This study investigated the predictive value of a physical examination, which is routinely applied in occupational medicine, in a sample of young workers with no or only limited antecedents of back disorders. The results can be readily summarized. Firstly, almost one out of eight workers developed low back complaints lasting one week or more in the one year of follow-up. Secondly, overall, tests were reliable, but only few abnormalities were observed. Thirdly, physical items proved not interesting in predicting LBP. Only the report of pain before and during the examination had predictive value.
Of particular interest in this study was the effect of physical variables on the development of LBP. Previous prospective studies have focused on workers with complaints and showed that pain increases during the examination were related to future LBP. To clarify the influence of physical abnormalities in the absence of the so dominant pain characteristics, we selected workers with no or only a limited history of back pain. Workers older than 30 years or with LBP lasting one week or more consecutively in the year before inclusion were excluded. Despite our efforts to select an asymptomatic population, about half of the workers indicated some back complaints, i.e. longer than one day but less than one week consecutively, in the year before intake and four of them reported back complaints at the day of the examination. Therefore, in the analyses, we made a distinction between workers without any history of back problems and workers with some, but a limited history in the year before intake.
Individual physical variables proved not useful at all for screening workers with no or limited back antecedents at risk for reporting LBP in the future. First, there were only a limited number of physical abnormalities. Second, none of the physical variables consistently predicted LBP one year later. This was not only the case for the abnormalities with low baseline prevalence (and consequently lower power to reach significance), but also for the abnormalities with higher baseline prevalence. Of the ranges of mobility, only the difference between lateral flexion to the right and to the left was statistically significant in workers with limited antecedents. However, this statistical difference proved of no practical significance. Other authors have also found side to side differences to be predictive for LBP problems. In workers of a Finnish forest industry enterprise [19
], a side difference ≥20° in the straight-leg-raising angle predicted sick leave more than 14 days. More specifically, the combination of side difference in straight-leg-raising angle ≥20° and pain below the knee and relief of pain when lying and severe trouble at work turned out to be highly predictive to identify a small subgroup (6%) of the study population at extremely high risk for sick leave more than 14 days. However, the study population, i.e
. workers visiting the occupational health service for medical advice for low back disorders, was quite different from the 'asymptomatic' and 'mildly symptomatic' population in our study. Nadler and co-workers [26
] showed that treatment for LBP in female collegiate athletes was predicted by the percentage difference between the right and left hip extensors. For men, this association was not significant.
In workers without any back history, only obesity was found of importance both in univariate and multivariate analyses. However, the epidemiological evidence for this factor remains unclear in literature [8
]. Furthermore, as several tests have been carried out, significant results may be due to multiple testing.
The variables that were most predictive for LBP one year later were the self-reports of pain before the start of the examination and the self-reports of pain provocation during the actual tests. Moreover, the self-report of pain before examination proved more important than any of the reports of pain during examination. Hence, while less time-consuming and intensive than a physical examination, a simple self-report of pain is more informative than a physical examination. Our results are in agreement with those of the Boeing study. In that study, Bigos and co-workers concluded that, once historical information about previous pain (treatment) was known, information gained from physical factors added no significant predictive value [27
]. Other than a history of back problems, the authors identified work perceptions and psychosocial factors to be most predictive of future reporting of back injury [28
The value of self-reports of pain in our population of 'mildly symptomatic' workers however remains limited for the purpose of screening. Of the four workers that reported LBP at baseline, three (75%) developed LBP one year later whereas one (25%) did not. Such a value is not enough for the identification of those at risk, as still 25% of the workers with pain at baseline did not develop LBP one year later. Furthermore, the low prevalence of self-reported pain at baseline (only 4 of the 692 workers reported pain at baseline) raises doubts about the costs-benefits ratio.
Because of the low prevalence of abnormal scores on each of the individual items separately, we have also constructed a combined variable. Workers that scored 'abnormal' in at least one individual item were scored deviant in the combined variable. A 'normal' score for the combined variable was given to the workers who scored 'normal' on all the individual items. By doing so, we increased the number of positive findings and consequently the power to show an effect. However, despite this effort, our physical examination was not significantly related to LBP one year later. This negative result may again be due to lack of power. Power calculations a posteriori indicated that, even for the combined variable, we would have needed 4981 workers with deviant scores and 4544 with overall normal scores in the group of the 'asymptomatic' workers to be able to show a significant difference between the proportions found with a power of 90% and alpha = 0.05 one-sided. For the group of the 'mildly' symptomatic workers, this numbers would have amounted to 523 405 workers with any deviant score and 620 148 workers with overall normal scores. Since so many workers are needed to show any effect, this finding stresses once more that, from a cost-benefit point of view, physical examinations as carried out by our protocol are not useful in screening workers with no or limited antecedents at risk for LBP.
Few prospective studies have investigated the predictive value of physical examinations. In most of these studies, subjects underwent a physical examination following the reporting of back disorders. Our study indeed lends support to a relationship between some pain provocation tests and LBP, but neither a limited range in the straight-leg raising tests, nor pain elicited in straight-leg raising was predictive for LBP. We should however be mindful that we had no workers with signs of root compression/inflammation at baseline either.