Although there is substantial evidence regarding patient preferences for treatment administration in other rheumatic diseases, there is a paucity of research on the treatment preferences of patients with AS. Many patients with AS have inadequately controlled disease despite the availability of effective treatments [8], with the introduction of effective treatment options varying in modes and frequencies of administration further highlighting the need to better understand patients’ treatment preferences. The current study, which incorporated multiple methodologies and included a multinational sample of patients who had a confirmed AS diagnosis, found that patients with AS endure a high disease burden with detrimental impacts, including pain, fatigue, and limited mobility. Of importance, experiencing pain and fatigue is associated with poor quality of life and physical functioning among patients with AS [25]. Furthermore, patients feel as though they have little control over their condition, and this perception is driven by the pain and unpredictability of AS. Patients expressed that they must plan their lives around the disease, with daily routines shaped by the need for medication refrigeration and daily administration. Patients with AS also commonly reported moderate to severe work impacts or not working at all due to their disease. Collectively, these findings were consistent with a recent analysis of pooled phase 2 and phase 3 tofacitinib clinical trial data in which greater AS pain and disease activity were associated with greater work productivity loss and impairment of non-work daily activities [3]. As such, to mitigate this burden, the optimal AS treatment for a given patient will need to reduce pain and provide better disease control.

The qualitative results further indicated that patients generally perceived less frequent administration and not requiring refrigeration as advantages. Patients who may prefer a pill over injection include those with active lifestyles who travel frequently or those who do not like needles. In addition, patients who have limited treatment experience may also be more likely to prefer oral over injection. These findings thus support the need to align the choice of AS treatment with the needs, lifestyle, and preferences of the individual patient.

Results from the quantitative survey demonstrated that, when making treatment decisions, patients take different attributes into consideration. Patients regarded ‘treatment taken once a month’, ‘treatment taken orally (by mouth) once a day’, and ‘treatment does not need to be taken at a doctor’s office or clinic’ as the most important attributes when making AS treatment decisions. Overall, there was a preference for pill over self-injection, but when there was a stark difference in dosing frequency between the two modes of administration, patient preferences shifted towards a treatment with less frequent administration (e.g., self-injection once a month preferred over a pill twice a day). This was consistent with findings from the qualitative interviews in which patients perceived less frequent administration as advantageous, as well as prior research showing that patients with endocrine or inflammatory diseases preferred oral over injectable mode of administration [26]. These results were likewise consistent with prior studies that examined treatment preferences in general inflammatory disease areas, including patients with AS, which have also determined that attributes related to the process of treatment administration and mode of administration are important to patients when making treatment decisions [9, 10, 12]. Findings were also in line with a prior patient preference study in which patients with r-axSpA more often preferred an oral pill than SC injection or infusion [11].

Study findings suggested that the importance of AS treatment attributes, as assessed via BWS choice tasks, are generally similar for older and younger patients despite increases in disease activity associated with age [1]. Furthermore, the preferences of male and female patients were mostly consistent. Prior research has shown sex differences in clinical characteristics and patient-reported outcomes [27], as well as differential impact on some aspects of physical function [28]. Hence, the similarity of preferences by sex observed in the current study is notable and suggests that preferences for attributes related to AS treatment administration generalize across sexes, although further research would need to confirm whether the findings generalize to patients who do not identify as cis-gender male or female.

Findings showed that preferences can differ considerably by treatment/disease status. Attributes, such as ‘treatment taken once a month’, ‘treatment does not need to be taken at doctor’s office or clinic’ were important across all three treatment/disease status groups. Nonetheless, differences in RI estimates for the other attributes varied by subgroup, indicating that the perceived value of AS treatment features may differ based on patients’ current AS treatment and the extent to which that treatment controls their disease.

There were several advantages of using a BWS exercise to elicit preferences. BWS can accommodate a larger number of attributes than alternative approaches, such as discrete choice experiments. BWS was an appropriate methodology for this study because we sought to elicit the perceived importance of each attribute relative to other attributes and not trade-offs between desirable and undesirable treatment features. Attributes are converted to the same scale in BWS, which enables evaluation of their overall influence, rather than being evaluated apart from the utility gains/losses associated with other attributes [16, 17]. Moreover, BWS obtains preference/importance scores for multiple items, such as treatment features and/or outcomes; this yields a rank ordering of the importance of selected treatment attributes, which is used to estimate the RI of these attributes.

Treatment preferences, as assessed via fixed-choice tasks, also differed by treatment/disease status in a manner generally consistent with the BWS data, thereby lending further credibility to the observed pattern of results. Specifically, those on advanced treatment with well-controlled disease strongly preferred injectable over any oral options, suggesting these patients were satisfied with AS treatment delivered by self-injection. Patients on advanced treatment who do not have well-controlled disease preferred self-injections over oral, but to a much lesser degree than those with well-controlled disease. In agreement with the qualitative interview findings, patients without advanced treatment experience preferred oral over injectable options.

Hence, findings from the subgroup analyses underscore the need to consider patient-specific factors, particularly disease control and current treatment, to inform AS treatment decision-making that aligns with patients’ preferences. Study results also provide potential directions for supporting shared decision-making for AS treatment and facilitating a more personalized approach to treatment selection for patients with AS. Given the evolving treatment landscape, future studies should identify independent predictors of preferences for the processes related to treatment administration to further understand which patients may be suitable candidates for currently available advanced therapies.

Overall, on average, patients with AS prefer oral options over self-injections, and they prefer less frequent administration. Approximately 50% of patients prefer a once-daily pill to a once-monthly injection. Patients on advanced therapy who have well-controlled disease appear to be satisfied with injectable treatments. However, a sizeable proportion of patients who do not have well-controlled disease, whether on advanced therapy or not, prefer oral options over self-injections. Therefore, offering a daily oral pill option to patients whose disease is not well controlled may not only help to improve adherence among those currently taking an advanced therapy but may also overcome a potential barrier to initiating advanced therapy.

The qualitative study findings are not intended to be generalizable; rather, they are meant to provide an in-depth depiction of patients’ perceptions of AS burden and treatment mode of administration. Nevertheless, efforts were made to recruit patients who represented the population of patients with AS in the US. It is possible that patients who participate in research more broadly are not representative, as they could be more educated or more vocal than those who choose not to participate in research. Also, given the small sample sizes in qualitative research and the means of recruitment used in the present study, there may be aspects of the experience of people with AS that are not well represented in the qualitative study results. The qualitative phase of this research was conducted nearly 3 years ago, although we do not expect that the findings from the qualitative interviews would change if taken in the current time.

Selection bias may limit the ability to generalize results to the extent that the preferences of respondents who agreed to participate in the quantitative survey differ from individuals who chose not to participate. Furthermore, online patient-reported surveys may underrepresent individuals who lack internet access or who are uncomfortable using digital platforms, along with older adults in poorer health, those living in institutional settings, and individuals with severe comorbidities and disabilities. Due to reliance on convenience sampling methods, certain subgroups of patients with AS may be overrepresented or underrepresented, which may reduce the generalizability of the findings to the entire population of patients with AS. Additionally, survey response rates could not be calculated; thus, the potential for non-response bias could not be ascertained. Self-reported data from respondents, including diagnosed comorbidities and treatment history, cannot be independently confirmed for accuracy.

While this study examined differences in preferences based on level of disease activity and treatment experience, observed differences in preferences could also be associated with other disease and socioeconomic characteristics. The BASDAI was used in the current study to assess disease activity. As such, we do not know whether the pattern of results would have differed had we used an alternative assessment to define disease activity, such as the AS Disease Activity Score (ASDAS), although the stark differences among treatment/disease status subgroups suggest that using a different assessment would not impact our conclusions. Furthermore, ASDAS calculation requires C-reactive protein measurement, which is often not available at the time of seeing the patient. As a result, ASDAS is collected less often than BASDAI in routine rheumatology practice, suggesting our study may better reflect real-world disease activity assessment.

Fixed-choice tasks did not present an option for self-injection once a week (e.g., etanercept). Nevertheless, the aim of the fixed-choice tasks was to explore how patients would view more frequently administered oral options relative to less frequently administered self-injection options, rather than to compare all possible administration frequencies for available oral and self-injection AS treatment options. Subgroup comparisons by treatment/disease status should be interpreted with caution, given the small sample of patients on advanced treatment with well-controlled disease. As most of the patients in this study did not have well-controlled disease, results may overrepresent the preferences of this patient subpopulation. While the BWS was designed to closely reflect clinical evidence regarding available AS treatments, it cannot capture all aspects involved in real-world AS treatment decisions. Lastly, while BWS can accommodate a larger number of attributes than other preference elicitation approaches, any unmeasured factors that influence AS treatment decisions may lead to error in the estimation of preferences.