This study analyzed secondary data from the Adelphi Real World SpA Disease Specific Programme (DSP)™. Data were collected in France, Germany, Italy, Spain, and the UK between March–November 2021 and June 2023–June 2024. Data were also collected in the US from June 2023 to June 2024. The DSP methodology has been previously described [21, 22], validated [23], and demonstrated to be representative and consistent over time.

A geographically representative sample of physicians (rheumatologists and internal medicine physicians in France) were recruited to participate in the DSP by local fieldwork agents, following completion of a short screening questionnaire. Physicians were eligible to participate in this survey if they were personally responsible for the management of patients with axial spondyloarthritis, saw at least five patients with r-axSpA (or AS) and five patients with nr-axSpA a month, and made treatment decisions for at least one patient with r-axSpA and one patient with nr-axSpA a month. Physician participation was financially incentivized, with reimbursement upon survey completion according to fair market research rates.

Patients were eligible for inclusion in the survey if they were ≥ 18 years of age, had a physician-confirmed diagnosis of r-axSpA or nr-axSpA, and visited the physician for consultation. Patients included in this analysis switched from an initial TNFi to UPA (TNFi-UPA), IL-17i (TNFi-IL-17i), or another TNFi (TNFi-TNFi), with the switch occurring at least 3 months prior to the most recent patient visit.

Physicians completed a patient record form for their next four consecutively consulting patients with r-axSpA and their next four consecutively consulting patients with nr-axSpA who visited them for routine care. Given the recent approval of UPA for treatment r-axSpA and nr-axSpA in Europe and the US, an oversample was included to ensure an adequate sample size for analysis; physicians completed one additional patient record form for patients with r-axSpA receiving UPA. In Germany, one additional patient record form was also completed for a patient with nr-axSpA receiving UPA. Patient record forms contained questions on demographic characteristics and treatment history, including patient age, sex, Charlson Comorbidity Index (CCI), time since diagnosis of axSpA, prior NSAID use, initial TNFi received, duration of initial TNFi use, reasons for switching from initial TNFi, and duration of most recent treatment. Clinical characteristics included physician-rated pain (rated 0–10, where higher scores indicate more pain) at initiation of most recent treatment and at the time of the most recent patient visit and physician reported number of affected joints by axSpA, based on physician assessment, at the most recent patient visit. Physicians indicated whether a patient had affected (inflamed or stiff) joints and were then able to select the number of joints affected as between 0 and 68.

Pain endpoints assessed retrospectively at treatment initiation and at the most recent patient visit included the proportion of patients achieving ≥ 50% reduction in physician-reported pain (based on “What is your overall assessment of the pain that this patient has experienced as a result of their r-axSpA/nr-axSpA?”) rated 0–10. In addition, the proportions of patients achieving a minimal clinically important difference (MCID; defined as ≥ 1-point reduction or ≥ 15% reduction), a "much better" improvement (defined as a ≥ 2-point reduction and ≥ 33% reduction) [24], and total pain resolution (defined as pain = 0 at most recent patient visit, where pain was not 0 at initiation) were assessed. Also, the proportions of patients with a physician-reported assessment of affected joints ≤ 1 and 0 at most recent patient visit were analyzed. Patient outcomes were stratified by (1) male vs female and (2) r-axSpA vs nr-axSpA.

Descriptive summary statistics, including mean ± standard deviation (SD) or median and interquartile range (IQR), were calculated for continuous variables. Frequency counts and percentages were calculated for categorical variables. All analyses were conducted in Stata v18 (StataCorp 2023)/UNICOM Intelligence Reporter version 7.5.1 (UNICOM Systems 2021). Missing data were not imputed; therefore, the base of patients for analysis could vary from variable to variable and was reported separately for each analysis.

Patient demographics and clinical characteristics were balanced using inverse-probability-weighted regression adjustment (IPWRA). IPWRA estimators use weighted regression coefficients to compute the predicted probabilities of the outcome, which are converted to percentages (× 100) for binary outcomes. The covariates balanced within the IPWRA were age, sex, CCI, and those with severe disease severity at treatment initiation (severity was determined as mild, moderate, or severe based on physician assessment) for the weighting and regression adjustment stage, and additionally treatment duration was used for the regression adjustment stage. The regression analyses were conducted separately for comparisons between TNFi-UPA and TNFi-TNFi as well as between TNFi-UPA and TNFi-IL-17i. A complete case analysis was conducted to ensure the same sample was assessed across outcomes for each comparison. Outcomes were modeled using IPWRA and reported as predicted percentages along with p-values for each treatment group, while average treatment effects (ATEs) were also calculated. Subgroup outcomes were assessed descriptively.

Using a checkbox, patients provided informed consent to take part in the survey. No identifiable protected health information was extracted during the survey. Data were collected in such a way that patients and physicians could not be identified directly, with data aggregated before being shared with the subscriber and/or for publication.

Data collection was undertaken in line with European Pharmaceutical Marketing Research Association guidelines and as such did not require ethics committee approval [25]. This research was submitted to the Pearl Institutional Review Board (study protocol nos. AG8929 for data from March to November 2021 and AG9361 for data from June 2023 to June 2024) and deemed to be exempt from ethical approval.

This research followed the principles of the Helsinki Declaration of 1964 and subsequent amendments. In addition, each survey was performed in full accordance with relevant legislation at the time of data collection, including the US Health Insurance Portability and Accountability Act 1996 and Health Information Technology for Economic and Clinical Health Act legislation.