Evaluating the effects of Return-to-Work program in Malaysia on post-injury employment outcomes

Background: Social security mechanisms are important policy tools to insure workers and their families against health shocks caused by injury. This study aims to evaluate the impact of the Social Security Return-to-Work (RTW) program in Malaysia, the first national disability management program in developing countries, on post-injury employment outcomes, with a special focus on causal effects of speed of enrollment on return to work outcomes.

Methods: Analysis is conducted on the existing data of retrospective cohorts of participants enrolled in the RTW program between 2010 and 2014. The data was collected by case managers of the RTW program, and includes information on participants’ demographic characteristics, socioeconomic status, injury characteristics, enrollment information, and employment outcomes. The outcome measure is time to return to work since program entry. Information beyond two-year program participation is administratively censored. The key exposure variable is enrollment speed, defined as time between enrollment and injury, and is categorized into 1, 2, 4 and 12 months, and beyond 1 year. Cox proportional hazards models in conjunction with the instrumental variable approach will be constructed to understand the association between enrollment speed and time to return to work. The instrumental variable will be the geographic distance between participant’s home and rehabilitation center where the participant would receive rehabilitation services once enrolled in the program. 

Results: Among the 6,922 RTW program participants enrolled between 2010 and 2014, more than 60% (4,171) suffered from commuting injuries. The median time of enrollment since injury was 3.5 months, and about 79% of, or 5,468, participants returned to work within two years in the program. The RTW program participants were likely to have lower socioeconomic status, being male and Malay, and having obtained high school education or lower. Results of the multivariable Cox proportional hazard model without the instrumental variable shows that participants would have lower hazards of return to work by 19% (CI: 15% to 30% lower), 37% (CI: 26% to 46% lower), and 59% (CI: 51% to 66% lower) if they were enrolled 2 months, 4 months and 12 months after injury as compared to people who enrolled within 30 days of injury. We will compare results of the Cox model only approach to the simultaneous Cox models including the instrumental variable. Sensitivity analyses will be conducted using propensity score matching method that matches participants who enrolled at different speed.

Conclusions: Social security mechanisms are designed to protect individuals in times of distress, and the results of this study would provide one of the first empirical evidence on the effectiveness of social security in enhancing the return of injured workers to economic productivity. Findings will also inform policy makers in Malaysia and other similar contexts to improve the program design.