Environmental Research 111 (2011) 12301235

Contents lists available at ScienceDirect

Environmental Research
journal homepage: www.elsevier.com/locate/envres

Cancer incidence among residents of the Three Mile Island accident

area: 19821995$
Yueh-Ying Han a, Ada O. Youk b, Howell Sasser c, Evelyn O. Talbott d,n
a

Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA
Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA
c
Department of Epidemiology and Community Health, School of Health Sciences and Practice, New York Medical College, Valhalla, NY, USA
d
Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 17 June 2010
Received in revised form
5 May 2011
Accepted 4 August 2011

Background: The Pennsylvania Department of Health established a registry of the Three Mile Island
(TMI) nuclear power plant accident in 1979. Over 93% of the population present on the day of the
accident within a 5-mile radius was enrolled and interviewed. We used the registry to investigate the
potential cancer risk from low-dose radiation exposure among the TMI population.
Methods: Cancer incidence data among the TMI cohort were available from 1982 to 1995. Because
more than 97% of the population were white and few cancer cases were reported for those younger
than 18 years of age, we included whites of age 18 years and older (10,446 men and 11,048 women) for
further analyses. Cox regression models were used to estimate the relative risk (RR) per 0.1 m Sv and
95% condent interval (CI) of cancer by radiation-related exposures. The cancers of interest were all
malignant neoplasms, cancer of bronchus, trachea, and lung, cancer of lymphatic and hematopoietic
tissues, leukemia, and female breast.
Results: Among men and women, there was no evidence of an increased risk for all malignant
neoplasms among the TMI cohort exposed to higher maximum and likely g radiation (RR 1.00, 95%
CI 0.97, 1.01 and RR 0.99, 95% CI 0.94, 1.03, respectively) after adjusting for age, gender, education,
smoking, and background radiation. Elevation in risk was noted for cancer of the bronchus, trachea, and
lung in relation to higher background radiation exposure (RR 1.45, 95% CI 1.022.05 at 8.08.8 mR/h
compared to 5.27.2 mR/h). An increased risk of leukemia was found among men exposed to higher
maximum and likely g radiation related to TMI exposure during the ten days following the accident
(RR 1.15, 95% CI 1.04, 1.29 and RR 1.36, 95% CI 1.08, 1.71, respectively). This relationship was not
found in women.
Conclusion: Increased cancer risks from low-level radiation exposure within the TMI cohort were small
and mostly statistically non-signicant. However, additional follow-up on this population is warranted,
especially to explore the increased risk of leukemia found in men.
& 2011 Elsevier Inc. All rights reserved.

Keywords:
Three Mile Island
Neoplasm
Incidence
Leukemia
Low-dose radiation

1. Introduction
The Three Mile Island (TMI) accident was a partial core meltdown of the Three Mile Island nuclear generating station on
March 28, 1979 in Dauphin County, Pennsylvania. The nuclear

Abbreviations: CI, condence interval; CLL, chronic lymphocytic leukemia;

OR, odds ratio; RR, relative risk; TMI, Three Mile Island; UNSCEAR, United Nations
Scientic Committee on the Effects of Atomic Radiation
$
Funding: This work was supported by a grant from the Three Mile Island Public
Health Fund. No competing interests are declared.
n
Corresponding author at: Department of Epidemiology, A526 Crabtree Hall,
University of Pittsburgh, Pittsburgh, PA 15261. Fax: 412 624 7397.
E-mail addresses: hany2@upmc.edu (Y.-Y. Han), ayouk@pitt.edu (A.O. Youk),
howell_sasser@nymc.edu (H. Sasser), eot1@pitt.edu (E.O. Talbott).
0013-9351/$ - see front matter & 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.envres.2011.08.005

power plant released about 370 P Bq of noble gases, mainly 133Xe

and 550 G Bq of 131I were released into the atmosphere
(UNSCEAR, 2011). Based on residential location and the amount
of time spent within a 5-mile radius area during the 10 days after
the accident, Gur et al. (1983) estimated that the average maximum and likely whole-body g-doses for individuals in this area
were 0.25 m Sv and 0.09 m Sv, respectively.
Although estimated radiation doses from the incident were
low, the Pennsylvania Department of Health created the TMI
population registry, which includes more than 35,000 residents
living near the TMI facility within two months after the accident
in order to monitor the possible health effects from low-level
short-term exposure to TMI radiation emissions. A 13-year
(Talbott et al., 2000) and a 20-year (Talbott et al., 2003) mortality
assessment of the TMI cohort have been reported. Based on the

Y.-Y. Han et al. / Environmental Research 111 (2011) 12301235

20-year follow-up, 19791998, maximum g whole-body exposure

was found to be signicantly related to cancer mortality of
lymphatic and hematopoietic tissues in men and breast cancer
in women (Talbott et al., 2003). Several studies have also
investigated radiation-related health consequences of the TMI
accident. Based on the TMI cohort, the standardized incidence ratio
for cancer indicated that the 6-year (19821988) age/gender
adjusted cancer incidence among those exposed to radiation and
psychological stress from the TMI accident was not different from
the entire state (Ramaswamy et al., 1991). A separate study that
ascertained cancer cases before and after the TMI accident among
residents living within 10 miles of the TMI plant by reviewing the
medical records at all local and regional referral hospitals during the
period of 19751985 was conducted by Hatch et al. (1991). Adjusted
for urbanization and social class, they reported that overall postaccident cancer incidence rates were higher among those living
within 10 miles of the TMI plant compared to those living farther
away (odds ratio, OR1.2, 95% condence interval, 95% CI1.0, 1.4).
Using a similar study protocol, the estimated percent increase in the
post-TMI accident cancer incidence rate (19751985) was 3.4% for all
cancers, 10.3% for lung cancer, and 13.9% for leukemia, considering a
two-year latency and adjusting for age, gender, and socioeconomic
characteristics (Wing et al., 1997).
The biological effects of low levels of radiation exposure have
been investigated within different populations. The epidemiologic
data suggest an increased cancer risk for a protracted exposure
above 50 m Sv (Brenner et al., 2003). Preston et al. reported a
linear increase in excess solid cancer risk in radiation exposures
5 m Sv and above among survivors of the atomic bombings in
Japan (Preston et al., 2008). They also found that both those
exposed in utero and in early childhood exhibited signicant
dose-related increases in incidence rates of solid cancers
(Preston et al., 2008). The seventh Biologic Effects of Ionizing
Radiation Report supports a linear-no-threshold risk model in
which cancer risk proceeds in a linear fashion to lower doses
without a threshold and that the smallest dose has the potential
to cause increased cancer risk, although the risk is small
(Committee on the Biological Effects of Ionizing Radiation, 2006).
Compared to the average annual radiation dose of 6.2 mSv
among the US population (NCRP, 2009), the radiation exposure
from the TMI accident was considered minimal. However, the
long-term health effects from low-dose radiation exposure
remain a public concern and a report on long-term cancer
occurrence is absent for the TMI cohort as the follow-up was
discontinued in 1998. Due to this unique opportunity, we report
13 years of follow-up data (19821995) on the overall and sitespecic cancer incidence rates of the TMI cohort to assess the
carcinogenic effects of the accident and to determine whether
continued follow-up of this population is necessary.
2. Materials and methods
2.1. Data collection
The Pennsylvania Department of Health established the TMI population
registry by enrolling and interviewing those present on the day of the accident
within a 5-mile radius. A total of 35,946 residents, over 93% of the population,
were initially enrolled in the TMI registry. Individual information on residential
history, demographic characteristics, medical history, occupation, smoking status,
and previous radiation exposures (medical treatment and occupational) was
collected. The nal TMI population registry consisted of 32,135 individuals after
excluding those (1) with unknown g- and b-exposure levels; (2) living outside of
the 5-mile radius; (3) born after March 28, 1979; (4) with unknown residential
history; (5) who established residency after March 28, 1979; and (6) who were
duplicated in the cohort. The TMI cohort was linked to the Pennsylvania cancer
registry to identify newly diagnosed cancer cases within the cohort. To maintain
the follow-up capabilities of the registry, the Pennsylvania Department of Health,
along with U.S. Postal Service, annually obtained the current address of persons in

1231

the registry. The TMI population registry was matched yearly against the cancer
registry les maintained by the Pennsylvania Department of Health to identify
those cohort members not previously diagnosed with cancer. Cancer incidence
data were available from 1982, when the statewide Pennsylvania cancer registry
started, through 1995, the most recent cancer data available. A total of 529
individuals were lost to follow-up (1.6%) by the end of 1996 when follow-up of the
TMI population was completed.
The natural environmental background exposure (excluding TMI radioactivity
releases) was based on a direct measurement recorded with a scintillation detector
and associated instrumentation from a 1976 airborne radon survey. Quartiles of
natural background radiation exposure dened as low (5.77.2 mR/h), low/medium
(7.37.9 mR/h), medium/high (8.08.7 mR/h), and high (8.8 Z10.5 mR/h) were
originally assigned to individuals residing within a 10-mile radius of TMI facility
for the analysis by Hatch et al. (1991), an area that covers the 5-mile radius of the
present TMI cohort. The natural background exposure estimates within each zip
code were then averaged and assigned a quartile of exposure. The highest quartile
lies largely outside of the 5-mile radius of the TMI facility and over three quarters of
the individuals within the cohort resided in low and low/medium dose areas (Talbott
et al., 2003; Talbott et al., 2000).
Estimates of individual average maximum and likely whole-body g-doses from
the TMI accident radioactivity release were calculated based on residential location
and the amount of time each person stayed in the 5-mile area during the 10 days
following the accident (Gur et al., 1983). This was accomplished by an estimate of
external dose from inert gases. The procedure for assigning these doses used
estimated time-dependent dose rate distributions in conjunction with the location
of residence and movements into and out of the ve-mile area. An assumption was
made that the emission rate during the rst 10-day period declined exponentially.
Hourly g dose was estimated by sector averaging the plume concentration around
the wind direction. The maximum possible dose assigned to individuals was the 10day integrated outdoor estimated dose in the location of residence with no shielding
or evacuation corrections and multiplied by a safety factor (S), which was derived
from the standard deviation of the ratio between estimated and measured doses in
the region. The likely dose assignment used skin dose (entrance dose) rather than
whole body dose, which is lower. It also considered time outdoors, residential
location, and shielding in the region. The average maximum and likely g-doses were
0.25 m Sv and 0.10 m Sv per individual, respectively.

2.2. Statistical analysis

All analyses shown in the tables were based on white cohort members who were
18 years of age and older on the day of the accident. This restriction was due to the
small number of nonwhites in the population and the lack of confounder information in children under the age of 18. Cox regression models (Breslow and Day, 1987;
Cox, 1972) were used to investigate the dependence of the cancer incidence rates on
the radiation exposure measures. The ve outcomes of interest were as follows: all
malignant neoplasms; cancer of bronchus, trachea, and lung; cancer of lymphatic
and hematopoietic tissues; leukemia; and female breast cancer. Chronic lymphocytic
leukemia and Hodgkins disease were excluded from the regression analyses as they
have rarely been linked to radiation exposure (Committee on the Biological Effects of
Ionizing Radiation, 1990). Cancer of the central nervous system and thyroid were not
analyzed due to the small case numbers of these cancers. Three exposure-related
covariates (background, maximum, and likely g-radiation exposures) and potential
confounding factors (age, gender, education, and smoking status) were investigated
for their inuence on the cancer site-specic outcomes. Previous radiation exposure
was not included in the models because of a large amount of missing data and
because less than 3% of the TMI cohort reported such experience. Natural background radiation was considered as a predictor variable as well as a confounding
variable in the maximum and likely g-radiation exposure models.
Cox regression analysis was performed separately for each of the ve time-toevent outcomes for men and women combined and again separately. In the
current analysis, any person not diagnosed with a cancer before the end of the
study (December 31, 1995) was dened as a censored observation. All individuals
who died during the study period (19821995) were censored from the follow-up.
A total of 20 subjects who died from cancer, but were not identied as incident
cases in the cancer registry, were not considered as events. Multiplicative relative
risk (RR) models of the form l(t) l0(t) exp[x(t) b] were tted to the internal
cohort rates, where x(t) is a vector of covariates, and b is the corresponding
parameter vector estimated by partial likelihood (Breslow and Day, 1987). The
proportional hazard assumption was tested by plotting log(t) versus log [  log l(t)
] for each group. The curve for each group was shown to be parallel indicating that
the predictors are proportional and meet the proportional hazard assumption.
The radiation-related exposure variables included background radiation,
estimated maximum and likely g exposures during the ten days following the
accident. Background radiation was treated as a categorical variable (low: 5.7
7.2 mR/h, low/medium: 7.37.9 mR/h, and medium/high 8.08.7 mR/h). Estimated
maximum and likely g exposures were treated as continuous variables and
linearity was examined. Individual RR estimates were expressed per 0.1 m Sv
and statistical signicance was examined based on the 95% CI. Models for
estimating cancer risk and maximum and likely g exposures were adjusted for

1232

Y.-Y. Han et al. / Environmental Research 111 (2011) 12301235

age and education (o 12 years, Z12 years) at the time of the accident, smoking
(never, former smoker, and current smoker), background radiation (5.77.2, 7.3
7.9 and 8.08.7 mR/h), and gender for white men and women combined. All tests
were done by SPSS Statistics 17.0 (Somers, NY) and no adjustment was made for
multiple comparisons.

3. Results
There were a total of 10,693 men and 11,376 women aged 18 and
older at the time of the accident in the TMI cohort. Demographic
characteristics of the study participants and newly diagnosed cancer
cases between 1982 and 1995 are shown in Table 1. The average age
of the participants was 40.2 years for men and 42.7 years for women.
Approximately 97% of the cohort was white (10,446 men and 11,048
women) so all remaining analyses were based on whites only. A total
of 1651 cancer cases were white among the 1687 cancer cases
reported in the adult TMI cohort. There were 30 cancer cases among
blacks and six cancer cases among other races/ethnicities (data not
shown). Among 797 cancer cases in white men, 158 cancers of
bronchus, trachea, and lung, 9 cancers of the central nervous system,
54 cancers of lymphatic and hematopoietic tissues, 32 cases of
leukemia, and 3 thyroid cancers were reported. Among 854 cancer
cases in white women, 66 cancers of bronchus, trachea, and lung, 10
cancers of the central nervous system, 44 cancers of lymphatic and
hematopoietic tissues, 23 cases of leukemia, and 5 thyroid cancers
were reported. A total of 278 new breast cancer cases were
diagnosed between 1982 and 1995.
A total of 44 cancer cases (11 boys and 33 girls) were reported
between 1985 and 1995 among 9250 individuals less than 18
years of age at the time of incidence (Table 2). Among girls, 24
were diagnosed with cervical cancer, two with ovarian, two with
Table 1
Demographic characteristics of the Three Mile Island cohort by gender, 18 years of
age and older.
Men (n 10,693)

Women (n 11,376)

Age (years)
1834
3560
460
Mean

4989
4126
1578
40.2 716.8

46.7
38.6
14.8

4918
4265
2193
42.7 7 18.3

43.2
37.5
19.3

Race
White
Black
Other

10,446
167
80

97.7
1.6
0.7

11,048
223
105

97.1
2.0
0.9

Education (years)
r 9
1012
Z 13
Unknown

1639
6251
2695
108

15.3
58.5
25.2
1.0

1829
7408
1875
264

16.1
65.1
16.5
2.3

Current marital status

Married
Others

7335
3358

68.6
31.4

7439
3937

65.4
34.6

Smoking status
Never
Former smoker
Current smoker
Unknown

3956
2610
4094
33

37.0
24.4
38.3
0.3

6318
1498
3482
78

55.5
13.2
30.6
0.7

Malignant neoplasms
All sites
Bronchus, trachea and lung
Central nervous system
Lymph/hematopoietic tissues
Leukemia
Breast
Thyroid

Total/white only
810/797
161/158
9/9
57/54
32/32
1/1
4/3

Total/white only
877/854
69/66
10/10
45/44
24/23
288/278
5/5

Table 2
Incident cases of all malignant neoplasms of the three mile island cohort, 18 years
of age and younger (n 9250), 19821995.
Malignant neoplasms

Boy
(n 4808)

Girl
(n 4442)

All sites
Brain
Non-Hodgkins lymphoma (NHL)
Chronic lymphocytic leukemia (CLL)
Soft tissue including heart
Trachea, mediastinum and other respiratory
organs
Colon
Testis
Cervix
Ovary
Corpus uteri

11
3
3
1
1
1

33
1
2
1
1
-

2
-

1
24
2
1

non-Hodgkins lymphoma, and one each with corpus uteri, brain,

colon, chronic lymphocytic leukemia, and soft tissue including
heart. Among boys, three were diagnosed with brain cancer, three
with NHL, two with testicular, and one each with chronic
lymphocytic leukemia, soft tissue including heart, and trachea,
mediastinum, and other respiratory organs. Adjusting for age and
gender, there was no signicant elevated risk for total cancers
associated with natural background, maximum or likely g radiation exposures among this young population (data not shown).
Results from the Cox regression models for cancer incidence in
white men and women are shown in Table 3. All models were
adjusted for age, education, and smoking status. Models containing
both white men and women were adjusted for gender. Maximum
and likely g models were also adjusted for background radiation
exposure. For white men and women combined, exposure to higher
maximum and likely g radiation was not found to elevate the risk
for all malignant neoplasms, cancer of bronchus, trachea, and lung,
cancer of lymphatic and hematopoietic tissues, and leukemia.
A signicant elevation in risk was noted for cancer of bronchus,
trachea, and lung in relation to higher background radiation
exposure (RR 1.45 at 8.08.8 mR/h compared to 5.27.2 mR/h).
No relationship was found between cancer risks and maximum or
likely g-radiation exposures among white women.
Among white men, exposure to higher levels of maximum and
likely g-radiation was found to signicantly increase the risk of
leukemia (RR 1.15, 95% CI 1.04, 1.29 and RR1.36, 95%
CI 1.08, 1.71, respectively). Although it was not statistically
signicant, the individual RR for natural background exposure
exhibited an increased risk for cancer of bronchus, trachea, and
lung after adjusting for age, education, and smoking status
(RRs1.10 at 7.37.9 mR/h and 1.49 at 8.08.7 mR/h). Elevated
RRs were also found between background radiation exposure and
the risk of cancer of lymphatic and hematopoietic tissues
(RRs1.31 at 7.37.9 mR/h and 1.42 at 8.08.7 mR/h), and leukemia (RRs 1.38 at 7.37.9 mR/h and 2.00 at 8.08.7 mR/h).

4. Discussion
The current study presents a 13-year follow-up of cancer
incidence among 32,135 individuals living within a 5-mile radius
from the TMI facility at the time of the accident. No increased
cancer risk was found among those younger than 18 years of age
by the end of 1995. There was no signicant increase in incidence
rates for all malignant neoplasms, cancer of bronchus, trachea,
and lung, cancer of lymphatic and hematopoietic tissues, and
leukemia by maximum g, likely g, or background radiation levels
among white women. Higher maximum and likely g radiation

Y.-Y. Han et al. / Environmental Research 111 (2011) 12301235

1233

Table 3
Summary of cox regression analysis for cancer incidence of the three mile island cohort, 18 years of age, 19821995.
Malignant neoplasms

Bronchus, trachea, and lung

Lymph and Hematopoietic Tissue

Leukemia

Breast

RRa

95% CIb

RR

95% CI

RR

95% CI

RR

95% CI

White men and women

Background (mR/h)c
5.27.2
580
7.37.9
703
8.08.7
360
Maximum g (m Sv)d 1643
Likely g (m Sv)d
1643

1.00
0.99
1.01
1.00
0.99

0.88,
0.88,
0.97,
0.94,

1.10
1.15
1.01
1.03

69
95
60
224
224

1.00
1.20
1.45
1.01
1.02

0.88,
1.02,
0.95,
0.90,

1.63
2.05
1.07
1.16

28
47
22
98
98

1.00
1.36
1.22
1.00
1.02

0.86,
0.70,
0.91,
0.85,

2.17
2.15
1.09
1.23

16
26
13
55
55

1.00
1.38
1.41
1.07
1.19

0.74,
0.68,
0.98,
0.97,

2.57
2.95
1.18
1.45

White men
Background (mR/h)e
5.27.2
7.37.9
8.08.7
Maximum g (m Sv)f
Likely g (m Sv)f

268
338
187
797
797

1.00
0.96
1.06
0.98
0.96

0.82,
0.88,
0.95,
0.60,

1.12
1.27
1.02
1.03

47
66
45
158
158

1.00
1.10
1.49
1.00
0.96

0.76,
0.98,
0.92,
0.83,

1.60
2.25
1.06
1.13

15
25
14
54
54

1.00
1.31
1.42
1.07
1.19

0.69,
0.68,
0.97,
0.95,

2.48
2.95
1.20
1.47

8
14
10
32
32

1.00
1.38
2.00
1.15
1.36

0.58,
0.79,
1.04,
1.08,

3.30
5.09
1.29
1.71

White women
Background (mR/h)e
5.27.2
7.37.9
8.08.7
Maximum g (m Sv)f
Likely g (m Sv)f

312
365
173
854
854

1.00
0.99
0.95
1.00
1.01

0.85,
0.79,
0.97,
0.94,

1.15
1.15
1.03
1.07

22
29
15
66
66

1.00
1.30
1.25
1.05
1.13

0.74,
0.64,
0.95,
0.92,

2.26
2.42
1.15
1.38

13
22
8
44
44

1.00
1.42
0.93
0.84
0.76

0.72,
0.37,
0.69,
0.51,

2.77
2.31
1.03
1.13

8
12
3
23
23

1.00
1.36
0.70
0.90
0.85

0.56,
0.18,
0.69,
0.53,

3.34
2.65
1.15
1.38

RR

95% CI

96
132
49
278
278

1.00
1.15
0.89
1.00
1.01

0.88,
0.63,
0.95,
0.90,

1.50
1.65
1.05
1.13

RR, relative risk.

CI, condent interval
c
Background models adjusted for age, gender, education, and smoking.
d
Maximum and likely g models adjusted for age, gender, education, smoking, and background radiation.
e
Background models adjusted for age, education, and smoking.
f
Maximum and likely g models adjusted for age, education, smoking, and background radiation.
b

exposure was found to signicantly increase the incidence of

leukemia among white men. Exposure to higher background
radiation was also found to increase the incidence of cancer of
the bronchus, trachea, and lung among white men.
Cancer of lymphatic and hematopoietic tissues, including leukemia, multiple myeloma, and non-Hodgkins lymphoma, is recognized or suspected to be associated with an increased risk from
radiation exposures and have a relatively short latency period
compared to solid tumors. Studies on cancer of lymphatic and
hematopoietic tissues and exposures to low-dose radiation from
nuclear power plant accidents and atomic weapons testing have
shown inconclusive results (Dainiak, 2002). In our study, the
maximum and likely g-radiation was found to signicantly increase
the risk of leukemia among white men in the TMI cohort. Leukemia
is the most signicant hematologic complication among atomic
bomb survivors and those at greatest risk are individuals exposed to
radiation during childhood. Combining data from seven cohorts in
three countries, Cardis et al. reported a signicant increase in all
leukemia (excluding chronic lymphocytic leukemia, CLL) mortality
rates among 95,613 workers who were exposed to low-level
radiation through employment of six months or longer in the
nuclear industry (Cardis et al., 1995). This result was supported
by a multinational retrospective cohort study combining nearly
6,00,000 workers in 15 countries, with an average cumulative
recorded dose at 19.4 m Sv, in which 90% of the workers received
cumulative doses less than 50 m Sv (Cardis et al., 2005). Continued
monitoring of the TMI cohort will help to elucidate hematopoietic
cancer risk among this population.
We found that higher natural background exposure led to a
statistically non-signicant increased risk of cancer of lymphatic and
hematopoietic tissues and leukemia among white men. The literature remains inconclusive with regard to natural background radiation and cancer risk. In a previous TMI study, a signicant elevated
mortality for cancer of lymphatic and hematopoietic tissues was

noted among women exposed to higher background radiation

(Talbott et al., 2003). However, incidence of cancer of lymphatic
and hematopoietic tissues among women in the same TMI cohort
was not reported to be associated with background radiation
exposure in our study. This may be due to the small number of
women (n3) exposed to the highest dose of natural background
radiation. These small numbers also produced a wide 95% CI
(RR0.7, 95% CI0.182.65). In another population-based study,
no cancer site was signicantly related to cumulative radiation dose
in India where the background radiation is known to be high
(median outdoor radiation levels are more than 4 m Gy/y) (Nair
et al., 2009). In the TMI cohort, a total of ve NHL cases were found
among 44 cancer cases who were less than 18 years of age. Among
those ve non-Hodgkins lymphoma cases, four children were
exposed to the third and fourth quartiles of maximum g and likely
g radiation. Most non-Hodgkins lymphoma cases occur in people
older than 60 years of age, and it is rarely seen among young people.
Whether TMI radiation exposure contributed to the non-Hodgkins
lymphoma occurrence among thee young children in our study who
had higher radiation exposure requires further investigation.
Higher natural background exposure was associated with a
signicant increased risk of cancer of bronchus, trachea, and lung
in white men after adjusting for age, education, and smoking
status. Based on the report from the United Nations Scientic
Committee on the Effects of Atomic Radiation (UNSCEAR), lung
cancer risk has been associated with external low linear energy
transfer radiation in survivors of atomic bombings, in persons
exposed to high doses for medical reasons, and nuclear plant
workers exposed to high doses. Adjustment for smoking status
and measurement of exposure uncertainty among studied populations should be considered accordingly (UNSCEAR, 2006). In the
TMI cohort, maximum and likely g exposures were not signicantly associated with cancer risk of bronchus, trachea, and lung.
It has been suggested that individual radon exposure should be

1234

Y.-Y. Han et al. / Environmental Research 111 (2011) 12301235

included as part of the overall radiation exposure assessment in

radiation-related lung cancer studies because the counties around
the TMI site have been reported to have high regional radon doses
(163266 Bq m  3) (US EPAs action level of radon is 150 Bq m  3)
(Field, 2005). A collaborative assessment of individual data from
13 European case-control studies reported the risk of lung cancer
increased by 8.4% per 100 Bg m  3 in measured residential radon
(Darby et al., 2005). Assessing and adjusting for the individual
level of radon exposure instead of using aggregated zip code level
data can better estimate the effect of radiation exposure and lung
cancer development.
Cancer of the thyroid was initially chosen due to its radiosensitive nature. A higher standardized incidence ratio for thyroid
cancer was found to relate to occupational exposure to ionizing
radiation based on the National Dose Registry of Canada (Sont
et al., 2001). Thyroid cancer patients who have previously been
exposed to radiation were found to have more aggressive disease
and worse outcomes than other patients (Seaberg et al., 2009).
Based on a recent UNSCEAR report and another epidemiologic
review after 25 years from the Chernobyl accident in 1986, a
considerable dose-related increase in thyroid cancer among
children and adolescents exposed to radioactive iodine isotopes
was noted, with the risk being greatest in those youngest at
exposure. However, it was suggested that there was little evidence to show that thyroid cancer risk increased among those
who were exposed as adults in the general population (Cardis and
Hatch, 2011; UNSCEAR, 2011). In our study, there were only eight
cases of thyroid cancer as of l995. We identied one case who was
exposed to maximum g at 0.39 m Sv and likely g at 0.17 m Sv
from the TMI incidence, but the average radiation level for the
eight thyroid cancer cases was 0.20 m Sv for maximum g exposure and 0.06 m Sv for likely g exposure. The limited number of
thyroid cancer cases in the TMI cohort precluded further statistical analysis. Based on the Pennsylvania Cancer Registry, an
increasing rate of thyroid cancer incidence over time was
observed in two of three at-risk counties surrounding the TMI
facility between 1990 and 2002 (Levin, 2008). This nding was
not adequate to prove a causal relationship between low-level
radiation exposure and thyroid cancer risk but further investigation, specic among those exposed at young age, is warranted.
Exposure to higher maximum and likely g-radiation was not
found to increase cancer risks among women in the TMI cohort.
Ionizing radiation exposure is a known risk factor for breast cancer
and no safe dose has been suggested. A doubled excess relative risk
(ERR) of radiation exposure was found for breast cancer occurring
after the age of 35 among atomic bomb survivors exposed under
the age of 20 during the World War II (Tokunaga et al., 1994), and
the ERR was observed to decline with the increasing age at
exposure (Land et al., 2003). Evidence from an epidemiologic study
linking low-dose radiation exposure and breast cancer is controversial. A signicant doseresponse relationship was found for
breast cancer risk among women who reported having had multiple
chest X-rays (P trend0.0007) (Ma et al., 2008). Ma et al. also found
that breast cancer risk increased among women who received
dental X-rays without lead apron protection before the age of 20
years (OR1.81, 95% CI1.13, 2.90). Higher risk of breast cancer
and leukemia excluding CLL has been reported among Taiwanese
residents exposed to long-term low-dose rate g-radiation in the
form of 60Co-contaminated steel in their buildings, with an average
excess cumulative exposure estimate of about 48 m Gy (Hwang
et al., 2008). However, these two populations were exposed to
higher radiation doses compared to the TMI cohort.
It has been suggested that radiation and estrogen may have a
synergistic effect on the development of breast carcinoma (Calaf
and Hei, 2000). Individual susceptibility and age at exposure to
radiation should also be considered in medical and occupational

radiation protection (Shore, 2009). Information on hormonal factors

including age at menarche, parity, menopause status, oral contraception and hormonal replacement therapy, and other risk factors
including familial history and breast cancer genes were not collected from the TMI cohort. Additional follow-up studies that
include other breast cancer risk factors and target women exposed
to radiation at a young age are needed to clarify breast cancer risk
caused by low-dose radiation exposure.
The major strength of the current study is the TMI cohort covered
over 93% of the exposed population with a comprehensive 13-year
follow-up, although the follow-up was terminated due to lack of
study funding. Several limitations are also noted. A 13-year survey of
radiation exposure and cancer occurrence may be inefcient due to
the long latency of cancer development. Given the low-dose exposure with inefcient follow-up, the statistical power was limited to
observe statistically signicant ndings. Additional radioactivity
exposure from the TMI accident, including b-radiation to the skin
and the internal dose from the inhaled and ingested radionuclides,
should be considered. However, b-radiation is less penetrative than
g-radiation and has a shorter range in air. Shelter, clothing, and other
shielding factors can reduce b-radiation to a small extent (Talbott
et al., 2000). The internal dose has been estimated among the TMI
cohort and it constituted less than 0.49.4% of the total whole-body
dose (Kemeny et al., 1979). Information on occupational and medical
radiation exposures, including x-rays and computed tomography
among the TMI cohort, was not available. On average, typical organ
radiation doses from computed tomography range from 0.005 to
20 m Sv in various radiologic studies (Brenner and Hall, 2007). Future
acquisition of personal occupational and medical radiation exposures
should be included to estimate the cancer risks attributable to TMI
radiation leakage among the exposed population.

5. Conclusion
No signicant increased risk was found for all malignant
neoplasms, cancer of bronchus, trachea, and lung, and female
breast cancer among the TMI adult cohort by the end of 1995.
Low-dose radiation exposure from the TMI accident may have
minimal effect on cancer risk, but the increased male leukemia
risk among the 5-mile radius residents cannot be ignored. Future
epidemiologic study is needed to further investigate hematopoietic cancer risk and low-dose radiation exposures in this cohort.
Due to the long latency periods in some types of radiationinduced cancers, additional follow-up should provide a more
comprehensive description of cancer risk related to low-dose
radiation exposure among the TMI cohort, especially for those
younger than 18 years of age at the time of the accident.

References
Brenner, D.J., Doll, R., Goodhead, D.T., Hall, E.J., Land, C.E., Little, J.B., et al., 2003.
Cancer risks attributable to low doses of ionizing radiation: assessing what we
really know. Proc. Natl. Acad. Sci. U S A 100 (24), 1376113766.
Brenner, D.J., Hall, E.J., 2007. Computed tomographyan increasing source of
radiation exposure. N. Engl. J. Med. 357 (22), 22772284.
Breslow, N.E., Day, N.E., 1987. Statistical methods in cancer research ,Vol. II: the
design and analysis of cohort studies. IARC Sci. Publ. (82), 178229.
Calaf, G.M., Hei, T.K., 2000. Establishment of a radiation- and estrogen-induced
breast cancer model. Carcinogenesis 21 (4), 769776.
Cardis, E., Gilbert, E.S., Carpenter, L., Howe, G., Kato, I., Armstrong, B.K., et al., 1995.
Effects of low doses and low dose rates of external ionizing radiation: cancer
mortality among nuclear industry workers in three countries. Radiat Res. 142
(2), 117132.
Cardis, E., Hatch, M., 2011. The chernobyl accidentan epidemiological perspective. Clin. Oncol. (R. Coll. Radiol.).
Cardis, E., Vrijheid, M., Blettner, M., Gilbert, E., Hakama, M., Hill, C., et al., 2005.
Risk of cancer after low doses of ionising radiation: retrospective cohort study
in 15 countries. Brit. Med. J. 331 (7508), 77.

Y.-Y. Han et al. / Environmental Research 111 (2011) 12301235

Committee on the Biological Effects of Ionizing Radiation, 1990. Radiogenic Cancer

at Specic Sites, Health Effects of Exposure to Low Levels of Ionizing Radiation:
BEIR V. National Academy Press, Washington, DC.
Committee on the Biological Effects of Ionizing Radiation, 2006. Health risks from
exposure to low levels of ionizing radiation: BEIR VII Phase 2. National
Academy Press, Washington, DC.
Cox, D.R., 1972. Regression models and life-tables. J. Roy. Stat. Soc. 34B (2), 1972.
Dainiak, N., 2002. Hematologic consequences of exposure to ionizing radiation.
Exp. Hematol. 30 (6), 513528.
Darby, S., Hill, D., Auvinen, A., Barros-Dios, J.M., Baysson, H., Bochicchio, F., et al., 2005.
Radon in homes and risk of lung cancer: collaborative analysis of individual data
from 13 European case-control studies. Brit. Med. J. 330 (7485), 223.
Field, R.W., 2005. Three Mile Island epidemiologic radiation dose assessment revisited:
25 years after the accident. Radiat. Prot. Dosimetry 113 (2), 214217.
Gur, D., Good, W.F., Tokuhata, G.K., Goldhaber, M.K., Rosen, J.C., Roa, G.R., 1983.
Radiation dose assignment of individuals residing near the Three Mile Island
nuclear station. Proc. Pennsylvania Acad. Sci. 57, 99107.
Hatch, M.C., Wallenstein, S., Beyea, J., Nieves, J.W., Susser, M., 1991. Cancer rates
after the Three Mile Island nuclear accident and proximity of residence to the
plant. Am J. Public Health 81 (6), 719724.
Hwang, S.-L., Hwang, J.-S., Yang, Y.-T., Hsieh, W.A., Chang, T.-C., Guo, H.-R., et al.,
2008. Estimates of relative risks for cancers in a population after prolonged
low-dose-rate radiation exposure: a follow-up assessment from 1983 to 2005.
Radiat. Res. 170 (2), 143148.
Kemeny, J.G., Babbitt, B., Haggerty, P.E., Lewis, C., Marks, P.A., Marrett, C.B., et al.,
1979. Report of the Presidents Commission on the Accident at Three Mile
Island. U.S. Government Printing Ofce, Washington, DC.
Land, C.E., Tokunaga, M., Koyama, K., Soda, M., Preston, D.L., Nishimori, I., et al.,
2003. Incidence of female breast cancer among atomic bomb survivors,
Hiroshima and Nagasaki, 19501990. Radiat. Res. 160 (6), 707717.
Levin, R.J., 2008. Incidence of thyroid cancer in residents surrounding the Three
Mile Island nuclear facility. Laryngoscope 118 (4), 618628.
Ma, H., Hill, C.K., Bernstein, L., Ursin, G., 2008. Low-dose medical radiation
exposure and breast cancer risk in women under age 50 years overall and
by estrogen and progesterone receptor status: results from a casecontrol and
a casecase comparison. Breast Cancer Res. Treat. 109 (1), 7790.
Nair, R.R., Rajan, B., Akiba, S., Jayalekshmi, P., Nair, M.K., Gangadharan, P., et al.,
2009. Background radiation and cancer incidence in Kerala, India-Karanagappally cohort study. Health Phys. 96 (1), 5566.

1235

NCRP. 2009. National council on radiation protection and measuurements, Report

no. 160Ionizing Radiation Exposure Of The Population Of The United States.
Preston, D.L., Cullings, H., Suyama, A., Funamoto, S., Nishi, N., Soda, M., et al., 2008.
Solid cancer incidence in atomic bomb survivors exposed in utero or as young
children. J. Natl. Cancer Inst. 100 (6), 428436.
Ramaswamy K., Tokuhata G.K., Hartman T. 1991. TMI population-based cohort
cancer incidence, Harrisburg, PA, July, 1982; June, 1989. in: Presented at the
Annual Meeting of the Society for Epidemiology Research, Buffalo, NY, June 13,
1991.
Seaberg, R.M., Eski, S., Freeman, J.L., 2009. Inuence of previous radiation exposure
on pathologic features and clinical outcome in patients with thyroid cancer.
Arch. Otolaryngol. Head Neck Surg. 135 (4), 355359.
Shore, R.E., 2009. Low-dose radiation epidemiology studies: status and issues.
Health Phys. 97 (5), 481486.
Sont, W.N., Zielinski, J.M., Ashmore, J.P., Jiang, H., Krewski, D., Fair, M.E., et al., 2001.
First analysis of cancer incidence and occupational radiation exposure based on
the national dose registry of canada. Am. J. Epidemiol. 153 (4), 309318.
Talbott, E.O., Youk, A.O., McHugh-Pemu, K.P., Zborowski, J.V., 2003. Long-term
follow-up of the residents of the Three Mile Island accident area: 19791998.
Environ. Health Perspect. 111 (3), 341348.
Talbott, E.O., Youk, A.O., McHugh, K.P., Shire, J.D., Zhang, A., Murphy, B.P., et al.,
2000. Mortality among the residents of the Three Mile Island accident area:
1979-1992. Environ. Health Perspect. 108 (6), 545552.
Tokunaga, M., Land, C.E., Tokuoka, S., Nishimori, I., Soda, M., Akiba, S., 1994.
Incidence of female breast cancer among atomic bomb survivors, 19501985.
Radiat. Res 138 (2), 209223.
UNSCEAR(United Nations Scientic Committee on the Effects of Atomic Radiation).
2006,Effects of Ionizing Radiation. Volume I: Report to the General Assembly,
Scientic Annexes A and B. Annex A: Epidemiological studies of radiation and
cancer.
UNSCEAR(United Nations Scientic Committee on the Effects of Atomic Radiation).
2011, UNSCEAR 2008 Report to the General Assembly with Scientic Annexes.
Volume II, Scientic Annexes C, D, and E.
Wing, S., Richardson, D., Armstrong, D., Crawford-Brown, D., 1997. A reevaluation
of cancer incidence near the Three Mile Island nuclear plant: the collision of
evidence and assumptions. Environ Health Perspect. 105 (1), 5257.