Conjoint Analysis in Healthcare Research

How Healthcare Uses Conjoint Analysis

Conjoint analysis in healthcare measures how patients, providers, and payers weigh trade-offs between treatment attributes like efficacy, side effects, dosing frequency, and cost. The method is called a discrete choice experiment (DCE) in health economics literature, though the underlying methodology is identical to choice-based conjoint used in commercial research.

Healthcare conjoint studies produce the same outputs as any conjoint: part-worth utilities showing how much each treatment attribute level contributes to preference, relative importance scores ranking which attributes drive decisions, and willingness-to-accept calculations that quantify how much efficacy patients would trade for fewer side effects or more convenient dosing.

What makes healthcare distinct is the stakes. Product decisions in CPG affect shelf placement. Treatment preference data in healthcare shapes clinical trial design, influences formulary decisions, and increasingly appears in regulatory submissions to the FDA and EMA.

Common Applications

Patient Treatment Preference Studies

The most frequent use case. Patients evaluate hypothetical treatment profiles that vary on clinical attributes (efficacy, side effects, dosing schedule) and practical ones (out-of-pocket cost, monitoring requirements, mode of administration). The output reveals which attributes matter most and where patients are willing to accept trade-offs.

A diabetes study, for example, might find that patients would accept a 0.5% reduction in HbA1c improvement to switch from daily pills to weekly injections. That kind of trade-off data directly informs formulation decisions and dosing strategy.

Medical Device Evaluation

Device manufacturers use conjoint to understand how clinicians and patients value features like accuracy, portability, battery life, and ease of use. A hospital purchasing committee might weigh accuracy and integration with EHR systems heavily, while patients prioritize comfort and at-home usability. Conjoint quantifies these differences so manufacturers can design for the right audience.

Formulary and Reimbursement Decisions

Health plans and pharmacy benefit managers use preference data to evaluate which treatments to cover. If conjoint shows that patients strongly prefer a weekly injection over a daily oral medication (all else equal), that preference data supports formulary placement for the injectable formulation. Payers increasingly expect quantitative preference evidence alongside clinical trial data.

Clinical Trial Design

Pharmaceutical companies use conjoint results before Phase III to identify which endpoints matter most to patients. If patients care more about symptom reduction than biomarker improvement, the trial's primary endpoint should reflect that. The FDA's Patient Preference Initiative explicitly recommends conjoint analysis and DCEs as methods for capturing patient input on benefit-risk trade-offs.

Study Design Considerations for Healthcare

Attribute Selection Requires Clinical Input

Attributes must be clinically meaningful and realistic. Including an efficacy level that no treatment in the class could achieve undermines respondent trust and produces artificial data. Work with clinicians to define attribute levels that map to plausible treatment profiles.

Common attribute categories for treatment preference studies:

Health Literacy Matters

Respondents in healthcare studies range from clinicians (who understand clinical endpoints) to patients (who may not). Attribute descriptions need to be accurate without being technical. Instead of "15% reduction in HbA1c from baseline," write "blood sugar control improves moderately." Test comprehension with a small pilot group before full fielding.

Visual aids help significantly. Risk grids (showing "5 out of 100 patients experience this side effect" with icons) produce more reliable data than percentage text alone. Several studies have shown that graphical risk presentations reduce misunderstanding and improve response consistency.

Ethical and Regulatory Considerations

Healthcare conjoint studies involving patients typically require Institutional Review Board (IRB) or Research Ethics Board (REB) approval, even when they're survey research rather than clinical trials. The key triggers are: you're asking patients about health-related decisions, the data might influence treatment availability, and respondents could experience psychological discomfort when evaluating serious health trade-offs.

For studies intended to support regulatory submissions, follow the ISPOR Conjoint Analysis Good Research Practices guidelines. These cover experimental design standards, sample size requirements, and analysis methods that regulators expect to see. The FDA has accepted conjoint-based preference data in several benefit-risk frameworks, particularly for medical devices through the Center for Devices and Radiological Health (CDRH).

Sample Size and Recruitment Challenges

Healthcare populations are harder to recruit than general consumer panels. Patients with specific diagnoses are a limited pool, and clinical confirmation of eligibility adds cost and time. Plan for smaller samples (200-400) and consider ACBC if your sample cap falls below 200, since adaptive conjoint extracts more information per respondent.

Recruitment channels in healthcare include: patient registries, specialty panels with diagnosis verification, physician office intercepts, patient advocacy organizations, and clinical site recruitment. Costs run 3-5x higher per completed respondent compared to general population studies.

Real-World Healthcare Example

A medical device company was developing a newer continuous glucose monitor (CGM) and needed to prioritize features for the launch product. They ran a CBC conjoint with 5 attributes:

• Accuracy (MARD): 8%, 10%, 12%
• Sensor wear time: 7 days, 10 days, 14 days
• Calibration: No calibration, 1x daily, 2x daily
• Smartphone integration: No app, Basic app, App with insulin dosing suggestions
• Monthly cost: $50, $100, $150, $200

With 350 Type 1 diabetes patients, the study found that calibration frequency was the most important attribute (29%), followed by sensor wear time (24%). The "no calibration" level had dramatically higher utility than even once-daily calibration. Smartphone app features mattered less than expected (12% importance), and the dosing suggestion feature had only marginally higher utility than the basic app.

The company made calibration-free the non-negotiable design requirement for launch, prioritized 14-day wear time in engineering trade-offs, and downscoped the app to basic functionality for V1.

Tips for Getting It Right

1. Partner with clinicians from the start. They keep your attribute levels realistic and your terminology accurate. A study that tests clinically impossible treatment profiles wastes respondents' time and produces meaningless data.

2. Pilot with patients, not just colleagues. Internal reviewers overestimate how clear your attribute descriptions are. Run 15-20 patient interviews before finalizing the survey to catch comprehension issues.

3. Plan for IRB/REB timelines. Ethics review adds 4-8 weeks to your project timeline. Start the application as soon as your study design is finalized.

4. Budget 3-5x for recruitment. Healthcare panel respondents cost significantly more than general population. A 300-person healthcare study might cost $30,000-$60,000 in sample alone, compared to $5,000-$15,000 for consumer research.

5. Document everything for regulatory use. If there's any chance your data will appear in a regulatory submission, follow ISPOR guidelines from the beginning. Retrofitting rigor is expensive and sometimes impossible.

Frequently Asked Questions

Is conjoint analysis the same as a discrete choice experiment?

In practice, yes. Healthcare researchers use "discrete choice experiment" (DCE) as the preferred term, and the methodology aligns with choice-based conjoint (CBC). The analysis techniques (conditional logit, mixed logit, hierarchical Bayes) are identical. The terminology difference is largely disciplinary: market researchers say "conjoint," health economists say "DCE."

Does the FDA accept conjoint data?

Yes. The FDA's Center for Devices and Radiological Health (CDRH) has accepted stated-preference data, including conjoint analysis, in benefit-risk assessments for medical devices. The Patient Preference Initiative, a public-private partnership, recommends conjoint and MaxDiff as appropriate methods. Pharmaceutical submissions have also included conjoint data, though it's more common for devices than drugs.

How many respondents do I need for a healthcare conjoint study?

Target 200-400 patients for a standard study. If you need segment-level analysis (e.g., comparing newly diagnosed vs. long-term patients), plan for 200+ per segment. Healthcare samples are expensive, so consider ACBC for studies where you can't reach 200 respondents.

Related Guides

• Conjoint Analysis: Complete Guide -- Full methodology overview
• Conjoint Analysis Examples by Industry -- Examples across CPG, SaaS, telecom, and automotive
• Conjoint Analysis for CPG -- Package design and product optimization
• How to Design a Conjoint Study -- Attribute selection and study setup
• Conjoint Analysis Sample Size Requirements -- Planning for healthcare's recruitment constraints
• Conjoint Analysis Survey Template -- Adaptable for healthcare studies

Build healthcare conjoint studies -- try Quali-Fi free for 14 days.