Predictive validity of presenteeism measures with dual answer keys in inflammatory arthritis

Background: Employment studies in arthritis have emphasized the importance of decreased productivity at work, or presenteeism. However, consensus is still lacking on how to best measure this construct. Instruments used include the Work Limitations Questionnaire (WLQ), which measures the amount of time people are limited, and the Workplace Activity Limitations Scale (WALS), which measures the degree of difficulty. We modified the response keys of the WLQ and WALS, creating a dual answer key assessing both degree of difficulty and amount of time with difficulty. Previous work by our group has shown that scores obtained from combining responses to the dual answer keys exhibit good internal consistency, criterion and construct validity. Objectives: The objective of this study was to evaluate the ability of the combined scores to predict future work cessation and number of work days missed due to arthritis. Methods: A longitudinal study used baseline and 6-month data from the RCT of an employment intervention, the 'Making It Work' Program. Inclusion criteria were: inflammatory arthritis, age 19-59, working at baseline, and concerns about arthritis affecting work. 327 participants were included (RA:173, PsA:48, SLE:42, AS:64). Combined scores were obtained by i) multiplying, and ii) adding, the scores of difficulty and time answer keys at the item level. We assessed the predictive ability of the combined scores, and the original scores, of the WALS and WLQ to predict work cessation due to arthritis within 6 months using binary logistic regression models, or number of work days missed due to arthritis using zeroinflated Poisson models. Results: WLQ and WALS original and combined scores were significant predictors of future work cessation and number of work days missed. The WLQ combined scores predicted work cessation better than the original score (AIC:117.6 & 118.2 for multiplicative and additive combinations, respectively vs. 125.4 for time) and work days missed (AIC:2179.8 & 2177.8 vs. 2188.7). WALS original score (ie. degree of difficulty) predicted work cessation better than combined scores (AIC:95.8 vs. 109.3 & 109.4). However, WALS combined scores predicted number of work days missed better (AIC:1696.9 & 1673.4 vs. 1795.2). Conclusions: Combined scores from dual answer keys applied to the WLQ and WALS significantly predicted future work cessation and number of work days missed. Using combined scores, rather than original answer keys, improved the ability of the WLQ to predict both work cessation and work days missed, but only improved the ability of the WALS to predict work days missed.