495. sures of work stress have not been found

6. Pearlin L. Schooler N. The structure of cop.

ween ex. ty Index to correlate with psychological sympto* ing. J health soc behavior 19 (1978) 2 -21. workers. matology (4. 7).

7. Rose R. Hurst M. Herd J. Cardiovascular Overall our findings suggest that the and endocrine responses to work and the risk of psychiatric symptomatology among impact of the Three Mile Island accident air traffic controllers. In: Barrett J. ed.

uth. (TMI== on workers at the plant was unremark-Stress and mental disorder. Raven Press, it workers) able 2.5 years af'ter the accident. We have New York, NY 1979. pp 237-250. ' "73 consistently failed to find that Three

8. Spanier G. Measuring dyadic adjustment.

J marriage fam 2 t 1976) 15-28. cumura. Mile Island workers are more sympto-tie Ra matic than other power plant employees. Although the accident may have precipi-0.32 tated short. term mental health effects in Received for pablication: 29 March 1983 0 39 0.23 GA8 0.30 0 39 0.37 0.39 0.36 0.39 0.39 345 l .fI y ^ b w _-._,..-..-_*w-..- y y-yw-w ww y1w.--e g q.g y7 m e

h @.Q I C5-S p MRICAN A l glh JOURNAL reprint Spontaneous Abortions After the Three Mlle Island Nuclear Accident: A Life Table Analysis MARILYN K. GOLDHABER, MPH, SHARON L. STAUB, BS, AND GEORGE K. TOKUHATA, DRPH, PHD Abstract: A study was conducted to determine whether the the first day of the last menstrual period, the estimated incidence of incidence of spontaneous abortion was greater than expected near spontaneous abortion tmiscarnage before completion of 16 weeks of the Three Mile Island tTMI) nuclear power plant dunng the months gestation) was 15.1 per cent for women pregnant at the time of the followmg the March 28.1979 accident. All persons hsing within fne TMI accident. Combining spontaneous abortions and stillbirths miles of TMI were registered shortly after the accident, and informa-(delivery of a dead fetus after 16 weeks of gcstation). the estimated tion on pregnancy at the time of the accident was col!ected. After incidence was 16.1 per cent for pregnancies after four completed one year, all pregnancy cases were followed up and outcomes weeks of ge<tation. Both incidences are comparable to baschne ascertained. Using the hfe table method. it was found that. given studies of fetal loss. (Am / Public Health 1983; 73:752 759.) pregnancies after four completed weeks of gestation counting from k ir & t h most health officials have predicted that no radiation health effects from the TMl accident would be detectable. Our primary focus in this report is on the incidence of Nevertheless, confusion and uncertainty continuc to spontaneous abortion (the natural expulsion of a conceptus surround the TMI incident. Thirty.five percent of the five-before completion of 16 weeks of gestation) among women mile area residents polled one and a half > ears after the who were pregnant sometime during the two-week " crisis accident stated that they believed the number of miscar-period" (March 28,1979-April 11,1979) following the riages and stillborns had increased since the accident.* accident at the Three Mile Island (TMI) nuclear plant. For Furthermore, the first few days after the accident were the sake of completeness, we also present the incidence of particularly stressful and disconcerting. On the third day of fetal loss over the full term (spontaneous abortion and the crisis, a precautionary advisory was announced by the stillbirth) although we are preparing a more thorough report Governor of Pennsylvania calling for the voluntary evacua-of pregnancy outcome after 16 weeks gestation, based on a tion of pregnant women and preschool children uithin a five-larger data base. mile radius from the plant. Over 60 per cent of the population Radiation exposure of pregnant females can cause a within five miles of the plant evacuated their homes." A variety of damage to the fetus including death or rejection series of psychological and emotional testing that followed from the womb. Biological effects have been shown to be the TMl accident found that the group of pregnant women directly associated with high lesels of ionizing radiation.' 2 and mothers of small children was the most highly stressed Levels below 10 rem (I rem = 1,000 millirem) have not of all groups tested." shown this association either because it does not exist at Stress has presiously been cited as an etiological factor these levels (threshold theory) or because it is too small to in the occurrence of spontaneous abortion'*'2 although

measure, quantification of the relation has not been made. Ani Radiation doses received near the TMI plant were quite it. crease in spontaneous abortion in the TMI area might more low. The University of Pittsburgh's Department of Radiation likely be related to stress than to radiation. In this report we Health estimated doses for each person living within five consider both hypotheses.

miles' of the plant and found that for pregnant females (and their fetuses) the average "likely" full body gamma dose was 8 millirem and the maximum "likely" full body dose Afarerials and Afethods was 50 millirem. Because of the low individual and total population doses estimated by the University of Pittsburgh The Data and, independently, by federal authorities 5 and others,"e In June of 1979, almost three months after the accident, researchers from the Pennsylvania Department of Health, the Centers for Disease Control, and the US Bureau of the Census conducted a census of nearly 36,000 persons within 'The Unisersity or Pittsburgh Department of Radiation Health was five miles of the TMI nuclear plant." The census data, contracted by the Pennsylvania Department of Health to calculate the knoWn as the TMI Population Registry. include an estimated f matimum and likely d ses f r each perwn in the Population Registry. 95 per cent of the population who resided in the area at the "These mclude files compiled by E. Brethauer in Las Vegas. Ns, under the Enuronmental Protection Agency. Dniuon of Radiation Programs from time of the March 28,1979 accident. %,e asked each house-t- D 1979 to 1980. hold respondent whether any females in the household (12 years of age and older) had been pregnant during the TMI Address repnnt requests to Manl > n K. Goldhaber. M PH. Chief. Nuclear accident or at any time since. For pregnancies which had Health Research Program. Dnisson of Epidemiological Research. Pennsyba. I nia Department of Health. P.O. Box 90. Room 10:3. Harnsburg. PA 17108. already terminated, we recorded dates and outcomes of Ms Staub n a Science Research Aswciate in the Program. Dr. Tokuhata n delivery, while for pregnancies still current, we recorded I' Director of the Dnision of Epidemiological Research. Thn paper, submitted expcCted due dates. to the Journal July 26. 1982. was revned and accepted for pubhcation At about the same time as the census was taken, the piernber 27. m Pennsylvania Department of Health began collecting data for - T cI'. mencan Journal of Pubhc Health 0090-00%113 $1.50 a larger, more extensive pregnancy study known as The TMI y a h 752 AJPH July 1983, Vol. 73, No. 7

W TMI STUDY OF SPONTANEOUS ABORTION TABLE 3 7.m._ a Outcome of Women Who were Pregnant during TABLE 1-Prognancy Outcome of Women Who Woro Pregnant during r the TMI Criele: TMI Population Regietry,1979 the TMI Crisie by Month of Last IAenetrual Period (LMP): TMI Population Reg 6etry,1979 Pregnancy Outcome Number Outcome Live Berth 436 Stutt>rth (beyond 16 weeks)* 3 Total Stdit>rth or Spontaneous Abortion (before 16 weeks)$ 25 Month and Women Lrve Spontaneous Other Ectope Pregnancies # 2 Year of LMP Pregnant Berth Abortion

• Abortiont Therapeute Abortion (induced)+

2 Electwo Abortion Onduced) M Jun 1978 15 15 0 0 TOTAL 479 Jul 1978 41 41 0 0 Aug 1970 46 46 0 0 Two of these cases were anginany caesaded as spontaneous atxwtions because of Sep 1978 53 53 0 0 Oct 1978 53 53 0 0 waoo*ne**o'oped tauses-t f these cases was esiminated tram the We tabia anatyss ha abon*" Nov 1978 42 41 1 0 occurred tetore the begmng of she fifth weeai of gestaten. Dec 1978 52 51 1 0

1. One of triese cases was in the taliopian t,ube. one in the abdominal cavity

% a, y,,,, c.,,,,,, ou, io ine.ct,an rom an suo. the other was inouced by em Jan 1979 38 35 3 0 patent s request because of entreme discomton Feb 1979 62 44 11 7 Mar 1979 59 42 10 7 Apr 1979 18 15 2 1 TABLE 2-Por Cent Dietribution of Women Who Were Prognent during TOTAL 479 436 28 15 the TM1 Criele Compared wM Women Delivering Live Births in the TMI Population Registry, Counties' Adjacent to TMI, . m one abanion beore m toginning W h tdm W gemmen and Penneytvania*, by Age, Marttal Status, Education, and tincludes eiecove abonions. therapeutic abomons, and ectopic pregnances Race,1979 Women Delivenng Lue Berths Only who cla.imed to have been pregnant at the time of the census. Women but were not found in the POS, were investigated through Pregnant Tu, personal interviews, followed up by reviews of medical dunng Population Ad acent i Population Cnsis Reg stry Counties

• Pennsytvania*

records pertaining to the pregnancies. We found that 37 of the 80 women had delivered live births and had either moved ^9' W'*) out of the ten mile area or refused to participate in the POS. The other 43 had had outcomes other than live birth as 19 1 1 1 1 shown in Table 1. There were no multiple live births (twins, 20-24 35 i 34 5 34 3 33 3 25-29 33 6 34 9 33 2 32 8 triplets, etc.) although there was one spontaneous abortion ]6 14j involving twins. We counted this as one case of spontaneous 1 9 abo m n. 40 + 04 00 07 06 During the follow-up interview, we asked all abortion Mantai Status Mamed 84 1 86 0 84 4 82.7 cases for the length of gestation of pregnancy, measured Unmarned 15 9 14 0 15 5 17.3 from the first day of the last menstrual period,(LMP) at the E time of delivery, the delivery date, and the delivery place. 2 16.7 16 3 26 2 19 4 For all cases that had appeared in the POS, we took these 12 58 2 58 9 48 0 52 2 13-15 11.9 11 0 12 8 14 0 data from hospital records.* After checking hospital rect 'ds, 16+ 13 2 13 8 12 2 13 4 we found two supposed spontaneous abortion cases 1 tat (' were greater than 16 weeks gestation. These were reclas:i. V[hrte 96 5 96 6 6 86 fied as stillbirths (Table 1). Other 35 34 95 14.0 Demographie ProAle mr MUe*n"n"s*yC.To".o*a'N*M. [e*s"o"**" '"' "'" " The demographic characteristics of the TMI women are shown in Table 2 compared to women in counties near TM1 and in Pennsylvania as a whole." The TMl women were Pregnancy Outcome Study (POS)."The POS study involved similar to women m both the surrounding counties and the all women who gave birth during the two years after the TMI state, except that slightly more TMI women were under 35 accident and lived within ten miles of the plant. Cases for Id, married, high school graduates (or higher), and years this study w ere identified through local hospitals for deliver-White. Although all differences are small, they favor a ies, live or dead, of 16 or more completed weeks of gesta. slightly better pregnancy outcome prognosis for the TMI tion'One year after the census, we returned to investigate the w men than for the general population of women in the e.e surrounding counties or sttte. outcomes of the undelivered cases and to re-assess all cases of outcome other than live birth. First we checked whether Life Table Method the pregnant women listed in the TMI Population Registry Accurate measures of the incidence of spontaneous appeared also in the POS. abortion have always been difficult to obtain. an ideal Four hundred women were found to be in both studies. measure being the proportion of all conceptions which Their outcomes meluded 399 live births and one stillbirth (at spontaneously terminate in death of the fetus. Much of the 40 weeks gestation). The outcomes of 80 remaining women difficulty lies in determining early conceptions, in particular '"in Penn9lvanu. Ictal deaths are reportable after 16 weeks of gesta-tTwo cases were unable to provide gestation penods and these we estimated as occumng dunng the eighth and ninth weeks. tion 753 AJDH July 1983. Vof 73. No. 7

l I j GOLDHABER, ET AL 8 TABLE 4-Charactortatice of the TMt Population Registry (Abortion Study) Compared to Four Other Spontaneous Abortion Studies l I TMI French" Shapiro's Taylor

• Hartap's Approxirnate year of study 1979 1955 1959 1964 1975 t.ocation PA HI NY CA CA Data Source TMI Census Kauas Ca*isus HIP (HMO)

Kaiser (HMO) Kaiser (HMO) Populahon Size (pregnant wornen orWy)* 478 3083 12000 5427 32182 Per Cent White 97 14 83 100 68 Median Age (in years) 25 26 28 26* 26 Per Cent Never Marned 9 9 Average Years of Education 12 6 13 It 13.3

• Sportaneous acomon cases termmateg tionare the bogeneg of the fifth weet of gestaten were onceuced.

8These figures were estimated from the ossenpton of the larger cohort n Taylorle eThese hgures were not avadable within two-to-four weeks after the LMP (two weeks after outcomes: live birth, spontaneous abortion or stillbirth, conception) where the woman has not yet missed a period. elective abortion, therapeutic abonion, and abortion due to Thus, spontaneous abortion rates are usually measured from ectopic pregnancy. The average daily probabilities for any a given confirmed gestational age. About the earliest time for given week of staying pregnant (p) or having one of the five reliable pregnancy confirmation is after four completed outcomes (q, r, s, t, u) are given by weeks of gestation from the LMP. In this study we attempt to measure the probability, or estimated incidence, of spon, p=1-q-r-s-t-u taneous abortion (before the completion of 16 weeks of number of live births gestation) and of fetal loss (over the full term), given that q= pregnancy extended into the fifth week of gestation, for number of woman 4ays

i women pregnant during the TMI crisis.

- {f When we asked during the census, "Has been number of spontane.as abortions or stillbirths pregnant at any time between March 28, 1979 and the r= g. present?", we defined a cohort of pregnant women whose number of woman-days , jg LMPs began as early as June 1978. Missing from our 1{ observation were women belonging to this cohort whose etc. pregnancies terminated before March 28,1979 (see Table 3). WeekiY probabilities are given by Although the missed cases would not contribute to a 7 r

p., p 1 Y "TMI effect" because they happened before the accident, they would contribute to a " baseline effect.' A crude 1 - p' I d proponion (fetal deaths after four weeks gestation, divided 9*"l-p 9 1 T by total pregnancies) would ignore the missed cases and i...

underestimate the true incidence. By using a life table , _ E, method, we were able to eliminate this problem and, at the r = r w same time, isolate a possible TMI effect by utilizing only 1 -- P ) pregnancy experiences after the accident. s A The life table determines the number of woman-days of etc. , j, observation available for each gestational week and relates < 4 the number of terminations occurring during the gestational And, lunar monthly probabilities are given by ' ? week to the number of woman-days at risk. Using a method p = poi p 2 p i p.4 ) f f, described by Taylor,we generated average daily risks for q = q.i + poi q.2 + pwl p.2 q.3 + p i p.2 p.3 q.4 j { y each week, yielding weekly risks, for each of five possible etc. i n e. TABLE 5-Eettmated incidence of Spontaneous Acortion or Stillbirth before End of Specified Week of '.l Geetatan per 100 Pregnancies beyond the 4th Week of Gestation: Women Who were Pregnant 2 during the TMI Criele in the TMl Population Regletry Compared to Pregnant Women in Four Other 4 Studies j I 1 .) b. Incidence Per 100

t 3

ml O Before Population 8 8 (.nd of Registry

• French 8 Shapiro8 Taylor Harfap

}

• p-Week 16 15 1 20 8 18 9 15 9 12 3

[ Full term 16 1 23 7 21 8 18 6 16 4 {n; h'

• Estimated incidence.as conved from the hfe tabie m Append C and excluoes one case of spontaneous aborten before the 3.

tegenrnng of the fifth ween of gestaten. 8 g.. Enoudes from the oenommator. omen who had educed etertens [ ~ 754 AJPH July 1983 Vol. 73. No 7 b l te.

( TMI STUDY OF SPONTANEOUS ABORTION TABLE 4-Cheeswed and Esposte6 (Somed en Pour Sr:sdisoF Numbers of Spanieneous Aber16ene or Simiruse in eseh Bestellen interwel ter Westen ushe sense Pregnant during the TIN Cetete: Tam Popedallen Regleery,1979 Observed Women-Number of FW Nurnber Genestonal Monihe Sporteneous treerval (l.uner) Abortone or (Weeks) et Rek Samurthe French Shepro Taylor Hartep 5-6 102.7 4 11.1 7.6 64 37 9-12 136.2 7 93 11.4 72 55 13-16 161.1 13 7.2 6.4 37 29 17-20 197.2 2 25 26 24 16 21-24 236 4 0 1.9 1.4 1.4 12 25-26 24.3 0 09 09 0.9 1.7 29-32 332.3 0 1.0 1.0 1.0 33-36 367.7 0 1.1 1.1 07 37-40 346 6 1 1.0 1.4 1.7 41-44 76.5 0 0.7 0.7 TOTAL 27 37.08 35 18 26.68 22 6' ostened dets an evemen er eso reervei 'The total eseected nummer of asentaneous shareens uns maand n a tie toets manner and need rates esameted horn the oIher i stuees in eio cases of numens dele Each woman was entered into the life table in the taneous, and the gestation or pregnancy to be the time gestational interval corresponding to the number of days between LMP and delivery, even if the fetus had died pregnant she was (from her LMP) on March 28,1979, the earlier. Arst day of the TMI crisis. The women were maintained Baseline rates varied somewhat from study to study, through each subsequent interval of the life table, contribut-This could be at least partially due to the association of ing to the denominator of " women-days at risk during the pregnancy outcome with certain demographic variables (age, week" until their pregnancies terminated, at which point race, parity, pregnancy spacing, etc.) which may be distrib- - they contributed to the numerators of" terminations during uted differently among the studies. Although it was theoreti-the week" as well. For mathematical consistency we consid-cally possible to " adjust" the cells of our life table by. ered all women to have entered at the beginning of the day variable distributions of a comparison population, this re-and all terminations to have happened at the end of the day. quired access to the comparison data base in appropriately A multiple decrement probability model was constructed by tabulated form (an N dimensional matrix for N comparison the life table for up to a 45-week gestation period (see variables for each gestational interval of the life table); such Appendices A, B, and C). data were not available to us. Benelise C _ -.w Seedles There is a scarcity of good baseline literature on the Resuhs subject of early fetal loss. Four large epidemiological studies Table 5 lists the incidences of spontaneous abocion have calculated incidence using the life table technique ' before 16 weeks gestation and fetal loss over the full term for These studies (referred to in this report by the name of the the TMI women compared to women in the other four primary author) are used as baseline data for comparison studies. The TMI risk was comparable to the more recent with the TMI study (Table 4). studies by Taylor and Harlap? incidences in all four studies were calculated given that Table 6 shows observed and expected numbers of fetal pregnancy extended into the fifth week of gestation. Cases losses for the TMI women per lunar month of gestation. The ending in elective abortions were excluded from the life expected numbers were based on life table rates from each of tables, hence the denominators, in the comparison stud-the four studies applied to the number of women-months at ies.** The incidences of both therapeutic abortions and risk in the TMI population. A low number of fetal losses after abortions due to ectopic pregnancies were very low in the 20 weeks of gestation among the TMI group is apparent, comparison studies and in our study, so that inclusion or balanced by a clustering in gestational weeks 13-16. exclusion of these cases would result in small differences in Although the incidences of spontaneous abortion and rates. Although the de6nition of a therapeutic abortion was fetal loss over the full term were not high, we wondered not speci6ed in any of the studies, we considered it as whether radiation exposure received during the TMI crisis related to problems of the uterine environment, not the or, conversely, stress encountered from the crisis were fetus. When dilation and curettage was performed because associated with adverse fetal outcome. At the time of the of a missed abortion or to terminate heavy bleeding signaling TMI census, data on individual evacuation behavior had ensuing miscarriage, we considered the abortion to be spon-been collected for each person for the 10 days following the March 28,1979 accident. Radiation doses were assigned to each person based on place of residence and time spent in

we round anat the effect of this on the calculation of spontaneous the Ave-mile area of the plant during the 10 days after the abomon and retal loss rates in our study were neeheible. Two studies (French accident. Previous studies have shown that evacuees (per-and shapirovw pecined ihai therapeutic abemons were also encloded and sons who left the area because of TMI) tended to be more s

Ir*re*[7

• [ec'*opi YNUs#

highly stressed than non-evacuees so that longer periods of t ie pon retas iosas and it is assusned that the other studies did hkewise. evacuation Can be Considered a rough maicator or TMI. AJPH July 1983, Vol. 73, No. 7 755

i GOLDHABER ET AL TAaLE 7-Ave ese seunter of Doye Evenusted and Averego Estesnoted were missed due to the time lag between the accident and the mecamen ones nomi m m Accesent by noenancy out. census or to reluctance of respondents to report the informa- "'E # '~ tion. This situation is thought to be minimal and, perhaps, over-reporting is more likely. We were unable to confirm 10 m of Average Days E of our reported spontaneous abortion cases because either Pregnancy Outcome Cases Eve meroms) the pregnancy was never reported to a doctor or medical records were unavailable. Nevertheless, all cases from the LNe Bwthe 436 7.1 83 census were included in our analysis except one where (o 13) (o 43) medical records showed that the woman was not pregnant. The incidence of spontaneous abortion among the TMI 28 62 94 to s6) (1.37) women compared favorably with the four baseline studies. Ectcpc Pregnancy 2 6.5 10 4 The different demographic characteristics of each of the four (o 50) (8.35) data bases account for some differences in incidences. Other

• • 'apeune Abonen facters which contribute are the legal status of induced

,,,,,,o 2 So 4.3 (5 00) (2 40) abortion and the rules for entry into the study. All four Elecove Aborton baseline studies entered women into the life table analysis in (Induced) 11 61 12 6 the gestatior.al interval in which they first reported for to 68) (5 02) -. % _ g-prenatal care. Because early reporters are not generally representative of all pregnant women, a self-selection bias can occur, allowing excess observation of early cases of related stress. Table 7 shows the average most likely total spontaneous abortion relative to early cases of healthy gamma exposure from the TMI accident and the average pregnancies. To avoid the bias, all studies systematically number of days evacuated over the 10-day crisis period for excluded cases first reporting for prenatal care because of women who were pregnant during the TMI crisis by preg. threatened miscarriage. Harlap excluded any case coming in nancy outcome. It appears that mean values of radiation with symptoms of threatened miscarriage whereas the other dose and days evacuated for each pregnancy outcome group investigators extuded cases that aborted on the day the were not different from the live birth group.* pregnancy was first reported or withm a few days after. Inclusion or exclusion of these early cases can alter the Discussion measurement of spontaneous abortion. It remains a problem to correct the bias without forcing it in the opposite direc-The subject of pregnancy outcome in the aftermath of tion. the TMI accident is of interest to both scientific and lay Although the incidence of spontaneous abortion for communities. In 1983 a report will be released which will women exposed to the TMI accident was not unusual, the fully examine the subject in terms of prematurity, immatu-incidence of fetal loss after 20 completed weeks of gestation rity, and congenital defects as well as fetal deaths for all was particularly low. We determined that this was not a deliveries of 16 or more weeks of gestation within a ten-mile problem of ascertainment since we cross-checked our file radius of TMI. The current study on spontaneous abortion against state fetal death certificate listings, which tend to be allowed us to examine spontaneous abortions ofless than 16 fairly complete after 20 weeks. Because fetal death inci-weeks of gestation in an area five miles from the nuclear dences show considerable instability in small regions over plant. short periods of time, it is likely that the scarcity of fetal Previously, we had estimated that census enumeration deaths after 20 weeks of gestation was a random fluctuation, was better than 95 per cent complete." Although we con-especially in light of the fact that we were not testing this ducted the census three months after the TMI accident, we particular hypothesis. Yet, the scarcity was balanced by an were able to track most persons who moved out of the area. excess number of spontaneous abortions in the 13-16-week We estimate that about one per cent of the population gestational period. It could be that some agent (emotional escaped enumeration through relocation."resulting in only stress or trauma from the accident) caused "dooined" four or five pregnant women lost, not enough to alter our fetuses, fetuses that would have eventually been lost, to findings. Also, when we matched our TMI Population Regis-abort earlier. This is a possibility that merits further investi-try against the POS, we found only two cases that were not gation in other cases of emotional trauma to pregnant reported in the census." Finally, the fertility rate, the females and will be investigated further in the POS. number oflive births disided by the number of women aged 15-44, of our TMI cohort was determined to be 65.4 per REFERENCES thousand as compared to 60.2 for Pennsylvania, and 68.0 for

1. Comnunce on the Biological Efects of loniang Radiations: The Efects the Un:ted States in 1979. alt appears from the above that on Populations of Espowre to Low Levels orlomnns Radiation.1980.

pregnancies were reported fairly completely and that con. washington. DC: Natiens! Academy Press. 1980; 477-495.

2. Brent RL: Efects of iomzing radiation on growth and development. be:

ception rates were normal. Contnbutions of Epidemiology and Biostatistics tVol I L Ness-Ziona.Tel-It is possible that some spontaneous abortion cases Aviv: Karger Pasel. 1979. 147-183.

3. Ad Hoc Population Dose Assessment Group: Population dose and health

• We did not take into account any of the important intersemng vanables impact of the accide'it at the Three Mile Island nuclear station. washing-such as age, race. pregnancy spacing. etc.. whrch are associated with ton. DC: US Govt Pnntmg Office.1979.

pregnancy outcome. Nor was the interaction of gestational age at time of

4. Woodward K: Assessment of ossite radiation doses from Three Mile caposure and radiation dose examined. Analysis of the POS data will preside Island unit 2 accident. Packard. Lowe, and Garnck (consultants to far better information on intervening vanables, stress measures, and the Metropolitan Edison Corp.).1979. (TDR-TMI.ll6. revision 0. August 31).

amount of radiation exposure as they relate to fetal death beyond 16 weeks

5. Auxier J A. Berger CF. Eisenhauer CM. Gesell TF. Jones AR. Masterson gestation and other types of adverse pregnancy outcome.

ME: Report of the public health and safety task force on health physics

• • By design, these were not included in our life table analysis because of and dosimetry. Washington. DC: US Govt Pnnting Ofice.1979. (Report possible bias.

to President's Commission on the Accident at Three Mde Island.) 756 AJPH July 1983. Vol. 73, No. 7

TMI STliOY OF SPONTANEOUS ABORTION

6. Houts PS. Miller RW. Tokuhata GK. Ham KS: Heahh-related behavioral Harnsburg. PA: Pennsylvania Department of Health 1980.

impact of the Three Mile Island nuclear incident. Parts 1.11.111. Hams.

16. Taylor WF: On the methodology of measunns the probabshty of fetal burg. PA: Pennsylvania Department of Health.1981 death in a prospective study. Human Biology 1964; 36.86_103.

7. Hu T. Slaysman KS: Health-related economic costs of the Three Mile

17. French FE. Bierman JM: Probabihties of fetal monahty. Pubhc Health Island accident. State College. PA: Pennsylvania State Universaty. Center Rep 1%2: 77.835_847.

for Research on Human Rewurces.1981.

18. Shapiro S. Levine HS. Abramomicz M-Factors anociated with early and K Flynn CB: Three Mile Island telephone survey: prehminary repon on later fetalloss. Advances in Planned Parenthood 1971; 6.45_61 pra edures and Andings. Washington. DC: US Nuclear Regulatory Com.

19. Harlap S. Shiono PH. Ramcharan S. A hfe tabic of spontaneous abortions minion.1979. IN U R EGTR.1093) and the effects of age, panty, and other ianables. Er Hook EB. Porter l

9. Bromet E: Prehmmary report on the mental health of Three Mile Island tedst Reproductive Loss. New York: NY Academy Prew.1980.

] reudents. Pittsburgh. PA: University of Pittsburgh School of Medicine.

20. National Center for Health Statistict Monthly utal statistics report.

Department of Psychiatry.1980. provisional statistics: annual summar) of the United Statet 1979. births.

10. Nuckolls KB, Kaplan BH. Cauel J: Psychosocial anett hfe cnsis and deaths, marnages, and divorces. Washington. DC: US Department of the prognous of pregnancy. Am J Epidemiot 1972: 95:438-441.

Health and Human Scruces. 1980. 28113). II. Monshima HO. Pedersen H. Finster M: The influence of maternal psychological stress on the fetus. Am J Obstet Gynecol 1978, 131.286_ 280 ACKNOWLEDGMENTS 12 Newton RW. Webster PA. Banu PS. Maskrey N. Philhps AB: Psychoso-The TMI Population Registry was developed jomtly by the Centers for Cial stress in pregnancy and its relation to the onset of premature labor. Disease Control. the National Institutes of Health. the US Bureau of the Bnt Med J 1979. 2.411-413. Census and the Pennsyisania Department of Health. Smcere appreciation is

13. Penn91sania Department of Health: The Three Mile Island population extended to the many people who helped, en particular. Jane Bratz from the registry. repon one a general desenption. Harnsburg. PA. Division of Dinuon of Epidemiological Research who prouded valuable information on Epidemiological Research.19NI.

the TMI Pregnancy Outcome Study and to Dr. Paul Sheche. TMl Adusors

14. Tokuhata GK: Pregnancy outcome around Three Mile Island. Harnsburg.

Panehst and Profenor at N.Y. Upstate Medical Center mho prouded PA. Pennulvania Department of Health. Division of Epidemiological consultation to the hfe table method. Special thanks are given to Theresa Research.1981. Brown Joyce Kim. James Lehman, and Vicki Newhn of the Diuwon who

15. Penantsania Health Data Center: Pennnivania utal staintics.1979.

helped collect and venfy the data and proof thn paper. APPENDIX A Number of Women Pregnant et Time of TMI by Week of Geetation and Outcome of Pregnancy: Life Table Analysis, TMI Population Registry,1979 (Excludes One Abortion Before the Fifth Wook of Gestation) A B. Be B B. Bs C D 3 Elective Therapeutic Women at Women-Beginning Women within Live Births Spontaneous Abort ons Abortions Ectopic Risk at Days at Gestational Gestational Week dunng Abortions dunng dunng dunng Beginning of Risk dunng Week i at Time of TMI Week dunng Week Week Week Week Week Week Before 5 76 5 17 0 2 0 0 0 76 M5 6 15 0 0 0 0 1 91 689 7 13 0 0 1 0 0 105 763 8 17 0 2 3 0 0 117 838 9 7 0 3 3 0 0 129 891 10 8 0 1 2 0 0 130 924 11 10 0 0 1 1 0 135 957 12 12 0 3 1 0 0 143 1.014 13 7 0 5 0 1 0 151 102 14 15 0 4 0 0 0 152 1.104 15 9 0 2 0 0 1 163 1.146 16 13 0 2 0 0 0 169 1.211 17 10 0 1 0 0 0 180 1.290 18 6 0 1 0 0 0 189 1.332 19 15 0 0 0 0 0 194 1.402 20 9 0 0 0 0 0 209 1.497 21 7 0 0 0 0 0 218 1.551 22 15 0 0 0 0 0 225 1.622 23 12 0 0 0 0 0 240 1.716 24 9 0 0 0 0 0 252 1.785 25 12 0 0 0 0 0 261 1.856 26 14 0 0 0 0 0 273 1.956 27 10 0 0 0 0 0 287 2.033 28 11 0 0 0 0 0 297 2.115 l 29 10 0 0 0 0 0 308 2.191 30 17 0 0 0 0 0 318 2.287 31 12 2 0 0 0 0 335 2.376 32 9 1 0 0 0 0 345 2.451 33 13 2 0 0 0 0 353 2.502 34 9 6 0 0 0 0 364 2.559 35 7 1 0 0 0 0 368 2.594 36 13 6 0 0 0 0 3 74 2.641 37 10 16 0 0 0 0 380 2 662 38 11 25 0 0 0 0 374 2.587 39 4 41 0 0 0 0 360 2.439 40 7 101 1 0 0 0 323 2.010 41 6 110 0 0 0 0 228 1.249 42 0 74 0 0 0 0 124 619 43 1 37 0 0 0 0 212 44 0 10 0 0 0 0 14 63 45 0 4 0 0 0 0 4 14 TOTAL 478 436 27 11 2 2 l l

GOLDHABER, ET AL APPENOlX 8 Rate

• Per 100 Prognent Women at R6ek Each Geotst6onal Week by Outcome of W

r. Ufo Tatde Anotyels, TMI Population Registry,1979 (Excludes One Abortton Before the Fifth Week of Gestation)

E. E, E E. E. F E ectne Therapeute Begmnmg Lwe Births Spontanaous Aborbons Abortons Ectope Gestabonal dunng A w ons dunng dunng dunng Still Pregnant Week 6 Week dunng Week Week Week Week End of Week 5 00 24 00 0.0 00 97 6 6 00 00 00 00 10 99.0 7 0.0 00 09 0.0 00 99 1 8 00 1.7 25 00 00 95 9 9 00 23 23 00 0.0 95 3 10 00 08 15 00 00 97.7 11 00 00 07 0.7 00 98 6 I 12 00 2.1 07 00 00 97.2 13 00 33 00 0.7 0.0 96.0 14 00 25 00 -e+ 6 U 00 97.5 15 00 1.2 00 00 06 98 2 16 00 12 00 00 0.0 98 8 17 00 06 00 00 00 99 4 18 00 06 00 00 00 99 4 19 00 00 00 00 00 1030 20 00 00 00 00 0.0 100 0 21 00 00 00 00 00 1@0 22 00 00 00 00 0.0 100 0 23 00 00 00 00 0.0 100 0 24 00 00 00 00 00 1@0 25 00 00 00 00 00 100 0 26 00 00 00 00 00 100 0 27 00 00 00 00 00 100.0 28 00 00 00 00 00 100 0 29 00 00 00 00 00 100 0 30 00 0.0 00 00 00 100.0 31 06 00 00 00 00 99 4 32 03 00 00 00 00 99 7 M 06 00 00 00 00 M4 M 16 00 00 00 00 M4 35 03 00 00 00 00 99.7 16 00 00 00 00 M4 37 41 00 00 00 00 95 9 38 66 00 00 00 00 93 4 39 11 2 00 00 00 00 88 8 40 30 2 0.3 00 00 00 69 5 41 47 6 00 00 00 00 52.4 42 59 0 00 00 00 00 41 Q 43 73 9 00 0.0 00 00 26 1 4 02 00 00 00 00 M8 45 90 5 00 00 00 00 95

• $ee test for formulas I

i l i l AJPH July 1983. Vol 73, No. 7 758 1

t TMI STUDY OF SPONTANEOUS ABORTION APPENDIX C Estimated incidence per 100 Pregnant Women at Risk from the 5th Week of Gestation by Gestatien and Outcome of Pregnancy: Life Table Analysis, TMI Population Registry,1979 (Encludes One Abortion Before the Fifth Week of Gestatkm) 45 g,, G ,F. g, E,.G. g, 4G.., g,E>G, g,E.G, gb, M, L,,1 H p 10.000 10.000 10.000 10.000 10.000 10.000 g Therapeute Spontaneous Beginning Lwe Births Spontaneous Electwe Abortons ADodons Gestatonal StJI Pregnant dunng Abortons Abortons dunng Ectopic dunng be' ore Week Week i End of Week Week dunng Week dunng Week Week Week 46 5 97 6 00 24 00 00 00 16 1 6 96 6 00 00 00 00 10 13 7 7 95 7 00 00 09 00 00 13 7 8 91 8 00 16 24 00 00 13 7 9 87 5 00 21 21 00 00 12 1 to 85 5 00 07 13 00 00 97 11 84 3 00 00 12 44-C. [ 00 93 12 81 9 00 18 06 00 00 93 i 13 78 6 00 27 00 06 00 75 ) 14 76 6 00 20 00 00 00 48 15 75 2 00 09 00 00 05 28 16 74 3 00 09 00 00 00 19 17 73 9 00 04 00 00 00 10 18 73 5 00 04 00 00 00 06 l 19 73 5 00 00 00 00 00 02 20 73 5 00 00 00 00 00 02 21 73 5 00 00 00 00 00 02 22 73 5 00 00 00 00 00 02 23 73 5 O CT 00 00 00 00 02 24 73 5 00 00 00 00 00 02 25 73 5 00 00 00 00 00 02 26 73 5 00 00 00 00 00 02 27 73 5 00 00 00 00 00 02 28 73 5 00 00 00 00 00 02 29 73 5 00 00 00 00 00 02 30 73 5 00 00 00 00 00 02 31 73 1 04 00 00 00 00 02 32 72 9 02 00 00 00 00 02 33 72 5 04 00 00 00 00 02 34 71.3 12 00 00 00 00 02 35 71 1 02 00 00 00 00 02 36 70 0 11 00 00 00 00 02 37 67 1 29 00 00 00 00 02 38 62 7 44 00 00 00 00 02 39 55 7 70 00 00 00 00 02 40 38 7 16 8 02 00 00 00 02 41 20 3 18 4 00 00 00 00 00 42 83 12 0 00 00 00 00 00 43 22 61 00 00 00 00 00 44 07 1.5 00 00 00 00 00 45 01 06 00 00 00 00 00 b l e-l 1 l l AJPH July 1983, Vol. 73, No. 7 759

l '* r, 4l at least partly responsible for the higher lesels of not clear. A large proportion of the plants suneyed health promotion services that table 4 shows that are subsidaries of, or have relationships with, na-the paper, petro, and mechan groups prosided. tional corporations that hase plants in many States, Predominance of employees of one sex, the socio-and these relationships may hase reduced the indu-economic levels of the workforce, or management's ence of purely regional factors. It is of interest that response to mandatory pressure might also hase Fielding (7) found lesels of health promotion ac-contributed to variations in the extent and type of tivity in California industry similar to those we ob-occupational health ser ices provided. served. Within the limitations of a cross-sectional suney, we were able to observe that medical senices, at References least up to mandated minimums, are probably as ail-able to most of the workforce in South Carolina Th" / J acobs. P. and Choul. A-Economa e aluation of type of survey does not permit conclusions as to the corporate medical programs j Occup Med M - 273-278 degree to which these senices were established in i l983 L e response to regulatory pressures The suney indi. 2 National occupational sursey DHEW Publication No. 78-114 N ti n 1 Insmute for Occupaional Safety and cated that industrial management in South Carolina h ealth. C incinnati. Ohio.19~4. s ol.1. was not yet taking full advantage of the possible cost-1 National Institute for Occupational Safety and Health benefits to be derived from olTering the secondary A nationwide sursey of the occ u pa tional safety and lesel of occupational health senices. Also. the na-health workforce DHEW Pubhcation No '8-164. Cin-tional sogue for health promotion and physical cinnati. Ohio,1978. 4 Omce of Statistical Standards. Executne Othce of the titness was not redected in the sersices prosided to President Bureau of the Budget: Standard industrial workers in the State through their employment. classification manual U S Gosernment Printing OtTice. The type of study we conducted permitted obser-Washington, D C. 1967. sation of interindustry differences in both the lesel 5. South Carolina State Deselopment Board South Carm and the type of services, but the reasons for these lina 1982 industrial directory. Columbia. S C.1982. ditTerences must remain speculatise. It is difficult to 6. Saltzman B E : Adequacy of current industrial hygiene nd ccup tional safety professional m a npow e r Am determine how relevant the findings are to other Ind Hyg Assoc 143: 254-260 (1982). States. South Carolina is an OSHA Agreement State, 7 Fielding. J E.. and Breslow, l.. Health promotion pro-but w hether this has caused services to develop grams sponsored by California employees Am J Public differently than in non-OSH A Agreement States is Health 73 - 5 38-542 ( 198 3 L The Three Mile Island director of the daision and Dr Stein as medical epidemiol-ogist. Special Studies Branch. Mr. Lutz. who is now retired. Populat. ion Registry was regional director of the U.S Bureau of the densus. Philadelphia. Dr Gur is associate professor of radiation health and radiologs at the Graduate School of Pubhc i MARILYN K. GOLDHABER, MPH Health. Unisecsity of Pittsburgh { GEORGE K. TOKUHATA, DRPH, PhD Tea sheet requests to George K. Tokuhata. DRPH. PhD. Dnisi n f Epidemiology Research. Pennsylvania De pa rt-l EDWARD DIGON, MPH ment f Health, P O. Box 90 Room 1013. Harrisburg. GmYN G. CALDWELL, MD, MPH Pa 17108. GARY F. STEIN, MD, MOH GUY LUTZ DAVID GUR, ScD SYNOPSIS shortly after the Afarch 28,1979, accident at the Ms. Goldhaber. Dr Tokuhata, and Mr Digon are with the Dnision of Epidemiology Research. Pennsyls ama De. Three Afile /3/and (TAffi nuclear plant outside Har-partment of Public Health--Ms Goldhaber as chief, Nu-risburg, Pa.. the Pennsylvania Department of Health. cicar Health Research Program. Dr. Tokuhata as director in conjunction with the Centers for Disease Control of the dwision and program manager Three Mile Island and the U.S. sureau of the Census, conducted a Research Program; and Mr. Digon as chief. Special Studies censu of the 35,930 persons residing within 5 miles Section Dr. Caldwell and Dr. Stein are with the Chronic Diseases Disision, Center for Ensironmental Health. C en-of the plant. With the help of 150 enumerators, ters for Disease Contrc.l. Atlanta-Dr Caldwell as deputy demographic and health-related information was November-December 1983. Vol. 98, No. 6 603 /I iw 1.

collected on each person to provide baseline data for studied yearly by matchmg the TAfi Population future short-and long-term epidemiologic studies of Registry with postal recordt cancer registry recordt the efects oI the accident. Individual radiation doses and death certificate data. Because the radiation dose = sere esnmated on the basis of residential locati<m from TAff was extremely small. any im rease in and the amount of time each person spent in the morbidity or mortality attributable to the ac cident 5-mile area during the 10 days af ter the accident would be so small as not to be measurable by present methods; however. adserse health ellects as a result ? Health and behavioral resurveys of the population of psychological stress may occur. Also, a temporary g will be conducted approximately every 5 years Pop-mcrease in reporting of disease could occur bec ause q talation mobdity, morbidity. and mortalits will be of increased surveillance and attention to heahh -9 ARLY ON WEDNFSDAY MoF NING, MARcu 28, sulting from TMI radiation-including all cases of i 1979, a minor plumbing problem occurred in the cancer (fatal and nonfatal) and genetic ill health elB cooling system of the Three Mile Island ( TMI > for all future generations-was estimated as two (5 L ] nuclear plant outside Harrisburg. Pa. Normally this Though the crisis passed, public apprehension 7 condition is easily remedied, but on this day the remained. The public questioned the salidity of the required intersentions. both human and mechanical. estimates of radiation dose to local residents and of -j were flawed A series of "unlikely esents" led to the the health risk from that dose. This apprehension d shutdown of the unit 2 reactor, as thousands of gal-was due. in part, to the general public's perception i lons of radioactise water spilled within the plant of radiation as being mysterious and dangerous. building and the reactor core dangerously oserheat-capable of causing cancer and other dread conds ed (I-3). An estimated 2.5 to 10 million curies of tions. In addition, lay persons remained dependent y radioactisity escaped into the atmosphere during a on the " experts" to ensure safety in the management tense week of worldwide concern over the fate of the of nuclear power and to evaluate risks in the esent nuclear plant and its surrounding population (3,4). of exposure The " experts" had almost failed in the The TMI accident has been called the worst to occur first case and were being questioned in the second. J in commercial nuclear power generation. Health authorities in both the Commonwealth of --7 By the end of the first week, gosernment authori-Pennsylsania and the Federal Gosernment agreed 3_- ties announced to a skeptical public that, although that because confusion and uncertainty had sur-the situation could have been very serious. it had rounded the TMI nuclear accident from the begin-been brought under control. They stated that the ning, and because the accident had been the first of 9 radiation dose receised during that week by persons its kind, the exposed population should be followed J living near the plant had been very small The maxi-up and studied for many years to monitor any posse J mum possible dose to someone standing unprotected ble changes in physical and mental health (-'t 1 anywhere on the border of the plant site for the 10-day duration of the accident was estimated at less Materials and Methods than 100 millirems f 5), the approximate equivalent of I vear's natural background radiation or of 3 -5 Population data. Within 3 months after the acci-du y chest X-ray exposures delivered to the whole body dent, a census of 35,930 persons living within 5 The average likely dose to persons living within 5 miles of the TMI nuclear plant was taken in a co-miles of the plant was estimated to be 9 millirems, operatise effort by the Pennsylvania State and Fed-4 an amount similar to the radiation a jet plane pass-eral Governments. The primary agencies insoked -l enger would receise from two roundtrip flights across were the Pennsylvania Department of Health, the the country (6). With respect to long-term effects Centers for Disease Control, and the U S. Bureau of E on health from TMI radiation exposure. it was pro-the Census. Jected that among the more than 2,164,000 persons All three organizations contributed to the plan-A living within 50 miles of the plant, one excess death ning and operation of the TMI census. The Centers from cancer would occur over the lifetimes of these for Disease Control secured funds and provided g residents. ( An estimated 325,000 of these permns onsite personnel: an epidemiologist. 2 managerial g will die from cancers unrelated to TMI.) The total staff persons. and 20 team leaders (public bealth g number of excess morbid or mortal cond tions re-advisors). The Bureau of the Census provided ex-

==i so4 evene w..iin si.e.,i. m-

.~. .y ~,., n...... .. = r >= . 4 }; ; - 1.. :y. y .s ,Y, i,- I a. '. y O. l., N /M 5 ' hJ T I' y ~ - - e

%. g -.

.r-D N N0Nc 'N Y W [UM d th 'N O U t-.i , {,( s =[. :..~. ' $,9 ^ pert consultation from its regional and national off ce 6~ ' ~ staff, an onsite demographer, and onsite training of e.Ip 'Though the crisis passed, public j.; census enumerators. A retired regional director of 7 the Bureau was brought in to oversee all data apprehension remained. The public 5' f f f.f collection actisities The Pennsyhania Department questioned the validity of the estimates 1,. } of Health provided facilities as well as its own epi-of radiation dose to local residents and h.f ,.? M, demiologist and numerous support staff-including of the health risk from that dose.' Q.- 150 newly hired local enumerators--and assumed y- --. ((f f responsibility for all followup studies conducted on the census population r.,, p, f. The information that was collected during the usited up to four times m an attempt to find some-19 .f ;f ' ' census will proude baselme data for future epidemio-one at home to be intersiewed if this failed. inter- }.: iogic studies of the effects of the nuclear accident siews were conducted by telephone. Once the mter- .i - '..;

• f T hese data-know n as the "TMl Population Regis-siews had been completed, the interuew forms were

-[ 9.f. ; '. :- try"-include demographic information on each reuewed sescral times for different purposes If the ), resident of the 5-mile area such as name, address. forms were found to hase missing or meonsistent 2 T ' -) ~ - age. sex, race, and a brief medical history of cancer information. callbacks to the household were made [ 3 P diagnoses, thy roid disorders, radiation treatment or in addition. seseral quality control measures were j$4 i ' > ' '- therapy, and prior exposure to ionizing radiation on performed. including standard Bureau of the Census (' ~ [.. ) the job Pregnancies at the time of the accident were measures of "between-household cos erage control" Y

..' l ' l

• i noted, and smokine histories were included for all and "w ithin-household cos erage control" ( A ). before

.g t . -l,[. teenagers and adults. In addition. each pe rso n's data were computerized and sahdated to eliminate 6 l c.M-s daily travel in and out of the 5-mile area during the sporadic and systematic errors. L h 10 days after the nuclear accident was recorded so The response of the TMI residents to the census [.' i iD[.' that crisis esacuation pa' terns could be esaluated was exceptionally good because of adequate media 9 l' and TMI-related radiation doses estimated for each coserage and a strong sense of commitment from

.. [ "

J.. - person in the registry both enumerators and residents. The w ell-t ested o' ^ ,T.~ By early June 1979. 150 enumerators had been methods of the Bureau of the Census and the epi-f ' '. 9..c..m hired and trained and were ready to collect data on demiologic expertise asailable to the project further 'I? {E 7; the TMI population On June 20. the enumerators ensured a solid data base. ,]

f. :g:'t:

began canvassing the 5-mile communities. register-Complete descriptions and frequency distributions l? $f mg each household, and conducting door-to-door of all the sariables in the data base are contamed j, ? '. T I intersiews with one adult respondent from each in another paper (9). Most variables were defined h 7 .3 y household. The respondent provided mformation on and coded as in the 1970 and 1980 U S. Bureau of

s.,

l ~ himself or herself and on all other members of the the Census reports. Missing values for each sariable (( { m ;.; ; -, Registration and mtersiewing proceeded intensely portant variables prosided subject matter for special household. averaged about 1 percent. Several of the more im- ..~,'...s 4 ,L for about I month. After 6 weeks. 92-93 percent reports: analyses of the incidence of spontaneous J.. 4

. ^ -

..y of the population had been successfully interviewed. abortion ( /0 ), prevalence of cancer in the TMI (.., ], and most of the interview forms had been coded and area before the accident (report to be released). and 7 1.: f.. 37,, edited by the enumerators themselves. When all the ensis evacuation fil). A O(:..N z.' data were collected, the registry was estimated as 7 7,. i.j ' 93-95 percent complete. Less than 2 percent of the Household location and radiation dose data. Some T."- ?.' households canvassed had refused to respond; less important variables were added to the data base (; / " jl than 2 percent of responses had been unohtamable after the TMI census was taken: location of resi-7S ls for other reasons; less than 2 percent of households dence (distance from TMI and direction ) and radia- - fl . [ had been missed during enumeration; and less than tion doses t " maximum possiSic" and "likely" w hole-Ag .) C 1 percent had remained uncontacted because the body gamma millirem doses) for each person in the y; .q residents had moved between the time of the acci-TM1 Population Registry. This work was carried out ] .AY;6 dent and the time of the census. by the University of Pittsburgh's Department of Ra-O In collecting and processing the data, the inter-diation Health. Residential locations were derived g ..,4 viewers made every attempt to maximize coserage from work maps of households canvassed during the 2 '. ).,.,. ] i and minimize incomplete data. Households were TMI census and were plotted by hand onto a com-M g y fh ,j-. November. December 1983, Vol. 98, No. 6 605 y ,.ca L- 'n ~c ~? e 7 _ y ;_ ; ~;.. _; ;....,; p.7 (;; ye. y.g .c 2

7..; y _.q

.. - ; _. 3.,,.> y, ; 3., .~

. _. _.. M puterized map of the 5-mile area. Indisidual radia-once every 3 years, the Pennsylvania Department of tion dose assignments were made on the basis of Health will match the Th1I population registry estimated time-dependent dose-rate distributions in against the Naticnal Death Index for those persons the 5-mile area (12), combined with housenold loca-who moved and died outside Pennsylvania. tion and individual whereabouts during the 10-day Periodically, age-adjusted Thil death rates will be period after the accident. compared with age-adjusted death rates in selected areas of the State. Age must be taken into account Post office update. To maintain the followup regis-because TMI registrants will grow older with time, try, it was necessary to have a way of keeping in whereas comparison populations remain a cross touch with registrants and obtaining vital data about section of the society at a given time. them. With the help of the U.S. Postal Service, the Pennsylvania Department of Health devised a Link to cancer registry data. Because ionizing radi-system to obtain annually the current adoresses of ation exposure has clearly been linked with cancer persons in the TMI Population Registry without con-(14), an optimum followup program for the TMI tacting the registrants directly. At about the same Population Registry includes ascertainment of new time each year, names and addresses of all persons cancer diagnoses after the TMI accident. On the in the registry aged 16 years or older are sent to the basis of studies from Hiroshima and Nagasaki, where local post office for address verification and update the average radiation exposure for survivors of the j (children under age 16 are assumed to move with bomb blasts was thousands of times higher than at I their mothers). In compliance with postal regula-TMI, no measurable increase in cancer incidence in tions, the names are sent on computer-generated the TM1 area has been predicted (5). Nevertheless, j cards sorted first by zip code and postal carrier route because of lingering public skepticism about esti-and then arranged alphabetically by name. The post mates of the magnitude of the radiation dose, and offices are obliged to respond, within 10 working because of voiced public concern about increased days after the request for update is received, by risk of cancer, it is important that the incidence of supplying all relevant forwarding address informa-cancer among the TMI population be carefully docu-tion on file. Many post offices also correct spelling mented. Given a reliable cancer incidence registry, errors, indicate deceased addressees, and supply this should not be difficult. other helpful information. Before 1982, no statewide cancer registry existed Each year, new current addresses are added to in Pennsylvania; however, under recently enacted the computerized registry file and followup codes legislation, all hospitals throughout the State are are assigned accordingly. In addition, previous moves required to report newly diagnosed cases of cancer are recorded by storing the zip code from which the to a central registry. Although the initial legislation move took place and the year in which the move was was drafted before the TMI accident, the 14 south-ascertained. Up to 10 previous moves can be stored central counties surrounding TMI were designated on the computer tape, allowing a systematic study in 1982 as the start-up area for the Pennsylvania of moving patterns of persons in the registry. A study Cancer Registry. By 1984 or 1985, the cancer re-of moving patterns during the first year after the porting system should become operative throughout accident has already been conducted (13). the State. Once the Pennsylvania Cancer Registry is fully Plans for Analysis operational, the Pennsylvania Department of Health plans to match the cancer incidence file annually Link to death certificate listings. Each year, the against the TMI Population Registry. As with all Pennsylvania Department of Health matches the " protected" State data, personal identifiers in both TMI Population Registry against death certificate registries will be kept completely confidential. Data listings in Pennsylvania. When an appropriate match will be released only in aggregate form. occurs on name, birthdate, and social security num-ber, the cause-of-death information (both underlying Special health surseys. Approximately every fifth and contributory causes) is extracted from the death year, the Pennsylvania Department of Health will files and added to the TMI Popu!ation Registry. This conduct special health surveys of the TMI Popula-M will allow analysis of deaths due to cancer, thyroid tion Registry. Either the total population of the 5-disease, or other causes that may be associated with mile area or a representative sample will be em-low-level ionizing radiation or stress. Approximately ployed, depending upon the nature of the survey. l e

Questions about reproductive experiences, major morbidity, and psychological and emotional prob-lems experienced since the accident will be asked. ' Health outhorities.. agreed that The surveys will also provide a means of ascertaining because confusion and uncertainty had opinions and attitudes of the public toward TMI and surroundedthe TMInuclear accident toward nuclear power in general. (For followup of from the beginning,and because the infants exposed in utero to the accident, a special accident had been the first of its kind, TMI Infant Registry has been devised from another the exposedpopulation should be data source, the TMI Pregnancy Outcome Study Ollowed MP andstudiedfor many (15), as a sampling frame for studying potential long-term health efiects.) years to monitor any possible changes \\ in physicaland mentalhealth.' Discussion q We believe that followup of the TMI accident is public health-that is, the actual excess number of of interest to both the scientific and the lay com-mortal or morbid conditions in a population. How-munities and has broad implications for the future ever, among persons who are skeptical of any calcu-of commercial nuclear power in this country and lation involving radiation, this controversy over the elsewhere. effects of low doses can evoke much apprehension. Since the 1950s, when nuclear power was first in-One major purpose of the registry is to document, troduced, decisions about the building and operation for a large number of human subjects, the effects on of nuclear powerplants had been made by experts, health of the low-level radiation exposure occasioned with little input or resistance from the general popu-by the TMI accident. On the basis of previous stu-lation. When, after 30 years of relative public con-dies of biological effects of ionizing radiation and fidence, the first major commercial nuclear accident the various dosimetric studies conducted after TM1, occurred, difficult but important questions were it is unlikely that any extra cases of cancer or genetic raised about the safety of nuclear energy. We are disorder will be detectable among the 35,930 persons ~ hopeful that the upsurge of public debate will help who lived within 5 miles of TMI at the time of the close the knowledge gap between the experts and accident and shortly after. Should increased mor-the public, as benefits of, risks from, and fears about bidity or mortality among the TMI population be nuclear energy become better understood and placed found that cannot be explained by sociodemographic in proper perspective. or other known intervening factors, exposure to Scientists already have a substantial amount of psychological stress and its sequelae could be impli-j knowledge about the biological effects of ionizing cated. radiation (14,16). From Federal funds alone, nearly Severalinvestigators have studied stress among the $5 billion has gone into research involving human TMI population (23-25). They found that c6nsid-and animal studies (17,18). Because of this massive crable levels of stress were experienced during the effort, more is known about the effects of radia-2-week crisis period and that moderate levels of tion-in terms of cancer, birth defects, chromosome stress were maintained for many months afterward. breakage, and other outcomes-than about the ef-Because the cleanup of the plant will continue for fects of any other environmental hazards (such as several years, it is likely that stress will remain with chemical carcinogens) on the human organism. the TMI population for years to come. We believe ~ There is general agreement among health physi-that it is important to continue to study psychological cists and radiation biologists about the effects of stress as an unavoidable outcome of a nuclear acci-high doses of ionizing radiation, but the consensus dent. is not as strong about the effects of low doses-Earlier studies showed that persons who evacu-especially doses below 10 rems (14,17,19-22). ated the 5-mile area during the accident had higher The question most often raised is whether the same stress indices, when evaluated several months later, dose-response relation seen with high doses also than persons who did not evacuate (23). Thus, both occurs with very low doses, whose health effects are TMI-related stress and TMI-related radiation dose difficult to measure. Although this is an important will be important to consider when the results of data scientific question, the various possible answers to it analysis are interpreted with respect to health effects. may yield relatively minor differences in terms of Other intragroup (internal) comparisons of various m e.r.o ,im.vi.n,we.e m

t a and Garrick (consultants to Metropolitan Edison Corp.), 1979 (TOR-TMI-I l6, resision O, August 31). 'One major purpose of the registryis

5. Ad Hoc Population Dose Assessment Group: Popula-to document,for alarge number of tion dose and health impact of the accident at the Three Mile Island nuclear station. U.S. Government human subj. ts, the egects on health Printing Office, Washington, D.C.,1979, p. 3.

ec of thelow-levelradiation exposure

6. wallace, R., and Sondhaus, C.: Cosmic radiation ex-occasioned by the TMI accident.'

posure in subsonic air transport, Aviat Space Environ Med 49: 610-623 (1978).

7. Committee on Federal Research into the Biological Effects of Ionizing Radiation: Follow-up studies on t

biological and health effects resulting from the Three MUC Island nuclear power plant accident of March 28, health parameters, as well as comparisons with con. 1979. NIH Publication No. 79-2065. U.S. Department trol groups selected from communities in Pennsyl-of Health, Education, and Welfare, Washington, D.C., vania with sociodemographic characteristics similar 1979. to the TMI population, will also be made. As indi-

8. U.S. Bureau of the Census: 1980 census of popul.ition cated earlier, age and other factors will be taken into and housing, Pennsylvania. U.S. Gosernment Printing Office, Washington, D.C.,1981.

account, particularly when the TMI population is

9. The Penns>ivania Department of Health: The Three compared with cross-sectional control populations.

Mile Island Population Registry, report one: a general It is possible that spurious increases in morbidity description. Division of Epidemiology Research, Harris-may occur because of heightened public awareness burg,1981. and increased surveillance activities. The TM1 popu-

10. Goldhaber, M. K., Staub, S. L., and Tokuhata, G. T.:

lation may perceive itself as being at greater risk of Spontaneous abortion after the Three Mile Island ac-cident: a life tabte analysis. Am J Public Health 73: diseases and, therefore, may report symptoms of 752-759, July 1983. s r disease to phys..icians and researchers earlier and

11. Goldhaber, M.

K., and Lehman, J. E.: Crisis evacu-more often than usual. If this occurs, a temporary ation during the Three Mile Island nuclear accident. increase in cancer incidence, for instance, may be The TMI Population Registry, Pennsylvania Depart-observed. However, earlier reporting will not neces-ment of Health, Harrisburg,1982.

12. Rao, G.

R., et al.: The TMI popu!ation: a closer sarily lead to increased mortality and, after a few I

k. In Proceedings of the Pennsylvania Academy of Iears, such fluctuations in morbidity and mortality Science, edited by Snyamal K. Majumdar. Vol. 56.

statistics should stabilize. This is another reason why Department of Biology, Lafayette College, Easton, it is important to conduct studies for a period of Pa.,1982, p. 44. many years to recognize any artifact that may occur

13. Goldhaber, M. K., Houts, P. S., and DiSabella, R.:

Moving after the crisis; a prospective study of Three in reporting. + MUe Isl nd aM population mobility. Environment As objective information accumulates from the and Behavior 15(1): 93-120 (1983). TMI Population Registry, the lingering uncertainty

14. Committee on the Biological Effects of Ionizing Radia-from the TMI accident should diminish and, perhaps, tion: The effects on populations of exposure to low future decisions about nuclear power as a domestic levels of ionizing radiation,1980. National Academy source of ene*gy can be more readily and confidently Press, Washington, D.C.,1980.

15. Tokuhata, G. K.: Pregnancy outcome around Three made' Mile Island. Division of Epidemiology Research, Penn-sylvania Department of Health. Harrisburg,1981.

References

16. Sources and effects of atomic radiation. United Na-tions, New York,1977.

1. Kemeny, J. G., chairman: Report of the President's

17. Problems in assessing the cancer risks of low lesel Commission on the Accident at Three Mile Island.

ionizing radiation exposure, sol.1. Report No. EMD-U.S. Government Printing Office, Washington, D.C., 81-I. U.S. Government Accounting Of5ce, Gaithers-1979. burg, Md.,1981.

2. Ford, D.: A reporter at large, Three Mile Island I-

18. Interagency Radiation Research Committee: Federal Class nine accident. The New Yorker 57: 49-120, strategy for research into the biological effects of Apr. 6,1981.

ionizing radiation. NIH Publication No. 81-2402. U.S. J. Nuclear Regulatory Commission Special Inquiry Government Printing Office, Washington. D.C.,1981. Group: Three Mile Island, a report to the Commis-

19. Dreyer, N. A., et al., editors: The feasibility of epi-sioners and to the public. Vol.1. U.S. Nuclear Regu-demiologic investigations of the health effects of low-latory Commission, Washington, D.C.,1980.

level ionizing radiation, final report. NUR EG < CR-

4. Woodward, K.: Astessment of offsite radiation doses 1728. U.S. Nuclear Regulatory Commission, Washing-from Three Mile island unit 2 accident. Pickard, Lowe, ton, D.C.,1980.

eee pene Hennn n perte

,t

20. Bross. I. D. J., Ball, M., and Falen, S.: A dosage re-and 3. Pennsylvania Department of Health, Harris-sponse curve for the I rad range: adult risk frons burg.1981.

diagrostic radiation. Am J Public Health 69: 130-136

24. Hu, T. W., and Slaysman, K. S.: Health-related eco-(1979).

nomic costs of tbg Three Mile Island accident. Center

21. Land, C. E.: The harards of fa!!out or of epidemio-for Research on Human Resources, Pennsylvania State logic research? N Engl J Med 300: 431-432 (1979).

University, State College,1981.

22. Boice, J. D., and Land, C. E.: Adult leukemia follow-

25. Bromet, E.: Preliminary report on the mental health of ing diagnostic X-rays? Am J Public Health 69: 137-Three Mile Island residents. Department of Psychiatry, 145 (1979).

University of Pittsburgh School of Medicine, Pitts-

23. Houts, P. S., et al.: Health-related behavioral impact burgh,1980.

, of the Three Mile Island nuclear accident, pts. I, 2, i New Partnership for Health? tistical data. It is in the implementation of plans that Business GrouUs on Health d C""#d ^"' "'"#' " P' 8""' "I'"d"8 '" anadequate community power base. and Health Systems Agencies The CMBGH,1 of more than 90 groups that have ROBERT C. BRADBURY,'PhD developed recently across the country to attack high health care costs, was formed in 1981 by business The paper is based on Dr. Bradbury's presentation at leaders to address these rising costs. The principal the lloth annual meeting of the American Public Hea'th strategy adopted by the CMBGH involves fostering Association on November 15, 1982, in Montreal. He is di-a competitive health care market by creating a criti-rector of the gonnt Clark Umvers ty Umv*rstty of Massa-cal number of competing health plans. The providers chusetts Medical School Health Admientration Program. l Tearsbeet requests to R. C. Bradbury, PbD, Graduate en each plan will then have incentives to provide School of Management, Clark University. 950 Main St., e8cctive care in an egicient manner to keep the Worcester, Mass. 01610. premium competitive and attract enrollees. 4 SYNOPSIS. Cooperation between the CMBGH and CMHSA is based on each organization's emphasizing its The experience of ^ the Central Massachusetts strengths. The CMHSA's data base and analyses Health Systems Agency (CMHSA) and the Central have been the primary resources used by the Massachusetts Business Group on Health (CMBGH) CMBGH to identify problems. Each organization y demonstrates the feasibility of cooperation between has developed its own set of goals and objectives, HSAs and BGHs. Objectives and strategies of the while keeping in mind those of the other organiza-two groups in carrying out community health plan-tion. The CMBGH adopted a subset of the ning and working for health systems change are CMHSA's goals-those that focus on hospital ca-compared. parity and utilization. Although the CMHSA's regu-latory strategies difer greatly from the CMBGH's Nearly two decades of government-sponsored competition strategies, they do not necessarily con-community health planning programs, first through flict. Actually, each organization is supporting the comprehensive health planning agencies and then other's strategies without deemphasizing its own. through HSAs, have had less impact than many had The CMBGH currently has a decisive advantage anticipated because neither the technical nor politi-over the CMHSA in implementing activities because ca! basis for such planning was suficiently estab-the business leaders are an integral part of the com-lished. The CMHSA experience is typical, dthough munity power structure. Also, their companies' will-it is credited with developing a hospital systems plan ingness to ofer additional health plans to their em-that is based on sound planning methods and sta-playees is the prime incentive to develop such plans. IN 1966 THE COMPREHENSIVE HEALTH PLANNING Health Act, called for consumers and providers of and Public Health Service Artendments Act (Pub-health care to join forces to plan improved health lic Law 89-749), referred to as the Partnership for systems. In the more than 15 years of federally Neomber. December 1983. Vol 98, No. 8 000

M% fo(. r PENNSYLVANIA DEPARTMENT oF HEALTH HEALTH REPORTER qu# P.O. BOX 90 HARRISBURG, PA 17108 m Vol. 4, No.12 pg n :e December 1983 THE THREE MILE ISLAND MOTHER-CHILD REGISTRY Byt Jane R. Brats B.S.S. Since many known and suspected factors affect preg-ProjectDirector, TMIMother-ChildRegistry nancy outcome, it was necessary to account for the Division ofEpidemiology Research influences of those factors in order to delineate the efrect of the TMI accident. This resulted in the ascertainment of data Within a few days after the Three hiile Island nuclear on a large number of independent or intervening variables accident (March 28,1979), the Pennsylvania Department of pertaining to the pregnant women themselves, as well as to Health conceptualized and developed a comprehensive plan the complex environment to which they had been exposed. for a variety of epidemiologic and social health studies The maternal characteristics considered as independent designed to assess the impact of the accident. Specific stud-factors included: (a) sociodemographiefactors such as race, les conceived during this critical period reflected the existing age, education, occupation, employment, marital status epidemiologic knowledge regarding biologic effects of low and residence; (b) behavioral attributes such as smoking, levelionizing radiation and of severe emotional stress. One drinking, pregnancy planning and prior practice of birth of the most important studies to develop was the Thil Preg-control; (c) medical-obstetric hirtories such as diabetes, nancy Outcome Study which considered both the effects of heart disease, hypertension, obesity, thyroid disorders, pre-radiation and maternal stress resulting from the accident on vious experiences of abortion, stillbirths, prematurity, the biologically sensitive fetus. immaturity and of birth defects; and (d) gravidity /index In this pregnancy outcome study, the health status of two birth order! cohorts of mothers and their infants / fetuses were evalu-The provider characteristics considered as independent ated. First, the study cohort included approximately 4,000 factors included: (a) medicalspecialty of the attending phy-pregnant women who delivered between hfarch 28, 1979 sician (obstetrician, family-general practitioner, and March 27,1980. These women resided in ten-mile Thil osteopath, etc.); (b) type of practice (solo versus group); communities and were pregnant at the time of the accident and (c) prenatal care in terms of trimester of initiation of or they became pregnant within a few months following the medical care, frequency of visits, special procedures or tests accident. done, instructions given, medications administered and x-In comparison, the control cohort consisted of approxi-ray pelvimetry during the index pregnancy, particularly mately 4,000 pregnant women who delivered exactly one during the last portion of gestation. year after the study cohort, i.e., March 28,1980 to March Maternalstress during the index pregnancy, one of the 27,1981. These women also resided in the same geographic two primary independent factors, was measured in terms of area (ten-mile TMI communities) at delivery as the study overt personal statements of " anxiety-fear" as experienced cohort of mothers; they, however were not pregnant at the by individual women during the crisis, as well as by actual time of the TMI accident. stress-coping patterns, such as taking tranquilizers and The objectives and methodology of the pregnancy out-sleeping pills during and after the crisis while pregnant. come study were discussed with in-state and out-of-state hospitals with obstetric services. Cooperative arrangements The other primary independent factor was maternal were made with these hospitals so that names and addresses radiation exposure during the 12-day crisis period following of obstetric patients who resided in the study area were the nuclear accident, as estimated by the Department of made available to the project staff. The completeness of Radiation Health of the University of Pittsburgh'. To deter-reporting was verified by means of regular cross-checks mine maternal whole body gamma radiation exposure, as with vital records compiled by the State Health Data Center well as maternal and fetal thyroid doses, several reliable and newspaper birth announcements. source data, including TLD recordings, residential distance l Shortly after delivery, each study participant was con-and direction from the damaged reactor and evacuation tacted at her home for an interview. Research associates, details compiled by various agencies including the State employed full-time on the project, conducted the mterviews Health Department's special census, were reviewed and con-by utilizing a standardized questionnaire developed for this solidated into an individualized estimate of" maximum pos-sible" and "most likely" doses. purpose. The dependent factors under study included the follow-At the completion of the interview, each mother was ing adverse pregnancy outcomes:(a)fetaldcath (born dead requested to sign a consent form which permitted the after 16 weeks gestation), (b) hebdomadal death (death abstraction of medical data from both her obstetric records within seven days after delivery), (c) neonataldeath (death and from the newborn's pediatric records at the hospital within 28 days after delivery), (d) perinatal death (a com-where the delivery occurred. In addition to the information obtained from home inter-bined measure of fetal and neonatal deaths), (e) immaturity (birthweight of 2,500 grams or less), (f) prematurity (gesta-views and from hospital medical records; vital records, i.e. tion less than 37 weeks), (g) congenital anomalies (one or birth and death (fetal and infant) certificates, were reviewed and select data from these sources were abstracted for use in more defects, major or minor, recognized during the neona-the study. tal period), and (h) low Apgar score (less than seven at one minute of delivery). (conrirmadon age 4) d l

7.- TMI MOTHER-CHILD REGISTRY ticipated in dita collection on two major research projects: (Continuedfron Page l) the first project evaluated the relationship of consumer products to accidential injuries; the other evaluated the rela-In order to evaluate the short term health effects of the tionship of prenatal care to pregnancy outcome. Th11 accident on pregnancy outcome, the baseline data are When the Th11 Pregnancy Outcome Study was initiated, currently being analyzed. In order to evaluate the long-term Jane was promoted to Science Research Associate 11 as effects (physical, psychological and behavioral), if any, the Operational hianager for the project. Within one year she baseline data have been consolidated into a Thil A1other-was promoted to Supervisor of the same project. As the Child Registry. This registry, containing approximately short-term health effects of the Th11 accident on pregnancy 3,800 mother-child pairs in each of the study and control outcome are currently being analyzed, she is planning long-cohorts, represents 94 per cent of the cases in the pregnancy term follow-up health studies for the mothers and children outcome study. included in this study. In order to locate the original partici-In order to update and maintain such a registry over the pants in the Thil Pregnancy Outcome Study, she has devel-next 10 years with current names, addresses, telephone oped, maintained and updated the Thil hiother-Child numbers, and vital status of each mother and child regis-Registry. trant, the following basic operations are necessary: hirs. Bratz earned her Bachelor of Social Sciences degree (1) yearly post office contact for possible address from the Pennsylvania State University at Capitol Campus change; in hiiddletown. She has completed post-graduate courses at (2) yearly link to the Pennsylvania death certificate, both Capitol Campus and University Park. cancer registry and marriage files; Jane is a member of the American Public Health Associa-(3) lost-to-followup retrieval system; and, tion, the Pennsylvania Public Health Association and the (4) every third year link to the National Death Index American Statistical Association, Harrisburg Chapter and for possible out-of-state deaths and search for out-has co-authored several papers relating to pregnancy out-of-state cancer diagnoses. come. By updating and maintaining the TN1151other-Child Reg-For most of her life Jane resided in the Harrisburg area. istry, a reliable data base is established for long-term follow-She and her husband are currently living in Susquehanna up health studies that are currently being planned. The Township. first morbidity survey is scheduled when the cohort children reach five years of age and again when they reach 10 years of age. Data collected in these follow-up studies will include PROGRAM hf 0NITORING the following variables: (a) demographic characteristics,(b) general health indicators, (c) maternal reproductive prob-lems, (d) selected specific diseases, (c) growth and develop. (continuedfrom Page2) ment, (f) behavior, (g) family social environment, and (h) stress symptom checklists. The results of other studies con-ments via regular attendance at SCA board meetings. Con-ducted under the aegis of the Th!I Health Effects Research sultation and information is provided to local boards Program will be incorporated into these follow-up studies. c neerning policies and procedures and issues as they relate The aim of the Th11 Health Effects Research Program to the drug and alcohol field. Periodically, regional meet-administcred by Ihe State Health Department is to provide ings are held by Field Offices with SCAs to convey informa-factualinformation based on such studies which are epide-tion and to address issues of concern at the locallevel. miologically sound and sociologically justified with respect Because of the Office of Drug and Alcohol Programs' to possible health effects of the Thil accident upon local res-monitoring system, individual local issues can be reJolved idents. Based on the available Thll radiation exposure data, before they become large problems for state and local gov-major adverse health effects are r.ot expected. However, the ernments. The system has also become a.nodel for other accuracy of radiation exposure has been questioned by hurrm mice agencies in developing tools for management some scientists. Because of the high levels of psychological a% i; ountabihty of public funds. stress experienced by the local residents during the crisis DIVISION STAFF period and the likelihood of distress continuing over the many years needed io cleanup the damaged reactor, psycho-logical health and its sequelae are considered as important The Western Field Monitoring Office is staffed by Rodger outcomes to monitor independently of the amount of radia. Beatty, Chief, Jon Dale, Al Mercik, and S. Kaye Gardner, tion exposure. These public health concerns should be field representatives and Dwis Ray, secretary. The Eastern addressed particularly with respect to pregnant women and Field Monitoring Office is staffed by Phil McIntyre, Chief, their in-utero exposed children who are much more sensitive Pat Raccine, Osborne Shamberger and Marie LeGrande, to such environmentalinsults thaa are other segments of the field representatives and Jyl Gillespie, secretary. Technical local population. Surwy3rt Unit's staff are Linda McKamey and Greg Curtis. INTRODUCING ABOUT TIIE AUTIIOR Jane R. Bratz, B.S.S. Sherry Migliore has been DDector of the Division of Program Monitoring Projector Director, TMI Mother-Child Registry since 1981 and has wori.ed in the Bureau of Community Assistance since Division of Epidemiology Research 1976. She began employment with the Commonwealth in 1970 as an ergpjolmrat counslot 3iththe_Depamnentaflabor and Industry and ' $ *o'f du b*n niv rsiIy [ Jane Bratz joined the Department of Health in August th a mt p o y an is o n 1969 as a research associate in the former Division of towards her master's degree in Public Health Administration at Penn State Research and Biostati tics. As a research associate, she par-University. 4}}