Exposure to the Thyroid from Intake of Radioiodine Isotopes after the Chornobyl Accident. Report I: Revised Doses and Associated Uncertainties for the Ukrainian-American Cohort

Abstract

Thyroid doses from intake of radioiodine isotopes (¹³¹I, ¹³²Te+¹³²I, and ¹³³I) and associated uncertainties were revised for the 13,204 Ukrainian-American cohort members exposed in childhood and adolescence to fallout from the Chornobyl nuclear power plant accident. The main changes related to the revision of the ¹³¹I thyroid activity measured in cohort members, the use of thyroid-mass values specific to the Ukrainian population, and the revision of the ¹³¹I ground deposition densities in Ukraine. Uncertainties in doses were assessed considering shared and unshared errors in the parameters of the dosimetry model. Using a Monte-Carlo simulation procedure, 1,000 individual stochastic thyroid doses were calculated for each cohort member. The arithmetic mean of thyroid doses from intake of ¹³¹I, ¹³²Te+¹³²I, and ¹³³I for the entire cohort was 0.60 Gy (median = 0.22 Gy). For 9,474 subjects (71.6% of the total), the thyroid doses were less than 0.5 Gy. Thyroid doses for 42 cohort members (0.3% of the total) exceeded 10 Gy while the highest dose was 35 Gy. Intake of ¹³¹I contributed around 95% to internal thyroid exposure from radioiodine isotopes. The geometric standard deviation of individual stochastic thyroid doses varied among cohort members from 1.4 to 4.3 with an arithmetic mean of 1.6 and a median of 1.4. It was shown that the contribution of shared errors to the dose uncertainty was small. The revised thyroid doses resulted, in average, in around 40% decrease for cohort members from Zhytomyr Oblast and an increase of around 24% and 35% for the cohort members from Kyiv and Chernihiv Oblast, respectively. Arithmetic mean of TD20 doses for the cohort was around 8% less than that estimated in TD10, 0.60 Gy vs. 0.65 Gy, respectively; however, global median of TD20 doses somewhat increased compared to TD10: 0.22 Gy vs. 0.19 Gy, respectively. The difference between TD10 and TD20 was mainly due to a revision of the individual ¹³¹I thyroid activity measured in the cohort members.

INTRODUCTION

The Ukrainian-American cohort study includes 13,204 subjects who were exposed between ages 0 and 18 years at the time of the accident (ATA) to Chornobyl (Chernobyl) fallout and who are followed-up for thyroid cancer and other thyroid diseases using a standardized screening protocol (1). The individual thyroid doses from intake of radioiodine isotopes were reconstructed two times for all cohort members using the so-called “Thyroid Dosimetry 2002 system” (TD02) that was revised in 2010 as “Thyroid Dosimetry 2010 system” (TD10) (2, 3). TD10 included the evaluation of shared and unshared uncertainties in the calculation of individual stochastic doses (3) and represented an important improvement over the previous version, TD02 (2). The TD10 doses were used in the latest analyses of the radiation-related risk of thyroid cancer and other thyroid diseases in this cohort (4–7).

However, it was “recognized” that there are limitations in the TD10, such as differences in thyroid doses normalized to the ¹³¹I ground deposition density between different areas and errors in instrumental measurements of ¹³¹I thyroid activity (3). To overcome these limitations, special studies were conducted to improve the TD10 system, including (i) revision of ¹³¹I thyroid activity measured in 146,425 individuals, including members of the Ukrainian-American cohort, and thyroid dosimetry system for the entire Ukraine that was used to calculate the ecological thyroid dose (8, 9); (ii) obtaining the age- and sex-specific thyroid-mass values for residents of the study area (10); and (iii) re-estimation of ¹³¹I ground deposition densities in Ukraine using a high-resolution meteorological model (11).

This paper describes the “Thyroid Dosimetry 2020 system” (hereinafter referred to as TD20), which incorporates all revisions and improvements described above, and provides the individual thyroid doses and associated uncertainties for the Ukrainian-American cohort members.

MATERIALS AND METHODS

Study Population

The Ukrainian-American cohort consists of 13,204 individuals who met the inclusion criteria: (i) residence ATA in one of three most contaminated after the Chornobyl accident oblasts in Ukraine: Kyiv, Zhytomyr, or Chernihiv; (ii) age ATA from 0 to 18 years old; and (iii) their ¹³¹I thyroid activity was measured between April 26 and June 30, 1986. The study was reviewed and approved by the institutional review boards of the participating organizations in Ukraine and the United States, and all study subjects or their guardians (for subjects who were 16 years of age or younger at screening) signed informed consent.

Thyroid Doses from ¹³¹I Intake

Figure 1 shows the scheme of the thyroid dose calculation for cohort members. Transfer of ¹³¹I from ground deposition to the thyroid gland via activity intake with inhaled air and due to ingestion of foodstuffs was evaluated using the ecological and biokinetic models and personal interview data on individual residential history and consumption. For each study subject k, two doses were calculated: (i) an “ecological” thyroid dose, $D_k^{ecol}$, based on the ¹³¹I thyroid activity, $Q_k^{ecol}(t)$, that was calculated for any time t after the accident using ecological and biokinetic models; and (ii) an “instrumental” thyroid dose, $D_k^{inst}$, based on the ¹³¹I thyroid activity measured at time t_m after the accident, $Q_k^{meas}\left(t_m\right)$. The instrumental dose is more reliable than the ecological dose because it is based on an individual measurement performed on each study subject. The ecological thyroid dose ($D_k^{ecol}$, Gy) for subject k was calculated as:

$$D_k^{ecol}=\frac{k_u\times e_{th,k}^{I-131}}{M_k}\cdot {\int }_0^T{Q}_k^{ecol}(t)dt$$ (1)

where k_u= 1.384 × 10⁻² is a unit conversion factor (Bq kBq⁻¹ g kg⁻¹ J MeV⁻¹ s d⁻¹); M_k is the thyroid mass that corresponds to age of subject k (g); $e_{th,k}^{I-131}$ is the mean energy absorbed in the thyroid gland per decay of ¹³¹I in the thyroid that corresponds to age of subject k (14) (MeV); $Q_k^{ecol}(t)$ is the “ecological” activity of ¹³¹I in the thyroid of study subject k at time t (kBq); T = 80 d is the upper limit of integration from the time of the accident on April 26, 1986 (t = 0) until July 14, 1986.

FIG. 1.

Scheme of thyroid doses calculation from ¹³¹I intake for the Ukrainian-American cohort.

To calculate the instrumental dose, the ecological ¹³¹I thyroid activity at time t_m, $Q_k^{ecol}\left(t_m\right)$, was replaced in Eq. (1) with the measured activity, $Q_k^{meas}\left(t_m\right)$. It was assumed that the relative shape of the variation of $Q_k^{ecol}(t)$ with time is correct, so that the adjustment at time t_m also applies to any other time after the accident. Under those conditions, the instrumental thyroid dose ($D_k^{inst}$, Gy) was calculated as:

$$D_k^{inst}=\frac{Q_k^{meas}\left(t_m\right)}{Q_k^{ecol}\left(t_m\right)}\cdot D_k^{ecol}=S{F}_k\cdot D_k^{ecol}$$ (2)

where SF_k is the scaling factor for study subject k (unitless).

A detailed description of the methodology to calculate the thyroid doses for the Ukrainian-American cohort can be found elsewhere (3).

Thyroid Doses from Intake of Short-Lived ¹³²Te+¹³²I and ¹³³I

Several short-lived radioiodine isotopes are produced in a nuclear reactor that have the same behavior as ¹³¹I in the environment and the human body. In addition, the radiotellurium isotopes, which are the precursors of the radioiodine isotopes, should also be considered. However, among them, only ¹³²I, which is progeny of ¹³²Te (half-lives of 2.3 h and 78.2 h, respectively), and ¹³³I (half-life 20.8 h) contributed substantially to the thyroid dose from inhalation (15), and, to a much lower extent, to the dose from ingestion (16). By the time of the beginning of wide-scale instrumental thyroid monitoring in Ukraine in the middle of May 1986, the ¹³²I and ¹³³I thyroid activities had decayed to negligible levels.

The ecological thyroid doses from intake of ¹³²Te+¹³²I and ¹³³I were estimated using the initial radioiodine isotopes ratio ATA (17) and daily ¹³¹I ground deposition density (9). The ecological thyroid dose due to inhalation and ingestion of ¹³³I was calculated in the same way as the dose from ¹³¹I, considering parameters specific for ¹³³I (3): energy absorbed in the thyroid gland (0.43 MeV per decay) and activity ratio of ¹³³I and ¹³¹I in the release from the reactor (1.6). The thyroid dose from ¹³²Te+¹³²I was based on ¹³²Te intake as its half-life is much longer than that of ¹³²I and these radionuclides are in radioactive equilibrium in the environment. The thyroid dose from inhalation of ¹³²Te was calculated in the same way as for ¹³¹I with an adjustment to 1.5 for the ratio of the released activities of ¹³²Te and ¹³¹I, radioactive decay rate of ¹³²Te, and the inhalation thyroid dose coefficients (18, 19). The intake of ¹³²Te with cow’s milk and milk products was not considered because transfer of ¹³²Te from feed to cow’s milk is much lower than that for ¹³¹I (20). The intake of ¹³²Te with leafy vegetables was calculated in the same way as the intake of ¹³¹I using the tellurium-specific parameter-values for the ecological model (3) and ingestion thyroid dose coefficients (21).

The individual instrumental thyroid dose (mGy) of cohort member k due to internal exposure from intake of short-lived radioiodine isotopes was calculated as:

$$D_{k,SL}^{inst}=(D_{k,I-132}^{ecol}+D_{k,I-133}^{ecol})\cdot S{F}_k$$ (3)

where $D_{k,SL}^{inst}$ is the instrumental thyroid doses of subject k from intake of ¹³²I and ¹³³I (Gy); SF_k is the scaling factor calculated for ¹³¹I [see Eq. (2)] (unitless); $D_{k,I-132}^{ecol}$ and $D_{k,I-133}^{ecol}$ are the ecological thyroid doses of cohort subject k from ¹³²Te+¹³²I and ¹³³I, respectively (Gy).

Estimation of Uncertainties

Monte-Carlo simulations were used to estimate the uncertainties in thyroid doses with separation of errors to unshared and shared Berkson and unshared classical. The classical and Berkson types of errors were evaluated separately because their influences on estimates of radiation-related risk of thyroid cancer following exposure to ¹³¹I were considerably different (22, 23). According to the method, the same value for each shared parameter was assigned in calculation of one i-th dose set for all cohort members for whom this shared parameter was applicable. Values of unshared parameters for each cohort member were sampled from their distributions and calculated one i-th dose realization for cohort member k, $D_{i,k}^{inst}$_. Set of doses from D_i,1 to D_i,13,204 represents i-th alternative cohort dose realization for 13,204 cohort members. Dose realizations for the same cohort member k, from D_1,k to D_1000,k, represent 1,000 individual stochastic thyroid doses of this individual. This method is consistent with the 2-dimensional Monte Carlo method (24).

Figure 2 shows the scheme of calculation of 1,000 sets of cohort thyroid doses. For each i-th (i = 1 to 1,000) simulation, sets of random values were generated for the model parameters that were the sources of shared and unshared errors, (boxes1–3, 7; Fig. 2). Fixed or precisely known parameters of the dosimetry model (box #4), including the radioactive decay constant of ¹³¹I and radiocesium isotopes, the fraction of ingested ¹³¹I transferred into the blood, the delay between milking and consumption of milk and milk products, etc. [see table A1 found in ref. (3)], were the same in all simulations. The values of shared parameters were the same in given simulation (boxes 1, 2, 7).

FIG. 2.

Scheme of stochastic doses calculation for i-th set (i = 1 to 1,000).

Uncertainties in the ecological thyroid dose (Fig. 2; box 8) were defined by mix of unshared and shared Berkson errors. The calculation of instrumental thyroid dose (Fig. 2; box 9) included, in addition to the ecological dose, the estimate of measured ¹³¹I thyroid activity (Fig. 2; box 7) that also associated with unshared and shared Berkson errors and unshared classical errors (8). So, the resulting uncertainty of individual stochastic thyroid dose was a mixture of unshared and shared Berkson errors and unshared classical error, which are described in the sections below.

Unshared and shared errors associated with measured ¹³¹I thyroid activity.

The following sources of classical and Berkson errors in measured ¹³¹I thyroid activity were considered (8): (i) unshared classical errors associated with device response related to the individual anthropometric characteristics of measured person, i.e., thyroid volume, thyroid location, etc.; (ii) unshared Berkson errors associated with the deviation of the thyroid detector’s position from the proper geometry of measurements; and (iii) shared Berkson errors associated with the corrections of the calibration coefficients and estimation of proper measurement geometry for the uncalibrated gamma-spectrometers or the SRP-68-01 devices. For the SRP-68-01 device, the overall relative unshared error associated with the device response during the measurements of ¹³¹I varied between oblasts from 0.29 to 0.50, while for gamma-spectrometers – from 0.10 to 0.50. The relative shared Berkson errors varied among devices used to measure members of the Ukrainian-American cohort in different oblasts from 0.12 for the SRP-68-01 device to 0.76 for a single-channel scintillation gamma-spectrometer, namely the GTRM-01ts. Further details of the uncertainty budget can be found in table 10 in ref. (8).

Shared errors associated with the parameters of the ecological model.

The ecological model parameters and their distributions are listed in table A2 in ref. (3). Errors were assigned to be shared among all cohort members, except the error in the ¹³¹I deposition in a settlement that was shared among subjects who resided in that settlement. In other words, in the process of calculating a dose realization for the cohort, the same value of ¹³¹I deposition density was applied to all cohort members who resided in the same settlement.

¹³¹I ground deposition density.

The ¹³¹I deposition density in Ukraine from Chornobyl fallout was calculated using the mesoscale atmospheric transport model (12, 13). The ¹³¹I atmospheric transport over the territory of Ukraine was simulated using the model for the period of deposition from April 26 to May 7, 1986, with the ¹³¹I source term that accounted for temporal variation in the rate and height of the release. Detailed fields of meteorological parameters calculated using the mesoscale weather forecast model (WRF) improved the simulation of radionuclides’ dispersion and deposition (11). The results of measurements of ¹³¹I in soil done in Belarusian settlements along the border with Ukraine (25) were used to improve the calculations of ¹³¹I deposition density in the Ukrainian settlements. The geometric standard deviation (GSD) of ¹³¹I ground deposition density was estimated, considering (i) variability of the ratios of the calculated to measured ¹³⁷Cs ground deposition density; and (ii) variability of the ratios of ¹³¹I to ¹³⁷Cs activity in the deposition, to be 2.6 for Zhytomyr, 2.9 for Kyiv, and 3.0 for Chernihiv Oblasts (11).

Unshared Errors in Dose Reconstruction Model

Thyroid-mass values.

This study used the results of ultrasound measurements of thyroid volume that were performed in children aged 5–16 years from Kyiv and Zhytomyr Oblasts by the Sasakawa Memorial Health Foundation (SMHF) (26). Autopsy measurements of thyroid mass in boys and girls under 3 years of age were used to supplement SMHF measurements. Figure 3 shows the age- and sex-dependent thyroid-mass values used in this study to calculate doses. Censored lognormal distributions were adopted for thyroid-mass values for all age and sex groups with parameters given in table 8 in ref. (10).

FIG. 3.

Age- and sex-dependent thyroid-mass values used in this study: arithmetic mean and 95% confidence interval [based on data from (10)].

Parameters of the biokinetic model.

Namely, the ventilation rate, the fraction of inhaled iodine transferred to blood, the thyroid uptake and the half-time of retention of ¹³¹I in the thyroid were sources of unshared errors because of their biological variability between individuals. Values of parameters of the biokinetic model and their distribution are given in tables A3 and A4 in ref. (3).

Questionnaire data.

Personal interviews with all 13,204 cohort members or, for those who were younger 10 years of age ATA, their relatives were conducted between 1998 and 2006 to collect information on the person’s residential history and foodstuff consumption during the first two months after the Chornobyl accident. Information on date of the beginning and duration of administration of stable iodine between April 26 and May 31, 1986, was also collected for 3,866 (29.3% of the total) cohort members who undertook it (3). For these individuals, the uptake of ¹³¹I by the thyroid gland was reduced by using a correction factor, which varies with the duration of administration and the time after intake of stable iodine [see table A1.5 found in ref. (2)]. Some respondents experienced difficulties recalling exact dates of foodstuff consumption. If a respondent did not recall the exact date of relocation or change of consumption habits, he or she was prompted to estimate the period during which the event occurred, e.g., “beginning of May” or “middle of June”. During the stochastic dose calculation, this imprecise response was replaced by the time interval e.g., “May 1×10” or “June 11–20”, and the date was sampled uniformly from this interval. For the answer “I did consume cow’s milk, but I do not remember how much milk I consumed”, the milk consumption rate was imputed from the censored lognormal age-, sex-, and type of settlement-specific distributions of cow’s milk consumptions that were based on the quantitative responses from other respondents (Supplementary Table S1; https://doi.org/10.1667/RADE-22-00125.1.S1). For the answer “I do not remember if I consumed cow’s milk”, it was assumed that the person consumed cow’s milk with a probability equal to the fraction of cow’s milk consumers, and the milk consumption rate was imputed as described above. The same rules were applied for consumption of milk products and leafy vegetables.

RESULTS

Table 1 shows the distribution of the arithmetic means of 1,000 individual stochastic thyroid doses from intake of radioiodine isotopes (¹³¹I, ¹³²Te+¹³²I and ¹³³I) calculated for the Ukrainian-American cohort by oblast of residence ATA. The arithmetic mean of the thyroid doses in the cohort was 0.60 Gy and the median was 0.22 Gy. For 9,474 subjects (71.6% of the total), the doses were less than 0.5 Gy, while for 42 individuals (0.3%) the doses exceeded 10 Gy. The highest thyroid dose in the cohort was 35 Gy. Cohort members living ATA in Zhytomyr Oblasts received the highest doses (mean of 0.84 Gy), followed by residents of Kyiv Oblast (mean of 0.72 Gy) and Chernihiv Oblast (mean of 0.42 Gy).

TABLE 1.

Distribution of Arithmetic Mean of 1,000 Individual Stochastic Thyroid Doses from Intake of Radioiodine Isotopes (¹³¹I, ¹³²Te+¹³²I and ¹³³I) According to Oblast of Residence at the Time of the Accident

Dose interval (Gy)	Zhytomyr Oblast			Kyiv Oblasta			Chernihiv Oblast			Entire cohortb
	N	%	Mean dose (Gy)	N	%	Mean dose (Gy)	N	%	Mean dose (Gy)	N	%	Mean dose (Gy)
<0.02	99	2.7	0.010	112	4.3	0.014	281	4.1	0.011	497	3.8	0.011
0.02–0.049	170	4.7	0.038	260	9.9	0.035	802	11.6	0.036	1,236	9.4	0.036
0.05–0.099	339	9.3	0.076	334	12.7	0.074	1,291	18.7	0.074	1,965	14.9	0.075
0.1–0.199	561	15.4	0.15	385	14.7	0.14	1,570	22.8	0.14	2,525	19.1	0.14
0.2–0.499	1,032	28.3	0.33	637	24.3	0.33	1,577	22.9	0.32	3,251	24.6	0.32
0.5–0.99	667	18.3	0.70	426	16.2	0.72	678	9.8	0.71	1,777	13.4	0.71
1.0–4.9	687	18.9	2.0	410	15.6	1.9	654	9.5	1.9	1,759	13.3	1.9
5.0–9.9	71	1.9	6.9	46	1.8	6.3	34	0.5	6.7	152	1.2	6.7
10+	19	0.5	16	14	0.5	15	8	0.1	15	42	0.3	15
Total	3,645	100.0	0.84	2,624	100.0	0.72	6,895	100.0	0.42	13,204	100.0	0.60

^aIncluding Kyiv-city (14 subjects).

^bIncluding 40 subjects who did not reside in any of the 3 oblasts at the time of the accident.

Figure 4 shows the distribution of the arithmetic means of 1,000 individual stochastic thyroid doses from intake of radioiodine isotopes (logarithm of value) among the cohort members. This distribution was narrower than in TD10 (not shown): doses varied from 4.0 × 10⁻⁴ up to 35 Gy in TD20 and from 4.0 × 10⁻⁴ up to 42 Gy in TD10, and the distribution has a GSD equal to 4.1 (TD20) vs. 4.8 (TD10).

FIG. 4.

Distribution of the arithmetic means of 1,000 individual stochastic thyroid doses from intake of radioiodine isotopes (logarithm of value) for the Ukrainian-American cohort.

Table 2 compares the arithmetic mean and median of 1,000 individual stochastic thyroid doses by oblast of residence and age ATA, TD20 vs. TD10. Essentially the same reduction of doses with age in all study oblasts was observed in both TD20 and TD10. However, the ratio (not shown) of the average thyroid dose in preschoolers (<7 years of age) to that in school-age children (7+ years of age) was the highest in Kyiv Oblast: 3.0 vs. 2.2 (Zhytomyr Oblast) and 1.7 (Chernihiv Oblast) because most school-age children from the high contaminated areas in Kiev Oblast were relocated during the first two weeks of May 1986 to the low contaminated areas in Ukraine, mainly Odessa Oblast and Crimea. This resulted into decrease of the doses among school-age children from Kyiv Oblast by 2–4 times compared with preschoolers and by 7 times with infants (Table 2), because preschoolers and infants stayed with their parents in the contaminated area.

TABLE 2.

Comparison of Arithmetic Mean and Median of 1,000 Individual Stochastic Thyroid Doses According to Age and Oblast of Residence at the Time of the Accident (ATA), TD20 vs. TD10

Oblast	Age ATA (years)	Na	Thyroid dose (Gy)
			Mean		Median
			TD20	TD10	TD20	TD10
Zhytomyr	0	218	1.7	3.3	0.64	1.4
	1–2	547	1.6	2.7	0.66	1.1
	3–4	449	0.92	1.5	0.51	0.76
	5–6	357	0.73	1.0	0.34	0.47
	7–8	447	0.57	0.85	0.30	0.38
	9–10	492	0.51	0.76	0.28	0.33
	11–12	490	0.53	0.76	0.24	0.30
	13–14	398	0.62	0.85	0.26	0.33
	15–16	189	0.55	0.79	0.23	0.36
	17–18	58	0.48	0.77	0.28	0.43
Kyivb	0	80	2.9	2.4	1.3	1.0
	1–2	258	1.5	1.1	0.75	0.63
	3–4	315	1.0	0.84	0.49	0.41
	5–6	290	0.86	0.69	0.42	0.32
	7–8	316	0.49	0.41	0.23	0.17
	9–10	397	0.38	0.30	0.17	0.13
	11–12	358	0.42	0.32	0.15	0.12
	13–14	354	0.37	0.28	0.16	0.12
	15–16	230	0.44	0.35	0.17	0.12
	17–18	26	0.36	0.28	0.095	0.061
Chernihiv	0	495	0.84	0.66	0.40	0.27
	1–2	1,122	0.60	0.50	0.27	0.21
	3–4	998	0.46	0.34	0.19	0.15
	5–6	839	0.34	0.23	0.14	0.098
	7–8	820	0.29	0.20	0.14	0.091
	9–10	833	0.33	0.22	0.13	0.089
	11–12	780	0.31	0.21	0.11	0.069
	13–14	697	0.31	0.21	0.11	0.069
	15–16	242	0.36	0.25	0.086	0.051
	17–18	69	0.17	0.12	0.043	0.024

^aExcluding 40 subjects who did not reside in any of the 3 oblasts at the time of the accident.

^bIncluding Kyiv-city (14 subjects).

Table 3 compares the distribution of arithmetic mean of 1,000 individual stochastic thyroid doses, TD20 vs. TD10, depending on age ATA and sex. The arithmetic mean of the TD20 doses was around 10% less than the TD10 doses, 0.62 Gy vs. 0.67 Gy for boys and 0.58 Gy vs. 0.63 Gy for girls, respectively. There was a marked difference between the doses for boys and girls at some ages. For example, at the age of 15–16 years of age, the thyroid dose in boys was about 50% higher than that in girls because of the larger fraction of cow’s milk consumers and higher consumption rates of milk among boys than girls, 72.7% vs. 67.1% and 0.86 L d⁻¹ vs. 0.49 L d⁻¹, respectively, for rural settlements (Supplementary Table S1; https://doi.org/10.1667/RADE-21-00125.1.S1). However, the cohort-averaged TD20 dose in boys was slightly higher than that in girls, arithmetic mean ± standard deviation (SD) of 0.62 ± 1.3 Gy (median = 0.23 Gy) vs. 0.58 ± 1.3 Gy (median = 0.21 Gy), respectively (P < 0.001 for log-values).

TABLE 3.

Distribution of Arithmetic Means of 1,000 Individual Stochastic Thyroid Doses from Intake of Radioiodine Isotopes Estimated in TD20 and TD10 According to Age at the Time of the Accident (ATA) and Sex

Age ATA (years)	N		Mean thyroid dose (Gy)
			TD20		TD10
	Boys	Girls	Boys	Girls	Boys	Girls
0	396	400	1.3	1.4	1.6	1.6
1–2	957	980	0.96	1.0	1.2	1.3
3–4	877	885	0.74	0.62	0.79	0.64
5–6	710	782	0.53	0.54	0.50	0.52
7–8	772	817	0.42	0.40	0.43	0.42
9–10	854	873	0.44	0.35	0.43	0.36
11–12	779	855	0.46	0.35	0.44	0.36
13–14	722	729	0.46	0.36	0.44	0.36
15–16	354	307	0.52	0.34	0.52	0.35
17–18	71	84	0.35	0.29	0.40	0.39
Entire cohort	6,492	6,712	0.62	0.58	0.67	0.63

Table 4 gives the doses according to the place of residence ATA, type of the settlement, and sex. The highest thyroid doses from radioiodine isotopes (mean = 1.2 Gy, median = 0.55 Gy) were received in Prypiat-town (Kyiv Oblast), which is in the vicinity of the Chornobyl NPP, and its population was evacuated the day after the accident. The cohort members evacuated from the 30-km zone (Kyiv Oblast) around the Chornobyl NPP received higher doses than those who resided in non-evacuated settlements, 0.80 Gy vs. 0.55 Gy for mean and 0.32 Gy vs. 0.20 Gy for median, respectively. Higher doses were observed in rural areas compared to urban areas, mean ± SD of 0.62±1.3 Gy (median = 0.23 Gy) vs. 0.52 ± 1.2 Gy (median = 0.18 Gy), respectively, (P < 0.001 for log-values), due to a higher contamination of cow’s milk with radioiodine isotopes in rural settlements compared milk from commercial trade networks in urban areas. Table 4 also shows the thyroid doses from ¹³¹I intake only. The arithmetic mean ± SD of the thyroid doses from ¹³¹I in the cohort was 0.53 ± 1.1 Gy (median = 0.20 Gy).

TABLE 4.

Comparison of Thyroid Doses from Intake of Radioiodine Isotopes (¹³¹I, ¹³²Te+¹³²I, and ¹³³I) and from Intake of ¹³¹I Broken Down According to Various Groupings of the Cohort Members

		Thyroid dose (Gy) from intake of
		all radioiodine isotopes (131I, 132T+132I, 133I)		131I separately
Parameter	N	Mean	Median	Mean	Median
Evacuated settlements
Prypiat-town	731	1.2	0.55	0.84	0.35
Other evacuated settlements	697	0.80	0.32	0.67	0.27
No evacuation	11,776	0.55	0.20	0.50	0.19
Type of settlement of residence a
Rural	10,486	0.62	0.23	0.55	0.21
Urban	2,698	0.52	0.18	0.41	0.15
Sex
Boys	6,492	0.62	0.23	0.55	0.21
Girls	6,712	0.58	0.21	0.50	0.18
Entire cohort	13,204	0.60	0.22	0.53	0.20

^aExcluding 20 study subjects with an unknown type of settlement of residence outside Ukraine.

Table 5 gives the median contribution of ¹³²Te+¹³²I and ¹³³I to the total thyroid dose by raion of residence ATA. The highest contribution of short-lived radioiodine isotopes (22–40% depending on the age) was estimated for the residents of Prypiat with mainly inhalation intake of radioiodine isotopes. A high contribution (8.4–11%) was also estimated for individuals living ATA in the raions closest to the Chornobyl NPP where the deposition of radionuclides occurred on the first day of the accident (Table 4). For the entire cohort, the median contribution of short-lived radioiodine isotopes to the total thyroid dose was around 5%.

TABLE 5.

Distribution of Median Contribution of Short-Lived Radioisotopes (¹³²Te+¹³²I and ¹³³I) to the Total Thyroid Dose for the Ukrainian-American Cohort Members by Age and Raion of Residence at the Time of the Accident (ATA)

	Contribution of 132Te+132I and 133I for the total thyroid dose for the age group (%)
Raion of residence ATA	0–2 years	3–7 years	8–13 years	14–18 years	All
Zhytomir Oblast
Narodychi	6.7	8.4	9.3	9.8	8.5
Ovruch	6.7	8.4	9.2	9.8	8.4
Kyiv Oblast
Ivankiv	2.1	3.8	4.5	5.2	4.2
Poliske	–	10	11	11	11
Chornobyl	7.5	8.0	11	11	9.2
Prypiat town	22	28	35	40	29
Chernihiv Oblast
Kozelets	1.4	2.6	3.0	3.3	2.4
Ripky	2.5	3.7	4.6	5.0	3.8
Chernihiv	2.2	3.7	3.9	4.6	3.5
Chernihiv City	2.1	3.8	4.4	5.5	3.6
Entire cohort	3.8	5.4	6.2	6.9	5.5

The fitted distribution of individual stochastic doses for each cohort member was approximately lognormal and the GSD of the distribution was used to characterize the uncertainty. The GSDs over all subjects varied from 1.4 to 4.3. Figure 5 compares the GSDs vs. geometric means of individual stochastic doses for the cohort members. For 92.7% of the total study subjects, the GSDs were less than 2.0. However, for a group of individuals from Chernihiv Oblast the GSDs exceeded 3.0 (the upper area on Fig. 5), since the largest errors associated with measurements of the ¹³¹I thyroid activity were in Chernihiv Oblast [table 10 from ref. (8)].

FIG. 5.

Subject-specific uncertainty (GSD) of individual stochastic thyroid doses vs. geometric mean of individual stochastic doses.

Figure 6a shows the cumulative distribution of 1,000 alternative ecological dose realizations for the cohort. The wide distribution with a range of medians across alternative dose realizations of 0.014–1.6 Gy [the coefficient of variation (CV) = 0.92] indicates that sources of shared errors were important contributors to uncertainty of the ecological doses, since almost all parameters of the ecological model were sources of errors that were shared among all cohort members. In contrast, Figure 6b shows a narrow distribution of 1,000 alternative instrumental dose realizations (the range of medians across cohort dose realizations was 0.092–0.27 Gy, CV = 0.10) indicating that the sources of shared errors have a little effect on the uncertainty of instrumental doses.

FIG. 6.

Cumulative percentage of 1,000 alternative realizations for the Ukrainian-American cohort: panel a) ecological dose and panel b) instrumental dose.

Table 6 illustrates the contribution of shared errors to the uncertainty of instrumental thyroid doses. The columns show the distribution of 1,000 alternative cohort instrumental dose realizations, i.e., the 2.5 and 25 percentiles, the median and the 75 and 97.5 percentile. All quantiles show small variation between minimal and maximal values of 1,000 cohort dose realizations (a factor of 2.6–3.1) with CV ranged 0.091–0.14, indicating that the contribution of shared sources to uncertainty of the estimated instrumental doses was small in the entire range of cohort dose realizations.

TABLE 6.

Characteristics of the Distribution of 1,000 Alternative Cohort Instrumental Dose Realizations

Parameter	1,000 alternative cohort instrumental dose realizations (Gy)
	2.5%	25%	Median	75%	97.5%
Minimum	0.0045	0.035	0.092	0.25	1.8
Median	0.0091	0.071	0.19	0.50	3.2
Mean	0.0091	0.070	0.18	0.49	3.2
Maximum	0.014	0.10	0.27	0.70	4.6
SDa	0.0013	0.0070	0.018	0.047	0.29
CVb	0.14	0.10	0.10	0.096	0.091

^aStandard deviation (SD).

^bCoefficient of variation (CV).

DISCUSSION AND CONCLUSIONS

Comparison with TD10 Doses

Figure 7 compares the arithmetic means of 1,000 individual stochastic thyroid doses due intake of radioiodine isotopes calculated in this study (TD20) with those estimated previously (TD10). The difference between two doses was mainly caused by the revision of measured ¹³¹I thyroid activity and the use of thyroid mass-values specific for Ukraine. The same questionnaire data that were collected in 1998–2006 for previous dose assessment (3), were used in this study.

FIG. 7.

Comparison of the arithmetic means of 1,000 individual stochastic thyroid doses due intake of radioiodine isotopes, TD20 vs. TD10. Dashed lines show a factor of 3 difference between the two sets of doses.

Table 7 compares mean and median of two doses, TD20 and TD10, by raion and oblast of residence ATA. The revised thyroid doses resulted, in average, in around 40% decrease for cohort members from Zhytomyr Oblast and an increase of around 24% and 35% for the cohort members from Kyiv and Chernihiv Oblast, respectively. Arithmetic mean of TD20 doses for the cohort was around 8% less than that estimated in TD10, 0.60 Gy vs. 0.65 Gy, respectively; however, global median of TD20 doses somewhat increased compared to TD10: 0.22 Gy vs. 0.19 Gy, respectively. The largest increase, by 1.6 times for mean, in the TD20 compared to TD10 doses was in Kozelets raion in Chernihiv Oblast, while the largest reduction in TD20 doses, almost twice, was in Narodychi raion in Zhytomyr Oblast. The difference between TD10 and TD20 was mainly due to a revision of the measured individual ¹³¹I thyroid activity. For the cohort members resided in Zhytomir Oblast the revised values of ¹³¹I thyroid activity (TD20) were, in average, by 35% lower as compared to those used for previous dose assessment in TD10, while for the cohort members who resided in Kiev and Chernihiv Oblasts, the revised values of ¹³¹I thyroid activity were higher than those used in TD10 by 16% and 28%, respectively (8).

Table 7.

Comparison of Thyroid Doses from Radioiodine Isotopes, TD20 vs. TD10, Averaged over Raion of Residence at the Time of the Accident

Oblast	Raion	N	Thyroid dose from intake of radioiodine isotopes (Gy)
			Mean			Median
			TD20	TD10	TD20/TD10	TD20	TD10	TD20/TD10
Zhytomyr	Narodychi	1,303	1.2	2.5	0.48	0.62	1.3	0.48
	Ovruch	2,313	0.62	0.71	0.87	0.27	0.33	0.82
	Entire oblast	3,645	0.84	1.3	0.65	0.36	0.51	0.71
Kyiv	Ivankiv	727	0.24	0.22	1.1	0.086	0.075	1.1
	Poliske	399	0.93	0.76	1.2	0.46	0.37	1.2
	Chornobyl	716	0.66	0.48	1.4	0.30	0.22	1.4
	Prypiat-town	731	1.2	0.94	1.3	0.55	0.44	1.3
	Entire oblasta	2,624	0.72	0.57	1.3	0.28	0.21	1.3
Chernihiv	Kozelets	1,962	0.22	0.14	1.6	0.10	0.063	1.6
	Ripky	1,653	0.40	0.30	1.3	0.16	0.11	1.5
	Chernihiv	2,206	0.72	0.53	1.4	0.31	0.21	1.5
	Chernihiv-city	1,036	0.21	0.20	1.1	0.11	0.11	1.0
	Entire oblast	6,895	0.42	0.31	1.4	0.16	0.11	1.5
Entire cohortb		13,204	0.60	0.65	0.92	0.22	0.19	1.2

^aIncluding Kyiv-city (14 subjects).

^bIncluding 40 subjects who did not reside in any of the 3 oblasts at the time of the accident.

Figure 8 shows the distribution of scaling factor (SF_k, logarithm of value) that is an indicator of the agreement between the dose estimated using the model and that derived from the thyroid measurement for the cohort member [see Eq. (2)]. This distribution was close to lognormal with median of 1.1 and GSD of 5.2. The range of SF_k-values was rather broad; 90% confidence interval was 0.13–12. The distribution of SF_k-values in this study had a lower GSD of 5.2 vs. 6.4, and the median was closer to one, 1.1 vs. 0.72 in TD20 and TD10, respectively. The narrower distribution of SF_k-values was the reason why the distribution of the arithmetic means of 1,000 individual stochastic thyroid doses in the present study was narrower than in TD10. Table 8 shows median and 25th–75th percentiles for SF_k-values of the thyroid doses estimated in 8 raions where most of the cohort members resided. The use of TD20 resulted in better than in TD10 agreement between ecological and instrumental doses, as median SF_k-values were consistent and close to 1.0 for all raions except Ivankiv. However, for 4,995 out of 13,204 (37.8% of the total) cohort members, SF_k-values were above 3.0 or below 0.33.

FIG. 8.

Distribution of scaling factors (logarithm of value) for the thyroid doses calculated for the Ukrainian-American cohort members.

TABLE 8.

Median and Range of 25th–75th Percentiles for Scaling Factor-Values of the Thyroid Doses Estimated for the Ukrainian-American Cohort Members Resided in Selected Raions

			Scaling factor, SFk
Oblast	Raion	N	Median	25th-75th percentiles
Zhytomyr	Narodychi	1,303	1.1	0.51–2.4
	Ovruch	2,313	1.1	0.54–2.5
Kyiv	Ivankiv	727	1.8	0.80–4.3
	Poliske	399	1.3	0.52–3.7
	Chornobyl	716	0.85	0.27–2.9
Chernihiv	Kozelets	1,962	1.1	0.49–2.4
	Ripky	1,653	1.0	0.51–2.3
	Chernihiv	2,206	1.2	0.53–2.9
	Chernihiv-city	1,036	1.3	0.60–2.9
Total		12,315	1.1	0.52–2.7

Comparison with Other Studies

The thyroid doses from ¹³¹I obtained in this study were compared with doses calculated for a similar Belarusian-American cohort, for which essentially the same dose reconstruction methodology was used (27). The thyroid doses from ¹³¹I were higher in the Belarusian-American cohort, an arithmetic mean ± SD was 0.68±1.5 Gy (median = 0.27 Gy) in Belarus vs. 0.53±1.1 Gy (median = 0.20 Gy) in this study. The stochastic doses estimated in this study are characterized by a smaller uncertainty with an arithmetic mean GSD of 1.6 (vs. 1.8 in Belarus), a geometric mean GSD of 1.4 (vs. 1.7 in Belarus) and a lower fraction of stochastic doses with a GSD greater than 2.0 (7.3% in this study vs. 10.6% in Belarus). Smaller uncertainties in our study are due to smaller errors in the estimates of ¹³¹I thyroid activity and exact known thyroid measurement dates, while in Belarus, dates were assigned for 43.4% of the cohort members (27).

Uncertainties in instrumental thyroid doses from ¹³¹I intake estimated in this study were lower than those reported for studies where thyroid doses were estimated using only modeling. In post-Chornobyl, the GSDs of stochastic thyroid doses ranged from 1.6 to 3.6 and a mean GSD of 1.9 in a case-control study of thyroid cancer among 2,239 Belarusian and Russian children (28). The GSDs of thyroid doses from ¹³¹I intake during residence in contaminated settlements among Ukrainian Chornobyl cleanup workers varied from 1.8 to 4.8 with a mean of 2.6 (29). A mean GSD of 2.2 (range 1.6–5.4) was reported for thyroid doses received by the participants of the Hanford thyroid disease study (30).

The contribution of shared errors to dose uncertainty in this study was small. A similar result was obtained for the Belarusian-American cohort (27) where all study subjects were also measured for ¹³¹I thyroid activity. In opposite, uncertainties in thyroid doses estimated, in the absence of thyroid measurements, using only modeling, were driven by shared errors in the parameters of the dose reconstruction model (28, 31, 32). If the shared parameter was a linear factor in the dosimetry model, then the deviation of the shared parameter from its mean value (mathematical expectation) led to the same deviation of thyroid doses for the cohort, which is the bias within a single simulation. Unshared parameters within a single simulation varied between the subjects and, therefore, did not resulted in the bias. The presence of shared parameters in the dosimetry model is more undesirable than unshared parameters since it leads to a bias in the estimates of radiation-related risk.

Geographical Pattern of Thyroid Doses

Figure 9 shows the geographical pattern of thyroid doses normalized to the ¹³¹I ground deposition density in the settlement of residence (Gy per MBq m⁻²). Normalized thyroid doses varied between settlements within two orders of magnitude. The lowest values, less than 0.1 Gy per MBq m⁻², were observed in Kyiv and Zhytomyr Oblasts, while the highest values, greater than 5.0 Gy per MBq m⁻², were observed in Chernihiv Oblast. One reason for this difference may be the timeliness and effectiveness of countermeasures to restrict exposure to radioiodine isotopes. Most of the settlements with normalized thyroid doses of <0.1 Gy per MBq m⁻² were evacuated shortly from the 30-km zone around the Chornobyl NPP after the accident in late April to early May 1986. Residents of settlements on the route of evacuation (Poliske raion in Kyiv Oblast and Narodychy and Ovruch raions in Zhytomyr Oblast) were also informed about the accident that led to self-restrictions in consumption of locally produced milk and milk products and, accordingly, to a decrease in the thyroid exposure. In Chernihiv Oblast, on the contrary, the untimely introduction of countermeasures, their low efficiency, and grazing of cows, which began earlier than in Kiev and Zhytomyr Oblast, resulted in high normalized thyroid doses (>5.0 Gy per MBq m⁻²), especially in the north-eastern part of Chernihiv raion. Another possible explanation could be larger errors in the estimates of the ¹³¹I ground deposition density-values in Chernihiv Oblasts compared to Zhytomyr and Kyiv Oblasts (11). However, it is difficult to justify what determined the variation of normalized thyroid doses between settlements located to the west, east or north-east of the Chornobyl NPP.

FIG. 9.

Geographical pattern of the thyroid doses normalized to the cumulative ¹³¹I ground deposition density in the settlement of residence of the Ukrainian-American cohort members (Gy per MBq m⁻²).

Strengths and Limitations of the Present Study

The strength of our study is the use of the revised thyroid dosimetry system, TD20 that is based on (i) revised ¹³¹I thyroid activities measured in Ukraine in April–June 1986 in 146,425 individuals; (ii) revised estimates of ¹³¹I ground deposition density in each Ukrainian settlement; and (iii) estimates of age- and sex-specific thyroid masses for the Ukrainian population (8, 9). Another significant advantage of the study is the assessment of the uncertainties in thyroid doses with account to new sources of shared Berkson errors and unshared classical and Berkson errors related to measured ¹³¹I thyroid activity.

However, this study has some limitations:

• 1To calculate the instrumental thyroid dose [see Eqs. (1) and (2)], the assumption was made that the relative shape of the variation of $Q_k^{ecol}(t)$ was precisely estimated using an ecological model and personal interview data. Therefore, the adjustment at time t_m by replacing the calculated ¹³¹I thyroid activity with the measured value [Eq. (2)], was also applied to the time after the measurement date. However, this is correct if (i) the measurement of ¹³¹I thyroid activity was assigned to the right person, and (ii) the answers provided during the personal interview were truthful and reliable.

The first assumption is reasonable when the SF_k-value (see Eq. (2)) is relatively close to 1.0 (i.e., between 0.1 and 10). However, for 97 out of 13,204 individuals (0.73% of the total), the SF_k-values were found to be very high (>100) or very low (<0.01) (Fig. 8). Uncertainties in the parameter values of the ecological model are not sufficient to explain these extreme SF_k-values. The most likely reason the SF_k-value to be either very high or very low is because a thyroid measurement was mistakenly assigned to a cohort member. For example, in a similar Belarusian-American cohort such situation was found in 2.3% of cohort members with thyroid doses from ¹³¹I intake over 5 Gy (33). Because of this limitation, the uncertainties associated with thyroid doses could be underestimated.

The second assumption is also reasonable if there is agreement between the instrumental and ecological doses. However, incomplete or inaccurate answers provided during the personal interviews may result in outliers for SF_k-values. The reliability of the questionnaire-based doses is limited, and imprecise answers could lead to underestimation or overestimation of the instrumental dose-values used in epidemiological analysis in this cohort, which in most instances does not exceed 2–3 times (34).

• 2This study considered the uncertainties in the parameters of the dose reconstruction model to be either shared or unshared. In particular, the thyroid mass was considered as a source of unshared error, and in the Monte Carlo calculations, a thyroid-mass value for each study subject was independently imputed by value randomly selected from the appropriate age- and sex-specific distribution. Although the same distribution was used for all cohort members of a given age and sex, there was some degree of shared uncertainty in the central values and the GSDs of the distribution. This was caused by an extrapolation process, since the thyroid-mass values for Chernihiv Oblast, where ultrasound measurements were not done by SMHF, were assigned to be the same as for Zhytomyr and Kyiv Oblasts. However, because the thyroid-mass values were classified into 33 age and sex categories, the inter-individual correlation associated with the shared uncertainty across the cohort was expected to be weak. Since around 1,000 or more measurements were done in each of 33 category and variability of thyroid-mass values were estimated with high precision (10), it was not necessary to consider shared uncertainty in the mean and GSD of the distributions of thyroid-mass values.
• 3The errors in parameters of the ecological model, e.g.,¹³¹I ground deposition density, were considered in this study as the shared Berkson type. However, these parameters, in addition to Berkson error, could involve bias in the estimates of the arithmetic and geometric mean of the distribution (24). Because these errors were much less than Berkson errors, they were neglected in the study.

In summary, 1,000 sets of thyroid doses from intake of radioiodine isotopes were calculated for the Ukrainian-American cohort using new developed TD20 dosimetry system. The thyroid doses were improved over TD10 as the distribution of SF-values in this study had a lower GSD of 5.2 vs. 6.4 in TD10, and the median was closer to one, 1.1 vs. 0.72 in TD10. The revised thyroid doses and associated uncertainties are being used to assess the radiation-related risk of thyroid cancer and other thyroid diseases in this unique cohort of individuals exposed during childhood and adolescence to Chornobyl fallout.