OBJECTIVES: Treatment selection is often associated with patient prognosis, leading to confounding when comparing effectiveness between treatments in clinical practice. We used NCO to assess residual confounding when comparing apremilast versus other PsO/PsA treatments.

METHODS: Adults ≥18 years old, with PsO/PsA, initiating apremilast, topicals, methotrexate, tumor necrosis factor inhibitor (TNFi), interleukin (IL)-12/23i, or IL-17i (index date) since 9/23/2016 from the Optum Clinformatics® DataMart database were included. Follow-up ended at treatment switch/discontinuation, NCO, end of enrolment or 9/30/2022. NCOs were assessed for frailty (accidents, decubitus ulcer, fractures); healthy user behavior (wellness visit, herpes zoster vaccine, colon cancer screening, pelvic screening); and treatment channeling for methotrexate (substance abuse, transfusions). NCO relative risks (RR) and 95% confidence intervals (CIs) for apremilast versus other treatments were estimated using inverse probability of treatment and censoring weights.

RESULTS: Treatment cohorts for PsO included 111,070, 5,105, 5,572, 7,779 and 2,359 new users of topicals, methotrexate, TNFi, IL-12/23i and IL-17i, respectively; the apremilast PsO cohort ranged from 6,663–7,264 depending on comparison treatment. RRs of most NCOs were comparable across treatments. Wellness visits and pelvic examinations were less likely among apremilast versus topicals users, RR (95% CI), 0.84 (0.71, 0.97) and 0.82 (0.66, 0.99), respectively. The difference in wellness visits was attenuated in individuals receiving ≥1 topical before index (0.89 [0.76, 1.03]). No significant differences were observed between apremilast versus methotrexate or IL-12/23i. Colon cancer screening was more likely in apremilast versus IL-17i users (1.43 [1.07, 1.79]), suggesting healthy user bias. NCO events for frailty were low and not informative. Results from the PsA cohort showed no residual confounding between apremilast and other treatments.

CONCLUSIONS: Eligibility criteria and weighting reduced residual confounding when comparing real-world effectiveness of apremilast with other treatments. NCOs should be integrated into comparative effectiveness/safety studies of PsO/PsA treatments to detect potential residual confounding and ensure robust results.