Abstract

Low back pain (LBP) is a common problem for drivers but the analysis of associations of occupational exposures with LBP is hampered by the absence of a standardized clinical examination. Therefore, LBP is commonly assessed by a questionnaire in which useful information is scattered over many items. Many one-item-at-a-time analyses may not adequately evaluate these associations. We applied item response theory-like model (IRT) with one latent variable to examine the association of occupational light vehicle driving with LBP in 406 workers in the mail and parcel delivery sector using a 14-items questionnaire on LBP in the last 12 months investigating the cumulative duration of LBP, pain intensity, level and radiation in the lower limbs, medical consultation, pain medication use and sick leaves. We normalized the latent variable by constraining its variance to one. The link between the latent variable and the LBP indicators was logistic for binary variables, ordered logistic for ordered categories and linear for quantitative variables. Ordinal variables and quantitative variables were ordered in ascending order of severity. The relationship between LBP and occupational exposure to driving was estimated by multiple linear regression models with the latent variable as dependent variable, adjusting for individual factors, other occupational physical constraints and psychosocial factors, separately in men and women. All the factor loadings between the latent variable and LBP indicators were positive (p<0.001) indicating that the latent scale is consistent with health impairment on the low back. In men, driving time (p=0.007) and high perceived driving-related physical constraints (p=0,024) were associated to LBP whereas in women no driving-related constraints were associated to LBP. In the absence of a standardized clinical tool, studies about association between occupational exposures and LBP could benefit from methods that are able to retrieve diffuse information broken out into ad hoc questionnaire.