Article Text

PDF

Original article
Work history and mortality risks in 90 268 US radiological technologists
  1. Jason J Liu1,
  2. D Michal Freedman1,
  3. Mark P Little1,
  4. Michele M Doody1,
  5. Bruce H Alexander2,
  6. Cari M Kitahara1,
  7. Terrence Lee1,
  8. Preetha Rajaraman1,
  9. Jeremy S Miller3,
  10. Diane M Kampa2,
  11. Steven L Simon1,
  12. Dale L Preston4,
  13. Martha S Linet1
  1. 1Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA
  2. 2Division of Environmental Health Sciences, University of Minnesota School of Public Health, Minneapolis, Minnesota, USA
  3. 3Information Management Services, Inc., Rockville, Maryland, USA
  4. 4Hirosoft International, Eureka, California, USA
  1. Correspondence to Dr Jason J Liu, Division of Cancer Epidemiology and Genetics, Radiation Epidemiology Branch, 9609 Medical Center Drive, Room 7E520, Rockville, MD 20850, USA; jasonjliu.jjl{at}gmail.com

Abstract

Objectives There have been few studies of work history and mortality risks in medical radiation workers. We expanded by 11 years and more outcomes our previous study of mortality risks and work history, a proxy for radiation exposure.

Methods Using Cox proportional hazards models, we estimated mortality risks according to questionnaire work history responses from 1983 to 1989 through 2008 by 90 268 US radiological technologists. We controlled for potential confounding by age, birth year, smoking history, body mass index, race and gender.

Results There were 9566 deaths (3329 cancer and 3020 circulatory system diseases). Mortality risks increased significantly with earlier year began working for female breast (p trend=0.01) and stomach cancers (p trend=0.01), ischaemic heart (p trend=0.03) and cerebrovascular diseases (p trend=0.02). The significant trend with earlier year first worked was strongly apparent for breast cancer during baseline through 1997, but not 1998–2008. Risks were similar in the two periods for circulatory diseases. Radiological technologists working ≥5 years before 1950 had elevated mortality from breast cancer (HR=2.05, 95% CI 1.27 to 3.32), leukaemia (HR=2.57, 95% CI 0.96 to 6.68), ischaemic heart disease (HR=1.13, 95% CI 0.96 to 1.33) and cerebrovascular disease (HR=1.28, 95% CI 0.97 to 1.69). No other work history factors were consistently associated with mortality risks from specific cancers or circulatory diseases, or other conditions.

Conclusions Radiological technologists who began working in early periods and for more years before 1950 had increased mortality from a few cancers and some circulatory system diseases, likely reflecting higher occupational radiation exposures in the earlier years.

Keywords
  • mortality
  • neoplasms
  • circulatory system diseases
  • radiologic technologists
  • work history

Statistics from Altmetric.com

Keywords

What this paper adds

  • This analysis of the largest medical radiation worker cohort addresses the previously little-studied topic of work history and long-term mortality risks from cancers, circulatory system diseases and a broad range of other serious non-malignant diseases among medical radiation workers.

  • Based on internal analyses, US radiological technologists who began working in earlier calendar years had increasing risks of dying from female breast cancer, stomach cancer, ischaemic heart disease and cerebrovascular disease.

  • Risks of breast cancer, leukaemia, ischaemic heart disease and cerebrovascular disease rose significantly with increasing number of years worked before 1950.

  • The findings suggest that protracted occupational exposure to radiation before the implementation of more stringent radiation protection measures in the late 1950s may be linked with mortality from specific cancers and circulatory system diseases.

  • To our knowledge, this is the only cohort of medical radiation workers with a large number of radiation-exposed female workers, questionnaire-derived work history and extensive information on potential confounding factors, and our findings contribute to the knowledge of long-term mortality risks among medical radiation workers, particularly women.

Introduction

There are more than 2 million medical radiation workers worldwide with potential occupational exposures to protracted low-dose ionising radiation.1 Clinical reports of leukaemia in early medical radiation workers established ionising radiation as a carcinogen more than 100 years ago.2 ,3 However, there have been relatively few large epidemiological studies of medical radiation workers with long-term follow-up, and most of these have not evaluated disease risks in substantial numbers of female workers. Previous studies have primarily focused on cancers,4–10 although some investigations have examined other outcomes.11–14 A few studies of medical radiation workers and other radiation-exposed populations have suggested that low- and moderate-dose radiation exposure may also increase mortality from diseases of the circulatory,14 ,15 digestive16 ,17 and respiratory16 ,17 systems.

The United States Radiologic Technologists (USRT) study represents the largest cohort study of medical radiation workers with fractionated low-dose rate radiation exposures that have likely cumulated to low-to-moderate levels, and one of only a small number of such cohorts that includes substantial numbers of female medical radiation workers.10 We previously conducted a mortality follow-up in this cohort through 1997 that examined only a few outcomes in relation to work history factors.7 ,14 We found that those who worked before 1950 had elevated mortality risks for breast cancer, leukaemia7 and circulatory system diseases, particularly cerebrovascular diseases.14 Using updated mortality data through 2008, we have extended our previous 16-year follow-up of the USRT to include an additional 11 years and have examined a broader range of cancers, as well as circulatory, digestive, respiratory and infectious diseases.

The USRT is the only cohort of medical radiation workers to have completed questionnaires inquiring about work history, job practices, and other risk factors for cancer and other diseases. Work history data obtained from the cohort are a surrogate for estimated occupational radiation exposure and captures the variability in work practices that contribute to differences in radiation exposures over time. In this paper, we assess the work history factors as a proxy for occupational radiation exposure.

Methods

Study population

Details of the study population, methods for tracing the technologists, content of the questionnaires, and information about the work history and job practices have been previously published7 ,14 ,18 and can be found online (<http://www.radtechstudy.nci.nih.gov>). A brief description of the study methods is provided below.

The USRT cohort consists of 146 022 US residents certified for two or more years by the American Registry of Radiologic Technologists (ARRT) during 1926–1982. The ARRT certifies technologists in radiography, nuclear medicine, radiation therapy and, more recently, in other subspecialties.

Follow-up is conducted through annual ARRT certification renewals. Technologists who drop their certification are linked with the records of the Social Security Administration and, if reported to be deceased, are linked with the National Death Index Plus to determine cause of death.

A baseline questionnaire was mailed to 133 298 cohort members who were believed to be alive and for whom a current address was available (93% of the target population) during 1983–1989. The 90 268 technologists (68% participation rate) who responded were included in the current analysis focusing on internal comparisons within the cohort. The baseline questionnaire included questions about work history and practices, cigarette smoking history, and other known or potential chronic disease risk factors. This study has been approved annually by the human subjects review boards of the National Cancer Institute and the University of Minnesota.

Work history factors

Work history information examined in this study was based on responses to the baseline questionnaire. We evaluated year first worked as a radiological technologist (<1940, 1940–1949, 1950–1959, 1960–1969, ≥1970), total years worked as of baseline questionnaire completion (<10, 10–19, 20–29, ≥30), age first worked as a radiological technologist (<20, 20–24, 25–29, ≥30 years old), number of years worked as a radiological technologist (0, 1–4, ≥5) before 1950, in the 1950s, in the 1960s, and in 1970 or later, and number of times the technologist held patients for X-rays (0, 1–9, 10–24, 25–49, ≥50).

Disease mortality follow-up and validation

A broad range of mortality outcomes with at least 45 deaths were evaluated (see outcomes assessed and corresponding International Classification of Diseases (ICD)-9 and ICD-10 codes in online supplementary table A1). We examined mortality from all causes, all and specific cancers as well as circulatory, digestive, respiratory and infectious diseases. We analysed specific cancers that have been commonly associated with radiation exposure in studies of exposed populations1 (table 2), as well as other types of cancer (table 3). The rationale for assessing risks for circulatory system diseases was that some of these were previously associated with work history in the US radiological technologists,14 and because these conditions have been linked with radiation exposure in the atomic bomb survivors and other populations with low-to-moderate radiation exposure.15 We conducted exploratory analyses of certain malignant and non-malignant diseases based on reports of elevated risks in the atomic bomb survivors (tables 3 and 4).16 ,17 Underlying causes of death were obtained from death certificates or the National Death Index Plus and coded according to the ICD revision in use at the time of death.19 ,20

Potential confounders

Information on potential confounders was obtained from baseline questionnaire responses. We adjusted for age during follow-up (continuous), birth year (<1930, 1930–1934, …, ≥1955), pack-years smoked cigarettes (0, 0–9, 10–19, 20–29, 30–39, ≥40), body mass index (BMI) in kg/m2 (0–18.4/underweight, 18.5–24.9/normal, 25–29.9/overweight, 30+/obese), race (white, other) and gender. We also considered other potential confounders for breast cancer mortality: age at menarche, parity and age at first birth. However, because the effect estimates changed minimally after adjusting for these disease-specific potential confounders, we adjusted for the same potential confounders for all outcomes.

Statistical analysis

The primary approach of our investigation was to conduct internal comparisons within the cohort of 90 268 technologists who completed the baseline questionnaire. For the internal analyses, we compared mortality risks across categories of each work history variable in relation to a referent category. In secondary analyses, we focused on external comparisons of mortality observed in the US radiological technologists versus that of the US general population.

Cox proportional hazards models21 were used for multivariable internal cohort analyses. Age during follow-up, which was defined by subjects’ ages at baseline questionnaire response and study exit, was used as the time scale for all analyses, with follow-up from the age at time of completion of the baseline questionnaire through the earliest of the age of death, age at the date of last known vital status or age at the end of 2008. Analyses were stratified by birth cohort to control for secular changes in mortality and were parametrically adjusted for pack-years of cigarettes smoked, BMI, race and gender. Analyses of year first worked were adjusted for total number of years worked, and vice versa. Analyses of number of years worked in a specific time period were adjusted for numbers of years worked in all other time periods. Any technologists who were at least 14 years old in a specific time period were considered eligible for work in that time period. To evaluate further secular changes in mortality from breast cancer22 and circulatory diseases23 among different birth cohorts and over the long time span of the follow-up, we conducted analyses that estimated mortality risks for two time periods (1983–1989 through 1997 (corresponding to the period of follow-up evaluated by Mohan et al7 and Hauptmann et al14) and 1998–2008). Technologists with unknown or missing information on work history variables (which ranged from 0.1% to 2.7% depending on the variable) were excluded from the analysis of the specific variables. Indicator variables were included for unknown values of potential confounding variables in all models. The HR and 95% CI were reported for the risk of disease mortality using categorical analyses. All p values were calculated from the partial likelihood ratios and were two-sided. Tests for linear trend were based on the underlying continuous variables. Analyses were performed using SAS V.9.3 software (SAS Institute, Cary, North Carolina, USA).

For the external comparisons, we carried out two sets of analyses, one involving the entire cohort of US radiological technologists (N=146 022) and a second restricted to the subset of 90 268 US radiological technologists who completed the baseline questionnaire. SMRs were calculated for the period 1950 (when nationwide mortality estimates became available) through 2008 for the entire population of 146 022 technologists after excluding 92 technologists with missing dates of birth or death, 138 who died before 1950, 55 who died within 2 years after certification and 11 who were older than 120 years of age at the end of follow-up (see online supplementary table A2). SMRs were also calculated for the period between the date the subject completed the questionnaire during 1983–1989 through 2008 for the subset of 90 268 technologists who completed the baseline questionnaire (see online supplementary table A2). To calculate the SMRs, the technologists were followed from the date of certification plus 2 years (for the full cohort) or the date of completion of the baseline questionnaire (for the questionnaire respondents) until the earlier of the date of death or 31 December 2008. There were 957 technologists in the entire cohort and 179 among baseline questionnaire respondents who were known to be deceased but with an unknown cause of death. These individuals were included in analyses of all-cause mortality but were excluded from analyses of specific causes of death. The numbers of expected deaths were calculated for age-, race-, gender- and calendar time-specific categories using US mortality rates. Expected deaths for technologists with unknown race were calculated using rates for whites. SMRs were computed as the number of observed deaths divided by the number of expected deaths, stratified by attained age (0–4, 5–9, …., 85+), race (white/non-white), gender and attained calendar year (1950–1954, 1955–1959, …, 2005–2008). Exact methods were used to calculate 95% CI.

Results

Among the 90 268 radiological technologists (69 502 women, 20 766 men) who completed the baseline questionnaire, 75 924 technologists were alive on 31 December 2008; 9745 were deceased with a known date of death; five were deceased with unknown date of death; and 4594 subjects were last known to be alive before 2008. Follow-up for the last group stopped at the date last known alive. The median follow-up from baseline questionnaire completion to study exit was 24 years, with 2 030 646 person-years accrued. For the 9745 who were deceased, 9566 had a known cause of death, with 3329 deaths from cancer and 3020 deaths from circulatory system diseases.

Demographic, lifestyle and work history characteristics for the 90 268 technologists who responded to the baseline questionnaire are shown in table 1. Participants were predominantly women (77%) and white (95%). Their mean age at questionnaire response was 39.0 years, with nearly 80% born in 1940 or later. In all, 50% of the technologists began working in 1970 or later, while 31%, 14% and 6% began working in the 1960s, 1950s and before 1950, respectively. The vast majority of technologists began working before age 25 (87%), while 42% began working before age 20. About half (51%) worked for 10 years or more. More than half reported holding patients more than 50 times during X-ray procedures. Slightly under half had never smoked cigarettes, while 15% had smoked 20 or more pack-years. The mean BMI at the time of the baseline questionnaire response was 23.7 kg/m2.

Table 1

Demographic characteristics, work history and mortality events of 90 268 US radiological technologists

We first reviewed cancer risks for total cancer and for radiogenic cancers (table 2). For mortality risks for all cancers combined and for bladder and colorectal cancers, we found no statistically significant associations with work history factors. Brain cancer was not associated with work history, except for an unexpected significant inverse association with total number of years worked (table 2).

Table 2

Relative risks for all cancers combined and for cancers consistently linked with radiation exposure among 90 268 US radiological technologists according to work history derived from baseline questionnaire

We observed excess mortality risks for female breast cancer among technologists who began working before 1940 and during the 1940s compared with mortality risks among technologists who began working in 1970 or later, and among those who worked an increasing number of years before 1950. Surprisingly, breast cancer mortality decreased significantly with increasing number of times held patients for X-rays. No other work history factors were linked with breast cancer (table 2).

Leukaemia risks were twofold higher in technologists who worked 30 or more years than in those who worked fewer than 10 years, and risk increased significantly with increasing number of years worked before 1950. The other work history variables evaluated were not related to leukaemia risk in the USRT (table 2).

The mortality risk for lung cancer in the technologists increased significantly with increasing number of years worked in the 1950s and with increasing number of times technologists held patients for X-rays. Otherwise, technologists’ work history was not associated with lung cancer mortality risks (table 2).

A nearly sevenfold increase in stomach cancer mortality risk was observed in technologists who began working before 1940 compared with others who first worked in 1970 or later, but mortality risk declined significantly with increasing number of times technologists held patients for X-rays. The remaining work history characteristics were not linked with stomach cancer risk in the technologists (table 2).

We found no associations between work history and mortality risks for the non-radiogenic cancers: melanoma, multiple myeloma, non-Hodgkin's lymphoma and ovarian cancer (table 3). There was a decrease in risk for oesophageal cancer with increasing number of years technologists worked in 1970 or later. We also found an increased risk of uterine cancer mortality among those who worked a greater number of years during 1960–1969, and a declining risk with increasing number of years worked in 1970 or later.

Table 3

Relative risks for other cancers among 90 268 US radiological technologists according to work history derived from baseline questionnaire

Table 4 includes results for all-cause mortality and mortality due to selected circulatory, digestive, respiratory and infectious diseases. For all-cause mortality we found no associations with year or age technologists first worked, total number of years technologists worked, number of years technologists worked before 1950 or during the 1950s or number of times technologists held patients, but mortality risk increased significantly with number of years worked during the 1960s, while a modest decrease in all-cause mortality was observed with increasing years worked during 1970 or later. Technologists who worked more years before 1950 and during 1960–1969, but not other time periods, experienced modest increases in mortality from total heart disease; no other reported work history characteristics were associated with this outcome. Technologists who began working in earlier years experienced increased mortality from ischaemic heart disease and cerebrovascular disease, and mortality from each of these circulatory system diseases rose significantly with increasing number of years worked before 1950. No other work history factors reported by the technologists were linked with excess risk of mortality from ischaemic heart disease or cerebrovascular disease. Work history factors were not associated with mortality risks in technologists for digestive, respiratory or infectious diseases except for an increasing risk of mortality from digestive disease among those working an increasing number of years during the 1960s.

Table 4

Relative risks for all causes of death combined and for circulatory, digestive, respiratory and infectious diseases among 90 268 US radiological technologists according to work history derived from baseline questionnaire

We examined further the associations between year first worked by follow-up period (1983–1989 through 1997 and 1998 though 2008) for selected mortality outcomes with at least 100 deaths (table 5). The notable and significant overall trend of higher mortality risk for breast cancer with earlier year first worked was strongly apparent during the follow-up period of 1983–1989 through 1997, but only a modest trend was observed for follow-up from 1998 to 2008. Trends in risk of ischaemic heart disease and of cerebrovascular disease according to year first worked by the technologists were similar for the two follow-up periods. Small numbers of outcomes limited our assessment of leukaemia and stomach cancer.

Table 5

Associations between year first worked and selected mortality outcomes by follow-up period

We found that overall mortality in the US radiological technologists was about 30% lower than the mortality in the general US population for both the entire cohort (N=146 022) and the subset of 90 268 technologists who completed the baseline questionnaire (see online supplementary table A2). The SMRs for the entire cohort of US radiological technologists compared with the general US population were 0.82 for all cancers and 0.70 for all circulatory system diseases (see online supplementary table A2). With few exceptions, SMRs were lower than 1 for most specific types of neoplasms (see online supplementary table A2).

Discussion

In our extended 27-year follow-up of work history and mortality among 90 268 radiological technologists in the USA, we found no association between most work history factors and mortality from all causes, all cancers combined or from most specific types of cancer. However, we found that those who first began working in earlier decades experienced mortality risks for female breast cancer, stomach cancer, ischaemic heart disease and cerebrovascular disease that increased with earlier year technologists began working. The significant trend with earlier year first worked was strongly apparent for breast cancer during follow-up from the baseline questionnaire through 1997 (p<0.001), but was only of borderline significance with follow-up from 1998 to 2008 (p=0.06). Risks were similar in the two periods for circulatory diseases; small numbers of leukaemia and stomach cancer deaths precluded comparisons for the two calendar year periods of follow-up. Radiological technologists who worked five or more years before 1950 experienced elevated mortality from breast cancer, leukaemia, ischaemic heart disease and cerebrovascular disease. For the external comparison analyses, our findings were consistent with a healthy worker effect in this cohort. Relative to the US population, US radiological technologists experienced significantly lower all-cause mortality and mortality from most diseases.

Among the work history factors evaluated in this analysis, we consider year first worked and number of years worked during different time periods to be the best proxies for the cumulative radiation exposure in radiological technologists owing to dramatic changes in occupational radiation exposure and protection over time. The literature on radiation protection measures for medical radiation workers8 ,10 and assessment of badge dose measurement readings over time24 provide confirmation that occupational radiation doses were likely to be higher in decades before the late 1950s when more stringent radiation protection measures were implemented in the USA. Although age first worked may be a proxy for sensitivity to radiation by body tissue maturity,25 ,26 we found little evidence of higher risks among technologists who began working at younger ages. We examined risks according to total years worked, as evaluated in other occupational studies, but found that for most outcomes (except leukaemia) there was no relationship with mortality risks. These findings are unsurprising because total number of years worked—without regard to the specific calendar years worked—is not a good indicator of cumulative occupational radiation exposure.

We found elevated risks with follow-up through 2008 for female breast cancer mortality with first working as a radiological technologist before 1940 (HR=2.51, 95% CI 1.24 to 5.05) and during 1940–1949 (HR=1.90, 95% CI 1.09 to 3.31) which were slightly lower than the HRs we observed in our earlier follow-up through 1997 (first working before 1940, HR=2.92, first working during 1940–1949, HR=2.44).7 The lower HRs for the full 2008 follow-up period reflected, at least in part, lower mortality risks according to year first worked in the later follow-up time period (1998 through 2008) than in the earlier time period (1983–89 through 1997). As in our earlier study, there was a significant association of breast cancer mortality with number of years worked before 1950, but not during later decades or with the total number of years worked.7 There have been few other studies reporting breast cancer risks in female medical radiation workers. Incidence risks were 70% increased among 5443 Chinese female diagnostic X-rays workers exposed before 1960 and 30% higher among those exposed from 1960 to 1969 and from 1970 to 1985 compared with risks among female surgeons, otolaryngologists and other physicians who worked at the same hospitals but were not exposed to radiation.27 No differences were observed in breast cancer incidence risks among 4200 (82% female) Danish radiotherapy workers employed from 1954 to 1982 compared with women in the general Danish population,4 or in breast cancer mortality risks among 43 982 female Canadian medical radiation workers employed from 1951 to 1987 compared with women in the general Canadian population.9 Female breast cancer is among the most radiogenic malignancies,1 and there is evidence of excess risk in a number of cohorts associated with highly fractionated exposures of the sort observed in the USRT data.28 ,29

Our findings for leukaemia mortality are generally consistent with reports from our earlier follow-up in which we found elevated risk of leukaemia mortality with increasing number of years the technologists worked before 1950. In most cohorts of medical radiation workers, significantly elevated mortality or incidence of leukaemia was seen in the early workers (eg, first registered, certified or born before 1950). This included male radiologists from the UK who registered with various radiological societies in 1897–1954 (SMR=6.15) compared with medical practitioners;30 US radiologists who joined the Radiological Society of North America in 1920–1939 (SMR=2.01) compared with non-radiation-exposed physicians;31 Chinese X-ray workers, who first worked before 1960 (RR=2.6), during 1960–69 (RR=3.0), and 1970–1985 (RR=1.3), compared with surgeons, otolaryngologists and other physicians who worked at the same hospitals but were not exposed to radiation;27 and Japanese radiological technologists born in 1933 or before (SMR=1.90) and those born in 1934 or later (SMR=1.10) compared with all Japanese workers.5 The findings from our study and that of other medical radiation workers are consistent with higher occupational radiation doses before 1950.

Although increased lung cancer risk, as we found for some work history factors, has been observed in the Japanese atomic bomb survivors32 and in patients treated with radiotherapy for Hodgkin's lymphoma33 and breast cancer,34 the only medical radiation worker studies that have found elevated risks were the UK radiologists in the 1897–1920 subgroup30 and the US radiologists in the 1940–1969 subgroup.31 The absence of information about smoking in the UK and US radiologists complicates interpretation of the elevated risk of lung cancer in these populations.

This extended mortality follow-up of US radiological technologists through 2008 revealed new excesses of stomach cancer, albeit based on small numbers, which was particularly notable among those first employed as radiological technologists before 1940. A significant increase in stomach cancer was reported among Chinese X-ray workers who were first employed before 1980,6 and modest increased risks were described in US radiologists first registered during 1940–196931 and UK radiologists first registered during 1897–1935.13 ,35

New findings for circulatory system disease mortality risks in this extended follow-up of the radiological technologists cohort include an increasing mortality risk from total heart disease with increasing number of years worked before 1950 and during the 1960s (but not during the 1950s), an increasing risk of dying from ischaemic heart disease with increasing number of years worked before 1950 and a significant increasing risk for ischaemic heart disease mortality with earlier year first worked. We confirmed our previous finding of an increased mortality risk from cerebrovascular disease in technologists who worked an increasing number of years before 1950 and a significant increasing risk of dying from cerebrovascular disease among those with earlier year began working.14 Elevated mortality risks for diseases of the circulatory system were found in the US radiologists,12 but not in the UK radiologists.13 Studies of other cohorts of radiologists or radiological technologists have not reported results for circulatory system diseases. Nuclear workers in the 15-country study coordinated by the International Agency for Research on Cancer36 and in the UK National Registry for Radiation Workers (NRRW) (third analysis)37 experienced modest, non-significant increases in mortality from cerebrovascular disease (both studies) and from ischaemic heart disease (UK NRRW) with increasing estimated doses of radiation, although there was a borderline significant trend for all circulatory system disease in the NRRW, and a generally significant increasing trend with radiation dose observed in the British Nuclear Fuels, Limited workforce,38 one of the constituents of the NRRW. A large meta-analysis of populations exposed to environmental, occupational and atomic-bomb radiation exposures found that low and moderate doses of ionising radiation were associated with circulatory system disease mortality and risks were particularly high for ischaemic heart disease and cerebrovascular disease.15 The most recent mortality follow-up of the Japanese atomic bomb survivors also reported increasing mortality with increasing radiation dose,17 although these are effects from a single acute radiation exposure rather than the chronic radiation exposure experienced by medical and nuclear radiation workers.

Some unexpected findings and inconsistencies in our results deserve mention. Breast cancer and stomach cancer both declined significantly with increasing number of times technologists held patients for X-rays, but these unexpected results were not observed for other cancers. Possible explanations may include chance or that physically fit technologists may be more likely to hold patients. Also surprising was the significant decrease in brain cancer risk with total number of years worked. With the exception of leukaemia (with a borderline trend of increasing risk with total years worked), the total number of years technologists worked was not linked with risk of most outcomes likely because this metric is not a good indicator of occupational radiation exposures unless the long duration of occupational exposure was in the early years when doses were higher.

Our findings on all-cause mortality appear inconsistent with an earlier study.11 While we found a modest increase in risk for all-cause mortality according to number of years worked during 1950–1959 (borderline significant) and 1960–1969 (significant), risk was not related to number of years worked before 1950 when exposures were likely higher. In the earlier study by Matanoski and colleagues, US radiologists had an excess risk from all-cause mortality for those who were first registered before 1940, with the excess still apparent when cancer deaths were excluded.11 In the UK radiologists, Berrington and colleagues13 described a reduced risk of death from all causes, but an increasing trend in risk of cancer mortality with time since first registration, with a 41% excess risk in those registered for more than 40 years. The differing results may reflect the previous studies’ assessment of mortality risks beginning with much earlier periods than our study, which began follow-up in the mid-1980s. Thus, we captured risks among those who had already survived several decades beyond their initial high exposures.

The current analysis has several important strengths. The large cohort size and median 24-year follow-up enabled us to examine rarer mortality outcomes, in addition to the broader range of cancers and other serious diseases than were evaluated in the previous analyses of this cohort.7 ,14 The current analysis includes about three times as many total deaths and three times as many cancer deaths (3329 in the current analysis vs 1283 in the previous analysis, including 518 vs 255 deaths from breast cancer, 790 vs 279 deaths from lung cancer and 119 vs 37 deaths from leukaemia other than chronic lymphocytic leukaemia). Deaths from non-cancer outcomes were also substantially increased (including 1558 deaths from ischaemic heart disease in the current analysis vs 633 in the previous analysis and 522 deaths from cerebrovascular disease in the current analysis vs 174 in the previous analysis). To our knowledge, this is the only cohort of medical radiation workers with a large number of radiation-exposed female workers and extensive information on potential covariates, such as cigarette smoking, to be studied for a broad range of causes of mortality in relation to work history. The nationwide study population allows the results to be generalised to a broad population of US radiological technologists.

The most important limitation of this study was absence of individual dose estimates. Although an historical dose reconstruction for this cohort is ongoing, dose estimates were not available for the current analysis. Since the cohort was assembled in the early 1980s and the baseline questionnaire administered during 1983–1989, a shortcoming of the study design is that technologists in the cohort who died before completing the baseline questionnaire could not be included in the assessment of questionnaire-derived work history and follow-up for mortality risk. This design limitation is an intrinsic limitation of many other occupational epidemiological studies that assess disease risks retrospectively. Our finding of increased mortality risks for several cancers and circulatory disease outcomes with increasing number of years worked in early time periods only reflects those who survived long enough to complete a baseline questionnaire. It is possible that work history factors and disease mortality associations may be different among those who died before questionnaire administration compared with those who survived long enough to complete the baseline questionnaire. We have tried to minimise confounding by age and calendar time using the proportional hazards model with age as the timescale and controlling for birth cohort by stratification. Any potential misclassification of self-reported work history data from questionnaire responses is likely to have been non-differential since the work history information was collected before any mortality events. While it is common to use the death certificate and the National Death Index Plus records in US occupational epidemiological studies, there may be some misclassification of disease mortality due to the lack of medical record validation, and this misclassification may have contributed to a few biologically implausible associations involving some work history factors and diseases.

In conclusion, we found that radiological technologists who began working before 1950 and worked more years before 1950 had increased risks of dying from some cancers consistently associated with radiation exposure (ie, breast cancer and leukemias other than chronic lymphocytic leukemia) and circulatory system diseases. These findings are consistent with occupational exposure to higher radiation doses before the implementation of more stringent radiation protection measures in the late 1950s and the requirement for use of lead aprons and film badge monitoring beginning in the 1950s and 1960s.10 ,39 ,40 Compared with our previous follow-up findings, the differences in the effect estimates and the new findings of increasing risks of death from lung and stomach cancer, as well as total and ischaemic heart diseases, may be due to substantially larger numbers of deaths or first manifestations after longer latency. Assessment of mortality risks in other populations of medical radiation workers are needed to confirm our new findings for stomach and lung cancer, and ischaemic heart disease. To evaluate further whether the results from our current analysis are related to the level of occupational radiation exposure, it will be important to examine disease mortality risks using high-quality estimates of occupational radiation doses to specific organs and tissues from our ongoing dose reconstruction effort.

Acknowledgments

We are grateful to the radiological technologists who have participated in this study; Mr Jerry Reid of the American Registry of Radiologic Technologists for continued support of the study; Ms Allison Iwan of the University of Minnesota for cohort follow-up and data management; Dr Ethel Gilbert of the National Cancer Institute for statistical advice; and Ms Ka Lai Lou and Ms Annelie Landgren of the National Cancer Institute for technical support. This study was supported by the intramural research programme of the Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health.

References

View Abstract
  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors JJL and MSL drafted the manuscript. JJL, JSM, and MPL conducted the statistical analyses. All authors participated in the interpretation of results and the critical review of this article.

  • Funding This study was funded in part through contracts N01-CP-21015, N01-CP-31018, HHSN261201000139C, and HHSN261201100007I with the National Cancer Institute, National Institutes of Health, US Department of Health and Human Services.

  • Competing interests None.

  • Ethics approval National Cancer Institute and the University of Minnesota.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.