ML20154P607
| ML20154P607 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 05/20/1988 |
| From: | Stello V NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO) |
| To: | Goodling B HOUSE OF REP. |
| Shared Package | |
| ML20154P612 | List: |
| References | |
| NUDOCS 8806030394 | |
| Download: ML20154P607 (33) | |
Text
~~
. i
+
May.20, 1988 p,g The Honorable Bill Goodling
-United States House of Representatives
- Washington, DC 20515
Dear. Congressman Goodling:
We are pleased to provide you with information that responds to.
Ms.--Jacqueline Rhen's request of March 1,1988. ;Ms. Rehn's inquiry involves potential health problems that may affect.the general population in the South Central Pennsylvania area as the result of the 1979-accident at Unit 2-of the Three Mile Island-(TMI) nuclear power plant. -Her concerns appear to be twofold; immediate health effects that may have occurred on the day of the-accident and the potential for long-term future health effects.
As indicated in the enclosed material, at least a dozen studies have.been initiated by various State, Federal and. scientific agencies that address health effects resulting from the TMI accident. Some of these studies have been i
completed and some are ongoing. The findings from these ' studies have been.
a generally accepted by the scientific conrnunity. The Pennsylvania Department of Health has been particularly active'in sponsoring many of these studies.
i Dr. George K. Tokuhata of that agency, who is also a Professor of Epidemiology and Biostatistics at the University of Pittsburgh, has written a number of papers on this subject. Two of these papers are enclosed for your information.
The two potential health effects addressed in these studies are radiation exposure and psychological stress.
Particular attention has been given to health effects on pregnant women and infants in these studies.
The studies completed to date have found no short-tenn evidence for-significant physical health effects from the 1979 accident. Several of the studies indicated that stress symptoms caused by the accident did affect the health-related behavior of area residents (e.g.,-increased use of alcohol, tobacco, and tranquilizers to cope with psychological stress). The highest i
"likely" radiation dose estimated for any individual offsite during and.after l
the accident was 80 millirem, less than the annual radiation dose received by 4
an average person in the United States from' natural causes. Therefore, there were no.ipinediate radiation effects such as skin turning red, and none would be expected from these low radiation, doses.
It is generally thought by the scientific consnunity that no "excess" deaths or illness 'will be detected as a result of the TMI accident. The NRC was a co-sponsor for some of the studies that have been completed. Some studies sponsored by the Pennsylvania Department of Health will continue for at least another 10 years. On the basis of the findings from the already completed and ongoing TMI-related health studies, the Nuclear Regulatory Commission' finds no need for any additional future studies of this subject.
)
i I
O OC PD P
The Honorable Bill Goodling
-2 l
We hope you will find this information, including the enclosures, responsive to your April 25, 1988, request. Should you have any additional questions, please contact us.
Sincerely, blainaleg.3g gE g,
gDctacqteng Victor Stello, Jr.
I Executive Director for Operations
Enclosures:
1.
History and Findings of TMI Health-Related Studies 2.
Description of TMI Health-Related Studies 3.
Dr. George K. Tokuhata and Edward Digon, "Cancer Mortality and Morbidity (Incidence) Around TMI," September 1985 4.
Dr. George K. Tokuhata, "Three Mile Island Nuclear Accident and Its Effect on the Surrounding Population," 1985
- See Previous Concurrence OFC :DRP:PDI-4
- DRP:PDI-4
- DRP:PDI-4
- TECH ED
- DRP:ADRI
- DRP:DIR
- ADPR NAME :SMNorris
- RWHernan:ah :JFStolz
- BABoger
- SAVarga
- FJMiraglia DATE :05/13/88*
- 05/fd/88* :05/13/88*
- 05/13/88*
- 05/ /88
- 05/ /88
- 05/ /88 i-V OFC :NRR:DIR
]l :E O
- 0CA l
NAME :TEMurley
- VS iel o DATE :05/16/88
- 05/%88
- 05/M /88 OFFICIAL RECORD COPY
)
(
~ DISTRIBUTION Docket File -
NRC PDR w/cy of incoming Local PDR w/cy of incoming EDO #0003676 ED0 RF W/0 ENCLOSURE:
TEMurleyJHSniezek' FMiraglia-
_SAVarga(2)
BABoger JFStolz
.SMNorris CA OGC-WF SECY (88-0376)
EBeckjord TMartin FGillespie W/ ENCLOSURE:
NRR Mailroom (E00-0003676) w/cy of incoming WTRussell,- RI RWHernan 1
s
/
'o UNITED STATES
~,,
NUCLEAR REGULATORY COMMISSION
[
g g
E WASHINGTON, D. C. 20555
%...../
HISTORY AND FINDINGS OF TMI HEALTH-RELATED STUDIES I
Shortly after the March 28, 1979 accident at the Three Mile Island (TMI) nuclear plant, the Gover'nor of Pennsylvania designated the Director of the Bureau of Health Research to coordinate and manage all health-related research activities relative to TMI. At the same time, a special Advisory
, Panel was comissioned by the Secretary of Health to oversee and guide all THI-related health studies administered by the Bureau of Health Research.
Within a few months after the accident, the Pennsylvania Department of Health initiated nearly a dozen epidemiological and other health studies to evaluate the possible health effects of the TMI accident.
A description of each of the completed or currently ongoing studies initiated by the Pennsylvania Department of Health is given in Enclosure 2.
0ne of the first projects initiated shortly after the accident was a special census of all persons living within 5 miles of TMI. This "TMI Population Registry" will provide a basic framework for future short-and long-term epidemiologic studies of the effects of the accident. One of the most important studies developed shortly after the accident was to detennine if the THI accident had any measurable impacts upon pregnancy outcome and infant health.
Findings from the completed pregnancy outcome portion of this study show that the impact of the TMI accident upon pregnancy outcome was negligible, if any.
m
. A study on the occurrence of hypothyroidism in infants near TMI documented one case of congenital hypothyroidism among infants living within a 10 mile radius of TMI during the year period following the accident. This incidence rate is well within the normal range of expectation for the infant population studied.
The Radiation Dose Assessment Study estimated that the maximum possible whole-body gama radiation dose to anyone offsite was no more than 175 mrem, while the average maximum"possible gama dose to those within 5 miles of the plant was approximately 25 mrem.
When shielding and evacuation corrections were applied, the highest "likely" dose assigned to an individual offsite was 80 mrem and the average "likely" gamma dose to those within 5 miles of the plant was 9 mrem.
Among the findings of the Health Behavioral (Stress) Study were that persons who are younger, more educated, married, and female were most distressed during the accident.
Also, the increased level of anxiety experienced following the accident persisted for approximately one year, but declined substantially by October 1980 (when the final survey of this study was performed).
The Mental Health Study found that, among the three "high-risk" groups near TMI, only TMI mothers had higher incidences of depression than the control g rou p..
The Infant Mortality Study found no evidence that the TMI accident had any significant impact upon infant mortality.
The Health Economics Study indicated that stress symptoms caused by the accident did affect the g
4 '
health-related behavior of area residents. Of the costs examined, the economic costs of work days lost and physician visits were the largest cost items.
Several of the studies mentioned above have been completed. Apart from the substantial psychological effects described above and adverse effects upon low birthy:eight of excess medications taken by pregnant women to cope with
_ their anxiety and stress, the study findings have not found any short-term
_c evidence for significant physical health effects from the 1979 accident.
.Nor are such significant physical health effects expected from the officially-reported low-levels of radiation released from the damaged TMI facility.
However, there is a possibility that the psychological stress from the accident will cause some adverse effects upon the local population. In order to measure the long-term effects of psychological stress and the need to study potential effects of very low dose radiation on humans, several of the ongoing studies are designed to run for a twenty year period or more.
In addition, the Pennsylvania Department of Health has recently initiated two new studies.,
one to study the cancer mortality and incidence rate of the TMI poptlation, and one to implement a baseline health information system on the populations living within 20 mile radii of the 6 nuclear plants in Pennsylvania.
On the basis of the findings from the already completed and currently ongoing
I
+
- j l TNI-related health studies, the Staff finds no need for any additional future studies on TMI-related health effects other than those studies which are already planned.
9 o
l 1
i l
i I
ENCLOSURE 2 T
a jo,,
UNITED STATES
[
g NUCLEAR REGULATORY COMMISSION g
- E WASHINGTON, D. C. 20555 Description of TMI Health-Related Studies TMICensus(Ref
- 42,64,70)
Time Span:
Initiated June 1979 Duration:
20 years or more Sponsors:
Pennsylvania Department of Health, Centers for Disease Control, U.S. Bureau of Census Primary
Purpose:
To develop a population profile (TMI Population Registry) which would provide a basic framework for future studies of morbidity and mortality.
==
Description:==
Shortly after the accident, the 3 sponsor agencies took a census of the 35,930 persons who were living within 5 miles of TMI at the time of the accident.
Data collected included infonnation on age, race, name, address, i
SSf, marital status, smoking habits, medical history, recent pregnancy experience, medical and occupational radiation exposure, and detailed whereabouts during the first 10 days following the accident.
,e 2-
)
Population mobility, morbidity, and mortality will be studied yearly by matching the TMI Population Registry with postal records, cancer registry records, and death certificate data.
Health and behavioral resurveys of the population will be conducted approximately every 5 years.
Duration:
The Pennsylvania Department of Health will monitor this population for 20 years or more to detect possible health effects of the TMI accident.
i I
.--,-------e.m,,
e ~
-,-------,--+--,w----.
e--
.e---
ew, - ~, -
.~--v-
.e 3-Pregnancy Outcome Study (Ref: 42,44,55,62,65,67,70,74)
Time Span:
Initiated August, 1979. Will be studied every 5 years Sponsors:
Pennsylvania Department of Health
~~
Primary Purpose-To determine if the TMI accident had any measurable impacts on pregnancy l
outcome near TMI.
==
Description:==
)
This study evaluated the health status of two cohorts of mothers and their infants / fetuses living within a 10-mile radius of the TMI plant during the time of the accident. The first cohort included approximately 4000 i
pregnant women who delivered between 3/28/79 and 3/27/80.
The control cohort consisted of approximately 4000 women in the same area who delivered exactly one year after the study cohort, i.e., 3/28/80-3/27/81.
,-,-yw-,- - -,
.y.-,
--m
, Short Term The effects of radiation exposure and psychological stress on the following pregnancy outcome measures were studied:
fetus death, hebdomadal death, neonatal death, pre-maturity, immaturity, congenital abnormalities, and low Apgar score.
Long-Te rm Approximately 3800 mother-child pairs in each of the groups have been consolidated into a TMI Mother-Child Registry. The physical, psychological, and behavioral effects of each of these groups will be studied every 5 years.
Findings:
Short-Te rm When pregnancy outcome measures were compared between the exposed study cohort and the unexposed control cohort, no significant differences were noted for any of the various outcome measures under study, indicating that _
the impact of the TMI nuclear accident upon pregnancy outcome was negligible, if any.
After adjusting for the influences of the many maternal and provider characteristics described earlier, the incidences of fetal and neonatal mortalities, congenital anomalies, prematurity,
5-immaturity, and of low Apgar score within the study cohort were not significantly different from those within the control cohort.
Long-Term The first 5 year study of the test cohort of 3800 mother-child pairs is currently underway (this study cost $300,000). The 5 year study of the control group will take place next year (5 years after the 3/80-3/81 period).
Both studies are 1 year late due to limited funding.
t-Resurveys of each cohort will be conducted every 5 years, I
e 3
r,- - - - - - - -
p-
--y---,
w--++ - - - - --
y
.m.-
-~,e w--,se,,-
w-e ew --
w----e,--e--
-m-.-
o--
-w--
6-Congenital / Neonatal Hypothyroidism Study (Ref: 42, 51, 56, 67, 70, 74)
Time Span:
Annually Sponsors:
Pennsylvania Department of Health f
Primary
Purpose:
.To examine the incidence of congenital / neonatal hypothyroidism among n'ewborn infants living within a 10-mile radius of TMI.
1
==
Description:==
j Radioactive iodine can cause hypothyroidism.
This study involves screening all infants living within a 10-mile radius of TMI for hypothyroidism (by Pennsylvania law, all newborn babies in the state must be screened for hypo-thyroidismafter7/78).
This survey will be done annually.
i Findings:
For the group born in the test area between March 28, 1979 and March 27, 1980, there was only one case of congenital hypothyroidism identified
among approxtmately 4000 newborn infants. This incidence rate is well within a normal range of expectation.
There was an apparent concentration of 7 cases of congenital hypothyroidism in Lancaster County (mostly beyond the 10-mile radius) in 1979.
This study concluded that these reported cases of congenital hypothyroidism were not related to the TMI nuclear accident, i.e., these types of anomalies are
, not expected to have resulted from direct or indirect exposure of the fetus to radioiodine. This conclusion was also supported by an independent
~
Hypothyroidism Investigative Committee organized by the State Health Department, which included expertise in the fields of epidemiology, pediatric endocrinology, obstetrics, medical genetics, biostatistics, a'd radiation physics.
O i
e-------
w
,,w--
v e
. Radiation Dose Asressment Study (Ref: 42,59,66,67,70)
Time Span:
1979-1984(Complete)
Sponsors:
Department of Radiation Health - University of Pittsburg Survey performed by:
Pennsylvania Department of Health, Centers for Disease Control, and U.S. Bureau of the Census Primary
Purpose:
Evaluate the extent to which local residents may have been exposed to radiation from TMI.
==
Description:==
Establishment of a radiation dose assessment for every resident within the 5-mile radius of TMI during the 10 days following the accident and every pregnant woman resident within the 10-mile radius during the same period.
The study has been done over a 4 year period.
~
. ~.,. - -..,
l
.g.
r' 1st yr - WBy-for people within 10-mile radius 2nd yr - pregnant women doses and fetal WBy mile radius 3rd yr - thyroid tissue dose to general population (within 5 miles) thyroid tissue dose to mother and child (10 miles) 4th yr - total skin dose ( y +8 ) (skin dose = 21/2 - 3x WBY)
For each of the doses (WBy, thyroid and total skin dose), the maximum possible
. dose (assuming no skielding and no evacuation over the 10-day period) and the,
~
likely dose (assuming evacuation and clothing) were calculated.
Study covered approximately 34,000 people within a 5-mile radius of TMI.
Findings:
\\
Based on the findings of this study, it is estimated that the maximum possible whole-body gamma radiation dose to anyone off site was no more than 175 mrem, and the average maximum possible gamma dose to those within 5 n:iles of the plant was approximately 25 mrem.
Because these estimates make no allowances for shielding, they are generally considered to represent over-estimates. The highest "likely" dose assigned to an individual off site was 80 mrem and the average "likely" gama dose to those within 5 miles of the plant was 9 mrem. Additional exposure of the population came from the beta radiation dose to the skin and from the inhalation dose to the lung.
It
.. is estimated that the total dose to the skin could have been much larger than the whole-body gama dose by a factor of 3 to 4 if the protective effects of shelter and clothing are neglected. The inhalation dose is estimated to have constituted no more than 3% to 7% of the dose to the whole body.
The findings from the total skin dose part of the study have not yet been published.
.The publication of the Beyea report in 1984 may prompt additional studies in this area.
l i
.- Health Behavioral (Stress) Study (Ref: 30,38,41,42,49,67,70)
Time Span:
3surveystaken(7/79,1/80,10/80) completed 1981 Sponsors:
Pennsylvania Department of Health
~
Primary
Purpose:
To describe how persons living in the vicinity of TMI reacted during the crisis, as well as 3, 9 and 18 months later.
==
Description:==
Several phone surveys were made at 3, 9, and 18 months after the accident to obtain the information needed to study the stress on people living near TMI. The surveys conducted were:
(1) 7/79 Penn State survey of 692 persons within 5 miles of TMI; 7/79 NRC survey of 1506 persons within 55 miles of TMI (2) 1/80 Penn State survey of same group, and (3) 10/80 Penn State survey of households within 5 miles and 41-55 miles of TMI.
Five distress indices studied were:
1.
How upset the respondent was about the situation at Three Mile Island.
2.
How serious a threat the respondent felt TMI was to safety.
.. 3.
Frequency of "behavioral symptoms" during atwoweekperiod(i.e., lack
~
of appetite, overeating, sleeplessness, feeling shaky, trouble thinking, irritabilityandanger).
4.
Frequency of "somatic symptoms" (i.e., stomachaches, headaches, diarrhea, frequent urination, rash, abdominal pain, and sweating spells) during a 2 week period.
5.
For those persons who reported either behavioral or somatic symptoms, whether they attributed those symptoms to the situation at Three Mile Island.
Findings:
The major impact of the accident was felt during the few weeks immediately following the accident. While some effects persisted over the following year, long-term effects were of a low magnitude. Overall, this study indicated (a) persons who are younger, more educated, married and females
~
were most distressed during the crisis; (b) those who reside within 15 miles of TMI had more stress than did those who reside farther out; (c) the use of sleeping pills and/or tranquilizers to deal with anxiety, as well as certain psychosomatic symptoms increased among certain individuals; (d) the increased level of anxiety experienced following the accident persisted
213-for approximately one year, but declined substantially. in October,1980; j
and (e) persons with more social support tended to be less distressed than others during the crisis.
a e
M-6 0
b 1
8 l
l
\\
l l
l
_ j
, Mental Health Study (Ref: 42,67,70)
Time Span:
1979 - 1980 completed 1980 Sponsors:
Western Psychiatric Research Institute f-Primary
Purpose:
To study the mental health impact of 3 "high-risk" groups near THI:
TMI employees, mothers with small children, and mental health clinic patients.
==
Description:==
Surveys of 3 "high-risk" groups near TMI:
1)TMIemployees,2) mothers with small children, and 3) mental health clinic patients, were taken at 9 months and one year following the accident. People residing near the Shippingport nuclear plant in western Pennsylvania were used as controls.
Findings:
TMI mothers had an excess risk of experiencing clinical episodes of anxiety and depression during the year after the accident.
They also reported more symptoms of anxiety and depression at subclinical levels.
. TMI workers were essentially similar to control workers (Beaver Valley Nuclear Plants) with respect to mental health indicators under study both at the clinical and subclinical levels.
Mental health clinic patients expressed similar level of symptoms or anxiety at both TMI area and Beaver Valley area selected as control.
, Overall TMI mothers had higher incidences of depression than the control
~
-~
There were little or no differences in the other two "high-risk" group.
, groups.
~ - _ _ _.. _ - - _ _
- _ _ _ _ _. Infant Mortality Study (Ref:
51,55,70)
' Time Span:
1979, 1980 & annually Sponsors:
Pennsylvania Department of Health Primary
Purpose:
.To determine if the TMI accident had any measurable influence upon infant mortality in the vicinity of the plant.
==
Description:==
The infant mortality data were analyzed by calender quarter for a 10-mile radius around TMI for the years 1977, 1978, and 1979. These data were compared with similar data for the State of Pennsylvania.
Findings:
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 3 years under consideration.
1
- ~.
-,a r,--
. 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 quarter of 1979 imediately 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 TMI 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 THI-related upward trend), particularly during the third r-and early fourth quarters.
Withinthe10-mileradiusofTMI,the1579infantmortalityrate(16.1)was not significantly different 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 imediately surrounding comunities 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 uncomon
~
with no particular epidemiologic significance.
For this reason, the 1978 infant mortality rate should not be used as a nonnal base for comparison.
18-Having considered both cross-sectional and temporal analysis of the
~
available vital statistics data ecmpiled by the State Health Department, there is no evidence that the THI nuclear accident has had any significant impact upon infant mortality.
e 4
O I
I 4
e o - -
g,.
~^~ ~
T
- -m a
w
4 -
Health Economics Study (Ref:
67a,70)
Time Span:
1979 - 1982 complete Sponsors:
Pennsylvania State University and the USNRC Primary
Purpose:
To estimate the economic costs incurred by individuals or communities as a
. result of a change in physical or mental health status and/or a change in health care services due to TMI.
==
Description:==
1 Three phone surveys were conducted soon after the accident of households within a 55-mile radius of TMI.
Data collected included information on social and psychological effects of the accident, evacuation costs, health care utiliz,ation patterns after the accident, and decisions to evacuate.
Findings:
The findings indicate that stress symptoms caused by the TMI accident did affect the health-related behaviors of area residents.
Based on regression analysis, it is estimated that the cost of the changes in health-related 1
. behaviors was about $178,419 for a period of 10 months within a 5-mile
~
ring of TMI. Of the costs examined, the economic costs of work days lost and physician visits are the largest cost' items.
The results also show that there are significant effects of the stress on the increased consumption of cigarettes, alcohol, and tranquilizers after the accident.
e e
m 4
4 e
W i
t i
i
. - - -. ~ -. - - ~ ~ - - - - - - - - - - - - - - ~ - - - - - - - -
.. i Residential Mobility Study (Ref:
60,67)
Time Span:
1979 1982 complete Sponsors:
Pennsylvania Department of Health
~
. Primary
Purpose:
_c To determine the effect of the 1979 TMI accident on residential mobility and subsequent population composition.
==
Description:==
The entire population living within 5 miles of TMI was registered shortly after the accident and traced one year later to identify movers.
Findings:
The rate at which people moved remained the same the year after the accident as before, and that approximately 15% of those who moved (changed address) gave TMI as the main reason for their decision to move.
The study also found that those moving because of TMI had attributes highly associated with mobility in general.
When those attributes were controlled in analysis, attitudes about TMI were virtually the same among movers and i
l l
22-nonmovers. On the other hand, demographic characteristics of new people moving into the area were not different from those who had moved out.
However, attitudes about THI were significantly more positive among the newly moved-in people than among the moved-out people.
e e
S
..-n.-
.--,-m
--e-----
,v-m-,--ar-,
e pw--
w+++
~
v o-+=v-~~-+9w"'e~~***&*W-WW"+^
^'*W"'*''^""~'*"*~*"
~
.. ~
Epidemiologif Surveillance in Pennsylvania (Ref: 71)
Time Span:
i985 - continuing Sponsors:
Pennsylvania Department of Health with coordination of PEMA, BRP, and NRC
~
Primary Purposes Develop and implement a baseline health information system which will ensure the continuous and systematic compilation, analysis, and interpretation of the health status of populations living within 20 mile radii of the 6 nuclear plants and selected control areas in Pennsylvania.
Decription:
Baseline health information will be collected and computerized from the six nuclear power plants in Pennsylvania and from selected control areas.
Data _
collected will include natality and mortality data, morbidity data, and population data. This data will be updated and monitored routinely. The Epidemiologic Surveillance System will be used to detect significant changes within or differences from norms in any health indicators under
.. 1 consideration.
It would also provide comparative data which could be used to assess the potential health effects of the TMI accident.
Finally, the System would provide health information that would be invaluable in the event of another nuclear accident or any other health threatening event.
i o
e I
e e
i
--- ----.- -,,-= --.- ~,- -,,,-,,,,,._ _,-.., _..,__..-.,--,,.-.,,-..,,-- -.-.-....-..-_-- -
REFERENCES *
- 30. Houts, P., et al., "Health-Related Behavioral Impact of the Three Mile Island Nuclear 'ncident," report submitted to the TMI Advisory Panel on Health Research Studies of the Pennsylvania Department of Health, Part I, Apr. 8,1980.
38.
Houts, P., et al., "Health-Related Rehavioral Impa ct of the Three Mile Island Nuclear Incident," report submitted to the TM1 Advisory Panel on Health Related Studies of the Pennsylvania Department of Healt h, Pa rt 11, Nov. 21,1980 39 U.S. Nucleer Regulatory Connission. NUREG/CR-1728, "The Feasihility of Epidemiologic Investigations of the Health Effects of Low-level lonizing Radiation," N. Dryers, et al., Health Systems Division, Nov.1980 41 Houts, P., et al., "Extent and Duration of Psychologi cal Distress of Persons in the Vi cinity of Three Mile Island," in Proceedings of the Pennsylvania Academy of Science, Vol 54, No 1,1980.
'4_2. Tokuhata, G., "Three Mile Island Health Ef fects Research Progran,"
in Proceedinos of the Pennsylvania Academy of science, Vol 54, No 1,19-21,1980.
44 Tokuhata, G., "Dregnancy Outcome Around Three Mile Island,"
presented at the conference:
Linking Public Health Social Worker and i
Publi c Social Servi ces for Comprehensive Care for Mothers and Children, University of Pittsburg, PA, Mar. 29 - Apr. 2,1981 49.
Houts, P., et al., "Health-Related Behavioral Impa ct of the Three Mile Island Nuclear Incident," report submitted to the TPI Advisory Panel on Health Research Studies of The Pennsylvania Department of Health, Part 111, May 12,1981.
51.
Tokuhata, G., and E. Digon, "Fetal and Infant Mortality and Congenital Hypothyroidisn Around THI," presented at the International Symposium on Health Impa cts of Dif ferent Sources of Energy, Nashville, TN, June 22-26, 1981.
55.
Tokuhata, G., "Impa ct of TMI Nuclear Accident Upon Pregnancy Outccme, Congenital Hypothyroidism and Infant Mortality," chapter i
prepared for Energy, Environment and the Economy published by the Pennsylvania Academy of Science,1981.
f
., 55a. Houts, P., et al., "Psychologi cal and social Ef feets on the Population Surrounding Three Mile Island After the Nuclear Accident on March 28, 1979," chapter prepared for Energy, Environment and the Economy published by the Pennsylvania Academy of Science,1981.
- 57. Goldhaber, M., and J. Lehnan, "Crisis Eva cuation During the Three Mile Island Nuclear Accident:
The TMI Population Registry,"
presented at the 1982 Annual meeting of the American Public Health Association, Mont real, Quehec, Nov. 16, 1982.
58a. Brmet, E., et al., "Mental Health of Residents Near the Three Mile Island Reactor: A Comparative Study of Selected Groups,"
in Journal of Preventive Psychiatry, Vol.1, No. 3,1982.
- 59. Rao, G., et a l., "The TM1 Population:
A Closer Look," in Proceedings of the Pennsylvania Academy of Science, Vol 56, No 1, 1982.
r-60 Goldhaber, M., et al., "Moving After the Crisis-A Prospective Study of Three Mile Island Area Population Mobility," Environment and Rehavior, Vol 15, No 1,93-120, Jan.1983.
- 62. Gol dhaber, M.,' et al., "Spont'aneous Abortions After the 'Three Mile Island Nuclear Accident:
A Life Table Analysis," Am Journ of Publi c Health, Vol 73, No 7, 752-759, July 1983.
64 Goldhaber, M., et al., "The Three Mile Island Population Registry,"
l Publi c Health Renorts, Vol 98, No 6, 603-609, No v. -De c. 19 83.
1 65.
Bratz, J., "The Three Mile Island Mother-Child Registry," Health Renorter, Vol 4, No 12,1,4, nec.1983.
66 Gur, D., et al., "Radiation Dose Assignment to Individuals Residing Near the Three Mile Island Nuclear Station," in Proceedings of the Pennsylvania Academy of Science, Vol 57, No 1 1983.
l 67.
Tokuhata, G., "Three Mile Island Nuclear Accident and its Ef fect l
on the Surrounding Population," Pennsylvania Department of Health, i
nivision of Epidemiology Research, Jan.1984, 67a. Hu, T., et al., "Health-Related Economi c Costs of the Three-Mile i
Island Accident," Socio-Economi c Planning Science, Vol.18, No. 3, 183-193, Jan. 21,1984 l
i l
1
-. 6R. Houts, P., et al., "Utilization of Medi cal Care Following the Three Mile Island Crisis," Am Journ of Public Health, Vol 74, No 2,140-142, Feb.1984 70.
Tokuhata, G., "Health Studies in the Three Mile Island Area,"
presented at the special session on Health Ef fects of Radiation at the annual meeting of the American Nuclear Society, Miami Reach, FL, June 7-12, 1981.
71.
Tokuhata, G., "Epidemiologi c Surveillance in Pennsylvania:
A Case of Nuclear Power Plants," Pennsylvania Department of Health, June 1984 74 Tokuhata, G., "Three Mile Island (TMI) Nuclear Accident and Pregnancy Outcome," to be presented at the Xll International Biomet ri c Conference, Sept. 2-7, 1984 75.
Tokuhata, G., "Three Mile Island:
An Environmental and Public Health Emergency," presented at the National Conference on Environnental Publi c Health, Oct. 16, 1084
- Reference numbers refer to Enclosure 1 (Chronologi cal Bibliography) of "Staf f Actions f rom the i
letter f ran W. Dircks to the Commissioners (M840815)," Aug. 31,19B4 August 15, 1984 Commission Meeting on TMI
)
O l
W N>
n cv r
c L' T.
ENCLOSURE 3 d
0=,<.
CANCER MORTALITY AND MORBIDITY(INCIDENCE) AROUND TMI George K. Tokuhata, Dr.P.H., Ph.D.*
Edward Digon, M.P.H.**
Division of Epidemiology Research Pennsylvania Department of Health September, 1985
- Dr. Tokuhata is Director, Division of Epidemiology Research; also Professor of Epidemiology and Biostatistics i
(adjunct), Graduate School of Public Health, University Pittsburgh of
- Mr. Digon is Chief, Special Studies Section, Division of Epidemiology Research provided by the following physicians and scientists
\\
s Dr. Robert Miller and Dr. Charles Land, National Cancer Institute; Dr. Matthew Zack, Centers for Disease Control; Dr. Paul Sheehe, Upstate Medical Center, College of Medicine, Syracuse, New York; and Dr. Troyce Jones Ridge Naticnal Laberatory.
, Oak p
SStatro2p
-e
--n-
4 i i TABLE OF CONTENTS 1
CANCER MORTALITY AND MORBIDITY (INCIDENCE) AROUND TMI PAGE I. Introduction.........................................
1
~
II.
Methods and Materials................................
3 III. Results..............................................
7, A.
General Population Characteristics...............
7 B.
Cancer Mortality.................................
10 1.
Overall Cancer Mortality by MCD:
20-Mile Radius...............................
10 2
Site-Specific Cancer Mortality:
10-Mile Radius...............................
11 3.
Overall Cancer Mortality by Direction from TMI:
MCD......................
12 4.
Four Selected Communities:
Downwind from TMI............................
13 5.
TMI Census Enumeration Distries (CED):
Downwind from TMI............................
15 C.
Cancer Morbidity: Incidence......................
16 D.
TMI Mother / Child Registry:
Cohort Follow-Up.................................
21 IV.
Summary and Conclusions..............................
23 AN ASSESSMENT OF A CANCER MORTALITY SURVEY IN THREE AREAS SELECTED BY LOCAL RESIDENTS NEAR TMI I. Background...........................................
1 II. Methods..............................................
3 III. Results..............................................
5 A.
Demographic Characteristics.......'...............
5 1.
Minor Civil Divisiens: MCDs..................
5 4
i
)
l l
-i-i
1 tm i
PAGE 2.
TMI Census Enumeration Districts: CEDs.......
6 3.
Survey Areas Selected by Local Residents.....
6 B.
Cancer Mortality in Three Selected Areas..'.......
8*
IV.
Discussion, Evaluation and Critique..................
11
=
9 i
e i
s a
f I
I i
r i
f=
=
1 1
g
,-,_.,w.
-,--__.,,-_-_7
,7--._,&
,-_,m9e
,,,p, 9.
,.,+3-9e9_,
,y,9 9-, - -
.9,,ew_.
.ww,7
.-,y, 79,--
i CANCER MORTALITY AND MORBIDITY o
ff JAOUND ThI I.
INTRODUCTION More than six years have elapseo slace the March 28, 1979 accident at the Three Mile Island. (TMI) nuclear power plant.
Since the accident, concerns about,the potential impact of radioactive substance releases on cancer incidence have been a subject of considerable attention.
These concerns have been fueled by a recent review of radiation dose assessment conducted at the request of the TMI Public Health Fund and a survey of cancer mortality carried out by a group of citizens who focused on sections of communities near TMI.fl)
Government reports published earlier of radioactive sub-stance releases and population exposures following the accident indicated that the radiation exposure of TMI area populations was very small and that future health effects would be minimal.
With respect to the long-term health effect, it was projected that the number of cancer cases and genetic abnormalities over the life-time of the some 2,000,000 persons living within 50 miles of the plant would be too small to be detectable. (2)
The purposes of the present study are to determine if there is sufficient epidemiologic evidence to indicate that cancer mortality and morbidity in communities around TMI show abnormal patterns recardless of the level of radiation releases, and that any substantial excess cancer cases, if observed, are consistent with the existing theory of radiation carcinogenesis.
i l
The Pennsylvania Drpartment of Health monitors cancer mortality for the 2,580 minor civil divisions (McDs) within the State and conducts epidemiologic investigations of potential problem areas.
Since the 1979 nuclear accident, th' TMI area ha's e
been the subject of more intensive monitoring for several reasons including, (a) the continuing health concerns of area residents, (b) a general widespread interest in the situation, and (c) offi-cial requests by the Advisory Panel for Decontamination of TMI Unit 2 and the TMI Public Health Fund Advisory Group fer further evaluation of cancer mortality and morbidity in certain TMI area communities investigated by local residents.
Now that the minimum latency periods for certain forms of cancer, such a.e leukemia, might have been reached, it is reasonable to consider possible long-term health effects of the accident, particularly the incidence of cancer in the area.
Furthermore, it was only recently that the cancer incidence data necessary for an acceptable assessment of cancer problems in the area, became available from the Pennsylvania Cancer Registry which was instituted in South Central Pennsylvania in July, 1982 The present study covered both cancer mortality and mo'rbid ity and the results presented include the following aspects (a) comparison of the observed and expected numbers of cancer deaths, (b) comparison of the observed and expected numbers of newly diagnosed cancer cases, and (c) assessment of the followup incidence (newly diagnosed cases) of cancer'within a defined i
population (cohort) presumably exposed to the~TMI accident.
radiation.
ConsidGred for the cohort entlysis et this time are pregnant women and their fetuses who were residing within the ten-mile area at the time of the accident for whom accurate cancer incidence data hav.e been made available.
II.
METHODS AND MATERI ALS The routine monitoring of cancer mortality in the TMI area has focused on all 35 minor civil divisions (MCDs) within ten miles of TMI.
The data used in monitoring cancer mortality in MCDs in the TMI area include:
(a) enumerated MCD populations by age, sex, and race from decennial United States censuses of population; (b) intercensal and postcensal population estimates (total popu-lations only) developed by the Pennsylvania Department of Health and other governmental agencies; and (c) cancer mortality data by age, sex, race and site from the State Health Data Center, Pennsylvania Department of Health, which is responsible for the state vital statistics registration system.
The time periods covered in this mortality study include a five-year pre-TMI period (1974-78), and a five-year post-TMI period (1979-83).a)
Death rates (such as number of deaths per 100,000 population) are not presented for individual geographic i
l areas because, when based on very small populations, as several are, single deaths can result in rates that,cannot be reasonably compared with norrative lates based on large populations.
a) For purposes of this report the pre-TMI period is January 1, 1974 through December 31, 1978 and the post-TMI period is January 1, 1979 through December 31, 1983. -
- . stead, 'oxpected" or theoretical numbers of deaths that would be expected under normal conditions were computed for com-parison with the observed numbers of deaths in each MCD.
These include expected numbers for eight major cancer categories The
- expected numbers of cancer deaths are those which would have been expected if the communities under st,udy had the same age-sex-specific cancer mortality experiences as Pennsylvania as a whole during the 1979-81 period.
More specifically, the expected numbers of cancer deaths are those which would have been expected if each age-sex population group in the communities under study had the same cancer mortality experiences as did the Pennsylvania population in that age-sex group between 1979 and 1981, Lnd if the community's population remained the same in size (and in age-sex distribution) as it was at the time of the census in 2:0.
It is important that the expected numbers be computed in this manner because cancer death rates vary according to age and i
The age-sex-specific population data needed to compute sex.
expected numbers of deaths specifically for the 1974-78 and 1
1979-83 periods were not available for the study populations.
Thus, period-specific expected numbers of deaths could not be computed.
Cancer morbidity data for a two-year period (July 1987-June 1954) have recently become available for 14 counties in South Central Pennsylvania, including the TMI areas.
As in the mortality analysis, the expected numbers of new cancer cases were
~
computed on the basis of 1980 U.S.
Census population data and on normative cancer incidence data (1978-1981) from the i
SurveillancG, Epidemiology and End Results (SEER) program of the National Cancer Institute.
SEER data are used as the "standard" because Pennsylvania Cancer Registry data are not yet available for the State as a whole.
Pcpulation data developed from the special 1979 census, which included all households within an approximate five-mile radius of TMI, are also available for use in special investiga-tiens.
These data permit the assessment of cancer morbidity and mortality in Census Enumeration Districts (CEDs), smaller areas within each minor civil division under study.
These special census data have been supplemented by popula-tion data from the 1979 school census in certain minor civil divisions.
While school census data do not provide the detailed information available from the special TMI Census, the informa-tien provided proved to be extremely valuable when the areas of interest lie beyond the five-mile limit of the TMI Census.
In order to utilize data from the TMI Census for studying areas which do not conform to the census Enumeration Districts (CEDc), field visits were necessary to determine the specific geographic boundaries of interest.
This is particularly true when rural areas are involved and street addresses are not available.
Mortality studies and investigations conducted in Pennsylvania, which are residence-specific, are complicated by the frecuent incorrect reporting of actual r'esidence on vital records.
i This problem, of minor significance at the county level, can, in many instances, be one of the most important i
4 factors to be considered in evaluating che mortalit t
y and morbid-ity rates for specific MCDs.
The problem, which tends to result inartificiallyinflatedyatesforcitiesandboroughs and reduced rates for townships, stems from a frequent confu mailing address, by the persons providing information s on of
, with actual place of residence.
Since post offices are usually located in cities or boroughs through which all mail s are delivered, individuals associate mailing address with place of residence and report it as the place of residence on vit l a
records.
The ascertainment and correction of incorrect information, although difficult and time consumi ence ng, is necessary in the assessment of health problems in small areas The statistital significance of the differences b t cbserved and expected cancer morbidity and mortality was e ween by computing standardized nortality and morbidity assessed ratios SMRs) and testing for significant departures from unity using th Poisson model.
e As previously stated, the expected numbers of ca ncer deaths are based on the Pennsylvania cancer mortality experienc 1979-81, e,
by age and sex, and the distribution (age-sex) of the study populations at the time of the 1980 Census of Popul ti\\
Thus, the expected numbers of cancer deaths a
on.
, based on normative data for the 1979-81 period can more appropriately b with observed numbers for the post-TMI period th
{
e compared an for the pre-TMI period.
The observed numbers of cancer deaths fo pre-TMI period are provided for reference purposes; how e
l differences between the pre and post-TMI observ d fi
- ever, e
gures should
-6 2
s l
not automatically be attributed to TMI.
As a matter of fact, the observed numbers for the pre-TMI period are likely to be lower 1
than the expected numbers.
In contrast, the observed numbers for the post-TMI period are likely to be higher than the expected i
~
1 numbers for the following reasons:
t l
]
-- The population in many of the areas under study increased' substt
' ally over this time period.
-- There is a continuing general increase in cancer mortality in Pennsylvania and elsewhere.
-- Diagnostic techniques and opportunities are improving.
-- The population is aging and with age comes an increase in
~
cancer incidence and, ultimately, mortality.
The expected numbers of new cancer cases (incidence) were based on incidence data by cancer site, age, and sex from the l
Surveillance, Epidemiology and End Results (SEER) program (1978-81) of the National Cancer Institute and population data from the 1980 census.
Since the observed numbers of new cancer cases in the study populations are for a later period, July 1982--June 1984, caution again must be exercised in interprating the data.
Factors such as population growth or decline; improved cancer reporting systems; improved diagnostic techniques and opportunities; and aging populations, as well as actual changes in cancer incidence can have an impact on the results.
III.
RESULTS A.
,G_ENERAL POPULATION CHARACTERISTICS The TMI ten-mile area which includes 35 minor civil 1
divisions and portions of five counties (Cumberland, 7-
)'
d 1
J.
'^
Dauphin, Lancaster, Lebanon and York) had an enumerated i.
population'of 289,930 in 1980.
The TMI five-mile a rea, with a population of 64,897 in 1980 includes 13 minor civil divisicos in.three counties (Dauphin, Lancast.er, and York)
The population of the TMI ten-mile area increased by 7.3 percent from 270,306 in'1970 to 289,930 in 1980 s
(the population of Pennsylvania increased by 0.6 percent).
The population in the five-mile area, however, increased by 31.3 percent, from 49,410 to 64, 897, while the population in the five to nine mile band increased by only 1.9 per-cent; from 220,896 to 225,033.
Only two of the'13 minor civil divisions in the' five-mile area lost population during the decade and these were relatively minor decreas t
1 es in small boroughs (Royalton and Goldsboro).
The popula-tions of seven of the 13 minor civil divisions increa y
more than 25 percent.
The population of East Manchester Township increased by 105.4 percent, from 1,735 to 3,564,-
while Newberry Township's population increased by 68 1 percent, from 5,978 in 1970 to 10,047 in 1980 The popula-tion of seven of the 22 minor civil divisions in the fiv to nine mile area decreased between 1970 and 1980
. e j
(six were boroughs, the other was Harrisburg City) but seven i
municipalities experienced population increases of 25 percent or more.
One, Conewago Township, increased by 119.8 percent, from 1,124 to 2,471.
Lewisberry Borough experienced the greatest relative population decrease, 36 9 percent (from 490 to 309), but Harrisburg City, whose
-e-m-
,ww
l population dacreased by 21.7 parcent, had the greatest absolute decline, 14,797 (from 68,061 to 53,264).
The population of the five to nine mile area, exclusive of Harrisburg, increased by 12.4 percent between 1970 and 19B0; with Harrisburg, the increase was only 1.9 percent.
In 1980, the median ages of the populations of the 35 minor civil divisions within ten miles of TMI ranged from 27.3 years for Conewago Township to 40.0 years for Paxtang Borough.
Both are in Dauphin County.
The median ages of the populations of ten of the 13 minor civil divisions in the five-mile TMI area and 17 of the 22 in the five to nine mile area increased between 1970 and 1980.
Since cancer morbidity and mortality generally increase with age, it might be expected that certain increases in cancer morbidity and mortality during the decade of the 1970's would be associated with the aging of the population.
In 1970, 25,822 or 9.6 percent of the 270,306 residents of the TMI ten-mile area were nonwhite.
By 1980, the num-ber had increased to 33,627; the percentage to 11.6.
Of the 33,627 nonwhites in the area, 25,074 (74.6 percent) resided in the City of Harrisburg; 2,079 (6.2 percent) in Lower Paxton Township; 1,833 (5.5 percent) in Steelton Borough; 1,619 (4.8 percent) in Swatara Township; 598 (1.8 percent) in Derry Township; and 571 (5.7 percent) in Middletown Borough.
Thus, only about five percent of the nonwhite population resided in the other 29 TMI ten-mile _ _ _ _ _ _ _ _ _ _
area communities.
More than 47 percent of the population of Harrisburg City and 28 percent of the population of Steelton Borough were nonwhite in 1980.
The only other communities in the ar'ea with nonwhite populations in excess-of five percent were Swatara Township (8.6 percent), Lower Paxton Township (6.0 percent), and Middletown Borough (5.6 I
percent).
All are located in Dauphin County.
B.
_ CANCER MORTALITY 1.
Overall Cancer Mortality by Minor Civil Division ~
(MCD) : 20-Mile Radius The monitoring of cancer mortality in the TMI area 1
by the Pennsylvania Department of Health, as previously stated, focuses on 35 minor civil divisions (MCDs) within ten miles of the nuclear facility, but periodi-cally includes communities in a 20-mile area.
This monitoring includes an annual comparison of observed and expected numbers or cancer deaths, total and selected cancer sites, in individual and grouped MCDs.
TABLE 1_ shows the numbers of MCDs, the popula-tions, and the observed and expected average annual numbers of cancer deaths for Pennsylvania and geo-graphic areas at specified distances from TMI during a pre-TMI five-year period (1974-1978) and a post-TMI five-year period (1979-1983).
The gr,ouped MCD data by distance from TMI provide no indication of excess cancer mortality in geographic areas up to 20 miles e
e
i.
- i distant from TMI.
The similarity between the observed and expected numbers of deaths, particularly for the post-TMI period, is remarkable.
The observed numbers for the pre-TMI period are somewhat lower than the expected numbers but are of no particular signifi-The observed average annual number of cancer,
cance.
i deaths in the 10-19 mile ring for the post-TMI (1979-1983) period was lower than expected, and the observed and-expected numbers for the 5-9 mile band and the five-mile radius areas wero quite similar.
While the average annual number of cancer deaths among Pennsylvania residents increased by 8.5 percent from the 1974-78 period to the 1979-83 period, in-creases such as this are not limited to Pennsylvania.
For example, the average annual number of cancer deaths of the United States increased by 12.1 percent during the same time periods, such increases in the actual numbers of cancer in recent years are a reflection of many factors, including environmental carcinogens in addition to size and characteristics, particularly aging, of the population.
The potential influence of the various cancer causing agents, however, are -not accounted for in the observed-expected methodology used in this study.
2.
Site-Specific Cancer Mortality: Ten-Mile Radius i
TMI ten-mile area cancer mortality data for eight major cancer site classifications are presented in 1.
r r=v
+
f
TABLE 2 The specified cancer sites are:
buccal cavity and pharynx; digestive-organs and peritoneum; respiratory system; bone, connective tissue, skin and breast; genitourinary system; other and unspecified sites; leukemia; and other lymphatic and hematopoietic tissue.
The variations from expectation, higher or lower, in these site-specific data are nominal for both time periods under study and have no particular epidem-iologic importance.
3.
Overall Cancer Mortality by Direction from TMI:
Minor Civil Divisions It has been alleged that, because of the wind direction at the time of the major radioactivity release from TMI, cancer mortality has increased in the population to the northwest (downwind) of the facility.
TABLE 3 shows cancer mortality data for TMI minor civil divisions grouped according to general direction quadrants from TNI.
For each quadrant, the observed average annual numbers of cancer deaths for both the pre-TMI and post-TMI periods approximate the expected numbers.
The average annual numbers of deaths for Goldsboro and Highspire Boroughs are higher than expected both before and af ter the accident.
On the other hand, the average numbers of deaths for other communities, such as Londonderry, Wes't Donegal, Conevago, Newberry, Fairview, and Lower Swatara, are lower than expected both before and after the _ -.
accident.
Such variations in mortality aro commonly observed when the population bases under study are relatively small.
4 Four Selected Communities: Downwind fr'om TMI Following the re' lease of results of the health survey conducted by local citizens, which reportedly t
showed excess cancer mortality in selected areas within York County, available cancer mortality data for four area minor civil divisions were reviewed to determine if'there was any evidence of a cancer mortality problem in these survey areas.
Using 1979 TMI Population Census data, the actual place of residence of each post-TMI cancer decedent was determined.
Presented in TABLE 4 are residence-corrected overall cancer mortality data pertaining to Fairview and Newberry Townships and Goldsboro and York Haven Boroughs for.the post-TMI period.
As shown in this table, the observed average annual number of cancer deaths in each MCD was l
remarkably close to expectation when necessary i
corrections were made on residential addresses.
The observed average annual number of deaths for Goldsboro Borough was 1.0 and the expected number was 0.9, while the respective observed and expected numbers for York Haven Borough were 1.0 and 1.2.
An annual average of 16.6 deaths was observed in Fairview Township, and.-.
16.5 was expected.
In Newberry Township, an average of 12.2 deaths per year was reported, while 11.9 was i
expected.
It was of fu'rther interest to examine.how overall,
cancer mortality fluctuated each year within these four' selected communities.
TABLE 5 shows the actual number of resident cancer deaths 'for the selected communities for each of the five years, beginning with 1979.
These numbers appear to reflect random-like variations in cancer mortality.
TABLE 6, which provides data for the combined pop-ulation of the same four MCD area according to cancer site, includes grouped data for both the pre-TMI and post-TMI periods.
As shown in this table, the site-specific number of cancer deaths generally approximate expectation within the limits of random variation.
The numbers of site-specific deaths, even for an entire five year period,.are sometimes very small and one should be aware that one or two deaths, more or less, can have a marked effect on mortality ratios, but be of no particular significance.
The fact that nine leukemia deaths were reported during the five-year post-TMI period in the four selected MCDs, compared to three leukemia deaths in the pre-TMI period, might appear to be of importance.
How-ever, neither nine nor three deaths are significantly _
~
different from the expected number of six for the area.
It should be noted that one of the nine patients actually died before the accident, and two others were r
i diagnosed as having leukemia prior to the acejdent (in 1976 and 1978), and consequently, these deaths cannot be attributed to TMI.
In still another case, the available clinical data s'uggest that the apparent onset of leukemia was noted by a physician well before the accident.
i 5
TMI Census Enumerati'on Districts (CED): Downwind
-f rom TMI f'
r The smallest geographic entities in the TMI area
\\
v
[
{
which can be studied with a reasonable degree of
[
precision (in a statistical sense) are the Census Enumeration Districts (CEDs) delineated in the special 1979 TMI Census which covered the entire population
'~
residing within approximately five miles of TMI.
There are 15 CEDs in Fairview and Newberry Townships, the E
downwind minor civil divisions of immediate concern L
Only two of these CEDs are in Fairview Township, the other 13 are in Newberry Township.
TABLE 7 presents population dats and average annual observed and b,
expected numbers of cancer deaths for the post-TMI (1979-1983) period for these small geographic areas.
The CEDs are grouped to show data for the two townships according to the directional quadrants in which they L
-1!-
)
lie.
The observed and expected average annual numbers of cancer deaths for all 15 CEDs combined are quite comparable (10.8 observed and 11.6 expected).
Similarly, the deviations in the figures for the Northwest (7 CEDs) and Southwest (8 CEDs) quadrants are small and negligible.
The mortality ratio for the Newberry Township CEDs in each of the two quadrants is 0.85; that is, observed mortality was less than expected.
The mortality ratio for the two Fairview Township CEDs, which had a combined population of only 710 residents, was 1.80 (nine deaths observed, an estimated five expected).
None of these ratios is indicative of a significant departure from expectation.
In summary, the data analyzed thus far provide no indication that cancer mortality in TMI area minor civil divisions, including much smaller areas within them, differed significantly from expectation during the five years following the March, 1979 accident at TMI.
B.
CANCER MORBIDITY: INCIDENCE Population-based cancer incidence data were not avail-able for Pennsylvania communities prior to July 1,1982.
It was then that the Pennsylvania Cancer Registry became operational in South Central Pennsylvania, which includes the TMI area.
While the Cancer Registry can now provide..
1 morbidity data, there is a data gap from the timo of the accident, March, l'579, through June, 1982.
However, this date gap may not be considered serious in view of the fact that radiogenic cancers are not expected to be observable within a few years post exposure.
Even leukemia, which is
~
known to have a short latency period, is not likely to develop until at least five to six years following the initial exposure.
Since six years have passed, early Registry data Dhould be examined for possible new cases of Cancer.
The se of Pennsylvania Cancer Registry data in epi-demiologic investigations is relatively new.
- However, cancer incidence data for the four MCDs (Fairview and Newberry Townships and Goldsboro and York Haven Boroughs) were made available for special analysis.
The results of the special analyses, which include the period July 1, 1982 through June 30, 1984, are presented in TABLES 8, 9, and 10.
The expected numbers of new cancer cases for this period are those which would have been expected if each age-sex population group in the communities under study had the same cancer incidence as did the SEER program population in that age-sex group during the years 1976-81, and if the community's population remained the same in size (and age-sex distribution) as it was at the time of the 1980 census, knnoinstancedidthe observed numbers of newly diagnosed cancer cases shown in these tables significantly exceed the expected numbers,.
m
i.e., the differences between the observed and expected numbers could have occurred by chance alone.
Specifically, the observed number of newly diagnosed cancer cases among residents of the combined four MCD area during thi's two-year post-TMI period was 133.as compared with 121.4 E
expected (TABLE 8).
The observed number of new cases was less than expected for Fairview Township (61 observed and a
65.6 expected) and there was virtually no difference-between the observed and expected numbers for Goldsboro Borough (4. observed and 3.9 expected).
The observed and expected numbers of newly diagnosed cancer cases for York Haven Borough were 8 and 4.5, respectively, while the respective numbers for Newberry Township were 58 observed and 47.8 expected.
Again, these differences in the incidence of cancer are within the limits of random variation and not statistically significant at the 5 percent level.
The exact place of residence of two cases could not be determined but they are included in the area total.
TABLE 9 shows, for the same July, 1982 through June, 1984 period, observed and expected numbers of cancer cases for 19 primary cancer sites (organ systems first affected)
[
among residents of the total four minor civil division area and Fairview and Newberry Townships, combined and separately.
The number of cancer cases of specific sites among residents of Goldsboro and York Haven Boroughs are shown in the "NOTES" section of the table.
The expected 1 _
numbers are not presented for those two boroughs because the numbers for specific primary cancer sites are too small for meaningful comparison.
Of 19 different cancer sites considered, the observed numbers for the entire four minor' civil division area were lower,than expected for 11 sites,
{
whereas the observed numbers were higher than expec'ted for,
eight others.
Among these eight cancer sites, the relative differences between observed and expected were nominal and could be demonstrated in any set of random numbers subjected to multiple teste of significance.
Similarly, the observed numbers for Fairview Township were lower than expected for 10 cancer sites and higher for nine others i
while for Newberry Township the observed numbers were higher for 12 sites and lower for seven others, some of the differences were very small and none was significantly different from expectation.
The occurrence of specific cancers among residents of Goldsboro and York Haven also' appears to be randomly distributed.
The 12 newly diagnosed i
cancer cases in these two communities include cancers of eight different primary sites.
The only cancer for which more than one new case was reported in either of these two j
boroughs was breast cancer.
Three York Haven women were diagnosed as having breast cancer during the two year period; this was not significantly more than expected.
I An additional analysis was made ok ten specific radiogenic cancers (TABLE 101 These are cancers, such
-le-
l as leukemia, which are more susceptibio to the injurious action of-radiation.
For purposes of this report, the following cancers were considered (although there is still-some disagreement regarding degree of radiogenicity of certain of these):
esophagus, stomach, colon, pancreas, lung, breast (female), thyroid,,non-Hodgkin's lymphomas, j
multiple myeloma, and leukemia.
The number of such cancers that would have been expected to be diagnosed among residents of tne four subjact minor civil divisions during the two-year period July 1982-June 1984 is 64.2; 70 cases were actually diagnosed.
Twenty-six residents of Newberry
)
Township were diagnosed as having these cancers; 25 were expected.
The respective observed and expected figures for Fairview Township were 36 and 34.9.
Goldsboro and York Haven Boroughs are too small in their populations for the computation of meaningful expected numbers of new cancers by specific sites but overall, seven were observed and about four to five were expected.
None of these differences was statistically significant.
These data do not provide any consistent indication of a possible radiation-related cancer problem.
According to the j
existing literature, leukemia is the most likely cancer to be detectable five to six years following exposure to high dose radiation.
Two leukemia cases were diagnosed in the
~
four minor civil division area and three to four were expected.
There was one case each in Fairview and Newberry Townships, but none in Goldsboro or York Haven Boroughs.
-2C-
t t
For some other forms of radiogenic cancer, more cases were reported than expected (e.g., colon, breast and non-Hodgkin's lymphomas) but for others there were fewer.
As already stated, these data are not indicabive of a' radiation related cancer problem in the area.
Although the post-TMI observed number of newly identified cancer' cases (133) is somewhat higher than the expected number (121.4) and although the observed number of radiogenic cancer cases (70) is somewhat higher than the expected number (64.2):
none,,of the differences, overall or for specific minor civil divisions or specific cancer sites, is statistically significant at the five percent level; leukemia, the type of cancer most apt to show any abnormal pattern the earliest, had in fact fewer cases observed than expected; l
there are several reasons why the observed numbers are apt to be higher than the expected numbers, including general increases in cancer, population growth, better diagnostic techniques, improved cancer reporting, and an aging population.
D.
TMI MOTHER /CRILD PEGISTRY: COHORT FOLLOWUP The Mother / Child Registry includes a study group of nearly 4,000 pregnant women who delivered between March 28, 1979 and March 27, 1980.
These women resided in the TMI
ten-mile communities and were pregnant at the time of the accident or they became pregnant within a few months of the accident.
The Mother / Child Registry also includes a com-parlson (control) cohort of nearly 4,000 pregnant women who delivered one year after the study cohort,-between March 28, 1980 and March 27, 1981.
The Registry is updated continuously to insure current information for each mother and child registrant on a number of variables for use in e variety of followup studies.
Linkages are being established with the Pennsylvania Cancer Registry, the Pennsylvania Mortality File, the National Death Index, and out-of-state cancer registries, to identify those regis-trants who have developed cancer since the 1979 accident.
During the period July 1, 1982 through December 31, 1983, four of the 3,582 mothers in the study group, most of whom resided in 14 South Central Pennsylvania counties, were diagnosed as having cancer.
Based on data from the Surveillance, Epidemiology, and End Results (SEER) program (1973-1977) for females in the 10-44 year age group, 3.9 mothers might have been expected to be so diagnosed during this time period.
The primary cancer sites of these four patients were two colon, one brain, and one skin (trunk).
During this same period, two of the study children under five years of age were diagnosed as having cancer; about one case would have been expected on the basis of the SEER data; this small difference can occur by chance alone.
IV.
SUMMARY
AND CONCLUSIONS More than six years after the 1979 accident at TMI, both cancer mortality and morbidity were examined.
Comprehensive analyses of the available data included comparisons of the observed numbers of cancer with the expected numbers for a series of geographic areas, fro'm a 20-mile radius down to small Census Enumeration Districts around TMI.
Because of the potential importance of wind direction and radioactive plume dispersion during the early days of the accident, data also were analyzed with emphasis on the northwest and southwest quadrants.
Also considered were the distribution by cancer sites for both living and deceased patients, as well as possible predilection for any radiogenic malignancies within expected latency periods.
Furthermore, the incidence of newly diagnosed cases of cancer was evaluated among those women who were pregnant at tre time of the accident and among their offspring.
While cancer mortality data are useful for certain pur-poses, they are severely limited when the time gap between exposure and observation point is shorter than the time gap between initial diagnosis and death.
In contrast, incidence data are far more appropriate in assessing possible linkage between environmental risk exposure and carcinogenesis.
The most effec-tive method to be used in such an investigation is the prospec-tive cohort followup study of the population actually exposed to the specific environmental. risk under study.
The results of our epidemiologic study, including both mor-tality and morbidity data as well as cohort followup analysis, do not provide evidence of increased cancer risks to residents near
~
the TMI nuclear facility.
In view of the known long laten'cy of carcinogenesis and the persistent dispute over the amount of' radioactive releases from the damaged TMI nuclear reactor, it is prudent to continue epi-demiologic surveillance around TMI.
The Pennsylvania Department.
of Health established the mechanism for such an effort shortly af ter the 1979' accident and several long-term followup studies are already in progress.
i
REFERENCES (1)Aamodt, M.M. and Aamodt, N.O. Aamodt motions for investigation of licensee's reports of radioactive releases dur-ing the initial days of the TMI-2 accident and postponement of restart decision pending resolution of this investigation.
Pre-sented to United States Nuclear Regulatory Commission, June 21, 1984.
(2)Ad Hoc Population Dose Assessment Group: Population dose, and health impact of the accident at the Three Mile Island nuclear station.
U.S. Government Printing Office, Washington, D.C.,
1979, p.3.
I
-2E-
,,,<,,.,..y m..,
TADLE 1:
NUmER OF MINNI CIVII. DIVISI0tG, POPULATION (19001, DDSERVED AND D'PECTED NUPEERS OF RESIDENT CANCER DEATHS:
PEhNQrLVANIA AN3 GEDGRMilIC AREAS SELECTED DISTAf.'ES FROM THREE MILE ISLAND (TMII, 1974-78 AND 1979-83
-..._,_m Five Year Total Average Annuel Ntsaber or Cancer Deathe Number of Cancer Deaths s
Mortellty Retto Number or Observed [3]
Observed [3]
(Observed /Espected)
Minor Civil Population re oeaphic Area Dtvletone 1900III Erpected[2]
gg74 78 1979-83 Espected(2) 1974-78 1979-83 1974-78 1979-83 r-~..tvente 2,500 11,863,835 129,122 120:581 130,873 25,824.3 24,118.2 28,174.8 0.93 1.01
- t r
.e D' '11 t e ned f ue 129 790,007 8,177 7,163 7,924 1,635.4 1,432.8 1,584.8 0.88 0.97 (10-19 Mite Bandl (941 (508,877)
(5,268)
(8.374)
(5,07.2)
(1 D53.7)
(874.8) (1,008.4)
[0.83)
[0.981 tn.riete nedIue 35 209,930 2,909 2,7BS 2,892 581.7 557.B 578.4 0.98 0.99
(".-9 MIts Bendl (22)
(225,033)
(2,348)
(2,275)
(2,360)
(409.8)
(455.h)
(472.0)
(0.97)
(1.01) 5 etite Ihdlus 13 64.897 581 514 532 112.1 102.8 106.4 0.92 0.85 NOTES: III Ents. orated poputettone,1980 United States Census of Population.
III Espected nebers or concer deaths beeed on the ege-een distributton of the respective poputettone,1980 U.S. Census of Poputetton, end everage er. nuel ege-sex spectric concer death rates, Pennsylvente, 1979-01.
(3) Source or observed numbers of doett si State Hestth Dete Center, Pennsylvente Department of Heetth.
These date oro not corrected for incorrect residence reporting nn doeth certf ricates.
(TADLE 1 UATA AnE DISCUSSED ON PAGES 10 AND 110F THE TEXT.)
TABLE 2 OBSERVED A2 DPECTID AVEMGE ANxuAL NueERS OF CANOER DEATHS BY SITE: TMI TEN-MILE AREA, 1974-78 AND 1979-63 Average Annual Nurber of Cancer Deatha Mortality Retto Cancer Site
\\0bserved[2)
[0bservevExpe~cted)
(Ninth Revision, Internettonal Classification of Diseases)
Code Nur.ber ExpectedIII 1974-78 1979-83 1974-78 1979-63 All Sitee 140-208 5B1.7 '
557.8 578.4 0.96 0.99 Buccal Cavity and Pharynx 140-149 10.8 11.2 11.0 1.C4 1.02 Digestiv's Organe end Peritoneum 150-159 165.5 1 48.2 155.6 0.68 0.94 Respiratory and Intrathoracic Organs 160-1ES 1 40.6 129.4 133.0 0.92 0.95 Bsne, Connective Tissue, Skin & Breast 170-175 66.1 64.8 73.2 0.98 1.11 Genitourtnery Orgens 179 ",E9 85.7 85.6 90.4 1.00 1.05 Other and Unspecified Sites 190 '99 64.7 68.4 64.2 1.06 0.99 Lsukemie 204-208 21.5 17.0 22.4 0.79 1.04 Other Lyrphetic & Heestcpotetic Tissue 200-203 29.5 35.2 28.6 1.19 0.97 HOTES: III Erpected nurrburs of concer deathw based on the ege-een distributions of TMI l
Ten +1te Area mir.or civit division populations,1980 U.S. Census of Poputetton, and everage annual ege-sex specific concer death rates Pennsylvente, 1979-61.
The expected everage annual number of deathe for concer of all sites (581.7) does not equel the stanse* ion of expected deathe for specific sites (584.4) because the all sites total mes computed on the basis of stenary population date (by age-sex) for the entire TNI Terr-Hite Area, se ses done for other erees in Table 1, while the espected nuwbers for specific sites were obtained by sumaing the espected ntebers for each of the 35 sfcor civit divisione in the eres.
(2) Source of cbserved cia:bers of deathat State Health Data Center, Pennsylvente Department of Health. These dets are not corrected fcr incorrect residence recerting on cesth certificates.
(TA3LE 2 CATA APE DISCUSSED ON FA:3ES 11 Ah? 12 0F WE TEX *.)
~
TABLE 3: DSSERVED NC EXPECTED AVERAGE ANY.LAL NLDEER OF RE3IDENT CANCER DEATHS FCR TMI AREA MIPOR CIY1L O! VISIONS BY GENERAL DIRECTION FROM TXI, 1974-78 AND 1979-63 m
Average Annual Neber of Cancer Deathe Mortality Ratio Observed (3)
(ObservevExponed)
General Otraction frove Population TFI/ Minor Civil Division 1980l1)
Espected(2) 1974-78 1979-83 1974-78 1979-83 TMI Area MCOs 64,B97 112.1 102.8 106.4 0.92 0.95 hortheast Quadrant 16.241 31.1 30.B 30.8 0.99 0.99 Londonderry T. (Deuphin Co.)
5,138 7.5 6.4 7.4 0.65 0.99 Middletcon B. (Deuphin Co.)
10,122 21.9 22.4 21.8 1.02 1.00 Royalton B. (Deuphin Co.)
- 981 1.7 2.0 1.6 1.18 0.94 i
Ssutheast Quadrant 11.481 27.9 24.0 27.4 0.26 0.S B Coney T. (Lancaster Co.)
2,309 3.7 3.8 3.8 1.03 1.03 West Donegal T. (Lancaster Co.)
4,262 16.4 12.6 14.B 0.77 0.90 East Manchester T. [ York Co.)
3,564 6.6 5.4 7.8 0.82 1.18 York Haven B. (York Co.)
746 1.2 2.2 1.0 1.83 0.B3 i
Southwest Quadrant Conewego T. (York Co.)
4.979 7.5 6.2 5.4 0.E3 0.72 Southwest /Northmest Quadrante Nesberry T. (York Co.)
10.047 11.9 11.6 11.4 0.97 0.96 Northwest Cuadrent 22.1 t.9 33.8 30.2 31.4 0.BS 0.93 Fairview T. [ York Co.]
11,941 16.5 11.4 16.2 0.69 0.98 Goldsboro B. (York Co.)
477 0.9 1.6 2.0 1.78 2.22 Highspire B. (Deuphin Co.)
2.959 5.9 B.O 6.8 1.36 1.15 Lower Seetere T. (Deuphin Co.)
6,772 10.5 9.2 6.4 0.88 0.61
==
h3TES: Ii) Enumersted populations,1983 United States Census of Population.
I2I Expected nurbers of concer deaths based on the ege-sex distribution of the respective populations,1983 U.S. Census of Population, and everege annual age-sex specific concer death rates, Pennsylvanie, 1979-81 I33 Scurce of observed numbers of doeths; State Health Date Center, Pennsylvania Department of Health. These date are not corrected for incorrect residence reperting en death certificates.
11ASLE 3 CATA ARE DISCUSSED ON PA3ES 12 AND 13 OF THE TEXT.)
l TABLE de POPULATION AND OBSERVED AND E>PECTED AVERADE ANNtLAL NU4ERS OF CANCER OEATHS: FOUR TMI AAEA MINOR CIVIL DIVISIONS (FAIRVIEW AND NEWBERRY TDYNCH3PS /JC 001.DS80M AND YORK HAVEN BOROUGis), 1979-63 Average Annual Nueber of Cancer Deaths Mortality'Retto Minor Civil Olvisions_(H00s)
PopulationIi}
ObservedIEI Espected(3)
[0bse rvevEmpec t ed)
Four Minor Civil Division Total 23,211 30.8 30.5 1.01 Feirvies/Neuberry Townships 21,928 28 B 28.4 1.01 Fairview Township 11,941 16.6 16.5 1.01 Neuberry Township 10,047 12.2 11.9 1.03 Coldebere/ York Haven Boroughs 1,223 2.0 2.1 0.95 Goldsboro Borough 477 1.0 0.9 1.11 York Haven Borough 746 1.0 1.2 0.83 nstest Ill Enteersted pcpulettens,1980 United States Census of Population.
III Observed niebers of concer deaths obtained from death certificates provided by the State Health Date Center, Pennsylvanie Deporteent of Health. Nuebers shcan for specific HODS say not eretch routinely reported ntebers because of correctiene of residence reported en death certificates.
(3) Espected nur hers of concer deaths based on ege-sex distributions of the respective populations,1990 U.S. Census of Population, and ege-sex specific concer death rates for Pennsylvente, 1979-61 (T/JLE 4 CATA ARE CISCUSSED ON PAGES 13 AND 14 0F THE TEXT.)
T MLE 5: POPULATION A)c OBSERVED AND E:@ECTED Nt#EERS OF CANCER DEATHS BY YEAA:
i i
CERTA 2H TMI AREA MDOR C8V8L DIVISIONS,1979-ft3 Observedl2) Annual Number of Cancer Deaths Expected Nurbe r of Cancer Deaths Minor Civil Divisions (HCDs)
Populationli}
1979 1980 1981 1982 1981 Per YearIH Four Minor Civil Division Totet 23,211 22 33 30 36 33 30.5 Fairvien/Newberry Townshipe 21,988 22 29 28 34 31 28.4 Feirview Tcunship 11,941 10 21 13 19 20 16.5 Neuberry Township 10.047 12 B
15 15 11 11,9 Goldsbere/ York Haven Boroughe 1,223 4
2 2
2 2.1 Goldsboro Borough 477 2
1 1
1 0.9 Yorkhoven Borough 746 2
1 1
1 1.2
!CIS III Enurereted populations,1980 United States Census of Population.
IEI Observed nu: bars of concer deaths obtained frori death certificates previded by the State Hestth Data Center, oennsylvanie Deporte.ent of Health. Huabers sheen for specific HODS rey not Petch routinely reported nuebers tocause of correcticns of residence reported on death certificates.
(3) Expected nurbers of concer deaths based on ege-een distributione of the respective populations,1980 U.S. Census of Poputetton, and ege-sex specific concer death rates for Pennsylvania, 1979-61
[TMLE 5 DATA ARE DISCUSSED ON PAGE 14 CF THE TEXT.]
TABLE 6: DBSERVED AND E:eECTED AVERA3E ANNUAL ktpeERS OF CANCER DEATHS BY SITE IN A
$ ELECTED FOUR MDOR C8VIL DIVIS80N AAEA (FA2W3EV TUwNSH2P, GOLDSB0m BOFCUGH, NEWERRf TUwNSH8P, YORK HAVEN B0f01.G1),1974-78 AND 1979-EO i
_=
Average Annual Neber of Cancer Deathe kortality Retto Cancer Site Observec(2)
[0b se rved/Expec tee)
(Ninth Revision, International Classification of Diseases)
Code Nurber
~Exp ect e' dliI 1974-78 1979-83 1974-78 1979-83 All Sites 140-208 30.5 26.9 30.8 0.E8 1.01 Buccel Cavity and Pharyna 140-143 0.6 0.8 0.4 1.33 0.67 Digestive Organs and Peritoneum 150-159 8.2 5.8 7.2 0.71 0.BB Respiratory and Intrathoracic Organs 160-155 7.9 5.8 6.4 0.73 0.81 Bone, Connective Tissue, Skin & Breast 170-175 3.6 4.2 4.0 1.17 1.11 1
Ceniteurinary Orgene 179-189 4.2 4.0 4.2 0.95 1.00 Other end Unspecified Sites 190-199 3.4 3.4 4.4 1.00 1.29 Leukerie 204-20B 1.2 0.6 1.8 0.50 1.50 DBher Lyephatic & Henatcpoietic Tissue 200-203 1.6 2.2 2.4 1.38 1.50 HOTES: III Espected nebers of concer deathe tesed en the age-sex distribution of the respectivs popi.:letions,1980 U.S. Census of Pcpulation, and everage annual age-sex specific cancer death rates, Pennsylvente, 1979-61 III Source of observed ntenbers of deathat State Health Data Center, Pennsylvania Depa rtment of Health. The 1974-78 date are not corrected for incorrect residence reporting on death certifiestest the 1579-63 date are corrected, i
(T/ME 6 DATA ARE DISCUSSED ON FACES 14 AND 15 0F THE TEXT.]
'a TCLE 7: POPULATION AND QBSERVED AN3 E)PECTED AYEMGE AhWUAL NUMERS OF CANCER DEATHS FOR FAIRVIEW AND NEirSERRY TOWNSHIP CENSLE ENUMERATION DISTRICTS WITHIN FIVE MILES OF TMI BY GENERAL DIRECTION FitM TMI, 1979-63 l
l Average Annual Number of Concer Deaths Census Enumeration Districts (CEOs)/
Population Mortality Rafte General Direction fronTHI 1979Ii}
Observed (2)
EspectedI3)
(Observes /Expectoc)
_ m.
N;sberry/Feirview Township CEDs in TNI Five Mlle Area 8.118 10.8 11.6 0.93 1
Northwest Quadrant 3,075 4.0 3.8 1.11 Feirview Township (2 CEDs) 710 1.8 1.0 1.60 Newberry Township (5 CEDs),
2,365 2.2 2.6 0.B5 S:uthmest Quedrent 5,0 43 6.0 8.0 0.B5 Newberry Township (8 CEDs) 5,0 43 6.8 0.0 0.B5
';TE : Iil Enumerated pcpulations,1980 United States Census of Population.
III Observed nurters of concer doeths obtained free death certificates provided by
)
the State Hestth Data Center, Pennsylvente Department of Health. Incerrect residence inforcatien on death certificates has been corrected.
(3) Expected nunbers of concer deaths based on ege-sex distributions of the respective pcpulations,1980 U.S. Census of Population, and ege-sez specific I
concer death rates for Pennsylvente, 1979-61.
i (TABLE 7 OATA A.AE DISCUSSED ON PACES 15 AND 16 0F THE TEXT.)
G
TELE 8: FCPULATION AND OBSERVED AND EXPECTED NtMERS OF 08 AG)CSED CMCCR CASES:
i, ftUR TXI AREA C8VIL DIVIS80NS, FAIRY 3EW AND NETBER;rf TmrNSH8PS AND GOLDSBORO AND YORK HAVEN BORDUGiS, JULY 1982 - JLmE 1984 l
New Concer Casse Retto Ninor Civil Divisions (HCDs)
PopulationIil Observed (2)
Espectedl3)
[0b se'rv e cVEmp acte d]
Hinor Civil Division Total 23,211
,133 1 21. 4 1.10 o
Fairvien/Neuberry Townshipe 21,968 119 113.4 1.05 Feirview Township 11,941 61 65.6 0.93 Neuberr,y Township 10,047 58 47.8 1.21 GoldsborvYork Haven Borcughs 1,223 12 8.0 1.50 Goldsboro Borough 477 4
3.9 1.03 Ycch Haven Borcugh 748 8
4.5 1.78 Unknown 2
H3TES: Ii) Entanerated populations,1983 United States Census of Populatic.
I2) Observed nLebers of diagnosed concer cases, Pennsylvanie Cancer Registry, Per.nsylvanie Department of Health. Incorrect residence inforestion on Cancer
- tegistry Report Fores has been corrected.
(3) Expected nurtbers of concer cases based on the ege-sez distributiens of the respective populations and ege-sex specific incidence rates from the Surveillance, Epideniology, and End Results (SEER) pecgree, 1978-81
[TMLE B OATA ARE DISCt:SSED ON PAUES 17 ANT 18 0F THE TEXT.)
- I t a R
l n
E.
25 1
"8 i
3"!
"N
". B. "* G. i E. R. '. *' i b
- 2. i 2. ". *"! " %"
t' 3
1a 5
e o o,. ew ue cv o cw e ew e e eo.
2.
o
..eu o
e o:
.s a
...c '
.t 8
o
.8 r
o an.
e t
5 N '. '1 ". N. * * * ! *
, cv o o o
,".l.'
l
!I73 2
~
u e.
~,
e e,eeos oe
. Sw t
lo g"-
o u
5
. *8 r----
g 203I s
1 9
t m
ee sn i uane i nun i nee ne s
3 2
n
. a. g e..
,E o
m.
on o
3 1
j 86 !=
L u
-.O o
e E
.C.
w D
4 O
3, es sa o
nae w n to n cn
.e as.
g 3
o -
n a
M. n eto. n. o. id. o coo. s. i cv. i i
- n... e.
o o
.e 81 W 6 m.
. i M. n. n.
m.
6 or.
go g
1 s.
5 g;g a
ocee ano e
eoen nuo o
_g,,,y 3
d, E
e e
o 8
Sg
-vas 3
ec6
=*
2313.0 5
E.
".4,
.I.
oo o.
cv
- = 9 o N " ! " '. N *. N *. ". *1 I. !. 5 8 e 57 w
a n
e cv e o cw c= e ew e io cw e e o ew.-
=
. =
c -
-o t; F.
a
- 3: l' :
E m.
- e. 4
. =.
.o 6
e Ch o.
k
- n..
2" ".. "e f-
.R e
u
- v. mn om ie i i me nn i un m
..=..e
.E.
. Il to e=
o.
.b..av c i o.)
E*
3 e
a
- o. f.
o e o omg kE
%2ou T.
Ft
'1 -
6 & o a o DE
- 32
- s. S. m.
s m o v e. n. e. m.
B.
s o.,
B h."
- 5 u
u a
o n
ne nosnee o.
- s. s.,. s. n. n. e.. no..
o n.
5
.w
-w.
. e o
e ~m
=
2Ji
-s.
o.
o e.
I,
- .8"iB 5*
1 6.
n,
,.e e
&.o a-o u
e.o 21
..e
.E 5
1.o o-.
e e
6 -
6
~.
" " "1 5 % % 5 *. "! N r1 5'!
- 6 r1 "
l g a m E
g Dq R.
w a o w a c s n n u, u m, a m. N ". "!
8' fL 3
%. m 2,
H 8g 3
o a
n e nn
-6 3
a e
35i.
i E
3.=
a 6
e-.e
- o..
.e E
w
=
v.
a u.
o<
e.
n cv e m e v s n s. anvenne n s cw n -
e u.
..e.
c v
e e
oa o<
8 3E*;E-o0
- F L-o o
o
-o.e s
. =.
e.
W
.-v a
S s g203 8 ?,
8,g g3 I
a
- *
- S. "rl % *
- 9
"! ' ** 8. '*. 2. *". "'. *"1
'P k. -
6
- R.
m *. '!
- gga-t "3 &
u g2 ooeeocwooe oe o o cw a cw ew o e
. n e o 85
. o o o %.3 4"
2C E.
e.
o 6
o
.o e.,
. 6 e
.e i.
.e-=
=
o.e
. = c 2 e
5
- n. !
'. o
- o * "! ul * ' u! *. N o * '. *. '! ut
,!F 8[*
g c.
.=
,n,. n c')
CD. c=),to cw g N,o. n N N e c) s.J n
=
6 6.
.g cy e
.E e
.i3 0
.a e b W
o2
- 6 r.,=
3 L
c") o
. n s1 0
44 g g).
3 L.
=
6 E
tj 6
O.
s o
N.
6 u
0 9 6.
t
.a o6 og W
n W N CD f.),= n ~ n e.
j g g fic 4 n cV n CD N CD C
O D r) e e
N
.e g.e.
E A
e.
=.
o 3
C o i va g
g g g O,.*=o 6.
o i
a I.
8 o o T o T.
6
= o e
o e U
l w
a t, l ' 3 4
- t. e
- t..
i w
e 1
=
a 43.
e.
g i
c
= w 6 o-w e E
s f
j e
e
~
n g
E 1=-
~
u.
- 5
=
oa a e
E*..
3 i. egu3*
s.;
oc-r,O n.
p
. g6
~
u e
e
=
g a-e
.a..
c u
a e e -
a e
, i, r
Eif.
- -'I
- ii t
.. r
=
e y
et..
.u t
s.f c
=
...I. t
.p :
t c -
i.
t...
.. c a.
ct uua-.
cw - cc: s c : :
s
=
] =
e a
r
. =
r
]
c p =
a a 1
n 0 s =
4 39 G4 3
1 f rt
[
e a_
85 e s d
0 I. 9 7 e
a s i
6 E =
p t / -
i e. -
1 0 1 1 0 1
0 t s e e t
ct e h
R e -
e e r n c
n ol s
b -
D t b o e
i n p i
h r
i n
0 =
w
[
=
s s p r
cet o
=
t m
o i r l T
=
s r e y c
f e u p e mL n
i s m 4
]
i c e I
c c.
3 L.
4241 54754 4 8
y h nf s e] d 9
r 5
0 1 4 1 7 8 0 1 0 1 s o p*
n R1 r
p 2
n cd 'n s( a e
U e
s w
ni h
OE b
E o c ak t
e e,
F N w
i g
l t ed U
e T
f
] d a
l t l 3J N
)
m yi i o e
at o 2 _
E 8
- 1 7 - 9 8 1 r c(
H e r T -
n y
2 r e f
e gh C
e e p n E2 b
L9 D
b s o
.o o
e n t m
l n
- i E 9 e r o t
e w,
S1 M o cd n
g o e f
n e
el u AY
]
L d
e a m
l g
N U p,
n s h t
d o e nf h I J o s =
3 3 1 3 8 00 8 3
a r t
)
4 e
3 e
a p
p t E =
0
- 6 36 7 20 3 -
5 wb f (
p e o E e i
t /.
=
=
1 0 1 0 0 1 1 2 0
e m o e
) h s T S h
e i u t d D
0 T e I P s
R e =
SI n
b =
v n e s e 0
r r a n e
9 t
w 0 =
)
Y O i d e ei t
1
. a o
[
=
a e c r m n
( 1 d R
T T
P A W F t nb r e
8 a N O w
I c o e
v e9 s X
E d e c et t
n1 u 3 -
9 681 8 2201 G. 9 I T e
n p t
e y
o -
T R
i 4
0 1 6 1 0 9 1 20 1 a s e ed e
t 8 e PY v
p E
3 1
ed m n
t 7 r H
1 r
s u
i E =
s u e e n
e9 e Yf T
BE a
h el rb e
t 1 -
u F
Dh c P
u F
u t n
,t p, d 3w o
i]
o o e o 0
)
2 E
I 8
- 1 8 1 8 1 1 5 - 1 r e 1 n yd p ei S N 2
1 1
o s s[
r e r r A
s 3
CD b
B u e d
t t t g e N
O a s nl e c c o p D
RA n c a o u l
e e r NA E
e e ct o g r j P 7 C W vb o c e r b
7 N E
)
a o c R o u] -
0 A I H e r
)
c sR 3 2
l o e n r
E 7 CV p
4 1 37 8 9 9 4 1
k b b h o e n eE 9 R
p o x r a st l c e h S 1 743 9 059 6
CI o
t E 0
9 ot d o
n e t (
I A T
t /.
1 01 0 0 1 0 1 0
Y l f c ob e
1 NF e
X y
R e -
e oo(
C f sh S
R r
b =
d h G e
ot t E
XJ r
1 =
nt e e e e l
I T e
(
a d r s t h nu r GA n e e a n
o s o D
b A A w
oit c c e s i e f P
r r n v m t R RT e
]
N A
N 3
9 0857 787603 od o o r t r u
s b e p c e y o b d t O
/
p 9
1 20 2 7 5 1 31 3 s t e c
eF i n e s
ESE e
s 5
1 1 1 d n r e n n
rE d D
s D T l
E l
e d o nt t
ES e r sd,
o5 c N A v
o s u
P o in s S
C R r
G e t
e p
d a u U
r A A l
I C
s p n n
, e
,h S
I P e
I.
2
- 251 7 7 1 7 -
e eio sR n
t DE F
I S
s 6
1 1 1 d t h a
e y D
F b
u oT sf s y e g,
l n
e o o r
- o1 OH O
E c
s ct el 0 R
T n e
.a e
s g o -
GI i r e c c ri et 8 A
W F
a n n
e g a7 A
E
]
a o n e c e n e 9 M1A p
e e e ed nR od 1 TA U E o s 9
7 8 3 0 7 G 1 f.
r h l v
f o
i a g r a e c r d p r D
0
- 6 53 925 1 -
5 N R t E u eH e e
eE o A
t /.
1 01 0 0 1 02 0
n o p r
d c n
f 0
D e
1 E N R e o r p k e n e,
a r f c o b e t t
o r E
T O i
b CI s
0 eD c o o oC c r L
Y n
s n o l
E S s
[
v nl n
g n e e p B
PI e
A i e ud o a o st e XV r
d v f el i
al r T
E I A
I
(
a g t o d n cl 040 88 881 8 D
3 2
1 l H n r t o
e R 3
L NL L
p 4
1 3 1 3 8 6 1 3 1 3 i
i o v f i r v E AI A
s 8
1 1 1 v k n pi ot or E V
E i r e ed a
uS DI c o e r s m eS Y s t
r r r
e E C r
e i
e o e e h v
)
=
od r v v
b f b h t f
R 2
=
f
=
0
- 2 8 1 7 1 1 E - 2 n n oi i
a n m t E
)
s m
ef f c
i i i
n a
i S
s -
7 1
1 2 t
9 I b
m p
n mi ot e r e
o 0 H O
r r l h o d c d r d.
u o e T n e n ef e e ob m i
v e t
d m 8
f s s m
rd c s u o 0
d el e el t rT 1
e l o) i-e a p t c o s
e h o o1 h b e s a n y E.
s G t
(
T D r E ri m x_
I t
i y
M t
I I
)
I 2
3 f
T p
I
(
m*
y e
)
L =
e n
S t
st E
a e
T
'n a
- i. m c
i 1y O
t N
r k
r -
=.
I
- p a a a.
t.
, i t
v -
p h o
t
. r -
e
. a r
n e
nt -
i., -
.- r p -.
- e. -.-.
s wt a.
y t
s
-i -
e n i t
t t
t' t
g t
t t
.j l
1 i{
I!
t
i s
AN ASSESSMENT OF A CANCER MORTALITY SDRVEY IN THREE TMI AREAS SELECTED BY LOCAL RESIDENTS DIVISION OF EPIDEMIOLOGY RESEARCH PENNSYLVANIA DEPARTMENT OF HEALTH S ept ert.ber, 19 85 1
b A
V O ( U / J ()+ 7,
i.
2 *.
BACKGROUND The results of a health survey conducted by a group of local residents were made public on June 21, 1984, with the release of a document, Aamodt Motions for Investication of Licensee's Reports of Radioactive Rel' eases,During the Initial Days of the TMI-2 Accident and Postpo'nement of R(start Decision Pendino Resolution of This Investigation. (1)
Three separate areas were selected for their survey.
The 6cor-to-door survey of the areas was conducted by a group of citizens who volunteered to interview local residents, but little information was provided in the report regarding quality control or the study design.
One of the areas was reported to be six miles northwest of the plant and another, three and one-half
. : to the southwest.
The third area, seven miles northwest of the plant, was reportedly chosen because of its high elevation and clear view of the TMI plant.
However, the specific 1ccations of the three areas are not provided in the report.
The demographic and cancer mortality data in the report are very limited.
No information regarding cancer site or dates of diagnosis and/or death is provided.
According to the text of the raport, however, the overall cancer mortality rate for the areas selected for the "five year period" since the accident is 5.2 to j
6.5 times higher than expected.
Following its release, the Centers for Disease Control (CDC), United States Public Health Service, reviewed the document at the request of the U.S. Nuclear Regulatory Commission.
In its
4 September 1984 response, CDC noted a number of epidemiologic deficiencies in the data presented and concluded that the survey did not present "convincing evidence of cancer incidence, cancer mortality, or adverse pregnancy outcome in TMI-area re'sidents following the accident."
In the meantime, the Pennsylvania Department of Health, which also reviewed the document, obtained some additional sum-mary surysy data.
This data permitted a more comprehensive evaluation of the data included in the document.
This evaluation cast additiona.1 doubt on the accuracy and utility of the survey.
It fcund not only a lack of information about the cancer deaths, but an actual difference in the numbers.
In one case, for example, nine deaths were reported, but eight were included on summary sheets.
Moreover, the only information reported for five of these eight deaths was as follows:
"not interviewed--cancer--
i died."
Concurrent with its review of the available survey data, the Pennsylvania Department of Health analyzed cancer mortality data for four York County minor civil, divisions in the vicinity of the three survey areas in question.
These analyses did not indicate the existence of an apparent cancer problem, but criti-cal evaluation of the allegations was not possible because the necessary detailed information, including the precise location of the survey areas, was not available.
The original survey data compiled by local residents indicated there was a total of 20 e
e
- -,.. ~
%4 cancer deaths in the three survey areas during the "five year period 1979-1984",(1) dates which actually cover a six-year period.
In tjovember, 1984, the Pennsylvania Department of Health
~
was requested by the Advisory Panel for Decontamination of TMI 2 to further evaluate cancer morbidity and mortality in the three We were able to ascertain more inform'ation regarding the areas.
reported cancer cases by directly contacting those responsible for reporting the residents' survey results.
The additional information they provided included the family names and the number of occupante in each household they surveye.d, information on the streets and roads surveyed, and th'e general boundaries of the survey areas.
However, the authors declined to provide other needed information, such as the age and sex of participants, health history, individual residential addresses, and length of residence, citing confidentiality as their reason.
The additional information did, however, enable the Pennsylvania Department of Health to initiate a more thorough assessment of cancer mortality in the same survey areas.
II.
METRODS The assessment of mortality in the three survey areas was complicated by the fact that the areas had no precise geographic boundaries and, being primarily rural areas, street addresses were nonexistent.
Fortunately, Area 2, the largest of the three areas, was included in a special 1979 TMI Census of Population which included all households within an approximate five-mile
i radius of the nuclear facility.
About one-half of Area 1, the second largest of the areas, also was included in the TMI Census.
Only Area 3, which included fewer than 20 households was
~
beyond the five-mile limit.
Thus, information on the resident population at the time of the accident could be produced for the majority of the persons included in the, subject survey areas.
Information on the populations outside the TMI Census area was obtained from a school census conducted in 1979 by the West shore School District and cross reference directories.
In order-to use the TMI and school census data, however, it was first necessary for us to delinente precisely the boundaries of the subject survey areas.
This required field visits to the three areas because (a) the Pennsylvania Department of Health was civen family names only, and these generally for those households contacted by local interviewers, and (b) the readily available residence information from the TMI Census, particularly for persons residing in Newberry Township, did not provide street addresses which would have facilitated the delineation of the '
survey areas.
During the field visits, housing unit counts were made, residents' names and Post Office Box numbers obtained, and maps developed to show area streets and the location of housing units.
On the basis of information obtained during the repeated field trips by Health Department personnel, Survey Area 2 was redefined and expanded because it was clear that not all appro-priate area streets were included in the original survey.
The data obtained in the visits to the three survey areas were W
O
o checked against and supplemented by data from the TMI and school censuses, cross reference directories, death certificates, and Cancer Registry files.
In addition, questionnaires were sent to physicians-who treated area cancer decedents to obtain additional diagnostic information.
These were followed up by telephone calls when necessary.
III.
RESULTS A.
DEMOGRAPHIC CHARACTERISTICS 1.
Minor Civil Divisions: MCDs All three of the areas surveyed by local residents are located within York County minor civil divisions; Areas 1 and 3 in Fairview Township and Area 2 in Newberry Township.
Both of these MCDs are situated to the west of the Susquehanna River.
More specifically, Fairview Township is predominantly west-northwest of Three Mile Island (TMI), while Newberry Township extends to the north and south of TMI.
The populations of both townships, Newberry in particular, increased substantially during the decade of the 1970s, an important consideration in evaluating time trends in morbidity and mortality.
The population of Newberry Township increased by 68.1 percent, from 5,978 in 1970 to 10,047 in 1980, while Fairview Town-ship increased by 29.1 percent, from 9,278 to 11,941.
The 1982 population estimates for Newberry and Fairview Townships are 10,275 and 12,337 rerpectively.
.c.
I e
The median ages of these predominantly white popu-lations (98.7 percent) were lower than the State median age of 32.1 years:
Fairview Township, 30.4 years and
_Newberry Township, 27.8 years.
2.
TMI Census Enumeration Districts: CEDs
~
Five TMI Census Enumeration Districts (CEDs) in Fairview and Newberry Townships generally correspond to the survey areas in question.
They had a total popula-tion.of 2,885 in 1979.
3.
Survey Areas Selected by Local Residents
}
Survey Area 1, about 4.5 to 6 miles northwest of TMI in Fairview Township, consists primarily of house-holds along a single rural road with a few housing units on connecting roads also being included.
Thirty-five households and an estimated 118 persons were reported in the survey by local residents for Area 1 (TABLE 1).
The Pennsylvania Department of Health esti-mates for Area 1, based on 1979 TMI and school census data supplemented by 1985 field visits, are 41 housing units and 140 persons (TABLE 2).
Survey Area 3, also in Fairview Township, but about seven miles north-northwest of TMI, likewise consists of a single road.
The count of households made by local residents and the estimate of total persons in Area 3 were 15 and 51, respectively (TABLE 1).
The corresponding Pennsylvania Deptr:.ent cf Hitlth fi?urer, based pred:cinan-13
- c..
.c-
1979 school census data supplemented by field visits are 17 housing units and 75 residents (TABLE 2).
It was in Area 2 (Newberry Township), however, where the greatest discrepancies were found.
During the course of field visits.to the area, it became evident that (a) the stated boundaries of the area did not coincide with the area actually surveyed, and (b) there were a number of streets in the immediate area which presumably should have been included in the survey but which were not.
Specifically, of about 14 streets / roads in the area which might have been sur-veyed by the residents' group, only four, which report-edly included 93 houses and an estimated 288 persons (TABLE 1), were stated to be in Area 2.
There was at least one cancer death on each of the four streets selected by local residents in Survey Area 2, but none on the streets not selected for the survey.
These no-cancer streets included 91 housing units with an esti-mated 325 persons in 1979.
The Pennsylvania Department of Health estimates for Area 2, thus, included a total of 201 housing units and 673 persons (TABLE 2).
These counts excluded housing units and persons not present at the time of the TMI accident.
Specifically, there were five housing units on one street, three of which were included in the June 1979 TMI Census, but unoccu-pied at the time of the TMI accident (Ma r ch, 19 7 9 ),
which ve:t ne: inclu6ed ir ou: analysir.
P B.
CAMCER MORTALITY IN THREE SELECTED AREAS The citizens' group reported that its data showed cancer mortality in the three survey areas to be 5.2 to 6.5 times higher than expected.
However, in view of (a) the results of the routine Pennsylvania Department of Health cancer mortality monitoring in the TMI area, (b) a special evaluation of cancer mortality data for the survey area MCDs, and (c) the obvious limitations of the survey con-ducted by local residents, such a contention seemed unten-able.
After proper adjustments were made for incorrect residences reported on death certificates, the observed number of cancer deaths for Fairview and Newberry Townships t
for the post-TMI period,144, was remarkably close to the expected number (age-sex adjusted) of 142.0 (TABLE 3).
Mortality data for 1984 were incomplete, thus they were not i
analyzed at the time our report was prepared.
Although the survey conducted by local residents resulted in a substantial under, count of the base population and a crude method, without adjustment for age-sex vari-ations, of estimating expected cancer deaths was used, their overall expected number of cancer deaths in the three survey areas (4.9) for the "5-year" post-TMI period was not as different from the Department of Health's expected number (6.3) as might have been anticipated.
The use of a rather high overall cancer death rate for Pennsylvania for 1979 as "standard", in comparison with the relatively e
m
__. _.. ~. _ _, _ _ _
s.
p younger population of the survey areas partially compen-sated for an under-estimation of the number of area residents.
~
+ tile the expected number of cancer deaths in the three survey areas was understated by'the local residents, the observed numbers were overstated.
First, the expected numbers presented by the residents' group are for a "five-year" (1979-83) period.
The 20 originally reported deaths, presumably for the 1979-83 period, however, included two who died in 1984; these should have been excluded.
- Second, one of the decedents died in 1978, i.e.,
prior to the March 1979 accident at TMI, and another person who died of a cause other than cancer was apparently confused with a relative who died of cancer prior to the TMI accident.
Third, two of the decedents were no longer residents of the area at the time of death, thus should not be included in this cross-sectional analysis.
However, because this is a cross-sectional mortality study, two decedents who were not survey-area residents at the time of the TMI accident are included in this analysis.
In the course of reviewing death certificate data for the Fairview/Newberry Township area, we located one survey area decedent who had cancer at death, but who had not been previously reported by local residents.
This decedent is included, as should be, in the TABLE 3 data compiled by the State Health Department.
.c.
1 The figures used by the Pennsylvania Department of j
e Health and those used by the local residents' survey group regarding the observed number of cancer deaths differ at
\\
times.
For example, four cancer deaths were repbrted in the residents' survey of Area 3; but our data (1979-83) show two.
One of the four was the previously reported non-cancer decedent apparently confused with a relative; i
and the second was a 1984 death.
While the residents' group claimed a 7.2 to 8.3 fold excess in cancer mortality for this,particular area, our data show two deaths in comparison with the expected 0.5.
Moreover, one of these two decedents was not a survey area resident at the time of the TMI accident.
According to our data, the age-sex adjusted (indirect method) expected number of cancer deaths for the modified, I
combined survey area was 6.3 for the 1979-83 period, as
^
compared with 15 observed (ZhBLE 3).
However, these 15 cancer deaths, as listed in TABLE 4 according to year of death and cancer site, include six decedents whose diag-nosis was made before the accident; one decedent who was not a survey area resident at the time of the accident; one decedent who was diagnosed within months of the accident as having a cancer which usually has a long latency period; and two lung cancer decedents who were long-term heavy smokers.
While five cancer cases were diagnosed in 1981, it should be noted that (1) three and possibly four cases were alt ciagn Eed ir 197E, Frier t: the accider,t; and C. tw;
-2 0-9
o of the five diagnosed in 1981 were long-term smokers who died of lung cancer.
It should also be noted that bias which already was introduced in the initial selection of the"three specific areas because of apparent knowledge of cancer by local residents' remains.
Thus, the results of subsequent statistical analyses' based on such biased data are invalid.
The observed and expected numbers of cancer deaths for the five combined TMI Census Enumeration Districts (CEDs) within fairview and Newberry Townships which encompass the three survey areas are 30 and 21.3 (TABLE 3).
The mortality ratio (1.41) is within a range of random variation.
IV.
DISCUSSION, EVALUATION AND CRITIODE Several deficiencies are inherent in the 1984 cancer survey conducted by local residents.
These are summarized in the nine points which follow:
1.
The study was based only on cancer mortality rather than cancer incidence data.
Mortality data have limited value when causal relationships are to be established between environmental exposure and cancer.
Many cancer patients live many years after initial diagnosis.
Reliable morbidity data, such as case incidence, would have been much more useful for this type of investigation.
2.
An insufficient latency peried had elapsed since the accident ::,ectify be conclusien reachef ir the
- curvey, Mer: form c' n.clienant; heve icne it : e r. )
i J
A e
periods, depending upon the type of cancer and the l
nature of-the environmental exposure.
In some cases, as long as 20 to 30 years may be required before the
-cancer manifests itself.
In the case of radiation-induced cancer, leukemia is known to have a relatively short latency with high dose exposure but, even in leukemia, the peak incidence may not be observed until five or six years after exposure.
Furthermore, many such patients can survive a number of years following the initial diagnosis.
Other forms of cancer induced by radiation may not be observable for 10 to 15 years or more after exposure.
Thus, one should not expect to observe any significant increase in cancer within a few years, even if radiation doses were much higher than previously thought.
3.
From the experience gained by observing the Japanese atomic bomb survivors and through other epidemiologic studies, certain specific forms of malignancies can be expected to occur following exposure to high-dose radiation.
Therefore, one might look for such specific types of cancer as leukemia, in relation to radiation.
l When many different types of cancer are observed, as is the case with the reported cancer deaths, it suggests an absence of a single causal relationship.
4.
The study conducted by the local r'esidents made a conclusion that a particular cause and effect rela-tienship had beer feund without ruling out other 12-
,,,.,,,-,,,_,,._.---y
w 4
possible causes.
Cancer is a group of diseases that can be caused by one or.:;re of a variety of environ-mental and genetic factors such as diet, tobacco, Jmic r o-organisins, radiation, food additives, c'ecupa-
~
tional/ industrial exposures, host susceptibility, etc.
Because of the complexity of cancer etiology, one should not draw quick conclusions about cause and effect relationships.
If higher mortality rates are found, it is tempting to attribute these to those potential causative agents that would seem to be most apparent in a given place or time.
The radiation releases from TMI are a case in point.
However, before one can make a responsible judgment about causation for a given form of cancer, it is necessary to search for and rule out other possible causes.
5.
The expected numbers of cancer deaths calculated by the local residents were based only on total population figures and overall cancer death rates for the stan-dard.
They did not take into account either the age or sex distributions of the populations being studied or the fact that cancer death rates' generally increase with age.
Thus, if the distribution of the population under study is quite different from that of the stan-dard, an expected number based on only total population and overall death rate may be totally inappropriate.
6.
For the purpose of establishing a causal relationship between any chronic disease with long latency such as _.
1 1
{
r cancer and radiation, crcss-sectional mortality and morbidity data at a given time in a given population are not completely adequate because:
- (a)
Some of the people originally exposed to the TMI related radiation have moved out of the area and some of them may have developed cancer or died elsewhere from it.
These cares would be lost from cross-sectional studier, but they should be included.
(b)" In contrast, cross-sectional population data do include a certain number of people who moved into the area after the accident but who may have developed car.:er before moving into the area and died from it at the new address.
Usually, these in-migrants are not identified in cross-sectional data, thus they are bound to be included in such studies, but they should not be.
(c)
It would be more appropriate to conduct a long-term follow-up study of the originally exposed population (cohort) regardless of the current address.
7.
Those who conduct any field investigation of this type should be aware that small area statistics are highly unstable and statistical significance tests applied to such data are difficult to interpret.
When small com-munity areas are considered, mortality and morbidity i
rates can fluctuate parkedly from one area to another.
i.
cnd from one year to another without any substantive
,f epidemiologic importance.
8.
It has been amply demonstrated and documented (3) tha-
_ "cancer clusters" are frequently observed in communith
~
settings such as in churches, schools, factories, or along certain streets.
This is because cancer is a common disease in the United States, with the life-time incidence of one case per every four to five people.
With such a high frequency, it is not at all difficult i
to, observe apparent "clustering" depending upon how the geographic boundary is delineated or how the small area population at risk is identified and selected.
i 9.
The most important and serious defect in the survey conducted by local residents is the rather obvious i
selection bias which was introduced in their data l
collection by the inclusion of only selected specific geographic areas (streets / roads), households, and individual residents in the survey.
There is evidence that such selection was influenced by the pre-existing knowledge of cancer deaths.
The statistical signifi-cance tests performed in any epidemiological studies are appropriate oniv if the selection of the study area (population) is uninfluenced by such bias.
Otherwise, the results of such tests are invalid and cannot be 1
t :epted.
While the authors of this small area survey claimed that cancer mortality has markedly increased around TF.:, and impli-
-1!-
.t.
cated the 1979 nuclear accident at TMI as being -responsible, the data presented in their survey do not support their conclusions for the reasons discussed in this critique.
Furthermore, the comprehe6sive epidemiologic studies of cancer mortality and morbidity conducted by the Pennsylvania Department of Health do not provide evidence that cancer mort'ality has increased significantly around TMI.
This would include the four selected downwind MCDs and the specific smaller areas within these York County MCDs where special in-depth analyses were performed.
k
~2 E-i
- - - -. - - - - - - - - - - - - ~, - - - -
REFERENCES h
(1)Aamodt, M.M. and Aamodt, N.O. Aamodt motions for investigation of licensee's reports of radioactive releases during the initial days of the TMI-2 accident and postponement of' restart decision pending resolution of this investigation.
Presented to United States Nuclear Regulatory Commission, June 21, 1984.
(2) National Center for Health Statistics: Advance Report, final mortality statistics, 1979. Monthly Vital Statistics Report, Vol. 31--No. 6, Supp. DHHS Pub. No. (PHS) 82-1120, Public Health Service, Hyattsville, MD. September 1982.
(3) Glass, A.G.; Hill, J.A.; and Miller, R.W. Significance of Leukemia Clusters. J. of Pediatrics 73, 1968.
i
~-
,g, T/JLE is ESTIETED N'MIEM OF ICUSING UNIT 3 AND POPUL.ATIONS IN THREE AREAS (STREITVICADS) SEl.fCTED BY L.DOAL RESIDENTS POR CANOER SUWEY MOUND T on w
survey A rea s (Streets /Acads]
Housin3 Units Population Estimates
~
Caebined Area (1, 2 8 3) 143 457 Area 1 35 118 Area 2 93 288 Area 3 15 51 MTES: Source of housing and reefdent date: Aerodt Motions for Investigation of Licensee's Reports of.Bedicactive Releases During the Initiet Ceys of the TMI-2 Accident and Postponeeent of Restert Decision Resolution of this Investigetton, June 21, 1984 I
)
I (TMLE i CATA ARE CISCUSSED CN PAGES 6 AND 7 0F TNE TEXT.)
{
t d
4
,7
,g, TABLE 2: ESTIKATII) WheEFE OF N0tSING UNITS AND POPULATICMS IN THREE AREAS (STREETS /
ROADS) SELECTED BY LDCAL RES8 CENTS FOR CANOER SURVEY A70VND TMI, AtVUSTED BY PDON FOR CMIS$10NS IN INITIAL SURVEY j
t n
Adjusted Survey Areas (S t re et s/ Roe de)
Estimated Housing Units Population Estientes l
Cortined Areas (i, 2 & 3) 259 gag 1
Ares 1 4g 343 Area 2 201(a) g [,)
Area 2
$7 NOTES: Sources of Pennsylvente Department of Hastth (PDOH) dets: The Three Hite Island (TMI) Pcpulation Re9 stry (1979 TMI Census) and/or a 1979 West Shore School District 1
Census supplemented by field visits to the survey areas by PDOM's staff.
I'I nittet survey conducted by local residents included only 4 streets /rceds, but did I
not include 10 other streetvreeds in the same general eras. There were no concer deaths on any of these ceitted strestvreeds during the 1979-63 period. There were five housing unite on one eres street, three of which were inctuced in the 1979 TMI Census, but unoccupied at the time of the TMI accident they are not included in this Tstle.
(TASLE 2 CATA AAE DISOUSSED ON PAGES 6 AhD 7 CF THE TEXT.)
\\
4'.
TABLE 3: PDPVLATIDN EST3KATES AND DSSEWED MD EXPECTED NtNBERS OF AGE-GEX AA CMCER DEATMS: SELECTEQ'Aa>USTED TMI AREAS, 1979-1983 Center Deaths Mortality Retio Geographic Areet PopulationI*I Dbserved(b)
Espected(c)
(Dbe./Exp.)
Feirvie dNo berry Townships (2 MODS) 21,988 144 142.0 1.01 TNI Census Districts (5 CEDs) 2.885 30 21.3 1.41 Adjusted Survey Areas (1, 2 & 3) 888 15 '
6.3 2.36' (a) Minor civit division (MOD) populations,1983 U.S. Census of Pcputationi Census NDTES:
Enumeretten District (CED) populations,1979 TKI Censust Areas (street-/rceds) inistetty selected by local residents for concer survey, which have been adjusted by State Health Departeent according to the 1979 TMI Census and 1979 School District Census.
(b)Dbserved number of concer doethe, Pennsylvente Department of Hestth vitet records. Numbers shown for H00s may not match routinely reported numbers because of residence corrections. Dbserved numbers do not include two forrer residents of the Survey Areas who were not residents of either Fairvies or Nesberry Townships et the time of dreth. Conversely, two Survey Area residents who were not residents of the eres et the tire of the TNI accident but who were residents et the time of death are included.
ICI spected ntribers of concer deaths for the post-TMI period (1979-1983)
E were based on estimates of the age-sex distributten of the respective populations and ogs-sex specirte everege annual concer death rates for Pennsylvente, 1979-81 used as stender'd (Indirect Method of Standardization).
'P <D.01 i
l (TMLE 3 DATA ARE DISCUSSED DN PAGES B - 11 DF THE TEXT.)
.9 o
TABLE 4:
C4CER DEATHS BY SITY, YEAR OF DEATH, M YEM OF DIAGOSIS:
AaJLSTTD i
SURYEY ARTAS, 1979-63 O
Yaer of
' Yea r of Death Cancer Site Diagnosta Romerke 1979 Breset
~
1969 1980 Leukonto 1976 i
Lyaphoes, tener opino 1977-78 Colon 1979 Prostete Un known 1981 Breast 1978 Pecetate 1978 Ne tanone, metastetic 1978 Overy Unkneen Death certificates onset one year prior to death.
Lung (oet cett) -
1981 Decedent sacked 30 cigarettee per day for 15 years.
iSB2 Adrenet, NCS 1981 Nottiple eyelome 1981
~~
' ' ' Lung 1981 Decodent seeked 30 cfgerettee per day for 30 years.
ketestetic (prisery unknown) 1981 Leukemia Unknown Decedent ses not a survey eres reefdent et the time of the THI occident.
NUTE: Sources of date includes death certificates, responese to e physician survey, and 1979 THI Census information.
i I
(TABLE 4 DATA ARE DISCUSSED DN PAGES 10 AND 11 DF THE TEXT.)
89 y
,.-yV
'-- ' ~