Commentary for the Then and Now Forum: The Healthy Worker Effect

Component Associations of the Healthy Worker Survivor Bias in Medical Radiation Workers

BACKGROUND

The healthy worker survivor bias may vary by sex. This study investigated three component associations necessary for this bias to determine the origins of sex differences in this bias among male and female workers.

METHODS

We analyzed a data set of 93,918 South Korean diagnostic medical radiation workers registered in the National Dose Registry from 1996 to 2011, linked with mortality and cancer incidence data. Component associations were assessed using Cox regression to estimate hazard ratios (HRs) and logistic regression with generalized estimating equations to estimate odds ratios (ORs).

RESULTS

A significant association between prior cumulative exposure and employment status was observed for all-cause mortality in male (HR 1.06, 95% CI 1.02-1.10), whereas an inverse association was noted in female workers (HR 0.82, 95% CI 0.78-0.87). Adjusted ORs for employment status and subsequent exposure for all-cause mortality, as well as HRs for employment status and survival time, demonstrated associations in the same direction in both males and females.

CONCLUSIONS

Our findings demonstrate that sex-specific differences in healthy worker survivor bias were primarily driven by the association between prior exposure and employment status. To improve bias mitigation in occupational cohort studies, sex-specific components should be incorporated.

What about the workers? An update

Epidemiological studies of nuclear industry workers are of substantial importance to understanding the risk of cancer consequent to low-level exposure to radiation, and these studies should provide vital evidence for the construction of the international system of radiological protection. Recent studies involve large numbers of workers and include health outcomes for workers who accumulated moderate (and even high) doses over prolonged periods while employed during the earlier years of the nuclear industry. The interpretation of the findings of these recent studies has proved to be disappointingly difficult. There are puzzling patterns of results involving the period of first employment and monitoring for radionuclide intakes, depending on the particular study examined. Explaining these patterns is crucial for a reliable understanding of results in terms of occupational radiation exposure. In this paper, an updated review of nuclear worker studies is presented in the context of these patterns of results, making use of the latest relevant results. It is apparent that the strikingly raised risks for mortality from solid cancers for workers hired in later years reported from the International Nuclear Workers Study (INWORKS) is effectively confined to workers at five nuclear facilities in the USA, and that the notable variation of risks in INWORKS between workers monitored or not for radionuclide intakes is driven by UK workers. These are the areas where effort must be concentrated before a confident derivation of radiation risk estimates can be obtained from these nuclear worker studies.

The effect of exposure to radiofrequency fields on cancer risk in the general and working population: A systematic review of human observational studies - Part II: Less researched outcomes

BACKGROUND

In the framework of the World Health Organization assessment of health effects of exposure to radiofrequency electromagnetic fields (RF-EMF), we have conducted a systematic review of human observational studies on the association between exposure to RF-EMF and risk of neoplastic diseases. Due to the extremely large number of included exposure types/settings and neoplasm combinations, we decided to present the review findings in two separate papers. In the first one we addressed the most investigated exposure-outcome pairs (e.g. glioma, meningioma, acoustic neuroma in relation to mobile phone use, or risk childhood leukemia in relation to environmental exposure from fixed-site transmitters) (Karipidis et al., 2024). Here, we report on less researched neoplasms, which include lymphohematopoietic system tumours, thyroid cancer and oral cavity/pharynx cancer, in relation to wireless phone use, or occupational RF exposure.

METHODS

Eligibility criteria: We included cohort and case-control studies of neoplasia risks in relation to three types of exposure to RF-EMF: 1. exposure from wireless phone use; 2. environmental exposure from fixed-site transmitters; 3. occupational exposures. In the current paper, we focus on less researched neoplasms including leukaemia, non-Hodgkin's lymphoma and thyroid cancer in mobile phone users; lymphohematopoietic system tumours and oral cavity/pharynx cancer in exposed workers. We focussed on investigations of specific neoplasms in relation to specific exposure sources (termed exposure-outcome pair, abbreviated E-O pairs), noting that a single article may address multiple E-O pairs.

INFORMATION SOURCES

Eligible studies were identified by predefined literature searches through Medline, Embase, and EMF-Portal. Risk-of-bias (RoB) assessment: We used a tailored version of the Office of Health Assessment and Translation (OHAT) RoB tool to evaluate each study's internal validity. Then, the studies were classified into three tiers according to their overall potential for bias (low, moderate and high) in selected, predefined and relevant bias domains.

DATA SYNTHESIS

We synthesized the study results using random effects restricted maximum likelihood (REML) models. Evidence assessment: Confidence in evidence was assessed according to the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach.

RESULTS

We included 26 articles, which were published between 1988 and 2019, with participants from 10 countries, reporting on 143 different E-O pairs, including 65 different types of neoplasms. Of these, 19 E-O pairs satisfied the criteria for inclusion in quantitative syntheses of the evidence regarding the risks of leukaemia, non-Hodgkin's lymphoma or thyroid cancer in relation to mobile phone use, and the risks of lymphohematopoietic system tumours or oral cavity/pharynx cancer following occupational exposure to RF-EMF. RF-EMF exposure from mobile phones (ever or regular use vs no or non-regular use) was not associated with an increased risk of leukaemia [meta-estimate of the relative risk (mRR) = 0.99, 95 % CI 0.91-1.07, 4 studies), non-Hodgkin's lymphoma (mRR = 0.99, 95 % CI = 0.92-1.06, 5 studies), or thyroid cancer (mRR = 1.05, 95 % CI = 0.88-1.26, 3 studies). Long-term (10 + years) mobile phone use was also not associated with risk of leukaemia (mRR = 1.03, 95 % CI 0.85-1.24, 3 studies), non-Hodgkin lymphoma (mRR = 0.99, 95 % CI 0.86-1.15, 3 studies), or thyroid cancer (no pooled estimate given the small number of studies). There were not sufficient studies of any specific neoplasms to perform dose-response meta-analyses for either cumulative call time or cumulative number of calls; individual studies did not show statistically significant associations between lifetime intensity of mobile phone use and any specific neoplasm. Occupational RF-EMF exposure (exposed vs unexposed) was not associated with an increased risk of lymphohematopoietic system tumours (mRR = 1.03, 95 % CI = 0.87-1.28, 4 studies) or oral cavity/pharynx cancer (mRR = 0.68, 95 % CI 0.42-1.11, 3 studies). There were not sufficient studies of any specific neoplasms to perform meta-analysis on the intensity or duration of occupational RF-EMF exposure; individual studies did not show statistically significant associations with either of those exposure metrics and any specific neoplasms. The small number of studies, and of exposed cases in some instances, hampered the assessment of the statistical heterogeneity in findings across studies in the meta-analyses. Based on the summary risk of bias, most studies included in the quantitative evidence syntheses were classified at moderate risk of bias. The most critical issue was exposure information bias, especially for occupational studies where the exposure characterization was rated at high risk of bias for all included studies. Outcome information bias was an issue in mortality-based occupational cohort studies investigating non-rapidly fatal neoplasms. Further, the healthy subscriber effect, and (at a lesser extent) the healthy worker effect, were identified as plausible explanations of the decreased risks observed in some studies. The association of RF-EMF exposure from wireless phone use, or workplace equipment/devices, with other important neoplasms was reported by only one or two studies per tumour, so no quantitative evidence syntheses were conducted on these outcomes. It is noted that there were generally no statistically significant exposure-outcome associations for any combinations, independently of the exposure metric and level, with a few studies reporting decreased risks (especially for smoking-related cancers). There was only one study which assessed the effect of RF-EMF exposure from fixed-site transmitters on less researched neoplasms and it reported no statistically significant associations between exposure from base stations and risk of lymphomas overall, lymphoma subtypes, or chronic lymphatic leukaemia in adults.

CONCLUSIONS

For near field RF-EMF exposure to the head from mobile phones, there was low certainty of evidence that it does not increase the risk of leukaemia, non-Hodgkin's lymphoma or thyroid cancer. For occupational RF-EMF exposure, there was very low certainty of evidence that it does not increase the risk of lymphohematopoietic system tumours or oral cavity/pharynx cancer. There was not sufficient evidence to assess the effect of whole-body far-field RF-EMF exposure from fixed-site transmitters (broadcasting antennas or base stations), or the effect of RF-EMF from any source on any other important neoplasms.

OTHER

This project was commissioned and partially funded by the World Health Organization (WHO). Co-financing was provided by the New Zealand Ministry of Health; the Istituto Superiore di Sanità in its capacity as a WHO Collaborating Centre for Radiation and Health; and ARPANSA as a WHO Collaborating Centre for Radiation Protection.

REGISTRATION

PROSPERO CRD42021236798. Published protocol: [(Lagorio et al., 2021) DOI https://doi.org/10.1016/j.envint.2021.106828].

Healthy worker hire and survivor effects in a cohort of medical radiation workers

BACKGROUND

The healthy worker effect may distort the association between exposure and health effects in workers. However, few studies have investigated both the healthy worker hire and survival effects simultaneously, and they are limited to mortality studies in male workers.

METHODS

We utilized a data set comprising South Korean diagnostic medical radiation workers registered in the National Dose Registry between 1996 and 2011, and merged it with mortality and cancer incidence data. Standardized mortality ratios (SMRs) and standardized incidence ratios (SIRs) were computed for comparison with the general population. To account for time-varying confounders influenced by prior occupational radiation exposure, we applied g-estimation using structural nested accelerated failure time models and compared the outcomes with those from Weibull regression.

RESULTS

A total of 1831 deaths and 3759 first primary cancer cases were identified among 93 918 workers. Both male (SMR = 0.44; 95% CI: 0.42, 0.46) and female workers (SMR = 0.53; 95% CI: 0.46, 0.60) showed lower mortality rates compared with national rates. In the SIR analysis, male workers exhibited reduced risks of solid cancer whereas female workers had increased risks. The g-estimation-derived hazard ratios (HRs) from radiation exposure exceeded those from Weibull regression estimates for all-cause death (HR = 2.55; 95% CI: 1.97, 3.23) and all-cancer incidence (HR = 1.96; 95% CI: 1.52, 2.55) in male workers whereas female workers showed the opposite results.

CONCLUSIONS

Comprehensive consideration of the healthy worker effect by sex is essential for estimating the unbiased impact of occupational exposure on health outcomes, notably in studies focusing on male mortality.

Association of Childhood Respiratory Status with Adult Occupational Exposures in a Birth Cohort

Rationale: People with better early-life respiratory health may be more likely to work in occupations with high workplace exposures in adult life compared with people with poor respiratory health. This may manifest as a healthy worker effect bias, potentially confounding the analysis of environmental exposure studies. Objectives: To evaluate associations between lung function in adolescence and occupational exposures at initial adult employment. Methods: The TCRS (Tucson Children's Respiratory Study) is a long-term prospective study of respiratory health beginning at birth. Associations between respiratory function at age 11 years and occupational exposures at first job at age 26 years were evaluated with logistic regression. We calculated percentage predicted values for forced expiratory volume in one second (FEV1), forced vital capacity (FVC), FEV1:FVC ratio, and forced expiratory flow from 25% to 75% of vital capacity at age 11. At the 26-year visit, participants self-reported occupational exposures to dust, smoke, and fumes/gas at first job in a standardized interview. Results: Forced expiratory flow from 25% to 75% of vital capacity and FEV1:FVC ratio at age 11 were positively associated with dust workplace exposures at the first job. Each 10% increase in percentage predicted prebronchodilator FEV1:FVC ratio was associated with 30% higher odds of workplace dust exposure (odds ratio for a 1% increase, 1.03 [95% confidence interval, 1.00-1.06; P = 0.045]). Similar associations were observed for FEV1 and FVC with workplace smoke exposures. We also observed modification by time at job: associations were stronger for those who remained in their jobs longer than 12 months. In addition, those with better function at age 11 were more likely to stay in their jobs longer than 12 months if their first jobs involved exposure to dust. Conclusions: Childhood lung function affects initial career choice. This study supports the premise of the healthy worker effect.

Association of respiratory symptoms and lung function with occupation in the multinational Burden of Obstructive Lung Disease (BOLD) study

BACKGROUND

Chronic obstructive pulmonary disease has been associated with exposures in the workplace. We aimed to assess the association of respiratory symptoms and lung function with occupation in the Burden of Obstructive Lung Disease study.

METHODS

We analysed cross-sectional data from 28 823 adults (≥40 years) in 34 countries. We considered 11 occupations and grouped them by likelihood of exposure to organic dusts, inorganic dusts and fumes. The association of chronic cough, chronic phlegm, wheeze, dyspnoea, forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1)/FVC with occupation was assessed, per study site, using multivariable regression. These estimates were then meta-analysed. Sensitivity analyses explored differences between sexes and gross national income.

RESULTS

Overall, working in settings with potentially high exposure to dusts or fumes was associated with respiratory symptoms but not lung function differences. The most common occupation was farming. Compared to people not working in any of the 11 considered occupations, those who were farmers for ≥20 years were more likely to have chronic cough (OR 1.52, 95% CI 1.19-1.94), wheeze (OR 1.37, 95% CI 1.16-1.63) and dyspnoea (OR 1.83, 95% CI 1.53-2.20), but not lower FVC (β=0.02 L, 95% CI -0.02-0.06 L) or lower FEV1/FVC (β=0.04%, 95% CI -0.49-0.58%). Some findings differed by sex and gross national income.

CONCLUSION

At a population level, the occupational exposures considered in this study do not appear to be major determinants of differences in lung function, although they are associated with more respiratory symptoms. Because not all work settings were included in this study, respiratory surveillance should still be encouraged among high-risk dusty and fume job workers, especially in low- and middle-income countries.

Applying the Bradford Hill Criteria for Causation to Repetitive Head Impacts and Chronic Traumatic Encephalopathy

Chronic traumatic encephalopathy (CTE) is a neurodegenerative disease associated with a history of repetitive head impacts (RHI). CTE was described in boxers as early as the 1920s and by the 1950s it was widely accepted that hits to the head caused some boxers to become "punch drunk." However, the recent discovery of CTE in American and Australian-rules football, soccer, rugby, ice hockey, and other sports has resulted in renewed debate on whether the relationship between RHI and CTE is causal. Identifying the strength of the evidential relationship between CTE and RHI has implications for public health and medico-legal issues. From a public health perspective, environmentally caused diseases can be mitigated or prevented. Medico-legally, millions of children are exposed to RHI through sports participation; this demographic is too young to legally consent to any potential long-term risks associated with this exposure. To better understand the strength of evidence underlying the possible causal relationship between RHI and CTE, we examined the medical literature through the Bradford Hill criteria for causation. The Bradford Hill criteria, first proposed in 1965 by Sir Austin Bradford Hill, provide a framework to determine if one can justifiably move from an observed association to a verdict of causation. The Bradford Hill criteria include nine viewpoints by which to evaluate human epidemiologic evidence to determine if causation can be deduced: strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, and analogy. We explored the question of causation by evaluating studies on CTE as it relates to RHI exposure. Through this lens, we found convincing evidence of a causal relationship between RHI and CTE, as well as an absence of evidence-based alternative explanations. By organizing the CTE literature through this framework, we hope to advance the global conversation on CTE mitigation efforts.

Patterns of Opioid Prescribing and Predictors of Chronic Opioid Use in an Industrial Cohort, 2003 to 2013

OBJECTIVE

To appreciate the impact of the opioid epidemic in workers, we described opioid prescription patterns in a US industrial cohort over a 10-year period and assessed predictors of chronic prescription.

METHODS

A multiyear (2003 to 2013) cross-sectional analysis of employer-sponsored health care claims for enrolled workers (N: 21,357 to 44,769) was performed.

RESULTS

The proportion of workers prescribed opioids nearly doubled in the 10-year period. The strongest predictor of chronic opioid prescribing was year, with an increase in prescriptions each year from 2003 to 2013 (odds ratio = 2.90, 95% confidence interval: 2.41 to 3.48). Additional predictors included older age, white race, hourly wage, low back pain, and osteoarthritis.

CONCLUSIONS

Opioid prescribing for industrial workers substantially increased from 2003 to 2013. Occupational health professionals should be aware of the potential for chronic opioid use among workers to assess job safety and appropriate treatment of work-related injuries.