Purpose To use a compositional data analysis approach and objective exposure assessments to study the association between the duration of arm elevation and the course of neck and shoulder pain (NSP) during a 2-year follow-up in physically demanding occupations.

Methods Construction (n=59) and healthcare (n=59) employees wore accelerometers on the dominant upper-arm during a full working day at baseline. Objective assessments using accelerometers addresses biases found in previous studies that estimate duration of arm elevation with self-reports. At baseline and every 6 months for two years, participants reported on NSP (scale 0–3). Duration of arm elevation within predefined ranges (<30°; 30–60°; ≥60) formed the parts of the composition. Compositional data analysis is a new statistical analysis method within occupational health and it is the correct way of analysing data with a compositional nature. The associations between the relative importance of the duration within the levels of arm elevation and the course of NSP during the 2-year follow-up were estimated with compositional linear mixed models, adjusted for confounders.

Results In non-adjusted analyses, only duration arm elevation <30° was associated with NSP at baseline (β = 0.37; p=0.015). Duration arm elevation <30° had a tendency to be associated with an improvement in NSP over the 2-year follow-up (<30°*time (β=-0.07; p=0.089)). Neither duration 30–60° nor ≥60 were associated with the course of NSP during follow-up. After adjusting for confounders, none of the durations of arm elevation were associated with the course of NSP over the 2-year period (<30° and NSP (β = 0.20; p=0.126); <30°*time (β=-0.06; p=0.097)).

Conclusion Among construction and healthcare personnel, duration of working in awkward arm elevation postures was not associated with the course of NSP over a 2-year period. Arm elevation alone, without considering force exertion, may not be sufficient to influence the course of NSP.