Imputation of missing values in a large job exposure matrix using hierarchical information

Self-reported occupational noise exposure and hearing protection device use among NHANES participants and the risk of hearing loss

Occupational noise exposure continues to be a prevalent hazard in many industries. While the proliferation of noise dosimeters and wearable devices has made it easier to assess a worker's exposure to noise, many employees exposed to hazardous (i.e., >85 dBA) levels of noise may go their entire career without ever having their personal noise levels measured. In contrast to other occupational exposures, noise is easily perceived by the individual exposed, allowing them to develop subjective judgments regarding its characteristics. To determine whether such self-reported exposures to occupational noise are associated with hearing loss, this analysis used audiometric data and self-reported occupational exposure to loud noise from the National Health and Nutrition Examination Survey (NHANES), which has collected such data from 1999 to May 2020. Linear and logistic regressions models found a statistically significant association between self-reported noise exposure and worsened hearing at the 3, 4, 6, and 8 kHz hearing frequency as well as an elevated odds ratio for the development of hearing loss greater than 25 dB at the 2, 3, and 4 kHz audiometric frequencies. The results of this analysis suggest that in the absence of exposure measurements, workers are likely able to detect exposure to hazardous levels of noise. In these instances, additional measurements should be collected to determine if the workers should be enrolled in a hearing conservation program.

Long working hours, sedentary work, noise, night shifts and risk of ischaemic heart disease

OBJECTIVE

Ischaemic heart disease (IHD) is a leading cause of death in Western countries. The aim of this study was to examine the associations between occupational exposure to loud noise, long working hours, shift work, and sedentary work and IHD.

METHODS

This data linkage study included all New Zealanders employed and aged 20-64 years at the time of the 2013 census, followed up for incident IHD between 2013 and 2018 based on hospitalisation, prescription and death records. Occupation and number of working hours were obtained from the census, and exposure to sedentary work, loud noise and night shift work was assessed using New Zealand job exposure matrices. HRs were calculated for males and females using Cox regression adjusted for age, socioeconomic status, smoking and ethnicity.

RESULTS

From the 8 11 470 males and 7 83 207 females employed at the time of the census, 15 012 male (1.9%) and 5595 female IHD cases (0.7%) were identified. For males, there was a modestly higher risk of IHD for the highest category (>90 dBA) of noise exposure (HR 1.19; 95% CI 1.07 to 1.33), while for females exposure prevalence was too low to calculate an HR. Night shift work was associated with IHD for males (HR 1.10; 95% CI 1.05 to 1.14) and females (HR 1.25; 95% CI 1.17 to 1.34). The population attributable fractions for night shift work were 1.8% and 4.6%, respectively. No clear associations with working long hours and sedentary work were observed.

CONCLUSIONS

This study suggests that occupational exposures to high levels of noise and night shift work might be associated with IHD risk.

Beware the Grizzlyman: A comparison of job- and industry-based noise exposure estimates using manual coding and the NIOSH NIOCCS machine learning algorithm

Recently, the National Institute for Occupational Safety and Health (NIOSH) released an updated version of the NIOSH Industry and Occupation Computerized Coding System (NIOCCS), which uses supervised machine learning to assign industry and occupational codes based on provided free-text information. However, no efforts have been made to externally verify the quality of assigned industry and job titles when the algorithm is provided with inputs of varying quality. This study sought to evaluate whether the NIOCCS algorithm was sufficiently robust with low-quality inputs and how variable quality could impact subsequent job estimated exposures in a large job-exposure matrix for noise (NoiseJEM). Using free-text industry and job descriptions from >700,000 noise measurements in the NoiseJEM, three files were created and input into NIOCCS: (1) N1, "raw" industries and job titles; (2) N2, "refined" industries and "raw" job titles; and (3) N3, "refined" industries and job titles. Standardized industry and occupation codes were output by NIOCCS. Descriptive statistics of performance metrics (e.g., misclassification/discordance of occupation codes) were evaluated for each input relative to the original NoiseJEM dataset (N0). Across major Standardized Occupational Classifications (SOC), total discordance rates for N1, N2, and N3 compared to N0 were 53.6%, 42.3%, and 5.0%, respectively. The impact of discordance on the major SOC group varied and included both over- and under-estimates of average noise exposure compared to N0. N2 had the most accurate noise exposure estimates (i.e., smallest bias) across major SOC groups compared to N1 and N3. Further refinement of job titles in N3 showed little improvement. Some variation in classification efficacy was seen over time, particularly prior to 1985. Machine learning algorithms can systematically and consistently classify data but are highly dependent on the quality and amount of input data. The greatest benefit for an end-user may come from cleaning industry information before applying this method for job classification. Our results highlight the need for standardized classification methods that remain constant over time.

Retrospective assessment of the association between noise exposure and nonfatal and fatal injury rates among miners in the United States from 1983 to 2014

BACKGROUND

Mining is a significant economic force in the United States but has historically had among the highest nonfatal injury rates across all industries. Several factors, including workplace hazards and psychosocial stressors, may increase injury and fatality risk. Mining is one of the noisiest industries; however, the association between injury risk and noise exposure has not been evaluated in this industry. In this ecological study, we assessed the association between noise exposure and nonfatal and fatal occupational injury rates among miners.

METHODS

Federal US mining accident, injury, and illness data sets from 1983 to 2014 were combined with federal quarterly mining employment and production reports to quantify annual industry rates of nonfatal injuries and fatalities. An existing job-exposure matrix for occupational noise was used to estimate annual industry time-weighted average (TWA, dBA) exposures. Negative binomial models were used to assess relationships between noise, hearing conservation program (HCP) regulation changes in 2000, year, and mine type with incidence rates of injuries and fatalities.

RESULTS

Noise, HCP regulation changes, and mine type were each independently associated with nonfatal injuries and fatalities. In multivariate analysis, each doubling (5 dB increase) of TWA was associated with 1.08 (95% confidence interval: 1.05, 1.11) and 1.48 (1.23, 1.78) times higher rate of nonfatal injuries and fatalities, respectively. HCP regulation changes were associated with 0.61 (0.54, 0.70) and 0.49 (0.34, 0.71) times lower nonfatal injury and fatality rates, respectively.

CONCLUSION

Noise may be a significant independent risk factor for injuries and fatalities in mining.

Fraction of acute work-related injuries attributable to hazardous occupational noise across the USA in 2019

INTRODUCTION

The contribution of hazardous noise-a ubiquitous exposure in workplaces-to occupational injury risk is often overlooked. In this ecological study, the fraction of US workplace acute injuries resulting in days away from work in 2019 attributable to hazardous occupational noise exposure was estimated.

METHODS

Using the NoiseJEM, a job exposure matrix of occupational noise, and 2019 Occupational Employment and Wage Statistics data, the proportion of workers experiencing hazardous occupational noise (≥85 dBA) was estimated for every major US Standard Occupational Classification (SOC) group. Population attributable fractions (PAFs) were calculated for each major SOC group using the relative risk (RR) taken from a published 2017 meta-analysis on this relationship.

RESULTS

About 20.3 million workers (13.8%) are exposed to hazardous levels of occupational noise. Nearly 3.4% of acute injuries resulting in days away from work in 2019 (95% CI 2.4% to 4.4%) were attributable to hazardous occupational noise, accounting for roughly 14 794 injuries (95% CI 10 367 to 18 994). The occupations with the highest and the lowest PAFs were production (11.9%) and office and administrative support (0.0%), respectively.

DISCUSSION

Hazardous noise exposure at work is an important and modifiable factor associated with a substantial acute occupational injury burden.

A Quantitative General Population Job Exposure Matrix for Occupational Noise Exposure

Occupational noise exposure is a known risk factor for hearing loss and also adverse cardiovascular effects have been suggested. A job exposure matrix (JEM) would enable studies of noise and health on a large scale. The objective of this study was to create a quantitative JEM for occupational noise exposure assessment of the general working population. Between 2001–2003 and 2009–2010, we recruited workers from companies within the 10 industries with the highest reporting of noise-induced hearing loss according to the Danish Working Environment Authority and in addition workers of financial services and children day care to optimize the range in exposure levels. We obtained 1343 personal occupational noise dosimeter measurements among 1140 workers representing 100 different jobs according to the Danish version of the International Standard Classification of Occupations 1988 (DISCO 88). Four experts used 35 of these jobs as benchmarks and rated noise levels for the remaining 337 jobs within DISCO 88. To estimate noise levels for all 372 jobs, we included expert ratings together with sex, age, occupational class, and calendar year as fixed effects, while job and worker were included as random effects in a linear mixed regression model. The fixed effects explained 40% of the total variance: 72% of the between-jobs variance, −6% of the between-workers variance and 4% of the within-worker variance. Modelled noise levels showed a monotonic increase with increasing expert score and a 20 dB difference between the highest and lowest exposed jobs. Based on the JEM estimates, metal wheel-grinders were among the highest and finance and sales professionals among the lowest exposed. This JEM of occupational noise exposure can be used to prioritize preventive efforts of occupational noise exposure and to provide quantitative estimates of contemporary exposure levels in epidemiological studies of health effects potentially associated with noise exposure.

Bayesian Hierarchical Modelling of Individual Expert Assessments in the Development of a General-Population Job-Exposure Matrix

The CANJEM job-exposure matrix compiles expert evaluations of 31 673 jobs from four population-based case-control studies conducted in Montreal. For each job, experts had derived indices of intensity, frequency, and probability of exposure to 258 agents. CANJEM summarizes the exposures assigned to jobs into cells defined by occupation/industry, agent, and period. Some cells may, however, be less populated than others, resulting in uncertain estimates. We developed a modelling framework to refine the estimates of sparse cells by drawing on information available in adjacent cells. Bayesian hierarchical logistic and linear models were used to estimate the probability of exposure and the geometric mean (GM) of frequency-weighted intensity (FWI) of cells, respectively. The hierarchy followed the Canadian Classification and Dictionary of Occupations (CCDO) classification structure, allowing for exposure estimates to be provided across occupations (seven-digit code), unit groups (four-digit code), and minor groups (three-digit code). The models were applied to metallic dust, formaldehyde, wood dust, silica, and benzene, and four periods, adjusting for the study from which jobs were evaluated. The models provided estimates of probability and FWI for all cells that pulled the sparsely populated cells towards the average of the higher-level group. In comparisons stratified by cell sample size, shrinkage of the estimates towards the group mean was marked below 5 jobs/cell, moderate from 5 to 9 jobs/cell, and negligible at ≥10 jobs/cell. The modelled probability of three-digit cells were slightly smaller than their descriptive estimates. No systematic trend in between-study differences in exposure emerged. Overall, the modelling framework for FWI appears to be a suitable approach to refine CANJEM estimates. For probability, the models could be improved by methods better adapted to the large number of cells with no exposure.