WHY HAS THE SEARCH FOR CAUSES OF LOW-BACK-PAIN LARGELY BEEN NONCONCLUSIVE

Study Design. Cross-sectional data were collected in a postal question naire within the framework of a 5-year randomized, controlled, prospec tive, population-based study. Objectives. To investigate to what exten t associations differ or concur when correlates of low back pain are t ested against various subdefinitions of low back pain. Summary of Back ground Data. Numerous factors have been suspected to cause low back pa in, but findings have not been constantly reproduced in epidemiologic studies. Methods. Data were collected on 746 people reporting nonspeci fic low back pain some time during the year preceding the survey. Six correlates of low back pain (age, sex, marital status, attitude to a h ealthy life-style, self-reported physical activity at work, and smokin g) were cross-tabulated against nonspecific low back pain and against four subgroups of low back pain. Results. There was only one statistic ally significant strong association between the potential risk indicat ors and the nonspecific definition of low back pain, but several emerg ed when the low back pain group was split into subgroups. Different su bgroups of low back pain did, indeed, relate differently to the variou s correlates. Conclusions. It is necessary to define some clinically r elevant subgroups of low back pain to accelerate the search for causal mechanisms.