Observational Cohort Design & Patient Selection Prompt
Prompt
You are an outcomes researcher designing an observational cohort study to generate real-world evidence. Given [PASTE: indication, patient population characteristics, treatment patterns, available data sources (EHR, claims, registry), and regulatory/payer questions], design the study: 1. Define cohort inclusion/exclusion criteria (diagnosis, age, disease severity, baseline comorbidities) 2. Specify primary and secondary outcomes (efficacy endpoints, safety events, healthcare utilization) 3. Design matching or adjustment for confounding (propensity scoring, inverse probability weighting) 4. Define analysis timeframe (follow-up duration, treatment duration, washout periods) 5. Specify statistical analysis plan (intention-to-treat, compliance thresholds, sensitivity analyses) Output: RWE cohort protocol (cohort definition | outcome definitions | confounding approach | analytical plan | power calculation | regulatory submission intent).
Why it works
Observational cohorts provide real-world efficacy/safety data complementing RCT evidence.
Watch out for
RWE is subject to unmeasured confounding, selection bias, and informative censoring. Observational outcomes are often incomplete or coded inaccurately in source systems.
Used by
Data Analysts