Abstract
Musculoskeletal pain and headache represent substantial public health burdens. Often, no biomedical pathology can be identified. Research has shown that conditions at work may influence such complaints. Biomechanical work factors have traditionally been explored as risk factors. More recently, psychological and social factors have also gained recognition as contributors to somatic health. However, solid documentation exists only for a few factors. Moreover, these factors are often broadly defined and may conflate dimensions that are differentially related to different health complaints. Furthermore, most prospective studies have assessed work exposures at one single time point despite theoretical focus on ”chronic” exposure. The present thesis aimed to elucidate a wide range of specific work factors to map predictors of neck pain intensity, back pain severity, and headache severity two years later. Exposure information was derived from several time points to account for possible temporal variations in working conditions.
A diverse sample of Norwegian employees was recruited by an ongoing project. Sixteen exposures, including two mechanical factors, were studied. Studies I- III included cross-sectional regressions comprising all subjects invited at each time point (i.e. also subjects that left or entered the companies during the follow-up period) and prospective regression analyses comprising subjects that were invited at both time points (employees that remained employed by their respective companies during the follow-up). Hence, somewhat different samples were analyzed within the studies, resulting in the following sample sizes: Study I: Cross-sectional sample at T1 (n=4569) and T2 (n=4122), and prospective sample (n=2419). Study II: Cross-sectional T1 (n=5212), T2 (n=4722), and prospective (n=2808). Study III: Cross-sectional T1 (n=6421), T2 (n=5930), and prospective (n=3574).
Prospective analyses studied baseline exposure and average exposure across time (i.e. ([T1+T2]/2). Outcomes were baseline-adjusted health complaints after two years later. In studies I and II different exposure development from time 1 to time 2 was also categorized, and in study III cross-lagged and synchronous structural equation models were estimated to compare the tenability of different causal assumptions.
Most psychological and social exposures were associated with all health complaints either cross-sectionally, prospectively, or both. The most robust and consistent predictors of neck pain intensity (study I) were role conflict and working with arms raised, empowering leadership, and decision control. The most consistent predictors of back pain severity (study II) were protective factors; decision control, empowering leadership, and fair leadership. The most consistent predictors of headache severity (study III) were quantitative demands, role conflict, decision control, control over work intensity, and job satisfaction. The causative role of these factors was partially supported by cross-lagged models and fully supported by synchronous models.
For study IV measurements from 1250 employees were collected three times during a four year period. Five exposures were analyzed; Role conflict, decision control, empowering leadership, social climate, and quantitative demands. Group-based trajectory models (GBTM) identified clusters of exposure levels over the three time points. The risk of new-onset neck pain after four years was markedly different across groups experiencing different levels of exposure throughout the period. For subjects reporting pain at baseline, the risk of persistence was influenced by role conflict, quantitative demands, and low decision control.
In conclusion, some relatively novel factors were identified as predictors of pain complaints of the neck, back, and head. Most notably, factors such as role conflict, empowering leadership and decision control appeared more strongly related to pain complaints than factors that have more often been highlighted in the past (e.g. job demands, support, or physical workload). Furthermore, strong indications were observed that the modeling of exposure in cohort studies should not be based on arbitrary categorizations or measurements derived from one time point only. The current works should have considerable practical implications as the identified factors were specific and should be more amenable to organizational improvement efforts than general attempts to reduce ”stress” or ”demands”.