Confirmation of diagnosis in patient preference studies

For regulatory-grade patient preference studies, confirmation of diagnosis is not optional. Self-report alone is insufficient - but full medical record verification is rarely feasible.

This article explains the spectrum of confirmation of diagnosis methods for patient preference studies, when each is appropriate, and how SurveyEngine manages COD in recruitment.

Ben White, 07.07.2026

Why diagnosis confirmation matters

Online panel respondents who claim to have a specific diagnosis may be unreliable. Studies comparing self-reported panel diagnoses with medical record verification have found error rates of 20–40% for many conditions. In a study designed to inform a regulatory decision about a treatment for a specific patient population, including 20–40% of non-patients invalidates the results.

The required level of confirmation depends on the regulatory context. For FDA submissions using patient preference information in a benefit-risk framework, the FDA guidance expects rigorous patient verification. For academic studies where regulatory submission is not the goal, a well-designed screener may be adequate.

The spectrum of confirmation methods

Self-report with screener questions is the simplest approach. A well-designed screener asks multiple questions about diagnosis, treatment, symptoms, and disease management that would be difficult for a non-patient to answer correctly. While imperfect, screeners can substantially improve sample quality compared to a single diagnosis question.

Patient Advocacy Groups (PAOs and PAG's) membership provides implicit verification - PAG members are typically genuinely affected patients. However, PAGE membership is not identical to confirmed diagnosis; carers and family members may also be members.

Physician attestation requires each recruited patient to provide a letter or code from their treating physician confirming the diagnosis. This is more reliable than self-report but adds time and cost - typically 3–7 days per respondent and significant logistical complexity.

Medical record extraction provides the highest confidence in diagnosis confirmation but is only feasible for studies conducted in clinical settings with appropriate data access agreements.

TLDR Quick links

SurveyEngine respondent sourcing

SurveyEngine health resources

Request a sample estimate

Implementing COD in SurveyEngine recruitment

Step 1: Design a multi-question screener. Include: the primary diagnosis question; questions about specific symptoms, treatments, or monitoring that are disease-specific; questions about when the diagnosis was made and by whom; and questions about current disease management that only a genuine patient would know.

Step 2: Calibrate the screener with clinical expert review. Have a clinician review the screener questions to confirm they would correctly distinguish genuine patients from non-patients. Screener questions that rely on publicly available information (e.g. disease name, common symptoms) are insufficient.

Step 3: Add PAO channel recruitment as a high-quality layer. Supplement panel recruitment with PAO channel recruitment to provide a verified sub-sample. The PAO sub-sample can be used as a gold standard to calibrate the panel screener results.

Step 4: Implement physician attestation for regulatory studies. For studies supporting FDA, EMA, or HTA submissions, implement physician attestation as the primary COD method. SurveyEngine's sampling team manages the attestation workflow.

Step 5: Document your COD approach for regulatory submission. Whatever method you use, document it in your study methods section. Regulators want to know how many respondents were recruited through each channel, what the screener failure rate was, and how the final sample was verified.

Worked example - COD for an FDA patient preference submission

A patient preference study for an FDA 510(k) submission requires 250 patients with moderate-to-severe plaque psoriasis. The COD approach uses a tiered system: all respondents complete a 12-question screener including physician-diagnosis questions, PASI score range, and current treatment type. Screener passers are then offered the choice of physician attestation (for a £15 incentive supplement) or PAO referral verification.

78% of screener passers opt for physician attestation; 22% are verified through PAO referral. The attestation process takes an average of 4.2 days. The final sample of 250 verified patients is accepted by the FDA as meeting the evidence standards for patient preference information.