Introduction: While osteoarthritis and/or musculoskeletal pain are known to interfere with ability to work and impair productivity, clinical studies may neglect to assess these outcomes. The aim of this study was to develop an imputation method for such studies, which quantified losses in work function and work productivity.

Materials and methods: Four research databases, representing a clinical severity spectrum, were selected to develop the method. Each database contained at least one health measure commonly used in clinical studies, (the SF-12 Physical Component and Mental Component Scores; the Western Ontario and McMaster Universities Osteoarthritis Index; and the Treatment Options in Pain Scale), and the Work Limitations Questionnaire (WLQ), a work productivity measure. First, each sample was characterized, as well as the combined sample, according to its age, gender, health status and WLQ scores. Next, for each database and the aggregate database, the relationship of any shared health status variable to each WLQ score was estimated using multivariate linear regression adjusting for the sample's age, gender and race. Beta coefficients corresponding to a 1-unit change in a specific health variable were obtained for the individual datasets. Beta estimates for the aggregate sample were obtained from the weighted average of the betas for a health variable from all the individual sample models.

Results: The coefficients from models containing the same dependent and independent variables yielded relatively consistent results as shown by the similarity in direction and magnitude of the coefficients. Based on results, imputation rules were written that help users identify the most appropriate coefficients and compute the imputed scores.

Conclusion: In the absence of a direct measurement of WLQ, this approach can offer a practical alternative in estimating WLQ scores when other health status measurements are available.