Developing explicit risk models for predicting low-back disability: A statistical perspective

Despite extensive research efforts, disability due to work-related low- back disorders continues to pose a significant economic burden to industry, and significant pain and economic loss to workers. Numerous epidemiological studies have been conducted, but many of these studies have resulted in qualitative guidelines rather than explicit quantitative guidelines. Furthermore, there is considerable disagreement and controversy concerning the validity of manual materials handling criteria based on biomechanical, physiological, and psychophysical theory and research. There is a pressing need for epidemiological field evaluation of such criteria, as well as epidemiologically-based models that provide estimates of low-back disability risk given a set of task, workplace and worker characteristics. A critical tool in such investigations is the statistical method chosen to model risk. The statistical appropriateness of previously used methods are reviewed and critiqued. A fairly comprehensive discussion of the statistical models available for modeling disability risk is then presented, with a focus on models that provide explicit estimates of risk. Recent advances in computer speed have significantly advanced the power and flexibility of statistical modeling techniques. These techniques have the potential to provide considerable insight into the etiology of low back disability and to provide explicit quantitative design criteria for the ergonomics practitioner. Relevance to industry. Statistical methods available for modeling the risk of low-back disability are discussed and previous methodologies used are critiqued. The methods discussed will aid researchers conducting epidemiological studies of low-back disorders in industry, and the discussion of these methods will aid practitioners faced with interpreting results of previous studies.