A Discrete Choice Experiment of Clinician Preferences Among Attributes of Approved Janus Kinase Inhibitors for Alopecia Areata

Abstract

Introduction

As the treatment landscape for alopecia areata (AA) expands, it is crucial to understand how treatment properties impact clinician decision-making. The objective of the study was to understand how clinicians treating AA make decisions and evaluate the relative importance (RI) of treatment attributes.

Methods

An online discrete choice experiment (DCE) was conducted among clinicians treating AA recruited through M3 Global Research’s US panel from February to March 2024. Treatment selection was evaluated using a series of 10 hypothetical binary choices. Preference weights from the DCE were estimated from conditional logistic regression models and were used to calculate willingness to trade off and attributes’ RI. For preference of prescription drug profile, clinicians chose among three blinded profiles of Janus kinase (JAK) inhibitor treatments approved for use in AA based on data from phase 2b–3 pivotal studies. Drug profile A was based on baricitinib 4-mg data, profile B on ritlecitinib 50-mg data, and profile C on deuruxolitinib 8-mg data. These blinded profiles were ranked as most and least preferred.

Results

Of the 249 screened clinicians, 155 were eligible for analysis (129 dermatologists, 15 physician assistants, and 11 nurse practitioners). Clinicians were predominantly White (75.5%) and male (57.4%), with a mean age of 47.3 years. Approximately 50% of the clinicians had been practicing for over 15 years, and the majority were in office-based practices (82%), seeing a mean of 38.8 patients with AA per month. The RI in decreasing order was scalp hair regrowth (52.7%), initial hair growth (31.5%), and medication frequency (15.8%). For the blinded treatment profiles, drug profile C was most preferred (72.2%) and drug profile B was least preferred (81.9%).

Conclusion

Our study demonstrated that, regarding JAK inhibitor treatment of AA, clinicians preferred options with higher efficacy and shorter time to onset regardless of once-daily or twice-daily dosing.

Key Summary Points

Why carry out this study?

Alopecia areata (AA) is an immune-mediated condition affecting approximately 2% of the population and is associated with significant psychosocial burden and variable treatment response.

The recent approval of Janus kinase inhibitors has transformed the AA treatment landscape, prompting the need to understand clinician decision-making in prioritizing attributes among emerging options.

What was learned from the study?

Clinicians ranked the extent of scalp hair regrowth (efficacy) as the most important attribute, followed by time to initial hair regrowth, whereas dosing frequency was least important; 72.2% of clinicians preferred the profile of deuruxolitinib. Findings highlight the central role of efficacy and speed of response in treatment choice and can inform the development of patient-centered therapies, clinical guidelines, and future comparative effectiveness research in AA.

Introduction

Recent approvals of the first therapies specifically indicated for alopecia areata (AA) [1–3]—a chronic, immune-mediated condition characterized by nonscarring hair loss [4, 5]—represent a significant advancement in the management of this burdensome disease [6, 7]. With a lifetime incidence estimated at approximately 2% [8], AA has historically lacked effective targeted treatments. The introduction of Janus kinase (JAK) inhibitors, including baricitinib, ritlecitinib, and deuruxolitinib, has changed the therapeutic landscape by offering options that directly address the immunopathogenic mechanisms underlying the disease [1–3].

This rapidly evolving treatment paradigm necessitates a deeper understanding of how various treatment attributes influence clinical decision-making. Although discrete choice experiments (DCEs) have been widely used in healthcare to quantify preferences and examine how clinicians and patients make trade-offs among treatment options [9–11], most existing studies in AA have focused primarily on patient preferences [12, 13]. One qualitative study examining physician decision-making in AA highlighted that treatment benefits and safety were key priorities; however, findings were limited by a small and geographically constrained sample [14].

To address this gap, the present study aimed to quantify the relative importance (RI) of multiple prominent treatment attributes from the clinician’s perspective using a DCE. Specifically, the study sought to identify how dermatology clinicians prioritize key treatment characteristics and to evaluate their willingness to make trade-offs when selecting JAK inhibitor therapies for patients with AA.

Methods

This was a cross-sectional, USA-based study employing a DCE administered via Qualtrics online survey platform. The DCE was designed and implemented in accordance with best practices as outlined by the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) guidelines [11].

Compliance with Ethics Guidelines

The study protocol was reviewed and determined to be exempt by the WIRB-Copernicus Group (WCG®) Institutional Review Board, with protocol number 70–1059-001. Informed consent was obtained online from participants prior to study participation.

DCE Design

The DCE methodology is based on the principle that medical treatments can be defined by a finite set of attributes and that preferences are determined by the specific levels within each attribute. In this study, a DCE was used to elicit clinician preferences for hypothetical AA JAK inhibitor treatment options.

Attribute selection was guided by a targeted literature review focusing on the safety and efficacy profiles of the JAK inhibitors baricitinib, ritlecitinib, and deuruxolitinib [15–18]. The development process followed best practices in stated-preference research [11] and was further refined through cognitive pilot interviews with 10 experienced clinicians treating AA. Three key JAK inhibitor treatment attributes were ultimately selected for inclusion in the DCE (Table 1): (1) extent of scalp hair regrowth, (2) time to initial hair regrowth, and (3) frequency of dosing. Each respondent completed 10 choice tasks, in which they were asked to select the treatment they would most likely prescribe between two hypothetical options that shared identical attributes but varied in the levels of those attributes (Fig. 1).

Table 1.

Discrete choice attributes and levels

Attributes	Levels
Percent of patients with scalp hair regrowth of 80% at week 24	23% of patients
	28% of patients
	33% of patients
The number of weeks until a patient will see initial hair growth	8–12 weeks
	16–24 weeks
	24 weeks
How many times per day the patient will need to take the medication	1 pill 1 time a day
	1 pill 2 times a day

Fig. 1.

Example of treatment comparison in the DCE. DCE discrete choice experiment

To isolate trade-offs among these attributes, side effect profiles were held constant across all hypothetical treatment options. Additionally, the adverse event profiles for baricitinib, ritlecitinib, and deuruxolitinib were not well differentiated across the treatments [15–18]. A Bayesian-optimized, statistically efficient orthogonal design was employed to generate the attribute-level combinations for each task. This approach was selected to reduce respondent burden while ensuring the ability to estimate precise and reliable preference weights.

Survey Content

Qualtrics XM was used to administer the survey, which included several sequential components. Initially, clinicians responded to questions about their demographics and clinical practice characteristics. Next, they were asked to rank seven treatment attributes in order of importance for therapeutic decision-making, with lower values indicating greater perceived importance (Table 2). This was followed by the DCE tasks.

Table 2.

Treatment attributes for ranking

Attributes
1. Extent of response/efficacy (e.g., percent of patients with scalp hair regrowth of 80% at week 24)
2. How quickly you will see initial scalp hair regrowth
3. How quickly you will see maximum meaningful scalp hair growth
4. The risk of a mild/minor side effect, such as a cold, runny nose, or headache
5. The risk of a rare but serious side effect, such as a serious infection, blood clot, or cancer
6. Patient’s satisfaction with the amount of scalp hair coverage
7. Frequency of administration (e.g., once daily, twice daily)

In the final section of the survey, clinicians were presented with three blinded treatment profiles designed to approximate the clinical characteristics of baricitinib 4 mg once-daily dosing, ritlecitinib 50 mg once-daily dosing, and deuruxolitinib 8 mg twice-daily dosing based on data from phase 3 clinical trials. Without disclosing the identities of the treatments, clinicians were asked to indicate their most and least preferred options among the three profiles.

Study Population

Eligible participants included board-certified dermatologists, physician assistants, and nurse practitioners specializing in medical dermatology and practicing within the USA. A convenience sample of clinicians was recruited through M3 Global Research, a healthcare research firm that maintains the world’s largest proprietary panel of verified healthcare professionals across more than 70 countries [19–23]. To qualify, participants were required to be licensed and practice in the USA, have a minimum of 5 years of postgraduate clinical experience (excluding training), treat at least five unique patients with AA per month, be able to read and understand English, and provide informed consent. When collecting demographic data for race, the Centers for Medicare & Medicaid Services policy was followed to suppress data for small cell size to protect individual privacy. As such, all participants who selected “Asian” in their response were grouped together in the “Asian” category under race.

Sample Size

There is no universally fixed sample size for a DCE, but several guidelines exist to help determine an appropriate sample [9, 24]. The optimal sample size depends on the study design, including the number of choice tasks, alternatives per task, and the number of attributes and levels. The present study aimed for 150 respondents. Using the Orme formula, $n\ge 500\times c/t\times a$ (n = minimum sample size, c = largest number of levels for any attribute, t = number of choice tasks per respondent, a = number of alternative tasks), the recommended minimum sample size was 75 respondents [25].

Analysis

Descriptive statistics were used to summarize clinician demographic and clinical practice characteristics. Categorical variables were reported as frequencies and percentages, while continuous variables were described using means, medians, and standard deviations (SDs), as appropriate.

Preference weights were estimated using conditional logistic regression models. These weights were then used to calculate the RI of each treatment attribute, reflecting its influence on clinician decision-making. Each attribute’s RI was calculated as the proportion of its individual utility range relative to the total span, expressed as a percentage. This approach provides a standardized metric to compare how strongly each attribute influenced choice behavior, allowing for direct interpretation of which treatment features clinicians valued most when making prescribing decisions.

Quality Control

Quality control measures in the Qualtrics system were enabled to prevent and identify multiple submissions, bots, and fraudulent responses (Q_RecaptchaScore < 0.5; Q_RelevantIDDuplicate = true; Q_RelevantIDDuplicateScore ≥ 75; Q_RelevantIDFraudScore ≥ 30).

Results

Demographics and Characteristics

Of the 249 clinicians who completed the screening survey, 155 met eligibility criteria and were included in the final analysis. The largest group was dermatologists (n = 129), with additional representation from physician assistants (n = 15) and nurse practitioners (n = 11) (Table 3). Over half of the participants were male (57.4%), with a mean (SD) age of 47.3 (10.0) years. Most clinicians identified as White (75.5%), while 18.7% identified as Asian and 9.7% as Hispanic. The majority of respondents practiced in office-based settings (81.9%), and approximately half reported more than 15 years of clinical experience. Clinicians in this sample had substantial experience managing AA, reporting a mean (SD) of 38.8 (35.0) patients with AA seen per month. Of these, a mean (SD) of 18.1 (23.7) patients presented with more than 50% hair loss, and 12.9 (19.7) were receiving JAK inhibitor therapy.

Table 3.

Clinician demographics and characteristics

Characteristic	N = 155
Age, mean (SD)	47.3 (10.0)
Sex, n (%)
Male	89 (57.4)
Female	65 (41.9)
Gender nonconforming	1 (0.6)
Race,a n (%)
Asian	29 (18.7)
Black	4 (2.6)
White	117 (75.5)
Other/prefer not to say	9 (5.8)
Hispanic or Latino, n (%)	15 (9.7)
Clinician type, n (%)
Physician	129 (83.2)
Nurse practitioner	11 (7.1)
Physician associate	15 (9.7)
Practice setting, n (%)
Academic medical center	13 (8.4)
Community medical center	15 (9.7)
Community office-based practice	127 (81.9)
Years in practice, n (%)
5–10	48 (31)
10–15	29 (18.7)
15+	78 (50.3)
Patients treated per month
Mean (SD)	421.6 (201.7)
Median (min, max)	400 (70, 990)
Patients with AA treated per month
Mean (SD)	38.8 (35.0)
Median (min, max)	25 (5, 200)
Patients with AA and > 50% hair loss treated per month
Mean (SD)	18.1 (23.7)
Median (min, max)	10 (0, 145)
Patients with AA treated with JAKi per month
Mean (SD)	12.9 (19.7)
Median (min, max)	5 (0, 135)

AA alopecia areata, JAKi Janus kinase inhibitor, max maximum, min minimum, SD standard deviation

^aClinicians could select multiple responses

Attribute Importance Ranking

When asked to rank seven treatment attributes by their importance in clinical decision-making, clinicians most frequently prioritized efficacy measures—specifically, the percentage of patients achieving scalp hair regrowth and the time to initial hair regrowth—along with rare but serious side effects (Fig. 2). Dosing frequency was ranked as the least important attribute when choosing treatments for AA.

Fig. 2.

Attribute importance ranking. AE adverse event

DCE Findings

The DCE revealed that clinician preferences were most strongly influenced by the percentage of patients achieving 80% scalp hair regrowth at week 24 (RI 52.7%) followed by time to initial hair regrowth (RI 31.5%; Fig. 3). In contrast, dosing frequency exerted the least influence on clinicians’ treatment choices (RI 15.8%), indicating a willingness to prescribe more frequent dosing if it is associated with greater efficacy.

Fig. 3.

DCE relative importance. DCE discrete choice experiment

The average level of utility analysis illustrates how clinicians evaluated specific levels within each treatment attribute (Fig. 4). These normalized utility values were derived from conditional logistic regression models and reflect the relative influence of each level on treatment selection. For the extent of scalp hair regrowth, the level representing 33% of patients achieving 80% scalp hair regrowth at 24 weeks had the highest positive utility (+ 0.597), whereas 23% regrowth was associated with a negative utility (− 0.689), indicating that clinicians strongly preferred profiles with higher efficacy. For the time to initial hair regrowth, clinicians preferred faster responses. The level of 8–12 weeks had a positive utility (+ 0.379), 16–24 weeks had a moderate negative utility (− 0.026), and 24 weeks had the lowest utility (− 0.353). Regarding dosing frequency, once-daily dosing had a small positive utility (+ 0.094), while twice-daily dosing was negatively associated (− 0.094). Although dosing frequency influenced clinician choices to a lesser extent than efficacy or onset, there was a slight preference for once-daily administration. These utility values align with the relative attribute importance findings, confirming that clinicians prioritize treatment efficacy and time to onset over the dosing convenience when evaluating JAK inhibitor therapies for AA.

Fig. 4.

DCE utility analysis. DCE discrete choice experiment

Blinded Treatment Ranking

Clinicians were asked to rank three blinded treatment profiles developed to represent US Food and Drug Administration approved JAK inhibitors using phase 3 data. Drug profile C, corresponding to deuruxolitinib 8 mg (33% scalp hair regrowth at week 24, onset of 8–12 weeks, twice-daily dosing), was most preferred by the majority of clinicians (Fig. 5). In contrast, drug profile B, which represented ritlecitinib 50 mg (23% scalp hair regrowth, 24-week onset, once-daily dosing), was least preferred. Drug profile A (baricitinib 4 mg) was selected as most preferred by a smaller proportion of respondents, reflecting its intermediate efficacy and onset characteristics.

Fig. 5.

Blinded treatment ranking. Profiles: drug profile A (baricitinib 4 mg) = 28% of patients with scalp hair regrowth of 80% at week 24, 16–24 weeks until initial hair growth, 1 pill once a day; drug profile B (ritlecitinib) = 23% of patients with scalp hair regrowth of 80% at week 24, 24 weeks until initial hair growth, 1 pill once a day; drug profile C (deuruxolitinib 8 mg) = 33% of patients with scalp hair regrowth of 80% at week 24, 8–12 weeks until initial hair growth, 1 pill twice a day

Discussion

DCEs are increasingly recognized as essential tools for understanding how healthcare providers make complex treatment decisions. By quantifying how clinicians value specific treatment features and the trade-offs they are willing to make, DCEs can inform evidence-based practice, guide clinical guideline development, and support formulary and policy decisions. In the context of AA, where treatment options are rapidly evolving, this methodology provides critical insight into physician decision-making in real-world scenarios. This study used a DCE to evaluate clinician preferences for attributes of JAK inhibitor treatments in the management of AA. The findings clearly indicate that clinicians prioritize treatment efficacy, both in terms of magnitude and speed of response, over other features such as dosing convenience. These results were consistent across the explicit attribute ranking, DCE analysis, and blinded treatment profile exercise.

Clinicians may favor treatments with both higher efficacy and shorter onset of action for AA due to the absence of validated biomarkers or diagnostic technologies capable of predicting treatment response or the timing of hair regrowth. In clinical practice, a more efficacious agent that produces visible results within a shorter time frame enables earlier assessment of treatment effectiveness, which is particularly valuable in a condition characterized by variable and often unpredictable response patterns. Early signs of hair regrowth can inform treatment decisions, support continued adherence, and reduce the need for prolonged therapeutic trial periods. Our findings align with prior qualitative research demonstrating that treatment benefit, often defined as the ability to achieve visible hair regrowth, is a core driver of clinical decisions [26]. In this study, the extent of scalp hair regrowth was the most important attribute across all evaluated dimensions. Time to initial regrowth was also highly valued, reflecting the clinical urgency often associated with restoring visible hair and improving patient quality of life.

Dosing frequency, by contrast, was the least influential factor. Although some clinicians in prior studies have cited treatment convenience as a consideration [10, 26], our results suggest that the potential for greater efficacy and faster results outweighs any concerns about prescribing medications with more frequent dosing schedules. This finding has implications for both clinical practice and product development, emphasizing the importance of treatment outcomes over convenience.

The specific drug profile choice of clinicians among profiles of approved JAK inhibitors was drug profile C, corresponding to deuruxolitinib 8 mg, based on data from the pivotal trials (33% scalp hair regrowth at week 24, onset of 8–12 weeks, twice-daily dosing) [16, 18]. The preference for drug profile C was consistent with the other main findings of the study; regarding JAK inhibitor treatment of AA, clinicians preferred options with higher efficacy and shorter time to onset regardless of once-daily or twice-daily dosing. In contrast, drug profile B, representing ritlecitinib 50 mg (23% scalp hair regrowth, 24-week onset, once-daily dosing) [17], was least preferred. Drug profile A (baricitinib 4 mg) was perceived as having intermediate efficacy and onset characteristics. Concerns about safety and side effects were controlled for in this study, as all hypothetical treatment profiles were assumed to have similar risk profiles. This design allowed for a focused analysis of how clinicians value efficacy and administration attributes without the confounding influence of safety variability. Nonetheless, it is important to acknowledge that in clinical practice, safety remains a key factor, particularly when treating patients with mild disease or comorbidities. Previous work demonstrates that physicians may be more cautious with systemic therapies when safety concerns are prominent [10, 26]. However, our treatment attribute ranking findings suggest that clinicians place high value on treatments that deliver meaningful clinical outcomes within a reasonable time frame while minimizing safety risks.

Strengths/Limitations

This study has several strengths. It followed rigorous best practices in DCE methodology, including the use of clinical trial data to define attributes and levels. However, limitations must be acknowledged. The clinician sample was drawn from a research panel, which may not be fully representative of the broader dermatology workforce in the USA. Participants may differ from the general population of clinicians in terms of practice patterns, access to treatments, and willingness to engage in research. Additionally, our study only included a limited set of treatment attributes, whereas real-world decisions are influenced by additional factors not captured in this experiment, such as adverse events, insurance coverage, patient preferences, and clinical comorbidities. Finally, this DCE focuses solely on preferences for JAK inhibitor treatment attributes and does not assess clinicians’ broader treatment preferences or their perspectives on the appropriateness, timing, or patient selection for initiating a JAK inhibitor.

Conclusions

This study offers important insights into clinician preferences for JAK inhibitor treatment attributes in the management of AA. Across multiple measures, clinicians demonstrated a clear preference for treatments that offer higher efficacy and faster onset of action, even when associated with more frequent dosing. These findings were consistent across the explicit attribute ranking, DCE task, and blinded treatment profile evaluation in which clinicians chose deuruxolitinib as the most preferred.

Understanding clinician preferences is critical in a rapidly expanding treatment landscape. The use of DCE methodology allows for the quantification of these preferences, supporting more informed clinical guidelines, product development, and shared decision-making frameworks. As AA remains a condition with a significant psychosocial burden, aligning treatment characteristics with both clinician and patient values can promote better adherence, greater satisfaction, and improved health outcomes.

Author Contributions

Kathleen L. Deering, Qing Harshaw, Victoria Barghout, and Thomas J. Ferro all contributed to the design, conduct, and analysis of the study and manuscript preparation and review. James Q. Del Rosso, David Pariser, and Neal Bhatia contributed to the interpretation and manuscript preparation.

Funding

Sponsorship for this study, rapid service fee, and Open Access fee were funded by Sun Pharmaceutical Industries, Inc.

Data Availability

The datasets generated and/or analyzed during the study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

James Q. Del Rosso has served as a clinical research investigator, consultant/advisor, and/or speaker for AbbVie, Aclaris Therapeutics, Almirall, Amgen (Celgene), AnaptysBio, Apogee Therapeutics, Arcutis Biotherapeutics, ASLAN Pharmaceuticals, Athenex, Bausch Health (Ortho Dermatologics), Beiersdorf, Biofrontera, Biopharmix, Biorasi, BlueCreek Pharmaceuticals, Botanix Pharmaceuticals, Brickell Biotech, Bristol Myers Squibb, Cara Therapeutics, Cassiopea, Dermata Therapeutics, Dermavant Sciences, Encore, EPI Health, Evommune, Ferndale, Galderma, Genentech, Incyte, Janssen, Jem Health, La Roche-Posay, LEO Pharma, Lilly (Dermira), MC2 Therapeutics, Nektar Therapeutics, Novan, Nutrafol, Pfizer, Ralexar Therapeutics, Regeneron Pharmaceuticals, Sanofi, Sente, Sol-Gel Technologies, Sonoma Pharmaceuticals (Intraderm), Sun Pharma, UCB, Verrica Pharmaceuticals, and VYNE Therapeutics (Foamix/Menlo); he is president of the American Acne & Rosacea Society. David Pariser is an investigator for AbbVie, Apogee Therapeutics, LEO Pharma, Lilly, MoonLake Immunotherapeutics, Novartis, Pfizer, Priovant Therapeutics, Regeneron Pharmaceuticals, and Veradermics, and he is on a data safety monitoring board for Bristol Myers Squibb. Neal Bhatia is an advisor, consultant, and investigator for AbbVie, Advanced Derm Solutions, Almirall, Arcutis Biotherapeutics, Beiersdorf, Biofrontera, Boehringer Ingelheim, Botanix Pharmaceuticals, Bristol Myers Squibb, Galderma, Johnson & Johnson, Journey, La Roche-Posay, LEO Pharma, Lilly, Novartis, Ortho Dermatologics, Pfizer, Regeneron Pharmaceuticals, Sagimet Biosciences, Sanofi, Skinfix, Soligenix, Sun Pharma, and Verrica Pharmaceuticals. Kathleen L. Deering is an employee of EPI-Q and received payment from Sun Pharma associated with the development and execution of this study. Qing Harshaw was an employee of EPI-Q at the time of the study and received payment from Sun Pharma associated with the development and execution of this study. Victoria Barghout is CEO of Viver Health LLC, which receives funding from Bristol Myers Squibb, Novartis, Sun Pharma, and Taiho Pharmaceutical, and received payment from Sun Pharma associated with the development and execution of this study. Thomas J. Ferro is a former employee of Sun Pharmaceutical Industries, Inc. and a current employment of Data Loft, Inc.

Ethical Approval

This study protocol was reviewed and determined to be exempt by WIRB-Copernicus Group (WCG®) Institutional Review Board (protocol number 70-1059-001). Informed consent was obtained online from participants prior to study participation.