OBJECTIVES: Estimate the impact of high-dose opioid use on sick leave among patients with osteoarthritis (OA) undergoing joint replacement, adjusting for endogeneity through propensity scores (PS).

METHODS: This was a non-interventional observational study using the Swedish Patient, Prescription Drug and sick leave registers in patients with hip/knee OA and a hip/knee joint replacement between 2011 and 2014. High-dose opioid use was defined as being dispensed ≥4500 oral morphine equivalents (OMEQ) (binary variable) during the exposure period from one month to one year after surgery. The outcome variable was the number of sick leave days in the year following the end of the exposure period. A PS model was estimated including age and sex at surgery, comorbidities (Elixhauser index and chronic pain related), NSAID and opioid use measured during three-year pre-surgery. Two OLS regression models were estimated, one naïve model regressing sick leave on opioid use, and one model also including the PS.

RESULTS: In total 19,901 patients were identified. The mean age at surgery was 53.6, 47% were male and 10% were dispensed ≥4500 OMEQs during exposure. In the naïve model, opioid use was associated with 7.3 more sick leave days (95% CI: 5.2-9.4). When adding the PS, use of opioids was associated with 6.9 fewer days on sick leave (95% CI: -10 - -3.7).

CONCLUSIONS: Endogeneity when studying the impact of opioids on sick leave is a potential large issue as being in pain will lead to higher sick leave as illustrated by the naïve model. When attempting to control for these endogeneity issues to isolate the impact of opioids alone, the treatment coefficient changed by 14 sick leave days and the treatment sign switched. These results can be interpreted as there exists a group of patients in which opioid treatment is well-managed and increase the ability to work.