Objectives
The positioning of secukinumab in the treatment of axial SpA (axSpA) and PsA is debated, partly due to a limited understanding of the comparative safety of the available treatments. We aimed to assess the risk of the key safety outcome infections during treatment with secukinumab and TNF inhibitors (TNFi).

Methods
Patients with SpA and PsA starting secukinumab or TNFi year 2015 through 2018 were identified in four Nordic rheumatology registers. The first hospitalized infection during the first year of treatment was identified through linkage to national registers. Incidence rates (IRs) with 95% CIs per 100 patient-years were calculated. Adjusted hazard ratios were estimated through Cox regression, with secukinumab as the reference. Several sensitivity analyses were performed to investigate confounding by indication.

Results
Among 7708 patients with SpA and 5760 patients with PsA, we identified 16 229 treatment courses of TNFi (53% bionaïve) and 1948 with secukinumab (11% bionaïve). For secukinumab, the first-year risk of hospitalized infection was 3.5% (IR 5.0; 3.9–6.3), compared with 1.7% (IR 2.3; 1.7–3.0) during 3201 courses with adalimumab, with the IRs for other TNFi lying in between these values. The adjusted HR for adalimumab, compared with secukinumab, was 0.58 (0.39–0.85). In sensitivity analyses, the difference from secukinumab was somewhat attenuated and in some analyses no longer statistically significant.

Conclusion
When used according to clinical practice in the Nordic countries, the observed first-year absolute risk of hospitalized infection was doubled for secukinumab compared with adalimumab. This excess risk seemed largely explained by confounding by indication.