CDRI : Cancer Data Registry of Idaho

THYROID CANCER IN IDAHO
1970-1996

Following release of the National Cancer Institute's report of its study to assess Americans' exposure to radioactive iodine-131 from atmospheric nuclear bomb testing in the 1950s and 1960s at the Nevada Test Site, and pursuant to requests from the public, media, and health officials, staff at the Cancer Data Registry of Idaho (CDRI) conducted several analyses of thyroid cancer. This report describes the analyses of thyroid cancer incidence rates in Idaho, 1970-1996, and the ratio of female-to-male thyroid cancer cases by age group. Because four of the five counties in the United States with highest estimated exposure to iodine-131 from atmospheric nuclear bomb tests at the Nevada Test Site are located in Idaho (Blaine, Custer, Gem, and Lemhi), and public health services are delivered at the health district level, analyses were conducted at both the county and health district levels of geography (see Appendix for listing of counties by health district).

METHODS

Established in 1969, CDRI is a population-based cancer registry that collects incidence and survival data on cancer patients who reside in the state of Idaho at the time of diagnosis or who are diagnosed and/or treated for cancer in the state of Idaho. All cases of invasive thyroid cancer diagnosed among residents of the state of Idaho between January 1, 1970, and December 31, 1996, were included in these analyses.

Thyroid Cancer Incidence in Idaho, 1970-1996, by Birth Cohort and Overall

A combination of direct and indirect age adjustment was used to compare the incidence rates of thyroid cancer among geographic areas and by birth cohort. The following provides an overview of the steps taken in the analysis.

Step 1.

The number of invasive cases of thyroid cancer diagnosed among residents of the state of Idaho, 1970-1996, were summed by sex, 5-year age group, county of residence, and year of diagnosis. The age groups were: 0-4, 5-9, 10-14, ..., 85+. Cases with missing information on sex, age at diagnosis, or county were not included in the analysis. If birth year was unknown, it was imputed from diagnosis date and estimated age at diagnosis. Birth cohort was defined by birth year: before 1948, 1948-1958, and after 1958. Given the years that atmospheric nuclear bomb tests were conducted at the Nevada Test Site, and the fact that children aged 0-5 are biologically most sensitive to iodine-131, persons born during the time period 1948-1958 are thought to have been at the risk of highest exposure. The other birth cohorts were chosen for comparison purposes.

Step 2.

Population data were obtained from the U.S. Bureau of the Census. Data for 1970-1989 were available by sex and 5-year age-group. Data for 1990-1996 were available by sex and single-year age group. Single-year age group estimates were derived from 1970-1989 data by dividing the population in each 5-year age group into 5 equal parts. For example, if 1,000 persons were in the 5-year age group 15-19, 200 persons each were assigned to ages 15,16, 17, 18, and 19. Separating the population data into single-year age groups was necessary to calculate age-adjusted rates by birth cohort.

Step 3.

Three birth cohorts were defined, again based upon differences due to age in estimated exposure to iodine-131 from atmospheric nuclear bomb tests at the Nevada Test Site: before 1948, 1948-1958, and after 1958. Population data were estimated for each of the birth cohorts, by county, sex, and 5-year age group. Two age breaks were defined and incremented by year of study. The lower age break was defined as: AGELOW = YEAR - 1958, and year was varied from 1970-1996, the years for which CDRI has reliable statewide case information. The upper age break was defined as: AGEHIGH = YEAR - 1948, and year was again varied from 1970-1996. AGELOW and AGEHIGH give the (truncated) ages a person would have been in each year, 1970-1996, given that they were born in the period 1948-1958. Thus, a person born from 1948-1958 would have been assigned ages 12-22 in 1970, and between 32 and 42 in 1990, etc. Population data were assigned to birth cohorts depending on single-year age group and year (1970-1996). For example, for the year 1970, persons aged 10 were assigned to the after-1958 birth cohort; persons aged 20 were assigned to the 1948-1958 birth cohort; and persons aged 30 were assigned to the before-1948 birth cohort. Finally, population data were collapsed over age to yield estimates by birth cohort, county, sex, and 5-year age group.

Step 4.

The age-adjusted rate of invasive thyroid cancer for the state of Idaho, 1970-1996, was calculated by the direct method, using the 1970 U.S. population as standard. Age-specific rates for the state of Idaho were calculated for use in direct age adjustment. The result was an age-adjusted incidence rate of thyroid cancer for the state of Idaho, 1970-1996, of 4.22 cases per 100,000 person-years.

Step 5.

Age- and sex-specific rates for the state of Idaho for the time period 1970-1996 were calculated for use as standard rates in indirect age and sex adjustment.

Step 6.

For all birth cohorts combined, the numbers of observed and expected cases were calculated for each health district and county. Expected cases were calculated by applying the age- and sex-specific rates for the state of Idaho to the population by age and sex in each health district and county. One-tailed p-values comparing the number of observed and expected cases were calculated using the Poisson probability distribution. Observed and expected cases, and p-values, were calculated separately for males, females, and both sexes combined.

Step 7.

The adjusted incidence rates for each health district and county were calculated as the standardized incidence ratio (observed/expected) multiplied by the age-adjusted rate from Step 4. For example, there were 314 cases observed, and 254.6 cases expected in Health District 4, yielding an adjusted incidence rate of (314/254.6) * 4.22 = 5.20 cases per 100,000 person-years.

Step 8.

Age- and sex-specific rates for the state of Idaho, 1970-1996, were calculated by birth cohort for use in indirect age and sex adjustment by birth cohort. (Step 5 was repeated by birth cohort.)

Step 9.

For each birth cohort, the number of observed and expected cases were calculated for each health district and county. Expected cases were calculated by applying the age- and sex-specific rates for the state of Idaho, by birth cohort (from Step 8), to the population by age and sex in each health district and county. One-tailed p-values comparing the number of observed and expected cases were calculated using the Poisson probability distribution. Observed and expected cases, and p-values, were calculated separately for males, females, and both sexes combined, by birth cohort. Statistical significance was set at a = .05.

Step 10.

The adjusted rates for each health district and county were calculated by birth cohort as the standardized incidence ratio for that birth cohort (observed/expected) multiplied by the age-adjusted rate for all birth cohorts (from Step 4). For example, there were 87 cases observed and 69.4 expected in Health District 4 for the birth cohort 1948-1958, yielding an adjusted incidence rate of (87/69.4) * 4.22 = 5.29. The overall rate from Step 4 was used as the reference rate in order to facilitate comparisons across birth cohort for individual health districts and counties, and to allow comparisons among health districts and counties within birth cohort.

Step 11.

The adjusted rates for the state of Idaho were calculated by birth cohort using the standardized incidence ratio for that birth cohort (observed/expected), with expected cases based upon age- and sex-specific rates for all birth cohorts (from Step 5). For example, there were 271 cases observed and 256.9 expected in the state of Idaho, 1970-1996, among the 1948-1958 cohort, yielding an adjusted incidence rate of (271/256.9) * 4.22 = 4.45 cases per 100,000 person-years. The state of Idaho rates by birth cohort were designed to be compared to each other, and to the overall state of Idaho rate of 4.22 cases per 100,000 person-years.

Ratio of Female-to-Male Thyroid Cancer Cases by Age Group and Birth Cohort

The overall ratio of thyroid cancer cases among females versus males differed by birth cohort, with a ratio of 2.7 in the before-1948 cohort, 5.2 in the 1948-1958 cohort, and 5.3 in the after-1958 cohort. In order to examine if the differences in female-to-male ratios by birth cohort were an artifact of differing age-specific rates by sex, cumulative ratios of female-to-male cases were calculated by age group. For all invasive cases of thyroid cancer diagnosed among Idaho residents, 1970-1996, the numbers of cases were summed separately for males and females by 5-year age group and birth cohort. The cumulative ratios of female-to-male cases were calculated by 5-year age group and birth cohort. For example, for the age group 35-39, the cumulative female-to-male ratio in the before-1948 birth cohort was 5.0 (75 cases among females aged 39 years and younger, and 15 cases among males aged 39 years and younger).

RESULTS

The overall age-adjusted incidence rate of invasive thyroid cancer in Idaho, 1970-1996, was 4.22 cases per 100,000 person-years (see Table 1). Incidence rates varied by geographic location, ranging from 3.28 cases per 100,000 person-years in Health District 2 to 5.20 cases per 100,000 person-years in Health District 4. There were significantly more cases of invasive thyroid cancer diagnosed among residents of Health District 4 than expected based upon rates in the state of Idaho (314 observed, 254.6 expected, p<.001), and the number of observed cases was higher than expected for both males and females. Ada, Custer, and Twin Falls Counties each had significantly more cases of invasive thyroid cancer than expected based upon rates in the state of Idaho. Of the four Idaho counties with highest estimated exposure to iodine-131, Custer was the only county to show an elevation in thyroid cancer cases from 1970-1996 (9 observed, 4.8 expected, p=.025).

Among the birth cohort born before 1948, the incidence rate of invasive thyroid cancer, 1970-1996, was 4.10 cases per 100,000 person-years (see Table 2). There were significantly more cases observed than expected in Health District 4, and Ada, Madison, and Twin Falls Counties. None of the four Idaho counties with highest estimated exposure to iodine-131 showed an elevation in thyroid cancer cases from 1970-1996 in the birth cohort born before 1948.

Among the birth cohort born 1948-1958, the incidence rate of invasive thyroid cancer, 1970-1996, was 4.45 cases per 100,000 person-years (see Table 3). There were significantly more cases observed than expected in Health District 4, and Ada and Butte Counties. None of the four Idaho counties with highest estimated exposure to iodine-131 showed an elevation in thyroid cancer cases from 1970-1996 in the birth cohort born 1948-1958. Although the incidence rate of invasive thyroid cancer, 1970-1996, was highest for the birth cohort born 1948-1958, the number of cases observed was not statistically significantly higher than that expected based upon rates for all birth cohorts.

Among the birth cohort born after 1958, the incidence rate of invasive thyroid cancer, 1970-1996, was 4.35 cases per 100,000 person-years (see Table 4). There were significantly more cases observed than expected in Blaine, Custer, Elmore, and Lincoln Counties. Two of the four counties with highest estimated exposure to iodine-131 showed an elevation in thyroid cancer cases from 1970-1996 in the birth cohort born after 1958. In Blaine County, there were 6 cases observed and 2.4 cases expected (p=.012). In Custer County, there were 2 cases observed, and 0.6 cases expected (p=.020).

Regarding the female-to-male ratios for invasive thyroid cancer cases, the differences in the overall female-to-male ratios by birth cohort appear to be due to the higher age-specific thyroid cancer incidence rates in younger females as compared with younger males. In all three birth cohorts, the cumulative age-specific ratios were similar for the age groups 25-29, 30-34, and 35-39 (the only age groups for which comparisons are available across all three birth cohorts, as CDRI has reliable statewide cancer incidence data since 1970).