To the Editor.—We find numerous severe faults with the article by Ms van Poppel and colleagues1 that sharply limit its contribution to the scientific understanding of the efficacy of lumbar supports. Complaint of back pain was not separated from the critical questions of whether the back was injured while the subject was at work and how to prevent such injuries. As fewer than half the persons assigned to wear the support actually wore it, the findings are necessarily suspect. Both compliance and days lost due to back pain were self-reported rather than being assessed objectively, whether back pain occurred while wearing or not wearing a support was not shown, and whether supports when worn were worn properly was not indicated. Statistical power was compromised by the large and unequal numbers of participants who lost interest or withdrew and the likelihood that this occurred nonrandomly. Moreover, it seems that the initial randomization of work groups did not remove nonrandom features among persons: for instance, median sick leave in the past year reported by the lumbar support intervention group was 27% higher than that reported by the next highest group. Thus, from this evidence, there is no way to know whether the effect (or lack of effect) can be attributed to correct full-time use of a lumbar support.
Kraus JF, McArthur DL. Preventing Low Back Pain in Industry. JAMA. 1998;280(23):1993-1994. doi:10-1001/pubs.JAMA-ISSN-0098-7484-280-23-jbk1216