Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Evaluation of the Validity of Job Exposure Matrix for Psychosocial Factors at Work

Evaluation of the Validity of Job Exposure Matrix for Psychosocial Factors at Work Objective: To study the performance of a developed job exposure matrix (JEM) for the assessment of psychosocial factors at work in terms of accuracy, possible misclassification bias and predictive ability to detect known associations with depression and low back pain (LBP). Materials and Methods: We utilized two large population surveys (the Health 2000 Study and the Finnish Work and Health Surveys), one to construct the JEM and another to test matrix performance. In the first study, information on job demands, job control, monotonous work and social support at work was collected via face-to-face interviews. Job strain was operationalized based on job demands and job control using quadrant approach. In the second study, the sensitivity and specificity were estimated applying a Bayesian approach. The magnitude of misclassification error was examined by calculating the biased odds ratios as a function of the sensitivity and specificity of the JEM and fixed true prevalence and odds ratios. Finally, we adjusted for misclassification error the observed associations between JEM measures and selected health outcomes. Results: The matrix showed a good accuracy for job control and job strain, while its performance for other exposures was relatively low. Without correction for exposure misclassification, the JEM was able to detect the association between job strain and depression in men and between monotonous work and LBP in both genders. Conclusions: Our results suggest that JEM more accurately identifies occupations with low control and high strain than those with high demands or low social support. Overall, the present JEM is a useful source of job-level psychosocial exposures in epidemiological studies lacking individual-level exposure information. Furthermore, we showed the applicability of a Bayesian approach in the evaluation of the performance of the JEM in a situation where, in practice, no gold standard of exposure assessment exists. Citation: Solovieva S, Pensola T, Kausto J, Shiri R, Helio¨ vaara M, et al. (2014) Evaluation of the Validity of Job Exposure Matrix for Psychosocial Factors at Work. PLoS ONE 9(9): e108987. doi:10.1371/journal.pone.0108987 Editor: James Coyne, University of Pennsylvania, United States of America Received January 31, 2014; Accepted September 6, 2014; Published September 30, 2014 Copyright:  2014 Solovieva et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The study was financially supported by the Finnish Work Environment Fund (project no 109364). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * Email: [email protected] diabetes [8] and musculoskeletal diseases [9]. The effects of the Introduction individual components of the job strain model on health have also During the past three decades, the effects of psychosocial factors been evaluated, although the results have often been inconsistent at work on health have received considerable attention in research. across the studies and health outcomes [7,9]. Psychosocial factors at work are numerous, with psychological job The interpretation of the observed associations between demands, job control (decision latitude), efforts and rewards [1,2] psychosocial factors at work and health mainly depends on the comprising the key dimensions. Another factor of importance is validity of the assessment methods of the risk factors. Self-reported social support at work [3]. questionnaires are widely used to measure psychosocial factors at The job strain model, introduced by Karasek in 1979 [4], is one work [10]. Self-reports provide subjective information representing of the most studied occupational stress models. According to the a worker’s perception of occupational stress and are therefore model, workers with a combination of high psychosocial job susceptible to reporting bias. The subjective assessment of demands and low control over a job (high job strain) have a higher psychosocial factors at work has been the largest concern in the risk of developing an illness as compared to workers with low debate on the interpretation of associations and on the possible psychosocial job demands and high job control (low job strain) [1]. causal role of these factors for illness. It has been suggested that The job strain model has been successfully used to predict the risk common source bias due to subjective measures of psychosocial of cardiovascular disease [5,6], major mental disorders [7], type II PLOS ONE | www.plosone.org 1 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix factors at work increases the likelihood of false positive findings, aged 30 years or over and 1894 subjects aged 18–29. The participation rates were 87% and 90%, respectively. A detailed particularly in cross-sectional studies with the self-reported health outcomes [11–13]. Workers having health problems are more comprehensive description of the methods and processes has been published elsewhere [29,30]. The sample of this study comprised likely to report certain psychosocial exposures than healthy workers. Such tendency might lead to differential misclassification, 4619 persons aged 18–64 who were working during the preceding 12 months and for whom information on occupational titles and which results either in an overestimation or underestimation of the true effect [14], particularly, when exposures and outcome are exposures were available. The age and gender distribution of the study population matched those of the employed persons in measured simultaneously. Finland in the year 2000. The assessment of psychosocial factors at work with a job The national Finnish National Work and Health Surveys have exposure matrix (JEM), where exposure level is assigned based on been conducted every third year since 1997 to collect information the job-specific average of exposure, is not prone to information on perceived working conditions and the health of the working-age bias and may therefore guarantee some degree of objectivity. The population, For the 1997–2003 Surveys, random samples of major advantage of the JEM in epidemiological studies is that it subjects aged 25–64 years independent of their working status can be applied to the populations with lacking exposure (e.g., working, unemployed, retired or student) were drawn from information. However, such method of exposure assessment the Finnish population register. For the 2009 Survey a random induces Berkson type error, which may not cause notable bias sample of subjects aged 20–64 years was drawn from Finnish on the effect estimates but weakens the precision of the estimates employment statistics. The sample size has varied between 2031 [15]. A JEM neglects both within worker (over time variation) and and 2355 persons from year to year with a response rate of 58– between worker (variation in tasks, activities and work processes) 72% [31]. At each survey, a phone number was not found for variation in a job [16] and therefore may result in false positive about 10–16% of subjects. The proportion of non-participants in and negative exposure assignments for a considerable proportion each survey was slightly higher among men than women and of the subjects. A non-differential misclassification bias induced by among subjects aged 24–34 years than among the older subjects. JEM will attenuate the observed associations towards null Age, gender, education, socioeconomic status and occupational [15,17,18]. Knowing the magnitude of measurement error (e.g. sector of the respondents were compared with the Census data. No sensitivity and specificity) and exposure prevalence, the extent of major differences were found. Thus, the respondents to the FWH non-differential bias can be estimated [15,19]. Surveys represent rather well the targeted population. The data Several psychosocial job exposure matrices have been devel- from all five surveys were combined. Hence, the total number of oped and used in epidemiological studies [20–26]. Even though the interviewed persons with information on occupation during the JEM measures are more objective than self-reported ones, they 1997–2009 was 11326. cannot be seen as a gold standard in the context of psychosocial The H2000 Study and the FWH Surveys have all obtained factors at work [13]. Therefore, the question of the reliability of ethical approval from the appropriate ethics committees. the associations between JEM-based exposures and health outcomes is always warranted. The validity of psychosocial JEM Classification of occupations measures in the absence of a gold standard method is challenging Occupations in both surveys were classified on the 4-digit level to evaluate and, as a result, has rarely been examined and reported [24–28]. Furthermore, the magnitude of misclassification bias of (including few occupations coded with 5 digits) according to the Classification of Occupations 2001 by Statistics Finland, which is psychosocial JEM measures on effect estimates has not been based on the International Standard Classification of Occupations examined so far. (ISCO-88). The classification is based on ten categories of The aims of the study were 1) to examine the accuracy of a professional skills. In total, the classification includes 444 job titles. developed gender-specific job exposure matrix in the assessment of psychosocial factors at work applying the Bayesian approach, 2) to evaluate the theoretical impact of exposure misclassification on Psychosocial exposures exposure-outcome associations and 3) to examine the ability of the Psychosocial exposures in the H2000 Study were measured with matrix to detect known associations between psychosocial factors a Finnish version of the Job Content Questionnaire (JCQ) [32]. at work and health outcomes. The JCQ has been shown to be a valid and reliable instrument to assess job stress and social support in many occupational settings worldwide [10,13]. Responses were given on a five point Likert- Materials and Methods scale from 1 (fully agree) to 5 (fully disagree). Study population Psychological job demands scale is the sum of the following five We utilized two large Finnish population samples. The Health items: ‘‘work fast’’, ‘‘work hard’’, ‘‘excessive work’’, ‘‘not enough 2000 (H2000) Study was used to construct the JEM and to time’’, and ‘‘hectic job’’. In the current study, Cronbach’s alpha examine the inter-method agreement, and the national Finnish for the scale was 0.76 for men and 0.81 for women. Job control Work and Health (FWH) Surveys were used to test the scale is the sum of two subscales. Decision authority was measured performance of the matrix. The study populations consisted of with three items: ‘‘allows own decisions’’, ‘‘decision freedom’’, and 18–64 year-old individuals, who had been working during the ‘‘a lot of say on the job’’), and skill discretion was measured with preceding 12 months. five items: ‘‘learn new things’’, ‘‘requires creativity’’, ‘‘high skill The Health 2000 Study is a large Finnish population-based level’’, ‘‘variety’’, and ‘‘develop own abilities’’. Cronbach’s alpha study carried out in 2000–01. The main objective of the study was for the scale was 0.85 for men and 0.86 for women. Since to obtain representative information on the current health status of monotonous (repetitive) work was weakly correlated with the other the whole non-institutional adult population in Finland. The five items of the skill discretion scale we treated it as a separate survey consisted of several questionnaires, a home interview, and a exposure. Job demands, job control and monotonous work were health examination. A nationally representative sample of the dichotomized using gender-specific median cut-off points. population was obtained using a two-stage stratified cluster Job strain was operationalized using the quadrant approach sampling design. The original samples consisted of 8028 subjects proposed by Karasek and Theorell [1]. It defines workers who are PLOS ONE | www.plosone.org 2 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix above the median on job demands and below the median on job Low back pain. In the H2000 Study information on low control as having a high strain job. Other categories are: low strain back pain was inquired with the following question: ‘‘Have you (low demands and high control), passive (low demands and low had pain in your back during the past month (30 days)?’’ (yes/no). control) and active (high demands and high control). Low strain In the FWH Surveys, data on low back pain were collected with an job was used as the reference category in the analyses. interview using the question: ‘‘Have you during the past month (30 days) had long-lasting or recurrent pain in the lumbar spine?’’ Social support at work was measured with four items: ‘‘support from supervisor’’, ‘‘supervisor appreciates’’, ‘‘support from co- (yes/no). workers’’, ‘‘discussion on work’’. Cronbach’s alpha for the scale was 0.80 for men and 0.82 for women. Social support was Data analyses dichotomized at a gender-specific median in order to define low In the H2000 Study, the inter-method agreement between self- and high support. reported and JEM measures was examined using intra-class correlation (ICC). Two-way mixed total ICC agreements were Development of the job exposure matrix (JEM) computed. In the FWH Surveys, the performance of the matrix We constructed a gender-specific matrix with exposure was evaluated by examining the accuracy of the matrix in the estimates at each intersection between rows (occupational groups) identification of exposed/non-exposed individuals, estimating and columns (psychosocial exposures). The exposure axis of the exposure misclassification error, and looking at the ability of the matrix included the above mentioned five psychosocial risk factors matrix to detect associations of psychosocial factors at work with at work. The occupation axis of the matrix was based on the one-month prevalence of depression or low back pain (predictive original job titles or occupational groups. validity) [40]. The accuracy of the JEM was evaluated using five Previous studies showed that ten individuals with the same job indicators: sensitivity (Se), specificity (Sp), Youden’s J index, title will be sufficient for a reliable estimation of exposures [33,34]. likelihood ratio positive (LR+) and likelihood ratio negative (LR2). The exposure estimates for job demands, job control, monotonous Sensitivity (ability of the test to identify positive results) and work and social support at work were calculated as a median score specificity (ability of the test to identify negative results) are usually of exposures in each occupation which included at least 10 subjects determined against a reference standard test (gold standard). in order to obtain reasonably precise estimates. The exposure Errors in measuring the sensitivity and specificity of a test will arise estimates for job strain were calculated as the proportion of if the reference test itself does not have 100% sensitivity and 100% exposed to passive, active and high strain work. The job titles with specificity. Since there is no gold standard measure for psycho- a small number (,10) of respondents were grouped based on the social factors at work, we estimated sensitivity and specificity using similarities of these job titles with regard to work tasks (including a Bayesian approach, proposed by Joseph et al. [41]. As the first supervising), work environment, and required educational level. step, the posterior distribution of sensitivity and specificity of the The gender differences in the exposures were also considered. If JEM measures was calculated using self-reported and JEM there was no reasonable way to merge the occupation with other measures of exposures from H2000 Study. For these analyses, occupations within the gender (such as female frontier guards), the the prior distribution of the parameters was derived based on the exposure estimates of both genders in that occupation/occupa- assumption that the self-reported measures have almost perfect tional group were combined. sensitivity and specificity and no prior information on sensitivity The sample size of the H2000 Study was large enough to enable and specificity of JEM measures is available. As the second step, us to develop a gender-specific job exposure matrix and to keep the posterior distribution of sensitivity and specificity of the JEM several job titles unmerged. Out of 444 possible job titles, measures was calculated using data from FWH Surveys. For these altogether 363 (300 among men and 267 among women) were analyses the prior distributions of the parameters were derived available in the Health 2000 Study. There were 61 job titles based on the posterior distributions obtained in the first step. At among men and 58 among women with at least 10 subjects. These each step, the posterior medians and their 95% Bayesian credible job titles covered 69% of the study sample. After merging the intervals were estimated using Gibbs sampler algorithm with smaller groups the number of job titles or occupational groups WinBUGS software version 1.4.3. reduced to 110 among men and 101 among women. The estimated sensitivity and specificity were used to calculate The exposure estimates for job demands, job control, monot- Youden’s J index as well as LR+ and LR2. The Youden’s J index onous work and social support at work were dichotomized using (J = Se+Sp21) has been used as a measure of the effectiveness of gender-specific median as a cut-off point. The categories of job the JEM to discriminate between exposed and non-exposed strain were obtained based on the dichotomized JEM-based job individuals. The possible range of the Youden’s J index value is demands and job control. between 0 (totally useless) and 1 (perfect). Likelihood ratio positive is the probability of an exposed person to be classified as exposed divided by the probability of a non-exposed person to be classified Health outcomes Based on the current evidence we chose two health outcomes as exposed. Likelihood ratio negative is the probability of an exposed person to be classified as non-exposed divided by the that are known to be associated with psychosocial factors at work. Both cross-sectional and longitudinal studies have shown that high probability of a non-exposed person to be classified as non- exposed. A likelihood ratio equal to 1 will indicate that the JEM level of psychological demands and job strain are associated with major mental disorders [7,35–37]. Suggestive evidence for a measure is unable to distinguish between exposed and non- exposed. A LR.1 will indicate that the JEM is likely to identify relationship of job demands, job control and monotonous work with low back pain has also been reported [9,38,39]. exposed and LR,1 will indicate that the JEM is likely to identify non-exposed. The higher LR+ value and lower the LR2 value, Depressive symptoms. In both studies, depressive symp- toms were assessed with the following question: ‘‘Have you had the better is the JEM performance. To estimate the theoretical magnitude of exposure misclassifi- melancholy or depression during the last month (30 days)?’’. The response categories ranged from 1 = not at all to 5 = very often. cation, biased odds ratios (OR9) were calculated based on the The occurrence of depressive symptoms was dichotomized as no obtained estimates of sensitivity (Se) and specificity (Sp) and (categories 1 and 2) or yes (categories from 3 to 5). assumed ‘‘true prevalence’’ (Pr) and ‘‘true odds ratios’’ (OR) using PLOS ONE | www.plosone.org 3 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix Table 1. Prevalence of self-reported and JEM-based psychosocial exposures. Men Women H2000 Study FWH Surveys H2000 Study FWH Surveys Exposures Self-reported JEM JEM Self-reported JEM JEM Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) High job demands 43.1 (41.1, 45.0) 33.1 (29.4, 33.3) 32.2 (31.0, 33.4) 44.2 (42.2, 46.3) 35.3 (33.4, 37.3) 36.0 (34.7, 37.3) Low job control 52.6 (50.5, 54.7) 49.8 (47.8, 51.9) 49.8 (48.5, 51.1) 56.1 (54.1, 58.1) 56.5 (54.5, 58.5) 55.4 (54.1, 56.7) Job strain Low strain job 26.2 (24.4, 28.1) 36.3 (34.3, 38.3) 36.3 (34.1, 36.6) 23.8 (22.1, 25.6) 24.6 (22.9, 26.4) 25.6 (24.4, 26.3) Passive job 30.7 (28.8, 32.7) 32.4 (30.5, 34.4) 32.5 (31.3, 33.8) 32.0 (30.1, 33.9) 40.0 (38.1, 42.1) 38.4 (37.2, 39.7) Active job 21.2 (19.5, 22.9) 13.9 (12.5, 15.4) 14.9 (14.0, 15.8) 20.0 (18.4, 21.7) 18.9 (17.3, 20.5) 19.0 (18.0, 20.1) High strain job 21.9 (20.2, 23.6) 17.4 (15.9, 19.1) 17.3 (16.3, 18.3) 24.2 (22.5, 26.0) 16.4 (15.0, 18.0) 16.9 (16.0, 18.0) Low social support 47.3 (45.3, 49.4) 48.5 (46.4, 50.1) 48.4 (47.1, 49.7) 44.2 (42.2, 46.2) 37.1 (35.2, 39.1) 37.5 (36.2, 38.7) Monotonous work 28.5 (26.6, 30.4) 17.2 (15.7, 18.8) 17.1 (16.1, 18.1) 31.8 (29.9, 33.7) 24.0 (22.3, 25.8) 22.1 (21.0, 23.2) H2000 Study- the Health 2000 Study; FWH Surveys- the Finnish Work and Health Surveys. doi:10.1371/journal.pone.0108987.t001 the following formula [19]: The Finnish Work and Health Surveys and are shown in the form of posterior medians and 95% Bayesian intervals (Table 3). The posterior estimates were very similar in both study populations. ((Se OR Prz(1-Sp) (1-Pr)) ((1-Se) PrzSp (1-Pr)) OR~ The specificity of JEM measures was higher than sensitivity for all ((Se Prz(1-Sp) (1-Pr)) ((1-Se) OR PrzSp (1-Pr)) exposures except job control among women. Specificity ranged from 0.62 to 0.90 in men, and from 0.68 to 0.86 in women. Sensitivity was the lowest for high strain job (0.46) in men and for The true prevalence was fixed at 0.50 for high job demands, low job control and low social support, at 0.33 for monotonous work low social support (0.52) in women. The best matrix performance and at 0.25 for high strain job. The true odds ratios were fixed at assessed by Youden’s J index and likelihood ratios was found for three values OR = 1.5, OR = 2 and OR = 3. The relative high strain job, particularly in women. The JEM was least effective difference between biased and true estimates was calculated in identification of men exposed to high demands (J = 0.17) and ((OR9-OR)/OR) and used as quantitative measure for the women exposed to low social support (J = 0.15). magnitude of exposure misclassification. The theoretical effect of exposure misclassification error on Logistic regression analyses with age, education and year of estimated ORs is shown in Table 4. In both genders, the smallest survey (the FWH Surveys) adjusted odds ratios (OR) and 95% misclassification error was observed for high job strain, followed by confidence intervals (CIs) were carried out to study the associations that for low job control. The largest misclassification error was between the JEM measures and one-month prevalence of found for low social support (both genders) and high job demands depression or low back pain. These analyses were performed (men). In general, when the true OR is equal to 1.5, the effect of using SAS version 9.1. The effect estimates were adjusted for misclassification error on point estimates is relatively small, though misclassification error using WINPEPI COMPARE2 program, there is a high likelihood of false negative findings. A statistically version 3.08 [42]. significant association can be detected only for low job control and All analyses were performed separately for men and women. high job strain in women. With the increase of true OR, there is a larger reduction in the biased odds ratios, but at the same time the Results likelihood of false negative findings is lowered. In both genders, the prevalence of high job demands, high job strain and monotonous work measured by job exposure matrix Table 2. Intra-class correlation (ICC) between individual- was statistically significantly lower than that assessed by self-reports based and group-based measures of psychosocial exposures (Table 1). In women, the prevalence of low social support was for men and women in the H2000 Study. lower for JEM measures than for self-reported measures. There were no differences in the distribution of exposures assessed by JEM between the two study populations, reflecting a similar job Men Women distribution in both surveys. In general, total agreement between self-reported and JEM measures assessed by ICC was slightly Job demands 0.31 0.40 better among women than men, with the largest ICC values Job control 0.53 0.60 observed for job control followed by monotonous work (Table 2). Monotonous work 0.45 0.51 Social support 0.36 0.33 Bayesian estimates of sensitivity and specificity and Job strain 0.21 0.29 magnitude of exposure misclassification error The Bayesian estimates of sensitivity and specificity were H2000 Study- the Health 2000 Study. doi:10.1371/journal.pone.0108987.t002 calculated based on the data from the Health 2000 Study and PLOS ONE | www.plosone.org 4 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix PLOS ONE | www.plosone.org 5 September 2014 | Volume 9 | Issue 9 | e108987 Table 3. Posterior medians and lower and upper limits of the posterior equally tailed 95% credible intervals (Bayesian confidence interval). H2000 Study FWH Surveys 1 2 3 Sensitivity Specificity Sensitivity Specificity J LR+ LR2 High demands Men 0.41 (0.38–0.44) 0.76 (0.74–0.78) 0.41 (0.39–0.44) 0.76 (0.74–0.78) 0.17 1.71 (1.53–1.91) 0.78 (0.73–0.82) Women 0.49 (0.46–0.52) 0.75 (0.73–0.78) 0.49 (0.46–0.52) 0.75 (0.73–0.78) 0.24 2.00 (1.77–2.25) 0.68 (0.63–0.72) Low control Men 0.67 (0.64–0.70) 0.69 (0.67–0.72) 0.67 (0.64–0.69) 0.70 (0.67–0.72) 0.37 2.20 (2.01–2.42) 0.48 (0.43–0.52) Women 0.75 (0.73–0.78) 0.68 (0.65–0.71) 0.76 (0.73–0.78) 0.68 (0.65–0.71) 0.44 2.35 (2.14–2.59) 0.36 (0.32–0.40) High strain job Men 0.58 (0.52–0.64) 0.87 (0.84–0.91) 0.60 (0.55–0.65) 0.82 (0.77–0.86) 0.42 3.34 (2.52–4.45) 0.49 (0.42–0.57) Women 0.77 (0.71–0.82) 0.88 (0.84–0.92) 0.78 (0.73–0.83) 0.86 (0.81–0.90) 0.64 5.75 (4.13–8.03) 0.25 (0.19–0.31) Low social support Men 0.60 (0.57–0.63) 0.62 (0.59–0.65) 0.60 (0.57–0.63) 0.62 (0.59–0.65) 0.22 1.59 (1.45–1.74) 0.65 (0.59–0.71) Women 0.48 (0.45–0.51) 0.72 (0.70–0.74) 0.44 (0.42–0.46) 0.71 (0.69–0.73) 0.15 1.52 (1.40–1.65) 0.79 (0.75–0.83) Monotonous work Men 0.35 (0.32–0.39) 0.90 (0.88–0.91) 0.36 (0.32–0.39) 0.90 (0.88–0.91) 0.26 3.39 (2.84–4.01) 0.72 (0.68–0.76) Women 0.46 (0.42–0.50) 0.86 (0.84–0.88) 0.46 (0.42–0.50) 0.86 (0.84–0.88) 0.32 3.31 (2.79–3.91) 0.63 (0.58–0.68) The sensitivity and specificity of JEM-based measures for men and women in the H2000 Study and the FWH Surveys. H2000 Study- the Health 2000 Study; FWH Surveys- the Finnish Work and Health Surveys; J - Youden’s index = sensitivity+specificity21. LR+ likelihood ratio positive. LR2 likelihood ratio negative. doi:10.1371/journal.pone.0108987.t003 Validity of Psychosocial Factors Job Exposure Matrix Table 4. Biased odds (OR9) ratios according to sensitivity and specificity of the job exposure matrix when the true odds ratios (OR) were assumed to equal 1.5, 2 or 3. OR = 1.5 OR = 2.0 OR = 3.0 Men Women Men Women Men Women High job demands 1.08 1.11 1.13 1.18* 1.21* 1.28* Low job control 1.16 1.20* 1.28* 1.35* 1.45* 1.58* Monotonous work 1.17 1.17 1.30* 1.31* 1.50* 1.52* Low social support 1.09 1.07 1.16 1.11 1.25* 1.17* High job strain 1.18 1.27* 1.34* 1.53* 1.60* 1.99* Prevalence of exposure is assumed to equal 0.50. Prevalence of exposure is assumed to equal 0.33. Prevalence of exposure is assumed to equal 0.25. *Statistical significance at the 5% level (two-sided test) of the biased odds ratios is calculated for a study population of 5000 men and 5000 women. doi:10.1371/journal.pone.0108987.t004 support was relatively low. The largest misclassification error was Predictive validity of the JEM measures found for low social support (women) and high job demands (men). The one-month prevalence of depression was statistically The difference between the odds ratios based on self-reports and significantly higher in the H2000 Study as compared with the JEM was larger for depression than for low back pain, especially in FWH Surveys, while the prevalence of low back pain during the women. Without correction for exposure misclassification, the preceding 30 days was similar (Table 5). In both study popula- JEM was able to detect the association between job strain and tions, women tended to report depression and LBP more depression in men and that between monotonous work and low frequently than men. back pain in both genders. The predictive ability of the matrix In the H2000 Study, associations between all self-reported substantially improved after correction for possible misclassifica- psychosocial factors at work and depression were statistically tion bias. significant in both genders (Table 5). In the FWH Surveys, the Although several psychosocial JEMs exist, their validity is poorly point estimates of associations between the JEM-based exposures explored. Most of the previous studies on the validation of JEMs and depression were reduced by 22–65% as compared with those examined their ability to detect known associations between JEM for self-reported exposures in the H2000 Study, particularly in measures and health outcomes (predictive validity) [24–28]. Few women. The smallest drop was found for low job control (men) studies evaluated inter-method agreement between JEM and self- and monotonous work (women), while the largest reduction in reported measures [24,43]. There are several parameters that can estimates was observed for low social support in women. After be used to evaluate the performance of an exposure assessment correction for exposure misclassification, the odds ratios obtained method, of which sensitivity, specificity, Youden’s J index and with JEM regained their statistical significance for low job control likelihood ratios are the most commonly applied. Considering all (both genders), monotonous work (women), and high job demands, performance indicators, the performance of our JEM was good for low social support and high strain job (men). However, women job control and job strain and was rather low for job demands and with high job demands or low social support assessed by JEM had social support. These findings are in line with the results of the reduced odds of depression. Similarly, monotonous work seemed previous studies that reported higher validity of the JEM measures to be associated with lower risk of depression in men. for job control and job strain than for job demands and social All self-reported psychosocial factors at work, except monoto- support [13,43,44]. The relatively low validity of job demands may nous work, were statistically significantly associated with LBP in suggest that variation of this factor between occupations is smaller women (Table 5). In men, high job demands, low job control and than that within occupation [20,21]. However, the poor perfor- low social support tended to increase the odds of LBP, although mance for social support may alternatively reflect that some the association was statistically significant for high job strain only. psychosocial factors are highly individually oriented in that a The estimated odds for JEM-based exposures were reduced by 6– particular job may be perceived as very strenuous for some 21% in men and by 12–32% in women as compared with those for whereas not for others. self-reported exposures. Unexpectedly, for monotonous work, the Among performance indicators, sensitivity and specificity are odds ratios obtained with JEM were increased by 21% as the key ones, because all others are calculated based on them. compared to odds ratios obtained with self-reports. After Theoretically, sensitivity and specificity should be determined correction for exposure misclassification error, all JEM-based against a reference test (gold standard). In practice, the sensitivity exposures in men and all except high job demands in women were and specificity of the JEMs are usually evaluated against self- statistically significantly associated with LBP. Women with low reports, even if it is well known that the self-reported exposures social support had a low prevalence of LBP. may be subject to information bias. In the current study, we used the Bayesian approach to estimate sensitivity and specificity of Discussion JEM measures. The similarity of estimates obtained in both of our We comprehensively validated a gender-specific job exposure study samples suggests their robustness. The sensitivity of the JEM- matrix that we constructed for the assessment of psychosocial based estimates for job control and high strain job was acceptable, factors at work. The matrix showed a good accuracy in while it was reduced for job demands, monotonous work and identification of individuals exposed to low job control and high social support. The specificity of all our JEM-based estimates job strain, while its performance for job demands and social varied from good to very good and was substantially higher, PLOS ONE | www.plosone.org 6 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix PLOS ONE | www.plosone.org 7 September 2014 | Volume 9 | Issue 9 | e108987 Table 5. The association of psychosocial exposures measured at individual (ind) level and at group level (job exposure matrix (JEM)) with one-month prevalence of depression and low back pain among men and women. High demands Low control Monotonous work Low social support High strain job Prevalence (95% CI) OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI Men Depressive symptoms H2000 (ind) 13.9 (12.3–15.5) 1.98 1.51–2.59 1.52 1.16–2.01 1.45 1.07–1.95 2.07 1.59–2.74 3.07 2.05–4.62 FWH (JEM) 10.6 (9.8–11.5) 1.15 0.96–1.37 1.19 0.98–1.43 0.95 0.75–1.21 1.11 0.93–1.31 1.34 1.04–1.72 FWH (JEM) 2.08 1.74–2.48 1.33 1.12–1.59 0.81 0.67–0.99 1.54 1.30–1.72 1.70 1.34–2.15 Low back pain H2000 (ind) 26.8 (25.0–28.7) 1.19 0.99–1.44 1.13 0.93–1.37 1.01 0.81–1.26 1.16 0.96–1.41 1.37 1.04–1.80 FWH (JEM) 29.8 (28.6–31.0) 1.04 0.92–1.18 1.06 0.94–1.21 1.22 1.05–1.42 1.06 0.94–1.19 1.10 0.92–1.30 FWH (JEM) 1.20 1.07–1.34 1.73 1.54–1.95 2.38 2.10–2.69 1.50 1.33–1.68 1.77 1.51–2.09 Women Depressive symptoms H2000 (ind) 18.7 (17.0–20.5) 1.55 1.22–1.95 1.49 1.17–1.90 1.29 1.00–1.67 2.75 2.16–3.50 2.38 1.68–3.37 FWH (JEM) 15.9 (14.9–16.9) 0.99 0.85–1.15 1.07 0.91–1.28 1.01 0.84–1.22 0.97 0.83–1.13 1.09 0.87–1.37 FWH (JEM) 0.88 0.76–1.01 1.33 1.15–1.54 1.24 1.05–1.45 0.79 0.69–0.92 1.20 0.96–1.50 Low back pain H2000 (ind) 29.7 (27.8–31.6) 1.29 1.08–1.55 1.27 1.05–1.54 1.16 0.95–1.43 1.28 1.07–1.54 1.68 1.28–2.20 FWH (JEM) 31.4 (30.2–32.) 1.00 0.89–1.13 1.12 0.99–1.27 1.19 1.03–1.38 1.00 0.89–1.12 1.14 0.95–1.38 FWH (JEM) 0.92 0.82–1.03 1.76 1.56–1.97 2.38 2.10–2.71 0.83 0.74–0.94 1.52 1.27–1.81 Odds ratios (OR) and their 95% confidence intervals (95% CI). H2000 – the Health 2000 Study; FWH- the Finnish Work and Health Surveys; ORs calculated based on self-reports and adjusted for age and education. ORs calculated based on JEM and adjusted for age, education and year of survey. ORs calculated based on JEM adjusted for exposure misclassification bias. doi:10.1371/journal.pone.0108987.t005 Validity of Psychosocial Factors Job Exposure Matrix especially in women, as compared to those found in a French study The ability of the JEM to detect known associations between [24]. risk factors and health outcomes primarily depends on the The studies that examined the predictive validity of the magnitude of misclassification error. Even though studies have examined the predictive validity of psychosocial JEM measures, psychosocial JEM measures have consistently reported weaker associations between JEM measures and health outcomes than none of them examined the effect of exposure misclassification on what has been found for the corresponding self-reported factors observed associations. Our results suggest that, due to misclassi- [24–28]. In general, the associations of JEM measures for job fication error, we were not able to observe associations between strain and job control with health outcomes were better job demands, job control and social support assessed by JEM with reproducible than the associations for job demands. However, either depression or low back pain. However, after correction for even unexpected results of a protective effect of high job demands misclassification bias, the ability of the matrix to detect the expected associations improved substantially. Furthermore, the assessed by JEM on anxiety disorders [25] and self-rated health [24] have been reported. bias-adjusted effect estimates for low job control and high job strain in our study were about the same as those reported in When JEM is used to study the association between an exposure and a health outcome, there is always some loss of information previous meta-analyses [7,9]. because the individual values are replaced with the group-based (job title) ones. Both self-reported exposures and JEM are prone to Conclusions classification errors whose consequences on effect estimates need Our results suggest that JEM more accurately identifies to be considered when interpreting the association between the occupations with low control and high strain than those with exposure and the outcome. The measurement error in exposures high demands or low social support. Although the JEM is a rather assessed by JEM is always of a Berkson type, while the error of self- crude exposure assessment method, it can be a useful source of reported measures is of a classical type. The group-specific average job-level psychosocial exposures in epidemiological studies lacking of exposures used in our JEM was obtained based on nationally individual-level exposure. Furthermore, we showed the applica- representative self-reported exposure data; therefore, the mea- bility of a Bayesian approach in the evaluation of the performance surement error of our JEM has both classical and Berkson of the JEM in a situation where, in practice, no gold standard of component, with the latter being dominant. The classical and exposure assessment exists. Berkson errors bias the effect estimates differently [15]. The Berkson error has almost no effect on the point estimate, while it Acknowledgments severely affects the estimate’s precision. In case of classical error, the direction and magnitude of bias are more difficult to assess. We We would like to thank Dr. Timo Kauppinen for his expertise in the observed a larger difference between the self-reported and JEM- development of the job exposure matrix. based exposures in the ORs for depressive symptoms than for LBP. This may suggest the presence of a higher common source Author Contributions bias in self-reported exposure measures among those reporting Conceived and designed the experiments: SS TP JK RS AB KH-P EV-J. depressive symptoms than among those reporting LBP. As a result, Performed the experiments: SS TP JK RS AB KH-P EV-J. Analyzed the for depressive symptoms, the risk estimates based on JEM data: SS. Contributed reagents/materials/analysis tools: MH EV-J. Wrote measures may be closer to the true risk than the risk estimates the paper: SS TP EV-J KH-P. Contributed substantially to the based on self-reports. These benefits support the use of the JEM in interpretation of the findings and critically revised the manuscript: SS epidemiological studies. TP JK RS MH AB KH-P EV-J. References 1. Karasek RA, Theorell T (1990) Healthy Work: Stress, Productivity, and the 13. Theorell T, Hasselhorn HM (2005) On cross-sectional questionnaire studies of Reconstruction of Working Life. New York: Basic Books. relationships between psychosocial conditions at work and health–are they reliable? Int Arch Occup Environ Health 78:517–22. 2. Siegrist J (1996) Adverse health effects of high-effort/low-reward conditions. 14. Blair A, Stewart P, Lubin JH, Forastiere F (2007) Methodological issues J Occup Health Psychol 1:27–41. regarding confounding and exposure misclassification in epidemiological studies 3. Johnson JV, Hall EM (1988) Job strain, work place social support, and of occupational exposures. Am J Ind Med 50:199–207. cardiovascular disease: a cross-sectional study of a random sample of the 15. Armstrong BG (1998) Effect of measurement error on epidemiological studies of Swedish working population. Am J Public Health 78: 1336–42. environmental and occupational exposures. Occup Environ Med 55:651–6. 4. Karasek RA (1979) Job demands, job decision latitude, and mental strain: 16. Tielemans E, Heederik D, Burdorf A, Vermeulen R, Veulemans H, et al. (1999) Implications for job redesign. Adm Sci Q 24:285–308. Assessment of occupational exposures in a general population: comparison of 5. Die `ne E, Fouquet A, Esquirol Y (2012) Cardiovascular diseases and psychosocial different methods. Occup Environ Med 56:145–151. factors at work. Arch Cardiovasc Dis 105:33–9. 17. Siemiatycki J, Dewar R, Richardson L (1989) Costs and statistical power 6. Kivima ¨ ki M, Nyberg ST, Batty GD, Fransson EI, Heikkila ¨ K, et al. (2012) Job associated with five methods of collecting occupation exposure information for strain as a risk factor for coronary heart disease: a collaborative meta-analysis of population-based case-control studies. Am J Epidemiol 130:1236–46. individual participant data. Lancet 380(9852):1491–7. 18. Bouyer J, Dardenne J, He ´mon D (1995) Performance of odds ratios obtained 7. Stansfeld S, Candy B (2006) Psychosocial work environment and mental health– with a job-exposure matrix and individual exposure assessment with special a meta-analytic review. Scand J Work Environ Health 32:443–62. reference to misclassification errors. Scand J Work Environ Health 21:265–71. 8. Cosgrove MP, Sargeant LA, Caleyachetty R, Griffin SJ (2012) Work-related 19. Flegal KM, Brownie C, Haas JD (1986) The effects of exposure misclassification stress and Type 2 diabetes: systematic review and meta-analysis. Occup Med on estimates of relative risk. Am J Epidemiol 123:736–51. (Lond) 62:167–73. 20. Schwartz JE, Pieper CF, Karasek RA (1988) A procedure for linking 9. Lang J, Ochsmann E, Kraus T, Lang JW (2012) Psychosocial work stressors as psychosocial job characteristics data to health surveys. Am J Public Health antecedents of musculoskeletal problems: a systematic review and meta-analysis 78:904–9. of stability-adjusted longitudinal studies. Soc Sci Med 75:1163–74. 21. Johnson JV, Stewart WF (1993) Measuring work organization exposure over the 10. Tabanelli MC, Depolo M, Cooke RM, Sarchielli G, Bonfiglioli R, et al. (2008) life course with a job-exposure matrix. Scand J Work Environ Health 19:21–8. Available instruments for measurement of psychosocial factors in the work 22. Mariani M (1999) Replace with a database: O*NET replaces the Dictionary of environment. Int Arch Occup Environ Health 82:1–12. Occupational Titles. Occup Outlook Quarterly 43:3–9. 11. Landsbergis P, Theorell T, Schwartz J, Greiner BA, Krause N (2000) 23. Kauppinen T, Toikkanen J, Pukkala E (1998) From cross-tabulations to Measurement of psychosocial workplace exposure variables. Occup Med multipurpose exposure information systems: a new job-exposure matrix. 15:163–88. Am J Ind Med 33:409–17. 12. Macleod J, Davey Smith G (2003) Psychosocial factors and public health: a 24. Niedhammer I, Chastang JF, Levy D, David S, Degioanni S, et al. (2008) Study suitable case for treatment? J Epidemiol Community Health 57:565–70. of the validity of a job-exposure matrix for psychosocial work factors: results PLOS ONE | www.plosone.org 8 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix from the national French SUMER survey. Int Arch Occup Environ Health pulmonary disease with job exposure matrices (ECOJEM). Occup Environ Med 57:126–32. 82:87–97. 34. Solovieva S, Pehkonen I, Kausto J, Miranda H, Shiri R, et al. (2012) 25. Wieclaw J, Agerbo E, Mortensen PB, Burr H, Tuchsen F, et al. (2008) Development and validation of a job exposure matrix for physical risk factors in Psychosocial working conditions and the risk of depression and anxiety disorders low back pain. PLoS One 7(11):e48680. in the Danish workforce. BMC Public Health 8:280. 35. Bonde JP (2008) Psychosocial factors at work and risk of depression: a systematic 26. Rijs KJ, van der Pas S, Geuskens GA, Cozijnsen R, Koppes LL, et al. (2013) review of the epidemiological evidence. Occup Environ Medicine 65:438–45. Development and Validation of a Physical and Psychosocial Job-Exposure 36. Netterstrøm B, Conrad N, Bech P, Fink P, Olsen O, et al. (2008) The relation Matrix in Older and Retired Workers. Ann Occup Hyg 2013 Dec 11. [Epub between work-related psychosocial factors and the development of depression. ahead of print], doi:10.1093/annhyg/met052. Epidemiologic reviews 30:118–32. 27. Theorell T, Tsutsumi A, Hallquist J, Reuterwall C, Hogstedt C, et al. (1998) 37. Niedhammer I, Sultan-Taı ¨eb H, Chastang JF, Vermeylen G, Parent-Thirion A Decision latitude, job strain, and myocardial infarction: a study of working men (2013). Fractions of cardiovascular diseases and mental disorders attributable to in Stockholm. The SHEEP Study Group. Stockholm Heart epidemiology psychosocial work factors in 31 countries in Europe. Int Arch Occup Environ Program. Am J Public Health 88:382–8. Health 2013 Apr 27. [Epub ahead of print], doi 10.1007/s00420-013-0879-4 28. Cohidon C, Santin G, Chastang JF, Imbernon E, Niedhammer I (2012) 38. Hartvigsen J, Lings S, Leboeuf-Yde C, Bakketeig L (2004). Psychosocial factors Psychosocial exposures at work and mental health: potential utility of a job- at work in relation to low back pain and consequences of low back pain; a exposure matrix. Occup Environ Med 54:184–91. systematic, critical review of prospective cohort studies. Occup Environ Med 29. Aromaa A, Koskinen S (2004) Population and methods. Health and functional 61:e2. capacity in Finland. In: Aromaa A. and Koskinen S, editors. Baseline result of 39. Macfarlane GJ, Pallewatte N, Paudyal P, Blyth FM, Coggon D, et al. (2009) the Health 2000 Helath Examination Survey. Helsinki: The National Public Evaluation of work-related psychosocial factors and regional musculoskeletal Health Institute pp. 11–23. pain: results from a EULAR Task Force. Ann Rheum Dis 68:885–91. 30. Laiho J, Djerf K, Lehtonen R (2006) Sampling design. In: Heistaro S, editor. 40. Bouyer J, He ´mon D (1993) Studying the performance of a job exposure matrix. Methodology report Health 2000 Survey. Helsinki: the National Public Health Int J Epidemiol 22 (Suppl 2):S65–71. Institute B26: 13–5. 41. Joseph L, Gyorkos TW, Coupal L (1995) Bayesian estimation of disease 31. Perkio ¨-Ma ¨ kela ¨ M, Hirvonen M, Elo A-L, Kauppinen K, Kauppinen T, et al. prevalence and the parameters of diagnostic tests in the absence of a gold (2010) Tyo ¨ ja terveys -haastattelututkimus. 2009; Helsinki: Tyo ¨ terveyslaitos (in standard. Am J Epidemiol 141:263–72. Finnish). 42. Abramson JH (2011) WINPEPI updated: computer programs for epidemiolo- 32. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, et al. (1998) The gists, and their teaching potential. Epidemiol Perspect Innov 8:1. Job Content Questionnaire (JCQ): an instrument for internationally comparative 43. Cifuentes M, Boyer J, Gore R, d’Errico A, Tessler J, et al. (2007) Inter-method assessments of psychosocial job characteristics. J Occup Health Psychol 3:322– agreement between O*NET and survey measures of psychosocial exposure 55. among healthcare industry employees. Am J Ind Med 50:545–53. 33. Le Moual N, Bakke P, Orlowski E, Heederik D, Kromhout H, et al. (2000) 44. Ostry AS, Marion SA, Demers PA, Hershler R, Kelly S, et al. (2001) Measuring Performance of population specific job exposure matrices (JEMs): European psychosocial job strain with the job content questionnaire using experienced job collaborative analyses on occupational risk factors for chronic obstructive evaluators. Am J Ind Med 39:397–401. PLOS ONE | www.plosone.org 9 September 2014 | Volume 9 | Issue 9 | e108987 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png PLoS ONE Unpaywall

Evaluation of the Validity of Job Exposure Matrix for Psychosocial Factors at Work

Download PDF
9 pages

Loading next page...
 
/lp/unpaywall/evaluation-of-the-validity-of-job-exposure-matrix-for-psychosocial-kHkfs0VXS7

References (90)

Publisher
Unpaywall
ISSN
1932-6203
DOI
10.1371/journal.pone.0108987
Publisher site
See Article on Publisher Site

Abstract

Objective: To study the performance of a developed job exposure matrix (JEM) for the assessment of psychosocial factors at work in terms of accuracy, possible misclassification bias and predictive ability to detect known associations with depression and low back pain (LBP). Materials and Methods: We utilized two large population surveys (the Health 2000 Study and the Finnish Work and Health Surveys), one to construct the JEM and another to test matrix performance. In the first study, information on job demands, job control, monotonous work and social support at work was collected via face-to-face interviews. Job strain was operationalized based on job demands and job control using quadrant approach. In the second study, the sensitivity and specificity were estimated applying a Bayesian approach. The magnitude of misclassification error was examined by calculating the biased odds ratios as a function of the sensitivity and specificity of the JEM and fixed true prevalence and odds ratios. Finally, we adjusted for misclassification error the observed associations between JEM measures and selected health outcomes. Results: The matrix showed a good accuracy for job control and job strain, while its performance for other exposures was relatively low. Without correction for exposure misclassification, the JEM was able to detect the association between job strain and depression in men and between monotonous work and LBP in both genders. Conclusions: Our results suggest that JEM more accurately identifies occupations with low control and high strain than those with high demands or low social support. Overall, the present JEM is a useful source of job-level psychosocial exposures in epidemiological studies lacking individual-level exposure information. Furthermore, we showed the applicability of a Bayesian approach in the evaluation of the performance of the JEM in a situation where, in practice, no gold standard of exposure assessment exists. Citation: Solovieva S, Pensola T, Kausto J, Shiri R, Helio¨ vaara M, et al. (2014) Evaluation of the Validity of Job Exposure Matrix for Psychosocial Factors at Work. PLoS ONE 9(9): e108987. doi:10.1371/journal.pone.0108987 Editor: James Coyne, University of Pennsylvania, United States of America Received January 31, 2014; Accepted September 6, 2014; Published September 30, 2014 Copyright:  2014 Solovieva et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The study was financially supported by the Finnish Work Environment Fund (project no 109364). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * Email: [email protected] diabetes [8] and musculoskeletal diseases [9]. The effects of the Introduction individual components of the job strain model on health have also During the past three decades, the effects of psychosocial factors been evaluated, although the results have often been inconsistent at work on health have received considerable attention in research. across the studies and health outcomes [7,9]. Psychosocial factors at work are numerous, with psychological job The interpretation of the observed associations between demands, job control (decision latitude), efforts and rewards [1,2] psychosocial factors at work and health mainly depends on the comprising the key dimensions. Another factor of importance is validity of the assessment methods of the risk factors. Self-reported social support at work [3]. questionnaires are widely used to measure psychosocial factors at The job strain model, introduced by Karasek in 1979 [4], is one work [10]. Self-reports provide subjective information representing of the most studied occupational stress models. According to the a worker’s perception of occupational stress and are therefore model, workers with a combination of high psychosocial job susceptible to reporting bias. The subjective assessment of demands and low control over a job (high job strain) have a higher psychosocial factors at work has been the largest concern in the risk of developing an illness as compared to workers with low debate on the interpretation of associations and on the possible psychosocial job demands and high job control (low job strain) [1]. causal role of these factors for illness. It has been suggested that The job strain model has been successfully used to predict the risk common source bias due to subjective measures of psychosocial of cardiovascular disease [5,6], major mental disorders [7], type II PLOS ONE | www.plosone.org 1 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix factors at work increases the likelihood of false positive findings, aged 30 years or over and 1894 subjects aged 18–29. The participation rates were 87% and 90%, respectively. A detailed particularly in cross-sectional studies with the self-reported health outcomes [11–13]. Workers having health problems are more comprehensive description of the methods and processes has been published elsewhere [29,30]. The sample of this study comprised likely to report certain psychosocial exposures than healthy workers. Such tendency might lead to differential misclassification, 4619 persons aged 18–64 who were working during the preceding 12 months and for whom information on occupational titles and which results either in an overestimation or underestimation of the true effect [14], particularly, when exposures and outcome are exposures were available. The age and gender distribution of the study population matched those of the employed persons in measured simultaneously. Finland in the year 2000. The assessment of psychosocial factors at work with a job The national Finnish National Work and Health Surveys have exposure matrix (JEM), where exposure level is assigned based on been conducted every third year since 1997 to collect information the job-specific average of exposure, is not prone to information on perceived working conditions and the health of the working-age bias and may therefore guarantee some degree of objectivity. The population, For the 1997–2003 Surveys, random samples of major advantage of the JEM in epidemiological studies is that it subjects aged 25–64 years independent of their working status can be applied to the populations with lacking exposure (e.g., working, unemployed, retired or student) were drawn from information. However, such method of exposure assessment the Finnish population register. For the 2009 Survey a random induces Berkson type error, which may not cause notable bias sample of subjects aged 20–64 years was drawn from Finnish on the effect estimates but weakens the precision of the estimates employment statistics. The sample size has varied between 2031 [15]. A JEM neglects both within worker (over time variation) and and 2355 persons from year to year with a response rate of 58– between worker (variation in tasks, activities and work processes) 72% [31]. At each survey, a phone number was not found for variation in a job [16] and therefore may result in false positive about 10–16% of subjects. The proportion of non-participants in and negative exposure assignments for a considerable proportion each survey was slightly higher among men than women and of the subjects. A non-differential misclassification bias induced by among subjects aged 24–34 years than among the older subjects. JEM will attenuate the observed associations towards null Age, gender, education, socioeconomic status and occupational [15,17,18]. Knowing the magnitude of measurement error (e.g. sector of the respondents were compared with the Census data. No sensitivity and specificity) and exposure prevalence, the extent of major differences were found. Thus, the respondents to the FWH non-differential bias can be estimated [15,19]. Surveys represent rather well the targeted population. The data Several psychosocial job exposure matrices have been devel- from all five surveys were combined. Hence, the total number of oped and used in epidemiological studies [20–26]. Even though the interviewed persons with information on occupation during the JEM measures are more objective than self-reported ones, they 1997–2009 was 11326. cannot be seen as a gold standard in the context of psychosocial The H2000 Study and the FWH Surveys have all obtained factors at work [13]. Therefore, the question of the reliability of ethical approval from the appropriate ethics committees. the associations between JEM-based exposures and health outcomes is always warranted. The validity of psychosocial JEM Classification of occupations measures in the absence of a gold standard method is challenging Occupations in both surveys were classified on the 4-digit level to evaluate and, as a result, has rarely been examined and reported [24–28]. Furthermore, the magnitude of misclassification bias of (including few occupations coded with 5 digits) according to the Classification of Occupations 2001 by Statistics Finland, which is psychosocial JEM measures on effect estimates has not been based on the International Standard Classification of Occupations examined so far. (ISCO-88). The classification is based on ten categories of The aims of the study were 1) to examine the accuracy of a professional skills. In total, the classification includes 444 job titles. developed gender-specific job exposure matrix in the assessment of psychosocial factors at work applying the Bayesian approach, 2) to evaluate the theoretical impact of exposure misclassification on Psychosocial exposures exposure-outcome associations and 3) to examine the ability of the Psychosocial exposures in the H2000 Study were measured with matrix to detect known associations between psychosocial factors a Finnish version of the Job Content Questionnaire (JCQ) [32]. at work and health outcomes. The JCQ has been shown to be a valid and reliable instrument to assess job stress and social support in many occupational settings worldwide [10,13]. Responses were given on a five point Likert- Materials and Methods scale from 1 (fully agree) to 5 (fully disagree). Study population Psychological job demands scale is the sum of the following five We utilized two large Finnish population samples. The Health items: ‘‘work fast’’, ‘‘work hard’’, ‘‘excessive work’’, ‘‘not enough 2000 (H2000) Study was used to construct the JEM and to time’’, and ‘‘hectic job’’. In the current study, Cronbach’s alpha examine the inter-method agreement, and the national Finnish for the scale was 0.76 for men and 0.81 for women. Job control Work and Health (FWH) Surveys were used to test the scale is the sum of two subscales. Decision authority was measured performance of the matrix. The study populations consisted of with three items: ‘‘allows own decisions’’, ‘‘decision freedom’’, and 18–64 year-old individuals, who had been working during the ‘‘a lot of say on the job’’), and skill discretion was measured with preceding 12 months. five items: ‘‘learn new things’’, ‘‘requires creativity’’, ‘‘high skill The Health 2000 Study is a large Finnish population-based level’’, ‘‘variety’’, and ‘‘develop own abilities’’. Cronbach’s alpha study carried out in 2000–01. The main objective of the study was for the scale was 0.85 for men and 0.86 for women. Since to obtain representative information on the current health status of monotonous (repetitive) work was weakly correlated with the other the whole non-institutional adult population in Finland. The five items of the skill discretion scale we treated it as a separate survey consisted of several questionnaires, a home interview, and a exposure. Job demands, job control and monotonous work were health examination. A nationally representative sample of the dichotomized using gender-specific median cut-off points. population was obtained using a two-stage stratified cluster Job strain was operationalized using the quadrant approach sampling design. The original samples consisted of 8028 subjects proposed by Karasek and Theorell [1]. It defines workers who are PLOS ONE | www.plosone.org 2 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix above the median on job demands and below the median on job Low back pain. In the H2000 Study information on low control as having a high strain job. Other categories are: low strain back pain was inquired with the following question: ‘‘Have you (low demands and high control), passive (low demands and low had pain in your back during the past month (30 days)?’’ (yes/no). control) and active (high demands and high control). Low strain In the FWH Surveys, data on low back pain were collected with an job was used as the reference category in the analyses. interview using the question: ‘‘Have you during the past month (30 days) had long-lasting or recurrent pain in the lumbar spine?’’ Social support at work was measured with four items: ‘‘support from supervisor’’, ‘‘supervisor appreciates’’, ‘‘support from co- (yes/no). workers’’, ‘‘discussion on work’’. Cronbach’s alpha for the scale was 0.80 for men and 0.82 for women. Social support was Data analyses dichotomized at a gender-specific median in order to define low In the H2000 Study, the inter-method agreement between self- and high support. reported and JEM measures was examined using intra-class correlation (ICC). Two-way mixed total ICC agreements were Development of the job exposure matrix (JEM) computed. In the FWH Surveys, the performance of the matrix We constructed a gender-specific matrix with exposure was evaluated by examining the accuracy of the matrix in the estimates at each intersection between rows (occupational groups) identification of exposed/non-exposed individuals, estimating and columns (psychosocial exposures). The exposure axis of the exposure misclassification error, and looking at the ability of the matrix included the above mentioned five psychosocial risk factors matrix to detect associations of psychosocial factors at work with at work. The occupation axis of the matrix was based on the one-month prevalence of depression or low back pain (predictive original job titles or occupational groups. validity) [40]. The accuracy of the JEM was evaluated using five Previous studies showed that ten individuals with the same job indicators: sensitivity (Se), specificity (Sp), Youden’s J index, title will be sufficient for a reliable estimation of exposures [33,34]. likelihood ratio positive (LR+) and likelihood ratio negative (LR2). The exposure estimates for job demands, job control, monotonous Sensitivity (ability of the test to identify positive results) and work and social support at work were calculated as a median score specificity (ability of the test to identify negative results) are usually of exposures in each occupation which included at least 10 subjects determined against a reference standard test (gold standard). in order to obtain reasonably precise estimates. The exposure Errors in measuring the sensitivity and specificity of a test will arise estimates for job strain were calculated as the proportion of if the reference test itself does not have 100% sensitivity and 100% exposed to passive, active and high strain work. The job titles with specificity. Since there is no gold standard measure for psycho- a small number (,10) of respondents were grouped based on the social factors at work, we estimated sensitivity and specificity using similarities of these job titles with regard to work tasks (including a Bayesian approach, proposed by Joseph et al. [41]. As the first supervising), work environment, and required educational level. step, the posterior distribution of sensitivity and specificity of the The gender differences in the exposures were also considered. If JEM measures was calculated using self-reported and JEM there was no reasonable way to merge the occupation with other measures of exposures from H2000 Study. For these analyses, occupations within the gender (such as female frontier guards), the the prior distribution of the parameters was derived based on the exposure estimates of both genders in that occupation/occupa- assumption that the self-reported measures have almost perfect tional group were combined. sensitivity and specificity and no prior information on sensitivity The sample size of the H2000 Study was large enough to enable and specificity of JEM measures is available. As the second step, us to develop a gender-specific job exposure matrix and to keep the posterior distribution of sensitivity and specificity of the JEM several job titles unmerged. Out of 444 possible job titles, measures was calculated using data from FWH Surveys. For these altogether 363 (300 among men and 267 among women) were analyses the prior distributions of the parameters were derived available in the Health 2000 Study. There were 61 job titles based on the posterior distributions obtained in the first step. At among men and 58 among women with at least 10 subjects. These each step, the posterior medians and their 95% Bayesian credible job titles covered 69% of the study sample. After merging the intervals were estimated using Gibbs sampler algorithm with smaller groups the number of job titles or occupational groups WinBUGS software version 1.4.3. reduced to 110 among men and 101 among women. The estimated sensitivity and specificity were used to calculate The exposure estimates for job demands, job control, monot- Youden’s J index as well as LR+ and LR2. The Youden’s J index onous work and social support at work were dichotomized using (J = Se+Sp21) has been used as a measure of the effectiveness of gender-specific median as a cut-off point. The categories of job the JEM to discriminate between exposed and non-exposed strain were obtained based on the dichotomized JEM-based job individuals. The possible range of the Youden’s J index value is demands and job control. between 0 (totally useless) and 1 (perfect). Likelihood ratio positive is the probability of an exposed person to be classified as exposed divided by the probability of a non-exposed person to be classified Health outcomes Based on the current evidence we chose two health outcomes as exposed. Likelihood ratio negative is the probability of an exposed person to be classified as non-exposed divided by the that are known to be associated with psychosocial factors at work. Both cross-sectional and longitudinal studies have shown that high probability of a non-exposed person to be classified as non- exposed. A likelihood ratio equal to 1 will indicate that the JEM level of psychological demands and job strain are associated with major mental disorders [7,35–37]. Suggestive evidence for a measure is unable to distinguish between exposed and non- exposed. A LR.1 will indicate that the JEM is likely to identify relationship of job demands, job control and monotonous work with low back pain has also been reported [9,38,39]. exposed and LR,1 will indicate that the JEM is likely to identify non-exposed. The higher LR+ value and lower the LR2 value, Depressive symptoms. In both studies, depressive symp- toms were assessed with the following question: ‘‘Have you had the better is the JEM performance. To estimate the theoretical magnitude of exposure misclassifi- melancholy or depression during the last month (30 days)?’’. The response categories ranged from 1 = not at all to 5 = very often. cation, biased odds ratios (OR9) were calculated based on the The occurrence of depressive symptoms was dichotomized as no obtained estimates of sensitivity (Se) and specificity (Sp) and (categories 1 and 2) or yes (categories from 3 to 5). assumed ‘‘true prevalence’’ (Pr) and ‘‘true odds ratios’’ (OR) using PLOS ONE | www.plosone.org 3 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix Table 1. Prevalence of self-reported and JEM-based psychosocial exposures. Men Women H2000 Study FWH Surveys H2000 Study FWH Surveys Exposures Self-reported JEM JEM Self-reported JEM JEM Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) Prev. (95% CI) High job demands 43.1 (41.1, 45.0) 33.1 (29.4, 33.3) 32.2 (31.0, 33.4) 44.2 (42.2, 46.3) 35.3 (33.4, 37.3) 36.0 (34.7, 37.3) Low job control 52.6 (50.5, 54.7) 49.8 (47.8, 51.9) 49.8 (48.5, 51.1) 56.1 (54.1, 58.1) 56.5 (54.5, 58.5) 55.4 (54.1, 56.7) Job strain Low strain job 26.2 (24.4, 28.1) 36.3 (34.3, 38.3) 36.3 (34.1, 36.6) 23.8 (22.1, 25.6) 24.6 (22.9, 26.4) 25.6 (24.4, 26.3) Passive job 30.7 (28.8, 32.7) 32.4 (30.5, 34.4) 32.5 (31.3, 33.8) 32.0 (30.1, 33.9) 40.0 (38.1, 42.1) 38.4 (37.2, 39.7) Active job 21.2 (19.5, 22.9) 13.9 (12.5, 15.4) 14.9 (14.0, 15.8) 20.0 (18.4, 21.7) 18.9 (17.3, 20.5) 19.0 (18.0, 20.1) High strain job 21.9 (20.2, 23.6) 17.4 (15.9, 19.1) 17.3 (16.3, 18.3) 24.2 (22.5, 26.0) 16.4 (15.0, 18.0) 16.9 (16.0, 18.0) Low social support 47.3 (45.3, 49.4) 48.5 (46.4, 50.1) 48.4 (47.1, 49.7) 44.2 (42.2, 46.2) 37.1 (35.2, 39.1) 37.5 (36.2, 38.7) Monotonous work 28.5 (26.6, 30.4) 17.2 (15.7, 18.8) 17.1 (16.1, 18.1) 31.8 (29.9, 33.7) 24.0 (22.3, 25.8) 22.1 (21.0, 23.2) H2000 Study- the Health 2000 Study; FWH Surveys- the Finnish Work and Health Surveys. doi:10.1371/journal.pone.0108987.t001 the following formula [19]: The Finnish Work and Health Surveys and are shown in the form of posterior medians and 95% Bayesian intervals (Table 3). The posterior estimates were very similar in both study populations. ((Se OR Prz(1-Sp) (1-Pr)) ((1-Se) PrzSp (1-Pr)) OR~ The specificity of JEM measures was higher than sensitivity for all ((Se Prz(1-Sp) (1-Pr)) ((1-Se) OR PrzSp (1-Pr)) exposures except job control among women. Specificity ranged from 0.62 to 0.90 in men, and from 0.68 to 0.86 in women. Sensitivity was the lowest for high strain job (0.46) in men and for The true prevalence was fixed at 0.50 for high job demands, low job control and low social support, at 0.33 for monotonous work low social support (0.52) in women. The best matrix performance and at 0.25 for high strain job. The true odds ratios were fixed at assessed by Youden’s J index and likelihood ratios was found for three values OR = 1.5, OR = 2 and OR = 3. The relative high strain job, particularly in women. The JEM was least effective difference between biased and true estimates was calculated in identification of men exposed to high demands (J = 0.17) and ((OR9-OR)/OR) and used as quantitative measure for the women exposed to low social support (J = 0.15). magnitude of exposure misclassification. The theoretical effect of exposure misclassification error on Logistic regression analyses with age, education and year of estimated ORs is shown in Table 4. In both genders, the smallest survey (the FWH Surveys) adjusted odds ratios (OR) and 95% misclassification error was observed for high job strain, followed by confidence intervals (CIs) were carried out to study the associations that for low job control. The largest misclassification error was between the JEM measures and one-month prevalence of found for low social support (both genders) and high job demands depression or low back pain. These analyses were performed (men). In general, when the true OR is equal to 1.5, the effect of using SAS version 9.1. The effect estimates were adjusted for misclassification error on point estimates is relatively small, though misclassification error using WINPEPI COMPARE2 program, there is a high likelihood of false negative findings. A statistically version 3.08 [42]. significant association can be detected only for low job control and All analyses were performed separately for men and women. high job strain in women. With the increase of true OR, there is a larger reduction in the biased odds ratios, but at the same time the Results likelihood of false negative findings is lowered. In both genders, the prevalence of high job demands, high job strain and monotonous work measured by job exposure matrix Table 2. Intra-class correlation (ICC) between individual- was statistically significantly lower than that assessed by self-reports based and group-based measures of psychosocial exposures (Table 1). In women, the prevalence of low social support was for men and women in the H2000 Study. lower for JEM measures than for self-reported measures. There were no differences in the distribution of exposures assessed by JEM between the two study populations, reflecting a similar job Men Women distribution in both surveys. In general, total agreement between self-reported and JEM measures assessed by ICC was slightly Job demands 0.31 0.40 better among women than men, with the largest ICC values Job control 0.53 0.60 observed for job control followed by monotonous work (Table 2). Monotonous work 0.45 0.51 Social support 0.36 0.33 Bayesian estimates of sensitivity and specificity and Job strain 0.21 0.29 magnitude of exposure misclassification error The Bayesian estimates of sensitivity and specificity were H2000 Study- the Health 2000 Study. doi:10.1371/journal.pone.0108987.t002 calculated based on the data from the Health 2000 Study and PLOS ONE | www.plosone.org 4 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix PLOS ONE | www.plosone.org 5 September 2014 | Volume 9 | Issue 9 | e108987 Table 3. Posterior medians and lower and upper limits of the posterior equally tailed 95% credible intervals (Bayesian confidence interval). H2000 Study FWH Surveys 1 2 3 Sensitivity Specificity Sensitivity Specificity J LR+ LR2 High demands Men 0.41 (0.38–0.44) 0.76 (0.74–0.78) 0.41 (0.39–0.44) 0.76 (0.74–0.78) 0.17 1.71 (1.53–1.91) 0.78 (0.73–0.82) Women 0.49 (0.46–0.52) 0.75 (0.73–0.78) 0.49 (0.46–0.52) 0.75 (0.73–0.78) 0.24 2.00 (1.77–2.25) 0.68 (0.63–0.72) Low control Men 0.67 (0.64–0.70) 0.69 (0.67–0.72) 0.67 (0.64–0.69) 0.70 (0.67–0.72) 0.37 2.20 (2.01–2.42) 0.48 (0.43–0.52) Women 0.75 (0.73–0.78) 0.68 (0.65–0.71) 0.76 (0.73–0.78) 0.68 (0.65–0.71) 0.44 2.35 (2.14–2.59) 0.36 (0.32–0.40) High strain job Men 0.58 (0.52–0.64) 0.87 (0.84–0.91) 0.60 (0.55–0.65) 0.82 (0.77–0.86) 0.42 3.34 (2.52–4.45) 0.49 (0.42–0.57) Women 0.77 (0.71–0.82) 0.88 (0.84–0.92) 0.78 (0.73–0.83) 0.86 (0.81–0.90) 0.64 5.75 (4.13–8.03) 0.25 (0.19–0.31) Low social support Men 0.60 (0.57–0.63) 0.62 (0.59–0.65) 0.60 (0.57–0.63) 0.62 (0.59–0.65) 0.22 1.59 (1.45–1.74) 0.65 (0.59–0.71) Women 0.48 (0.45–0.51) 0.72 (0.70–0.74) 0.44 (0.42–0.46) 0.71 (0.69–0.73) 0.15 1.52 (1.40–1.65) 0.79 (0.75–0.83) Monotonous work Men 0.35 (0.32–0.39) 0.90 (0.88–0.91) 0.36 (0.32–0.39) 0.90 (0.88–0.91) 0.26 3.39 (2.84–4.01) 0.72 (0.68–0.76) Women 0.46 (0.42–0.50) 0.86 (0.84–0.88) 0.46 (0.42–0.50) 0.86 (0.84–0.88) 0.32 3.31 (2.79–3.91) 0.63 (0.58–0.68) The sensitivity and specificity of JEM-based measures for men and women in the H2000 Study and the FWH Surveys. H2000 Study- the Health 2000 Study; FWH Surveys- the Finnish Work and Health Surveys; J - Youden’s index = sensitivity+specificity21. LR+ likelihood ratio positive. LR2 likelihood ratio negative. doi:10.1371/journal.pone.0108987.t003 Validity of Psychosocial Factors Job Exposure Matrix Table 4. Biased odds (OR9) ratios according to sensitivity and specificity of the job exposure matrix when the true odds ratios (OR) were assumed to equal 1.5, 2 or 3. OR = 1.5 OR = 2.0 OR = 3.0 Men Women Men Women Men Women High job demands 1.08 1.11 1.13 1.18* 1.21* 1.28* Low job control 1.16 1.20* 1.28* 1.35* 1.45* 1.58* Monotonous work 1.17 1.17 1.30* 1.31* 1.50* 1.52* Low social support 1.09 1.07 1.16 1.11 1.25* 1.17* High job strain 1.18 1.27* 1.34* 1.53* 1.60* 1.99* Prevalence of exposure is assumed to equal 0.50. Prevalence of exposure is assumed to equal 0.33. Prevalence of exposure is assumed to equal 0.25. *Statistical significance at the 5% level (two-sided test) of the biased odds ratios is calculated for a study population of 5000 men and 5000 women. doi:10.1371/journal.pone.0108987.t004 support was relatively low. The largest misclassification error was Predictive validity of the JEM measures found for low social support (women) and high job demands (men). The one-month prevalence of depression was statistically The difference between the odds ratios based on self-reports and significantly higher in the H2000 Study as compared with the JEM was larger for depression than for low back pain, especially in FWH Surveys, while the prevalence of low back pain during the women. Without correction for exposure misclassification, the preceding 30 days was similar (Table 5). In both study popula- JEM was able to detect the association between job strain and tions, women tended to report depression and LBP more depression in men and that between monotonous work and low frequently than men. back pain in both genders. The predictive ability of the matrix In the H2000 Study, associations between all self-reported substantially improved after correction for possible misclassifica- psychosocial factors at work and depression were statistically tion bias. significant in both genders (Table 5). In the FWH Surveys, the Although several psychosocial JEMs exist, their validity is poorly point estimates of associations between the JEM-based exposures explored. Most of the previous studies on the validation of JEMs and depression were reduced by 22–65% as compared with those examined their ability to detect known associations between JEM for self-reported exposures in the H2000 Study, particularly in measures and health outcomes (predictive validity) [24–28]. Few women. The smallest drop was found for low job control (men) studies evaluated inter-method agreement between JEM and self- and monotonous work (women), while the largest reduction in reported measures [24,43]. There are several parameters that can estimates was observed for low social support in women. After be used to evaluate the performance of an exposure assessment correction for exposure misclassification, the odds ratios obtained method, of which sensitivity, specificity, Youden’s J index and with JEM regained their statistical significance for low job control likelihood ratios are the most commonly applied. Considering all (both genders), monotonous work (women), and high job demands, performance indicators, the performance of our JEM was good for low social support and high strain job (men). However, women job control and job strain and was rather low for job demands and with high job demands or low social support assessed by JEM had social support. These findings are in line with the results of the reduced odds of depression. Similarly, monotonous work seemed previous studies that reported higher validity of the JEM measures to be associated with lower risk of depression in men. for job control and job strain than for job demands and social All self-reported psychosocial factors at work, except monoto- support [13,43,44]. The relatively low validity of job demands may nous work, were statistically significantly associated with LBP in suggest that variation of this factor between occupations is smaller women (Table 5). In men, high job demands, low job control and than that within occupation [20,21]. However, the poor perfor- low social support tended to increase the odds of LBP, although mance for social support may alternatively reflect that some the association was statistically significant for high job strain only. psychosocial factors are highly individually oriented in that a The estimated odds for JEM-based exposures were reduced by 6– particular job may be perceived as very strenuous for some 21% in men and by 12–32% in women as compared with those for whereas not for others. self-reported exposures. Unexpectedly, for monotonous work, the Among performance indicators, sensitivity and specificity are odds ratios obtained with JEM were increased by 21% as the key ones, because all others are calculated based on them. compared to odds ratios obtained with self-reports. After Theoretically, sensitivity and specificity should be determined correction for exposure misclassification error, all JEM-based against a reference test (gold standard). In practice, the sensitivity exposures in men and all except high job demands in women were and specificity of the JEMs are usually evaluated against self- statistically significantly associated with LBP. Women with low reports, even if it is well known that the self-reported exposures social support had a low prevalence of LBP. may be subject to information bias. In the current study, we used the Bayesian approach to estimate sensitivity and specificity of Discussion JEM measures. The similarity of estimates obtained in both of our We comprehensively validated a gender-specific job exposure study samples suggests their robustness. The sensitivity of the JEM- matrix that we constructed for the assessment of psychosocial based estimates for job control and high strain job was acceptable, factors at work. The matrix showed a good accuracy in while it was reduced for job demands, monotonous work and identification of individuals exposed to low job control and high social support. The specificity of all our JEM-based estimates job strain, while its performance for job demands and social varied from good to very good and was substantially higher, PLOS ONE | www.plosone.org 6 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix PLOS ONE | www.plosone.org 7 September 2014 | Volume 9 | Issue 9 | e108987 Table 5. The association of psychosocial exposures measured at individual (ind) level and at group level (job exposure matrix (JEM)) with one-month prevalence of depression and low back pain among men and women. High demands Low control Monotonous work Low social support High strain job Prevalence (95% CI) OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI Men Depressive symptoms H2000 (ind) 13.9 (12.3–15.5) 1.98 1.51–2.59 1.52 1.16–2.01 1.45 1.07–1.95 2.07 1.59–2.74 3.07 2.05–4.62 FWH (JEM) 10.6 (9.8–11.5) 1.15 0.96–1.37 1.19 0.98–1.43 0.95 0.75–1.21 1.11 0.93–1.31 1.34 1.04–1.72 FWH (JEM) 2.08 1.74–2.48 1.33 1.12–1.59 0.81 0.67–0.99 1.54 1.30–1.72 1.70 1.34–2.15 Low back pain H2000 (ind) 26.8 (25.0–28.7) 1.19 0.99–1.44 1.13 0.93–1.37 1.01 0.81–1.26 1.16 0.96–1.41 1.37 1.04–1.80 FWH (JEM) 29.8 (28.6–31.0) 1.04 0.92–1.18 1.06 0.94–1.21 1.22 1.05–1.42 1.06 0.94–1.19 1.10 0.92–1.30 FWH (JEM) 1.20 1.07–1.34 1.73 1.54–1.95 2.38 2.10–2.69 1.50 1.33–1.68 1.77 1.51–2.09 Women Depressive symptoms H2000 (ind) 18.7 (17.0–20.5) 1.55 1.22–1.95 1.49 1.17–1.90 1.29 1.00–1.67 2.75 2.16–3.50 2.38 1.68–3.37 FWH (JEM) 15.9 (14.9–16.9) 0.99 0.85–1.15 1.07 0.91–1.28 1.01 0.84–1.22 0.97 0.83–1.13 1.09 0.87–1.37 FWH (JEM) 0.88 0.76–1.01 1.33 1.15–1.54 1.24 1.05–1.45 0.79 0.69–0.92 1.20 0.96–1.50 Low back pain H2000 (ind) 29.7 (27.8–31.6) 1.29 1.08–1.55 1.27 1.05–1.54 1.16 0.95–1.43 1.28 1.07–1.54 1.68 1.28–2.20 FWH (JEM) 31.4 (30.2–32.) 1.00 0.89–1.13 1.12 0.99–1.27 1.19 1.03–1.38 1.00 0.89–1.12 1.14 0.95–1.38 FWH (JEM) 0.92 0.82–1.03 1.76 1.56–1.97 2.38 2.10–2.71 0.83 0.74–0.94 1.52 1.27–1.81 Odds ratios (OR) and their 95% confidence intervals (95% CI). H2000 – the Health 2000 Study; FWH- the Finnish Work and Health Surveys; ORs calculated based on self-reports and adjusted for age and education. ORs calculated based on JEM and adjusted for age, education and year of survey. ORs calculated based on JEM adjusted for exposure misclassification bias. doi:10.1371/journal.pone.0108987.t005 Validity of Psychosocial Factors Job Exposure Matrix especially in women, as compared to those found in a French study The ability of the JEM to detect known associations between [24]. risk factors and health outcomes primarily depends on the The studies that examined the predictive validity of the magnitude of misclassification error. Even though studies have examined the predictive validity of psychosocial JEM measures, psychosocial JEM measures have consistently reported weaker associations between JEM measures and health outcomes than none of them examined the effect of exposure misclassification on what has been found for the corresponding self-reported factors observed associations. Our results suggest that, due to misclassi- [24–28]. In general, the associations of JEM measures for job fication error, we were not able to observe associations between strain and job control with health outcomes were better job demands, job control and social support assessed by JEM with reproducible than the associations for job demands. However, either depression or low back pain. However, after correction for even unexpected results of a protective effect of high job demands misclassification bias, the ability of the matrix to detect the expected associations improved substantially. Furthermore, the assessed by JEM on anxiety disorders [25] and self-rated health [24] have been reported. bias-adjusted effect estimates for low job control and high job strain in our study were about the same as those reported in When JEM is used to study the association between an exposure and a health outcome, there is always some loss of information previous meta-analyses [7,9]. because the individual values are replaced with the group-based (job title) ones. Both self-reported exposures and JEM are prone to Conclusions classification errors whose consequences on effect estimates need Our results suggest that JEM more accurately identifies to be considered when interpreting the association between the occupations with low control and high strain than those with exposure and the outcome. The measurement error in exposures high demands or low social support. Although the JEM is a rather assessed by JEM is always of a Berkson type, while the error of self- crude exposure assessment method, it can be a useful source of reported measures is of a classical type. The group-specific average job-level psychosocial exposures in epidemiological studies lacking of exposures used in our JEM was obtained based on nationally individual-level exposure. Furthermore, we showed the applica- representative self-reported exposure data; therefore, the mea- bility of a Bayesian approach in the evaluation of the performance surement error of our JEM has both classical and Berkson of the JEM in a situation where, in practice, no gold standard of component, with the latter being dominant. The classical and exposure assessment exists. Berkson errors bias the effect estimates differently [15]. The Berkson error has almost no effect on the point estimate, while it Acknowledgments severely affects the estimate’s precision. In case of classical error, the direction and magnitude of bias are more difficult to assess. We We would like to thank Dr. Timo Kauppinen for his expertise in the observed a larger difference between the self-reported and JEM- development of the job exposure matrix. based exposures in the ORs for depressive symptoms than for LBP. This may suggest the presence of a higher common source Author Contributions bias in self-reported exposure measures among those reporting Conceived and designed the experiments: SS TP JK RS AB KH-P EV-J. depressive symptoms than among those reporting LBP. As a result, Performed the experiments: SS TP JK RS AB KH-P EV-J. Analyzed the for depressive symptoms, the risk estimates based on JEM data: SS. Contributed reagents/materials/analysis tools: MH EV-J. Wrote measures may be closer to the true risk than the risk estimates the paper: SS TP EV-J KH-P. Contributed substantially to the based on self-reports. These benefits support the use of the JEM in interpretation of the findings and critically revised the manuscript: SS epidemiological studies. TP JK RS MH AB KH-P EV-J. References 1. Karasek RA, Theorell T (1990) Healthy Work: Stress, Productivity, and the 13. Theorell T, Hasselhorn HM (2005) On cross-sectional questionnaire studies of Reconstruction of Working Life. New York: Basic Books. relationships between psychosocial conditions at work and health–are they reliable? Int Arch Occup Environ Health 78:517–22. 2. Siegrist J (1996) Adverse health effects of high-effort/low-reward conditions. 14. Blair A, Stewart P, Lubin JH, Forastiere F (2007) Methodological issues J Occup Health Psychol 1:27–41. regarding confounding and exposure misclassification in epidemiological studies 3. Johnson JV, Hall EM (1988) Job strain, work place social support, and of occupational exposures. Am J Ind Med 50:199–207. cardiovascular disease: a cross-sectional study of a random sample of the 15. Armstrong BG (1998) Effect of measurement error on epidemiological studies of Swedish working population. Am J Public Health 78: 1336–42. environmental and occupational exposures. Occup Environ Med 55:651–6. 4. Karasek RA (1979) Job demands, job decision latitude, and mental strain: 16. Tielemans E, Heederik D, Burdorf A, Vermeulen R, Veulemans H, et al. (1999) Implications for job redesign. Adm Sci Q 24:285–308. Assessment of occupational exposures in a general population: comparison of 5. Die `ne E, Fouquet A, Esquirol Y (2012) Cardiovascular diseases and psychosocial different methods. Occup Environ Med 56:145–151. factors at work. Arch Cardiovasc Dis 105:33–9. 17. Siemiatycki J, Dewar R, Richardson L (1989) Costs and statistical power 6. Kivima ¨ ki M, Nyberg ST, Batty GD, Fransson EI, Heikkila ¨ K, et al. (2012) Job associated with five methods of collecting occupation exposure information for strain as a risk factor for coronary heart disease: a collaborative meta-analysis of population-based case-control studies. Am J Epidemiol 130:1236–46. individual participant data. Lancet 380(9852):1491–7. 18. Bouyer J, Dardenne J, He ´mon D (1995) Performance of odds ratios obtained 7. Stansfeld S, Candy B (2006) Psychosocial work environment and mental health– with a job-exposure matrix and individual exposure assessment with special a meta-analytic review. Scand J Work Environ Health 32:443–62. reference to misclassification errors. Scand J Work Environ Health 21:265–71. 8. Cosgrove MP, Sargeant LA, Caleyachetty R, Griffin SJ (2012) Work-related 19. Flegal KM, Brownie C, Haas JD (1986) The effects of exposure misclassification stress and Type 2 diabetes: systematic review and meta-analysis. Occup Med on estimates of relative risk. Am J Epidemiol 123:736–51. (Lond) 62:167–73. 20. Schwartz JE, Pieper CF, Karasek RA (1988) A procedure for linking 9. Lang J, Ochsmann E, Kraus T, Lang JW (2012) Psychosocial work stressors as psychosocial job characteristics data to health surveys. Am J Public Health antecedents of musculoskeletal problems: a systematic review and meta-analysis 78:904–9. of stability-adjusted longitudinal studies. Soc Sci Med 75:1163–74. 21. Johnson JV, Stewart WF (1993) Measuring work organization exposure over the 10. Tabanelli MC, Depolo M, Cooke RM, Sarchielli G, Bonfiglioli R, et al. (2008) life course with a job-exposure matrix. Scand J Work Environ Health 19:21–8. Available instruments for measurement of psychosocial factors in the work 22. Mariani M (1999) Replace with a database: O*NET replaces the Dictionary of environment. Int Arch Occup Environ Health 82:1–12. Occupational Titles. Occup Outlook Quarterly 43:3–9. 11. Landsbergis P, Theorell T, Schwartz J, Greiner BA, Krause N (2000) 23. Kauppinen T, Toikkanen J, Pukkala E (1998) From cross-tabulations to Measurement of psychosocial workplace exposure variables. Occup Med multipurpose exposure information systems: a new job-exposure matrix. 15:163–88. Am J Ind Med 33:409–17. 12. Macleod J, Davey Smith G (2003) Psychosocial factors and public health: a 24. Niedhammer I, Chastang JF, Levy D, David S, Degioanni S, et al. (2008) Study suitable case for treatment? J Epidemiol Community Health 57:565–70. of the validity of a job-exposure matrix for psychosocial work factors: results PLOS ONE | www.plosone.org 8 September 2014 | Volume 9 | Issue 9 | e108987 Validity of Psychosocial Factors Job Exposure Matrix from the national French SUMER survey. Int Arch Occup Environ Health pulmonary disease with job exposure matrices (ECOJEM). Occup Environ Med 57:126–32. 82:87–97. 34. Solovieva S, Pehkonen I, Kausto J, Miranda H, Shiri R, et al. (2012) 25. Wieclaw J, Agerbo E, Mortensen PB, Burr H, Tuchsen F, et al. (2008) Development and validation of a job exposure matrix for physical risk factors in Psychosocial working conditions and the risk of depression and anxiety disorders low back pain. PLoS One 7(11):e48680. in the Danish workforce. BMC Public Health 8:280. 35. Bonde JP (2008) Psychosocial factors at work and risk of depression: a systematic 26. Rijs KJ, van der Pas S, Geuskens GA, Cozijnsen R, Koppes LL, et al. (2013) review of the epidemiological evidence. Occup Environ Medicine 65:438–45. Development and Validation of a Physical and Psychosocial Job-Exposure 36. Netterstrøm B, Conrad N, Bech P, Fink P, Olsen O, et al. (2008) The relation Matrix in Older and Retired Workers. Ann Occup Hyg 2013 Dec 11. [Epub between work-related psychosocial factors and the development of depression. ahead of print], doi:10.1093/annhyg/met052. Epidemiologic reviews 30:118–32. 27. Theorell T, Tsutsumi A, Hallquist J, Reuterwall C, Hogstedt C, et al. (1998) 37. Niedhammer I, Sultan-Taı ¨eb H, Chastang JF, Vermeylen G, Parent-Thirion A Decision latitude, job strain, and myocardial infarction: a study of working men (2013). Fractions of cardiovascular diseases and mental disorders attributable to in Stockholm. The SHEEP Study Group. Stockholm Heart epidemiology psychosocial work factors in 31 countries in Europe. Int Arch Occup Environ Program. Am J Public Health 88:382–8. Health 2013 Apr 27. [Epub ahead of print], doi 10.1007/s00420-013-0879-4 28. Cohidon C, Santin G, Chastang JF, Imbernon E, Niedhammer I (2012) 38. Hartvigsen J, Lings S, Leboeuf-Yde C, Bakketeig L (2004). Psychosocial factors Psychosocial exposures at work and mental health: potential utility of a job- at work in relation to low back pain and consequences of low back pain; a exposure matrix. Occup Environ Med 54:184–91. systematic, critical review of prospective cohort studies. Occup Environ Med 29. Aromaa A, Koskinen S (2004) Population and methods. Health and functional 61:e2. capacity in Finland. In: Aromaa A. and Koskinen S, editors. Baseline result of 39. Macfarlane GJ, Pallewatte N, Paudyal P, Blyth FM, Coggon D, et al. (2009) the Health 2000 Helath Examination Survey. Helsinki: The National Public Evaluation of work-related psychosocial factors and regional musculoskeletal Health Institute pp. 11–23. pain: results from a EULAR Task Force. Ann Rheum Dis 68:885–91. 30. Laiho J, Djerf K, Lehtonen R (2006) Sampling design. In: Heistaro S, editor. 40. Bouyer J, He ´mon D (1993) Studying the performance of a job exposure matrix. Methodology report Health 2000 Survey. Helsinki: the National Public Health Int J Epidemiol 22 (Suppl 2):S65–71. Institute B26: 13–5. 41. Joseph L, Gyorkos TW, Coupal L (1995) Bayesian estimation of disease 31. Perkio ¨-Ma ¨ kela ¨ M, Hirvonen M, Elo A-L, Kauppinen K, Kauppinen T, et al. prevalence and the parameters of diagnostic tests in the absence of a gold (2010) Tyo ¨ ja terveys -haastattelututkimus. 2009; Helsinki: Tyo ¨ terveyslaitos (in standard. Am J Epidemiol 141:263–72. Finnish). 42. Abramson JH (2011) WINPEPI updated: computer programs for epidemiolo- 32. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, et al. (1998) The gists, and their teaching potential. Epidemiol Perspect Innov 8:1. Job Content Questionnaire (JCQ): an instrument for internationally comparative 43. Cifuentes M, Boyer J, Gore R, d’Errico A, Tessler J, et al. (2007) Inter-method assessments of psychosocial job characteristics. J Occup Health Psychol 3:322– agreement between O*NET and survey measures of psychosocial exposure 55. among healthcare industry employees. Am J Ind Med 50:545–53. 33. Le Moual N, Bakke P, Orlowski E, Heederik D, Kromhout H, et al. (2000) 44. Ostry AS, Marion SA, Demers PA, Hershler R, Kelly S, et al. (2001) Measuring Performance of population specific job exposure matrices (JEMs): European psychosocial job strain with the job content questionnaire using experienced job collaborative analyses on occupational risk factors for chronic obstructive evaluators. Am J Ind Med 39:397–401. PLOS ONE | www.plosone.org 9 September 2014 | Volume 9 | Issue 9 | e108987

Journal

PLoS ONE – Unpaywall

Published: Sep 30, 2014

There are no references for this article.