Association between cumulative radiation dose and haematological parameters among Korean radiation workers: a retrospective cohort study

Abstract

Objective

To evaluate the association between cumulative radiation dose and haematological parameters among radiation workers and compare the prevalences of abnormalities in blood counts with those in the general population.

Design

Retrospective cohort study.

Setting

Nationwide radiation dose registry in Korea, linked with occupational health examination data. Cumulative doses were estimated using annual personal dose equivalent (H_p(10)) records from 1984 onward.

Participants

The study included 20 414 radiation workers, comprising 17 651 men (86.5%) and 2763 women (13.5%), with baseline survey data, dosimetry records and at least one complete blood count (CBC) record between 2014 and 2019.

Primary and secondary outcomes

The primary outcome was the continuous haematological parameters, including white blood cell (WBC), platelet (PLT) and haemoglobin (Hb) counts, in relation to cumulative radiation dose. Associations were evaluated using linear mixed-effects models incorporating repeated measurements and adjusting for age, smoking status and body mass index. The secondary outcome was the prevalence of abnormal blood counts among radiation workers.

Results

Most haematological parameters among radiation workers were within normal ranges. In male workers, cumulative radiation dose was associated with increased Hb levels (β=0.5 mg/dL per 1 mSv; 95% CI 0.006 to 0.9) after adjusting for age, smoking status and body mass index. No significant associations were observed between cumulative dose and WBC or PLT counts in either sex. Overall, compared with the general population, radiation workers had significantly lower standardised prevalence ratios for abnormal WBC and PLT counts.

Conclusion

No substantial adverse changes in haematological parameters were found among radiation workers exposed to prolonged low-dose radiation. The findings suggest that cumulative doses at occupational levels may not substantially affect CBC profiles, although continued monitoring and follow-up are warranted.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study utilised a large, nationwide cohort of Korean radiation workers with repeated haematological measurements over a 6-year period.

• Individual cumulative radiation doses were assessed using registry-based dosimetry data spanning from 1984 to 2019.

• Linear mixed-effects models enabled repeated measures analysis while adjusting for key covariates, including smoking, body mass index and age.

• The follow-up period for haematological data was relatively short (2014–2019), and questionnaire data were collected during 2016–2017, which limits long-term inference.

• Residual confounding from unmeasured factors influencing complete blood count (CBC), such as inflammation or infection, cannot be ruled out.

Introduction

The haematopoietic system is highly sensitive to radiation, and peripheral blood counts can be used to indicate the effects of radiation exposure.¹ Complete blood count (CBC) testing, which measures red blood cells (RBC), haemoglobin (Hb), white blood cells (WBC) and platelets (PLT), is routinely used to assess general health status, detect infection or inflammation and diagnose haematologic and bone marrow diseases.² Since 2013, nuclear workers in South Korea have undergone annual occupational health examinations, including CBC testing, in accordance with the Nuclear Safety Act.

Previous research has investigated whether changes in CBC parameters reflect radiation doses and their potential utility as biological indicators of radiation exposure. Epidemiological studies on A-bomb survivors have demonstrated that elevated total leucocyte and differential cell counts are associated with radiation exposure^{3 4}; however, the effects of low-dose radiation on WBC count remain unclear.⁴ Among populations occupationally exposed to low-dose radiation in China, nonlinear relationships have been reported between RBC, PLT, Hb and cumulative radiation dose, with a range of 0.101–4.908 mSv and a median of 3.313 mSv.^{5 6} A study conducted in Italy highlighted a significantly lower WBC count among radiation-exposed hospital workers compared with controls,⁷ while a study from Serbia observed increased erythrocyte and monocyte counts in periodic compared with baseline check-ups among nuclear medicine professionals.⁸ Overall, the scientific evidence regarding the effects of protracted exposure to low-dose radiation on haematological parameters is limited and inconsistent. In addition, previous studies have focused on medical radiation workers, were constrained by relatively small sample sizes^{6 9} and often lacked adjustments for confounders, such as smoking and body mass index (BMI) in a cross-sectional study design.⁷ Therefore, this study aimed to (1) compare the prevalence of abnormalities in these parameters between radiation workers and the general population, and (2) examine the dose–response relationship between cumulative radiation dose and continuous haematological parameters (WBC, PLT and Hb counts) among workers exposed to low-dose radiation from 2014 to 2019, while accounting for potential confounders.

Methods

Data sources

The data used in this study were derived from the Korean Radiation Workers Study. The study design and population have been described in detail elsewhere.^{10 11} Briefly, nationwide surveys of active radiation workers registered with the Nuclear Safety and Security Commission (NSSC) were conducted through mandatory radiation safety education programmes between 2016 and 2017. During this period, our research team visited each education site across the country to recruit participants. A self-administered questionnaire was distributed at each site, and written informed consent was obtained prior to their enrolment. In total, 42 067 radiation workers were approached and 35 789 (85%) responded to the survey. After excluding duplicate responses, foreign workers and participants for whom personal identification information could not be verified or who declined participation, 20 608 workers (50% of those approached) were enrolled in the final cohort (figure 1).¹¹

Figure 1. Study population selection process. CBC, complete blood count.

For this study, registry-based records of radiation doses from 1984 to 2019 and CBC measurements from 2014 to 2019 were obtained from the Central Registry for Radiation Workers Information, maintained by the NSSC and the Korea Foundation of Nuclear Safety. CBC measurements have been collected annually as part of the legally mandated occupational health surveillance programme under the Nuclear Safety Act of Korea. For comparison with the general population, data from the Korea National Health and Nutrition Examination Survey (KNHANES) were used to derive reference distributions of haematological parameters and define abnormality cut-offs. KNHANES is a population-based, nationwide cross-sectional survey of approximately 10 000 non-institutionalised Korean civilians aged ≥1 year.¹² KNHANES was conducted annually using a multi-stage, stratified probability sampling design and included standardised health interviews and examinations (including CBCs) performed by trained medical staff with regular equipment calibration. In addition, external quality-control programmes for all survey processes—covering data collection, laboratory analyses and data processing—have been implemented in collaboration with the Korea Centres for Disease Control and Prevention and related academic societies.

Variables

Demographic variables included sex, year of birth and education level. Lifestyle factors comprised smoking status (never, former, current), smoking intensity (≤10, 11–20 and ≥21 cigarettes per day) and BMI, derived from self-reported height and weight. Occupational characteristics included facility type (nuclear power plants, industry, industrial radiography, medical institutes, educational institutes, research institutes, public institutes and the military), and the use of protective equipment. Radiation exposure variables included annual personal dose equivalent (H_p(10)) and cumulative dose. The primary outcomes were continuous haematological parameters—WBC, PLT and Hb levels. Secondary outcomes were binary indicators of abnormality for each haematological parameter, defined using clinical reference ranges.

Measurements

Haematological parameters included WBC (×10⁹ cells/L), PLT (×10⁹ cells/L) and Hb (g/L). Abnormal values were defined according to the Korean guidelines for special medical examinations: WBC 3.5–11 ×10⁹ cells/L; PLT 100–400×10⁹ cells/L; Hb 120–160 g/L in women and 130–180 g/L in men.¹³ Each participant contributed between one and six CBC measurements during 2014–2019, with a mean of 4.4 measurements; 7% of participants had only one measurement. Personal radiation dose equivalent (H_p(10)) for all radiation workers has been measured quarterly since 1984 using individual dosimeters. Cumulative dose was calculated as the sum of annual H_p(10) values from the first monitoring year to the year preceding each CBC test.

Data linkage and analytic study population

Questionnaire data for 20 608 workers were linked with information on radiation doses and CBC measurements using identifiers, including worker’s name and date of birth.

20 414 workers were included in the present analysis if they completed the questionnaire with informed consent and had both radiation dose and at least one CBC record available for the period 2014–2019 (figure 1). Accordingly, 194 workers from the original cohort who lacked CBC measurements during this period were excluded. No restriction was applied regarding length of employment. Based on facility information, the occupations of the study participants were classified accordingly. Among 20 414 workers, the distribution by occupational facility type was as follows: nuclear power plants (30.3%, n=6184), industry (19.2%, n=3914), industrial radiography (17.2%, n=3510), medical institutes (14.3%, n=2913), educational institutes (9.6%, n=1955), research institutes (5.5%, n=1118), public institutes (3.2%, n=649) and the military (0.8%, n=171). This study received ethics approval from the Institutional Review Board of the Korea Institute of Radiological and Medical Sciences (IRB No: KIRAMS-2021-03-004). This study was performed in accordance with relevant guidelines/regulations and the Declaration of Helsinki.

Statistical analyses

Age- and sex-standardised prevalence ratios (SPRs) and 95% CIs were calculated to compare the prevalence of abnormalities in haematological parameters between radiation workers and the general population. The SPRs were calculated as the ratio of observed to expected cases. The expected cases were estimated by multiplying the number of the study population in each sex-, age- and calendar year-specific stratum by prevalence rates in the general population collected from the KNHANES from 2014 to 2019. Participants were divided into 10-year age groups, and analyses were restricted to those aged 20–69 years (28 participants excluded). To compare the proportion of abnormal values in haematological parameters between radiation workers and the general population, we calculated sex-specific age-SPRs and 95% CI. To assess changes in haematological parameter values according to cumulative radiation dose, haematological parameters collected annually during 2014–2019 were analysed using the linear mixed-effects model described by Laird and Ware,¹⁴ which accounts for within-person correlations in repeated measurements. Considering that previous studies have shown that CBC parameters vary by sex and age,15,17 we first explored age-related trajectories in our data by plotting individual CBC measurements against age. This exploratory assessment revealed that longitudinal changes in WBC differed markedly by sex, with older men showing a pronounced and curvilinear increase, whereas women exhibited relatively stable, linear patterns. Integrating the literature-based rationale with these data-driven observations, all analyses were stratified by sex. To adequately model the curvature observed among men, we included a linear–quadratic age term (age+age²) for men only, while a linear age term was sufficient for women.

After determining the sex- and age-related functional forms, we proceeded with model specification. Likelihood ratio tests indicated that including age as both a random intercept and a random slope provided the best model fit, allowing individual variability in age-related changes to be captured. As our descriptive analyses suggested differences in CBC parameters across occupational groups, occupation, BMI (kg/m²), and smoking intensity were included as fixed-effect covariates. Participants with missing information on smoking status or BMI (n=1668) were excluded from the covariate-adjusted mixed-effects models, resulting in a final analytic sample of 18 746 workers. No imputation was performed for missing values. With cumulative radiation dose as the primary exposure, each covariate was initially fitted in a univariable model and subsequently included jointly in multivariable models. For categorical variables with more than two levels (smoking status and facility), joint significance was evaluated using Wald χ² tests. Several covariance structures were evaluated, and the independent structure was selected for both men and women because it provided a slightly lower AIC and a more parsimonious and stable model than the unstructured form. All statistical analyses were performed using Stata version 17.0 (StataCorp LLC, College Station, TX, USA).

Patient and Public Involvement

Radiation workers participated in the survey component of the study. They were not involved in the study design, analysis, interpretation or dissemination plans.

Results

Table 1 presents the baseline characteristics of the radiation workers included in 2014. Most radiation workers (89.2%) were men, and approximately two thirds of the participants were under 40 years of age. Approximately 40% of the workers began employment after 2011 and had worked for 5 to 9 years, with an average employment duration of 11.6 years. Approximately two thirds of the participants worked in nuclear power plants, industry and industrial radiography. Among them, 31.5% of males were employed in nuclear power plants, whereas 48.8% of females worked in the medical field. The mean annual dose for radiation workers included in this study decreased from 0.9 mSv in 2014 to 0.4 mSv in 2019, whereas the mean cumulative dose showed some variation by year due to differences in the included participants, but increased slightly from 12.1 mSv in 2014 to 13.2 mSv in 2019 (online supplemental table 1). In 2014, the mean haematological values were 6.7×10⁹ cells/L for WBC in men and 6.4×10⁹ cells/L in women; 243.5×10⁹/L for PLT in men and 260.6×10⁹/L in women; and 153 g/L for Hb in men and 130 g/L in women. Over the study period (2014–2019), annual mean values for WBC ranged from 6.5 to 6.7×10⁹ cells/L, PLT from 245.4 to 255.6×10⁹/L and Hb remained around 150 g/L. Regarding lifestyle characteristics, 42.5% of the workers were current smokers (47.4% among men, 1.9% among women), and 37.2% had never smoked. Among females, 97.1% were non-smokers. The overall proportion of obesity (BMI ≥25 kg/m²) was 36.2%, with a notably higher prevalence in men (39.9%) than in women (5.4%). The proportion of missing data for BMI was 7.5% overall and 17.8% among women, and no adjustment or imputation was performed for these missing values. Meanwhile, the mean age increased from 37.1 to 40.1 years during the study period. The characteristics of radiation workers from 2015 to 2019 are shown in online supplemental table 1.

Table 1. Characteristics of the study population in 2014.

	Total (n=11 626)	Male (n=10 371, 89.2%)	Female (n=1255, 10.9%)
	n (%)	n (%)	n (%)
Age
<30	2563 (22.1)	2030 (19.6)	533 (42.5)
31–40	4776 (41.1)	4369 (42.1)	407 (32.4)
41–50	3141 (27.0)	2893 (27.9)	248 (19.8)
51–60	961 (8.3)	902 (8.7)	59 (4.7)
≥60	185 (1.6)	177 (1.7)	8 (0.6)
Occupation
Public institutes	337 (2.9)	319 (3.1)	18 (1.4)
Education	924 (8.0)	600 (5.8)	324 (25.8)
Military	107 (0.9)	100 (1)	7 (0.6)
Industrial radiography	2404 (20.7)	2354 (22.7)	50 (4.0)
Industry	2131 (18.3)	2061 (19.9)	70 (5.6)
Research	634 (5.5)	548 (5.3)	86 (6.9)
Nuclear power plants	3349 (28.8)	3262 (31.5)	87 (6.9)
Medical	1740 (15.0)	1127 (10.9)	613 (48.8)
Year first worked
1984–1990	349 (3.0)	339 (3.3)	10 (0.8)
1991–2000	2373 (20.4)	2257 (21.8)	116 (9.2)
2001–2010	4241 (36.5)	3870 (37.3)	371 (29.6)
2011–2019	4663 (40.1)	3905 (37.7)	758 (60.4)
Length of employment
<5	1400 (12.0)	1157 (11.2)	243 (19.4)
5–9	4806 (41.3)	4130 (39.8)	676 (53.9)
10–14	2313 (19.9)	2128 (20.5)	185 (14.7)
15–19	1283 (11.0)	1203 (11.6)	80 (6.4)
≥20	1824 (15.7)	1753 (16.9)	71 (5.7)
Smoking status
Never-smokers	4322 (37.2)	3103 (29.9)	1219 (97.1)
Former smoker (unknown)	1894 (16.3)	1881 (18.1)	13 (1.0)
Former smoker (≤10)	112 (1.0)	112 (1.1)	–
Former smoker (11–20)	217 (1.9)	217 (2.1)	–
Former smoker (≥21)	36 (0.3)	36 (0.4)	–
Current smoker (unknown)	550 (4.7)	550 (5.3)	–
Current smoker (≤10)	1374 (11.8)	1364 (13.2)	10 (0.8)
Current smoker (11–20)	2719 (23.4)	2718 (26.2)	1 (0.1)
Current smoker (≥21)	307 (2.6)	307 (3.0)	–
Non-response	95 (0.8)	83 (0.8)	12 (1.0)
Body mass index (kg/m2)
Underweight (<18.5)	207 (1.8)	67 (0.7)	140 (11.2)
Normal	6345 (54.6)	5521 (53.2)	824 (65.7)
Obese (≥25)	4207 (36.2)	4139 (39.9)	68 (5.4)
Not measured	867 (7.5)	644 (6.2)	223 (17.8)
	Mean (SD) Median (IQR)	Mean (SD) Median (IQR)	Mean (SD) Median (IQR)
Cumulative radiation dose (mSv)	12.1 (0.3)	13.4 (28.4)	2.0 (7.2)
	0.9 (11.4)	1.3 (14.1)	0.0 (0.8)
Annual radiation dose (mSv)	0.9 (0.02)	1.0 (2.6)	0.2 (0.8)
	0.0 (0.5)	0.0 (0.6)	0.0 (0.0)
White blood cells (×109 cells/L)	6.7 (1.6)	6.7 (1.6)	6.4 (1.6)
	6.5 (2.1)	6.5 (2.2)	6.3 (2.1)
Platelets (×109 cells/L)	245.4 (51.4)	243.5 (50.7)	260.6 (55.0)
	240 (66)	240 (65)	253 (70)
Haemoglobin (g/L)	151 (12)	153 (10)	130 (11)
	152 (15)	154 (13)	131 (13)
Age	37.1 (9.1)	37.5 (9.0)	33.4 (8.8)
	36 (13)	36 (12)	32 (14)
Length of employment	11.6 (7.2)	12.0 (7.3)	8.7 (5.3)
	9.3 (9.3)	9.8 (9.8)	7.0 (5.1)
Body mass index (kg/m2)	24.4 (3.3)	24.8 (3.1)	20.9 (2.6)
	24.2 (4.0)	24.5 (3.8)	20.5 (2.8)

Most radiation workers (87.2–99.5%) had haematological parameters (ie, CBC test results) within the normal ranges during the study period (2014–2019) (table 2). The prevalence of abnormal WBC and PLT counts was approximately 3% or lower in most years, with no notable year-to-year variation. The prevalence of abnormal Hb counts among female workers was higher than that of other haematological parameters, ranging from 11% to 13%, and was primarily driven by low Hb values, although no distinct temporal trend was observed. Figure 2 shows that the vast majority of participants had CBC values within the normal reference ranges, and no clear trends were observed in haematological parameters with increasing cumulative radiation dose. Figure 3 presents the sex-specific age-SPRs of abnormalities in CBC levels among radiation workers compared with the general population. Overall, the sex-specific results support the finding that radiation workers had lower prevalence of haematological abnormalities compared with the general population, regardless of sex. Among men, the SPRs for both high and low WBC counts were consistently below 1.0 throughout the study period, indicating lower prevalence of abnormalities compared with the general population. Similar patterns were observed for PLT counts, with SPRs for both high and low PLT abnormalities remaining low and stable across years. For Hb, high Hb abnormalities among men showed reduced SPRs ranging from 0.2 to 0.4, whereas the SPRs for low Hb abnormalities fluctuated without a clear pattern. In women, the SPRs for WBC and PLT abnormalities also remained below or around 1.0, although the estimates were less stable due to the small number of cases. Notably, high Hb abnormalities in women exhibited wider year-to-year variation, reflecting small sample sizes and limited statistical precision. In contrast, low Hb abnormalities among women showed a slight increase in 2014 and 2015.

Table 2. Distribution of haematological parameters (WBC, PLT and Hb) among radiation workers by sex, 2014–2019.

Haematological parameters	2014	2015	2016	2017	2018	2019
	n (%)	n (%)	n (%)	n (%)	n (%)	n (%)
Men
WBC (×109 cells/L)
<3.5	28 (0.3)	31 (0.2)	44 (0.3)	47 (0.3)	40 (0.3)	58 (0.5)
3.5–11	11 010 (98.5)	12 674 (98.5)	14 920 (98.4)	14 909 (98.7)	12 879 (98.8)	11 457 (98.7)
>11	144 (1.3)	158 (1.2)	203 (1.3)	148 (1)	118 (0.9)	99 (0.9)
PLT (×109 cells/L)
<100	5 (0.0)	6 (0.1)	8 (0.1)	7 (0.1)	4 (0.0)	5 (0.0)
100–400	11 115 (99.4)	12 798 (99.5)	15 054 (99.3)	14 995 (99.3)	12 931 (99.3)	11 498 (99.1)
>400	68 (0.6)	64 (0.5)	103 (0.7)	104 (0.7)	94 (0.7)	105 (0.9)
Hb (g/L)
<130	116 (1.0)	123 (1.0)	176 (1.2)	149 (1.0)	165 (1.3)	163 (1.4)
130–180	11 035 (98.7)	12 700 (98.7)	14 935 (98.5)	14 908 (98.7)	12 818 (98.4)	11 403 (98.2)
>180	33 (0.3)	43 (0.3)	48 (0.3)	46 (0.3)	47 (0.4)	43 (0.4)
Women
WBC (×109 cells/L)
<3.5	16 (1.2)	19 (1.1)	32 (1.3)	37 (1.6)	21 (1.2)	26 (1.9)
3.5–11	1257 (97.7)	1650 (98.2)	2346 (98.1)	2215 (97.2)	1680 (97.9)	1321 (97.1)
>11	14 (1.1)	11 (0.7)	13 (0.5)	27 (1.2)	16 (0.9)	13 (1.0)
PLT (×109 cells/L)
<100	0 (0.0)	0 (0.0)	1 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
100–400	1265 (98.2)	1655 (98.4)	2354 (98.5)	2241 (98.3)	1661 (96.9)	1334 (98.2)
>400	23 (1.8)	27 (1.6)	36 (1.5)	38 (1.7)	54 (3.2)	24 (1.8)
Hb (g/L)
<120	164 (12.8)	216 (12.9)	277 (11.6)	278 (12.2)	191 (11.1)	154 (11.3)
120–160	1121 (87.2)	1460 (86.9)	2111 (88.3)	2000 (87.7)	1521 (88.7)	1203 (88.5)
>160	1 (0.1)	5 (0.3)	3 (0.1)	2 (0.1)	3 (0.2)	3 (0.2)

WBC, white blood cell; PLT, platelets; Hb, haemoglobin.

Figure 2. Scatter-plot of haematological parameters by cumulative radiation doses among radiation workers. The normal range for each parameter is indicated by the red lines. (a) White blood cells, (b) platelets and (c) haemoglobin.

Figure 3. Sex-specific age-standardised prevalence ratio (SPR) and 95% CIs for abnormalities in haematological parameters among radiation workers, 2014–2019. (a) White blood cells, (b) platelets, (c) haemoglobin (left: males, right: females).

We found that male workers had a significant increase of 0.5 mg/dL (95% CI 0.006 to 0.9) in Hb levels per 1 mSv cumulative radiation dose after adjusting for age, smoking, BMI and facility (table 3). Although a positive association between Hb levels and radiation dose was observed in female workers, it was not statistically significant (β=3.3 mg/dL, 95% CI −2.7 to 9.3). No significant association was found between radiation dose and PLT or WBC counts in either sex. Significant variations in the dose–response coefficients for WBC, PLT and Hb were observed among male workers according to smoking status and occupational category.

Table 3. Change in haematological parameters from the linear mixed model by sex (n=18 746).

	Men			Women
	Coefficient* (95% CI)		P value	Coefficient* (95% CI)		P value
White blood cell
Cumulative dose (mSv)	0.7	(−0.005 to 1.4)	0.05	−8.1	(−16.3 to 0.2)	0.06
Age	7.0	(−4.6 to 18.5)	0.24	−31.6	(−37.6 to −25.7)	<0.001
Age2	−0.1	(−0.3 to −0.005)	0.04	–
BMI (kg/m2)	89.8	(83.5 to 96.1)	<0.001	139.1	(117.6 to 160.5)	<0.001
Smoking (status/number of cigarettes per day)†
Former smoker (unknown)	177.7	(117.7 to 237.8)	<0.001	−603.6	(−1170.6 to −36.7)	0.04
Former smoker (≤10)	−57.2	(−253.8 to 139.4)	0.57	1136.3	(−1388.5 to 3661.2)	0.38
Former smoker (11–20)	185.8	(34.9 to 336.8)	0.02	−962.5	(−3587.3 to 1662.3)	0.47
Former smoker (≥21)	246.8	(−110.3 to 604.0)	0.18	–
Current smoker (unknown)	756.1	(649.6 to 862.6)	<0.001	–
Current smoker (≤10)	370.7	(308.7 to 432.6)	<0.001	23.8	(−475.2 to 522.8)	0.93
Current smoker (11–20)	842.6	(790.0 to 895.1)	<0.001	−40.3	(−1058.4 to 977.9)	0.94
Current smoker (≥21)	1189.7	(1065.9 to 1313.5)	<0.001	1384.3	(−349.1 to 3117.7)	0.12
	Wald χ2=1297.72		<0.001‡	Wald χ2=8.18		0.23‡
Facility§
Public institutes	−315.9	(−426.6 to −205.1)	<0.001	213.8	(−207.5 to 635)	0.32
Education	202.9	(118.1 to 287.7)	<0.001	242.0	(31.6 to 452.3)	0.02
Military	−263.9	(−475.7 to −52.1)	0.02	23.6	(−818.3 to 865.5)	0.96
Industrial radiography	280.0	(219.7 to 340.3)	<0.001	231.4	(−74.5 to 537.3)	0.14
Industry	196.5	(141.2 to 251.9)	<0.001	229.5	(−40.6 to 499.5)	0.10
Research	−294.3	(−386.8 to −201.8)	<0.001	−82.4	(−348.4 to 183.5)	0.54
Medical	51.1	(−20.3 to 122.5)	0.16	252.0	(51.2 to 452.8)	0.01
	Wald χ2=256.18		<0.001‡	Wald χ2=15.46		0.03‡
Random effects
Random slope (age)	11.1	(9.3 to 13.3)		13.8	(9.4 to 10.3)
Between-subject (intercept)	1070.9	(1036.4 to 1106.5)		945.5	(850.5 to 1051.1)
Residual (within-subject)	1018.7	(1012.8 to 1024.6)		1137.9	(1118.7 to 1157.4)
Platelet
Cumulative dose (mSv)	0.004	(−0.02 to 0.03)	0.77	0.01	(−0.3 to 0.3)	0.94
Age	0.4	(0.3 to 0.4)	<0.001	−0.1	(−0.3 to 0.1)	0.47
BMI (kg/m2)	1.0	(0.8 to 1.3)	<0.001	4.3	(3.4 to 5.2)	<0.001
Smoking (status/number of cigarettes per day)†
Former smoker (unknown)	−1.0	(−3.2 to 1.3)	0.40	−2.0	(−25.5 to 21.4)	0.87
Former smoker (≤10)	−19.2	(−26.7 to −11.7)	0.00	37.9	(−60.5 to 136.3)	0.45
Former smoker (11–20)	−11.9	(−17.6 to −6.1)	<0.001	−19.8	(−121.6 to 82.0)	0.70
Former smoker (≥21)	−11.8	(−25.5 to 2.0)	0.09	–
Current smoker (unknown)	5.6	(1.6 to 9.5)	0.01	–
Current smoker (≤10)	2.2	(−0.1 to 4.4)	0.06	−4.7	(−24.9 to 15.4)	0.65
Current smoker (11–20)	4.9	(3.0 to 6.8)	<0.001	−25.5	(−68.1 to 17.2)	0.24
Current smoker (≥21)	3.0	(−1.6 to 7.6)	0.20	35.5	(−33.0 to 104.0)	0.31
	Wald χ2=94.02		<0.001‡	Wald χ2=3.37		0.76‡
Facility§
Public institutes	−7.4	(−11.5 to −3.3)	<0.001	5.6	(−11.4 to 22.5)	0.52
Education	3.6	(0.6 to 6.7)	0.02	1.2	(−7.4 to 9.9)	0.78
Military	−6.1	(−13.7 to 1.5)	0.12	−11.3	(−46 to 23.5)	0.53
Industrial radiography	4.4	(2.2 to 6.6)	<0.001	14.2	(1.6 to 26.8)	0.03
Industry	2.1	(0.04 to 4.1)	0.05	4.4	(−6.6 to 15.5)	0.43
Research	−8.1	(−11.6 to −4.7)	<0.001	−2.5	(−13.3 to 8.4)	0.66
Medical	1.2	(−1.4 to 3.9)	0.37	2.6	(−5.6 to 10.9)	0.53
	Wald χ2=71.23		<0.001‡	Wald χ2=8.57		0.28‡
Random effects
Random slope (age)	0.6	(0.6 to 0.7)		0.5	(0.3 to 0.8)
Between-subject (intercept)	38.4	(37.2 to 39.7)		44.5	(41.4 to 47.7)
Residual (within-subject)	24.5	(24.4 to 24.7)		28.4	(27.9 to 28.9)
Haemoglobin
Cumulative dose (mSv)	0.5	(0.006 to 0.9)	0.05	3.3	(−2.7 to 9.3)	0.28
Age	−15.6	(−16.9 to −14.3)	<0.001	−2.5	(−6.8 to 1.7)	0.24
BMI (kg/m2)	46.7	(42.7 to 50.7)	<0.001	24.1	(8.8 to 39.4)	<0.001
Smoking (status/number of cigarettes per day)†
Former smoker (unknown)	8.9	(−29.8 to 47.6)	0.65	350.3	(−50.0 to 750.7)	0.09
Former smoker (≤10)	35.4	(−94.3 to 165.1)	0.59	310.9	(−1385.7 to 2007.5)	0.72
Former smoker (11–20)	8.5	(−91.4 to 108.4)	0.87	−113.8	(−1937.0 to 1709.4)	0.90
Former smoker (≥21)	4.2	(−233.4 to 241.8)	0.97	–
Current smoker (unknown)	251.2	(183.1 to 319.3)	<0.001	–
Current smoker (≤10)	135	(95.5 to 174.6)	<0.001	181.5	(−167.0 to 529.9)	0.31
Current smoker (11–20)	214.8	(181.2 to 248.3)	<0.001	797.3	(82.2 to 1512.4)	0.03
Current smoker (≥21)	279.3	(199.1 to 359.5)	<0.001	607.7	(−553.2 to 1768.7)	0.31
	Wald χ2=239.86		<0.001‡	Wald χ2=9.67		0.14‡
Facility§
Public institutes	−64.8	(−136.1 to 6.6)	0.08	−356.6	(−658.1 to −55.2)	0.02
Education	−32.1	(−85.6 to 21.4)	0.24	−237.0	(−384.9 to −89.1)	<0.001
Military	−256.6	(−389.9 to −123.3)	<0.001	−56.5	(−644.4 to 531.5)	0.85
Industrial radiography	23.4	(−15.1 to 61.9)	0.23	−315.1	(−530.4 to −99.9)	<0.001
Industry	76.0	(40.5 to 111.5)	<0.001	−37.8	(−228.8 to 153.1)	0.70
Research	−50.0	(−110.1 to 10.1)	0.10	−162.9	(−350.4 to 24.7)	0.09
Medical	55.1	(9.1 to 101.0)	0.02	−70.2	(−211.6 to 71.3)	0.33
	Wald χ2=55.27		<0.001‡	Wald χ2=25.76		<0.001‡
Random effects
Random slope (age)	10.4	(9.6 to 11.3)		13.7	(11.2 to 16.8)
Between-subject (intercept)	641.7	(619.5 to 664.7)		643.0	(575.1 to 718.8)
Residual (within-subject)	579.0	(575.7 to 582.4)		665.2	(653.9 to 676.7)

^*Adjusted for occupation, BMI, smoking intensity and cumulative doses; Age included as fixed (linear-quadratic for males, linear for females) and random slope.

^†A group of never-smokers was used as a reference.

^‡Wald χ2 tests were used as joint test to assess overall effects of categorical variables.

^§A group of nuclear power plants was used as a reference.

BMI, body mass index; CI, confidence interval.

To improve interpretability of the regression coefficients from the linear mixed model, white blood cell count and haemoglobin measurement were expressed in cells/μL and mg/dL, respectively.

Regarding the effects of confounding factors on haematological parameters, former and current smoking were associated with significantly higher WBC counts in men, compared with non-smokers. In men, a clear dose-response relationship was observed between the number of cigarettes smoked per day and WBC counts among current smokers, with increases of 370.7 cells/µL, 842.6 cells/µL and 1189.7 cells/µL for those smoking ≤10 cigarettes, 11–20 cigarettes and ≥21 cigarettes, respectively, compared with non-smokers. Additionally, WBC counts significantly increased by 89.8 cells/µL in men and 139.1 cells/µL in women for each one unit increase in BMI. The PLT levels significantly increased with increasing BMI. Similarly, Hb levels increased with BMI and smoking but decreased with age.

When fitted separately, several variables showed statistically significant associations with haematological parameters. For example, cumulative dose was significantly positively associated with WBC counts among men, while age, smoking status/intensity and BMI were also significant determinants of WBC levels. In men, platelet counts showed a significant negative association with cumulative dose. For haemoglobin, unlike the multivariable model, a negative association with cumulative dose was observed in the univariable model (online supplemental table 2).

Discussion

In our study, approximately all radiation workers (87–99%) had normal WBC, PLT and Hb counts. The SPRs for abnormalities in WBC, PLT and Hb levels were generally lower among radiation workers than among the general population between 2014–2019. Additionally, among male workers, Hb levels tended to increase with cumulative radiation doses.

In 2014, the mean WBC, PLT and Hb levels in men were 6.7×10⁹ cells/L, 243.5×10⁹/L and 153 g/L, respectively. In women, the corresponding values were 6.4×10⁹ cells/L, 260.6×10⁹/L and 130 g/L, respectively. These findings are consistent with those of previous studies that reported higher WBC and Hb levels in men and higher PLT levels in women.^{4 6 15 18} In the A-bomb survivor cohort, the mean WBC count ranged from 6.47 to 7.21×10⁹/L for men and from 5.39 to 6.35×10⁹/L for women.⁴ A study of Chinese radiation workers reported average WBC and PLT counts of 6.37×10⁹/L and 226.23×10⁹/L for men and 6.04×10⁹/L and 246.09×10⁹/L for women, respectively.⁶ Sex differences are generally caused by genetic or hormonal (androgen and oestrogen) influences, which either stimulate or lower each haematological parameter.^{15 18}

In this study, the CBC values among radiation workers were mostly within normal ranges. This may reflect the considerably low doses of radiation exposure and the effectiveness of radiation protection measures in place for Korean radiation workers. In Korea, radiation exposures are monitored monthly using dosimeters assigned to each radiation worker. The Korea Institute of Nuclear Safety investigates workers and work sites that exceed the dose limit of 5 mSv per quarter or 20 mSv per year in accordance with the guidelines of the International Commission on Radiological Protection (ICRP).¹⁹ However, the mean effective dose received by Korean radiation workers in our study ranged from 0.94 to 0.42 mSv and 99.86–99.97% were exposed to <20 mSv per year over 6 years, which is considerably below the limits set by the ICRP. Age-adjusted means of haematological parameters by exposure levels (<100, 100‒200 and ≥200 mSv) indicated no substantial changes in haematological parameters by exposure level among male workers (online supplemental table 3). Although WBC counts were relatively increased in the 100‒200 mSv group (6.8×10⁹ cells/L) than in the <100 mSv group (6.5×10⁹ cells/L), they appeared lower in the ≥200 mSv group (6.5×10⁹ cells/L), with a p value of 0.038 among male workers. The average values of all parameters remained within the normal range, and no significant differences were observed for PLT (p value=0.436) or Hb (p value=0.450). Moreover, few female workers are exposed to ≥100 mSv, thereby making it difficult to determine a statistical trend. The maintenance of haematological parameters within the range of normal values could be explained by the fact that the radiation doses were considerably low to induce measurable alterations in the blood count.¹

Although most of the parameters were normal in 98–99% of the workers, only 87–89% of the female workers exhibited normal Hb levels, suggesting that 11–13% of them had Hb levels below 120 g/L. Average Hb concentrations can vary throughout life and differ between men and women.²⁰ Several studies on premenopausal women in Asia, Europe and North America reported cut-offs lower than the WHO recommendation of 120 g/L.²⁰ This is likely due to common iron deficiency resulting from menstrual bleeding and pregnancy. The relatively higher proportion of female workers with Hb levels below the cut-off in our study may be explained by their relatively young age, with a mean of 34.8 years, and more than 90% under 50 years.

Overall, significantly lower SPRs of abnormalities for haematological parameters were observed in Korean radiation workers compared with the general population. The lower risk of non-cancer health effects compared with the general population, a typical phenomenon in occupational cohorts, was similar to that of radiation workers in other occupational studies.21,24 The SPRs for some parameters, such as low PLT count (range of SPR during 2014–2019=0.10–0.30) and high Hb count (range of SPR during 2014–2019=0.23–0.40) among male workers, showed the lowest estimates. The magnitude of the healthy worker effect may vary depending on several factors, such as disease type, age at observation and time since initial employment.²⁵ It tends to be stronger for non-cancer diseases, younger populations and shorter time since employment.^{22 23} In our cohort, the relatively young age (mean age, 37.1 years) of the workers and short follow-up (mean 4.4 years) may have contributed to the markedly low SPR. In addition, as our study population consisted of actively employed workers, the healthy worker effect may have had a greater impact on our results.

A few studies have compared the means of haematological parameters between workers exposed and not exposed to radiation.^{9 26 27} Most of these investigations have focused on healthcare workers and have yielded inconsistent results. A recent study reported no statistically significant differences between the mean WBC, PLT and Hb levels in radiation workers and the control group.²⁷ Moreover, contrary to our results, previous studies observed decreased WBC count and Hb levels in hospital workers exposed to radiation.^{9 26} These discrepancies could be attributed to several factors, including the study design, number of study participants and exposure characteristics of the participants.

Our study applied a mixed-effects model to examine the relationship between repeatedly measured radiation doses and haematological parameters. We found a significant increase in Hb levels (0.5 mg/dL per 1 mSv (95% CI 0.006 to 0.9)) with cumulative radiation dose among male workers. In contrast, the A-bomb cohort study reported a significant decrease in Hb levels in the 1 Gy exposure group compared with the non-exposed group.²⁸ However, our population was exposed to substantially lower doses, with a mean cumulative dose of 18.2 mSv in the exposed group, compared with 680 mGy in the A-bomb cohort.²⁸ A study of industrial irradiation workers exposed to low doses (0.101–4.908 mSv) showed that Hb counts initially increased and subsequently decreased as the cumulative radiation dose increased.⁵ Considering these results, where non-linear patterns were observed in the low-dose range, our study suggests the need for continuous observation of changes in haematological indicators according to cumulative dose in the future. In contrast to previous studies on the A-bomb cohort, which reported a linear dose-response association wherein mean WBC counts increased per Gy,²⁹ no significant association was observed between cumulative radiation dose and WBC counts among the radiation workers in our study (β=0.7 cells/µL per 1 mSv (95% CI −0.005 to 1.4) for men; β=−8.1 cells/µL per 1 mSv (95% CI −16.3 to 0.2) for women). The elevated WBC count in the A-bomb cohort may have been due to persistent inflammation after acute exposure to high-dose radiation.^{30 31} However, radiation effects on WBC counts were apparent only at high doses (>2 Gy) and not at low doses, especially <1 Gy⁴ in the A-bomb cohort. Similarly, in a study of industrial irradiation workers exposed to low-dose radiation, no significant differences were observed in WBC counts according to the cumulative radiation dose categories or the dose-response relationship with the cumulative radiation dose.⁵ Consistent with these observations, our findings suggest that radiation doses in this study were markedly low to induce a significant increase in WBC counts.

Although an increase in Hb levels was observed in this low-dose radiation-exposed population, this pattern may be more plausibly attributable to other factors such as smoking—an established factor with a stronger impact on Hb elevation—rather than to radiation exposure itself. Smoking has been shown to affect Hb counts.³² A study on the A-bomb cohort indicated that smoking has a greater effect on elevating Hb levels than radiation.²⁸ In our study, male workers, comprising approximately 90% of our population, had smoking rates of 46.5–48.0%, which were higher than those in the general population, at 35.7–43.2%.³³ Moreover, smokers received higher radiation doses (14.8 mSv) than non-smokers (6.4 mSv) did. Although smoking was adjusted for in the analysis of the association between radiation doses and CBC parameters, residual confounding may remain. Quantitative effects of smoking, such as the number of smoking years, may influence haematological parameters within the categories classified by smoking status (ie, former and current) and the number of cigarettes smoked (ie, ≤10, 11–20, ≥21).

Although several studies have examined haematological parameters in radiation-exposed populations, most were based on cross-sectional data. While some analyses of A-bomb survivors employed repeated measures and mixed-effects models, direct comparisons with our findings remain limited due to key differences in exposure patterns (acute vs chronic), radiation dose levels and the timing of outcome assessment. Studies focusing on occupational populations exposed to chronic low-dose radiation with repeated haematological measurements remain scarce. In this context, our study fills an important gap in the existing literature and provides new insights into the long-term haematological effects of low-dose occupational radiation exposure.

This study had some limitations. Information on smoking and BMI was obtained from a self-administered questionnaire; however, the overall responses from the self-administered questionnaire showed moderate-to-high reliability in our study participants.³⁴ BMI was missing for 7.4% of participants (17.8% among women), representing the largest proportion of missing covariate data. Patterns of cumulative radiation dose, WBC and platelet counts differed between participants with and without BMI data, whereas haemoglobin levels did not differ significantly. In linear regression analyses, however, the estimated associations of cumulative dose with WBC, platelet count and haemoglobin were similar regardless of whether participants with missing BMI were included or excluded, with no meaningful changes in statistical significance, suggesting minimal impact of BMI missingness on the dose–response estimates. Since this study utilised existing data sources, detailed information on other potential confounders relevant to studies on radiation and haematological markers, such as dietary intake, inflammatory conditions or environmental background radiation, was not available. Moreover, this study focused on occupational external radiation exposure, measured as personal dose equivalent H_p(10), while internal exposure was not considered due to data limitations. Despite these limitations, the findings provide valuable fundamental evidence for future research focused on thoroughly elucidating haematological marker changes caused by low-dose occupational exposure. Additionally, the relatively short follow-up of 4.4 years is a limitation in interpreting our results. Therefore, confirming these findings in future studies with longer follow-up, including retired workers, is necessary. Continued monitoring is also warranted, given the relatively short observation period and the potential for long-term or delayed haematological effects that could not be fully captured in this study. The relatively small number of female workers resulted in wide CIs, and thus the sex-specific results for females should be interpreted with caution. Furthermore, despite the simplicity and low cost of CBC tests, their effectiveness in evaluating the health effects of low-dose exposure remains debatable. A review study raised concerns about the efficacy of CBC in detecting low-dose radiation effects, suggesting the need for more precise tests like cytogenetic assays.³⁵ Studies on chronic low-dose radiation have shown inconsistent effects on WBC counts and other blood parameters. Nevertheless, consecutive blood cell count results are valuable for monitoring the health of radiation workers and ensuring adherence to safety standards.³⁶

This is the first longitudinal study to demonstrate changes in haematological parameters with cumulative radiation doses among Korean radiation workers. This study had a larger sample size (approximately 20 000 workers) than did previous studies. We used the occupational health examinations provided by the government for all radiation workers, minimising selection bias. In addition, major confounding factors, such as smoking and BMI, were included. In particular, BMI is a significant risk factor for increased WBC count.⁴ In this study, an increase of 1 kg/m² in BMI was associated with an increase in WBC counts by 89.8 cells/µL in men and 139.1 cells/µL in women. Thus, by adjusting for these confounding factors, the association between radiation dose and CBC was more accurately assessed.

Conclusion

Radiation workers have healthier haematological parameters than did the general population. We found a limited impact of prolonged exposure to low-dose radiation on haematological parameters in the workplace among Korean radiation workers. However, because actively employed workers were targeted in this study, relatively stronger healthy worker effects might have been observed. Thus, further follow-up is needed with a more specific endpoint in blood measurements (eg, leucocyte subsets and inflammation markers) or inflammatory diseases with exposure to low-dose radiation. In addition, exploring the mechanisms underlying the changes in the haematopoietic system that may occur following exposure to low levels of radiation is warranted.

Data availability statement

The individual raw data underlying this study cannot be publicly available as they contain personal and sensitive information about radiation workers. Proposals for possible collaboration in further analyses of the cohort should be addressed to Songwon Seo (seo@kirams.re.kr).