2E-05 - Use of alternative study designs in evaluating comparative risks of infections among patients with moderate-to-severe psoriasis treated with newer vs. older immunomodulatory treatments

Background: The new-user design is generally preferred in observational studies evaluating drug safety. However, for relatively uncommon conditions such as psoriasis, this design can substantially reduce precision by excluding a large number of patients with prior exposure to existing treatments. Despite these limitations, alternative study designs have not been widely explored in this context.

Objectives: To evaluate the risk of infection associated with interleukin-17 inhibitors (IL-17is: ixekizumab, secukinumab) compared to tumor necrosis factor inhibitors (TNFis: adalimumab, certolizumab pegol, etanercept, infliximab) among patients with psoriasis, using three complementary study designs: a new-user cohort (Cohort 1), a prevalent new-user cohort (Cohort 2), and a switcher cohort (Cohort 3).

Methods: We used the MarketScan claims database (2016-2023) to identify adult psoriasis patients (≥1 diagnosis coded by a dermatologist) treated with an IL-17i or a TNFi. Cohort entry date (CED) was the first qualifying dispensing date of a study drug. Cohort 1 included new users with a 1-year washout period for both drug classes. Cohort 2 allowed prior TNFi use in the IL-17i group. Cohort 3 required ≥1 TNFi dispensing one the year before the CED, with TNFi users switching from a different TNFi agent. We excluded patients with other biologics, autoimmune diseases, immune-compromising conditions, chronic/congenital/opportunistic infections any time before the CED. Study outcomes included serious infections requiring hospitalization and oral candidiasis (used as a positive tracer outcome). Follow-up began one day after the CED and continued until the earliest of the following: outcome occurrence, death, disenrollment, 180 days of follow-up, or end of data availability. Baseline characteristics were balanced using 1:1 propensity score matching. Risk ratio (RR)s and their 95% confidence intervals (CIs) were reported.

Results: We identified 6284 TNFi and 3429 IL-17i users in cohort 1, 6281 TNFi and 4295 IL-17i in cohort 2, and 596 TNFi and 1371 IL-17i in cohort 3. There was a secular trend where IL-17i users shifted from being "switchers" to "new users," while TNFi use declined over time. After matching, the RRs (95% CI, in IL-17i vs TNFi) for serious infection were 1.71 (0.68-4.35), 2.11 (0.96-4.66), and 2.00 (0.37-10.87) for cohorts 1 to 3, respectively. RRs for oral candida were 4.60 (1.75, 12.08), 4.80 (1.83, 12.57), and 2.00 (0.37, 10.87) for cohorts 1 to 3, respectively.

Conclusions: The relative risks of serious infection and oral candidiasis were similar in Cohorts 1 and 2, suggesting consistent findings across the new-user and prevalent new-user designs. However, risk estimates in Cohort 3 were difficult to interpret due to the limited sample size. These findings should be interpreted with caution, as residual confounding from unmeasured factors may still influence the results.