The Inequality Paradox: Workplace Safety Intervention Decreases Overall Worker Injuries, but Widens Socioeconomic Disparities

           Background and purpose: The “inequality paradox” posits that interventions, even if they improve overall population health, may increase disparities by improving outcomes primarily in more-advantaged groups. This phenomenon is rarely tested empirically.

            An evaluation of a comprehensive, hospital-wide intervention that taught hospital patient care workers how to safely lift patients in 2013 found that, compared with workers at a control hospital, those at the hospital who received the intervention had improved perceptions of workplace norms around safe patient handling and reduced workplace injuries. The present study tests whether the intervention improved safe patient handling behavior and reduced injury for everyone, or whether changes occurred primarily in higher-wage workers (nurses), who as a group have lower overall injury rates than low-wage workers (patient care associates).

            Methods: For these analyses, we used data from the Boston Hospital Workers Health Study, a cohort study of approximately 8,500 nurses and nursing assistants at two large Boston-area hospitals. We also had longitudinal survey data from 2012 and 2014 on a subset (n=1,595) of workers across the two hospitals. First, using the survey data and comparing high-wage and low-wage workers, we examined changes in perceptions of workplace health and safety, and in self-reported pain, before and after the intervention. We also had employer-provided injury data and payroll data for the entire cohort, which we used to examine changes in injury rates in the two hospitals, again comparing high-wage and low-wage workers.

            Results:  In the intervention hospital, low-wage workers showed more improvements than high-wage workers in self-reported safe patient handling behaviors and bigger reductions in number of lifts per shift, comparing pre-intervention and post-intervention surveys. While both high-wage and low-wage workers had improvements in self-reported overall pain, only high-wage workers reported reductions in shoulder/neck pain, pain interference with work, and moderate to high pain severity; low-wage workers actually reported (non-significantly) higher levels of the preceding three variables post-intervention. In preliminary analyses, among high-wage workers, administratively-reported injury rates decreased from 14% to 12% pre- to post-intervention, but for low-wage workers, injury rates remained steady at 20% both pre- and post-intervention. Therefore, while the overall injury rate decreased following the intervention, the disparity in injury rates between high- and low-wage workers actually grew wider.

            Conclusion and implications: These results have implications for both intervention development and analysis. Low-wage workers in this study received the same training in safe patient handling as their high-wage counterparts, and they did report that their handling practices improved, but they did not see the concomitant reductions in pain and injury that were seen in high-wage workers. Further analyses may reveal the explanation for this disconnect, and could point to ways that interventions could be delivered in a way that has similar uptake for diverse working populations. From an analytic perspective, this paper demonstrates the necessity of analyzing interventions for disparities in intervention effectiveness, and of collecting data on relevant social and demographic characteristics so that researchers are able to test for such disparities.