Real-world evidence (RWE) analytics and post-market surveillance
The systematic collection, integration, and analysis of data from outside traditional clinical trials (e.g., electronic health records, claims data, patient registries) to evaluate drug/device safety, effectiveness, and value in real-world clinical practice.
It is highly valued because it reduces the time and cost of evidence generation compared to RCTs, directly informing regulatory decisions, payer reimbursement, and lifecycle management, thereby accelerating market access and optimizing commercial strategy. The impact is a demonstrable link to improved patient outcomes and sustained market position.
1Careers
1Categories
9.0 Avg Demand
15%
Avg AI Risk
How to Learn Real-world evidence (RWE) analytics and post-market surveillance
1. Grasp core data sources: Understand the structure, strengths, and limitations of EHR, claims, registry, and patient-generated data. 2. Learn fundamental study designs: Focus on cohort studies, case-control studies, and the concept of causal inference vs. association. 3. Study regulatory frameworks: Familiarize yourself with key guidelines (e.g., FDA RWE Framework, EMA guidance) and the concept of data quality (e.g., the OHDSI Data Quality Dashboard).
1. Master data wrangling: Practice harmonizing disparate data sources using common data models (e.g., OMOP CDM, PCORnet CDM) and handling missingness. 2. Apply advanced analytics: Implement propensity score methods (matching, weighting), survival analysis, and basic predictive modeling for RWE questions. Avoid the common mistake of confusing correlation with causation without proper confounding control.
1. Architect RWE programs: Design multi-study evidence generation plans that align with regulatory submission pathways (e.g., FDA Real-World Evidence Program) and payer evidence requirements. 2. Lead cross-functional evidence teams: Integrate RWE insights with clinical, commercial, and HEOR strategies to shape product labeling and market access. 3. Develop and mentor: Establish organizational data governance and analytical best practices.
Practice Projects
Beginner
Project
Conduct a Descriptive Epidemiology Study Using Public Claims Data
Scenario
Using the CMS Limited Data Set or a similar public claims database, answer a basic question like: 'What is the incidence rate of a specific adverse event among new users of a given drug in the US Medicare population?'
How to Execute
1. Acquire access to a relevant public or simulated dataset. 2. Define clear inclusion/exclusion criteria based on claims codes (ICD, CPT, NDC). 3. Use SQL or Python/R to create an analytic cohort and calculate crude incidence rates. 4. Write a structured summary of methods and results.
Intermediate
Project
Perform a Comparative Effectiveness Analysis Using a Common Data Model
Scenario
Using an OMOP CDM database (e.g., from OHDSI), compare the risk of hospitalization for heart failure between two second-line diabetes drugs in a real-world cohort.
How to Execute
1. Design a retrospective new-user cohort study within the OMOP framework. 2. Implement and assess covariate balance using propensity score stratification/weighting. 3. Conduct a Cox proportional hazards analysis to estimate hazard ratios. 4. Perform sensitivity analyses (e.g., negative control outcomes) to test robustness.
Advanced
Case Study/Exercise
Develop a Post-Market Surveillance (PMS) Strategy for a Novel Biologic
Scenario
A biologic is approved for a rare autoimmune condition. The company must design a global PMS plan to satisfy EMA post-authorization safety studies (PASS) and gather effectiveness data for value-based contracting.
How to Execute
1. Define key safety and effectiveness questions in collaboration with medical affairs and regulatory. 2. Map required evidence to potential data sources (e.g., mandatory registry, large EHR networks, targeted chart reviews). 3. Design a hybrid study protocol leveraging both prospective and retrospective data. 4. Draft a cost-benefit analysis and risk-management plan for stakeholder review.
Tools & Frameworks
Data Platforms & Common Data Models
OMOP Common Data Model (OHDSI)PCORnet Common Data ModelIBM MarketScanIQVIA OpenClaims/EMR
Apply OMOP or PCORnet when conducting network studies across institutions to ensure standardization. Use IBM or IQVIA for large-scale, longitudinal US claims or EHR data for regulatory-grade analyses.
Use R/Python for advanced propensity score methods, survival analysis, and machine learning. SQL is non-negotiable for data extraction and manipulation from relational databases. SAS is standard for regulatory submission-ready analyses.
Regulatory & Methodological Frameworks
FDA RWE Framework & GuidanceEMA Guideline on Good Pharmacovigilance Practices (GVP)ISPOR RWE Task Force ReportsSTROBE & RECORD-PE Reporting Guidelines
The FDA and EMA frameworks dictate study design and evidence standards for regulatory submissions. STROBE/RECORD-PE ensure study transparency and are required for publication. ISPOR provides the conceptual foundations for value assessment.
Interview Questions
Answer Strategy
Structure using PICOTS (Population, Intervention, Comparator, Outcomes, Timing, Setting) framework. Emphasize rigorous confounding control (e.g., target trial emulation) and prospective protocol design. Sample: 'I would emulate a target trial using a large EHR/claims network, defining a new-user cohort with strict eligibility. Primary analysis would use propensity score weighting to address channeling bias. Key regulatory considerations include pre-specifying all analytic choices in a protocol, ensuring data provenance, and engaging with the FDA's RWE Program early to align on endpoint definitions and data quality standards.'
Answer Strategy
Tests methodological rigor and understanding of bias. Use a framework like the Bradford Hill criteria for causation. Sample: 'I would assess biological plausibility, strength and consistency of association, temporality, and dose-response. Critically, I would investigate potential biases: was there detection bias (more monitoring in one group)? I'd run sensitivity analyses with different comparators and control outcomes. If the signal persists across robustness checks and aligns with preclinical data, it's more likely true. I would then escalate through pharmacovigilance channels for formal assessment.'
Careers That Require Real-world evidence (RWE) analytics and post-market surveillance