Module 1: Why Real-World Evidence Matters

Randomised controlled trials are not always feasible, affordable or appropriate. Real-world evidence (RWE) has become an increasingly accepted component of health technology assessment (HTA) — but only when the underlying studies are designed to a standard that payers will trust. This module introduces the RWE landscape and explains why generating credible observational evidence requires a structured methodology.

• what counts as real-world evidence — retrospective database studies, prospective registries, claims data analyses and post-market surveillance
• the growing regulatory and HTA acceptance of RWE
• why published RWE is often poor — single-arm studies with inconsistent endpoints, no comparator and insufficient follow-up
• why well-designed observational studies can approach the credibility of trials and why poorly designed ones are worse than no evidence at all

Module 2: Data Sources for Real-World Evidence

The quality of any RWE study is constrained by the data it draws on. This module surveys the main data sources available for such research and covers how to assess whether a data source is fit for the research question at hand.

• national administrative datasets — Hospital Episode Statistics, NHS Digital and equivalents in other health systems
• disease and procedure registries — national clinical registries, specialty databases and working with registry holders
• single-centre and investigator-maintained datasets — opportunities and risks including completeness, standardisation and generalisability
• linking datasets — combining data sources and navigating the practical, ethical and governance challenges involved

Module 3: Study Design for Observational Research

The choice of study design determines the strength of causal inference an observational study can support. This module covers the main observational designs and provides a framework for selecting the right design based on the research question and available data.

• retrospective cohort studies — the workhorse of RWE, defining exposed and unexposed groups and the importance of a clearly specified index date
• prospective registries and cohort studies — when prospective data collection is warranted and designing registry protocols for HTA-grade evidence
• case-control and cross-sectional designs — their limited but sometimes useful role, particularly for rare outcomes and safety signals
• target trial emulation — designing observational studies as if they were trials to strengthen both design and analysis

Module 4: Protocol Development and Pre-Registration

An observational study without a pre-specified protocol is an invitation to data dredging — and HTA bodies know it. This module covers how to write a robust study protocol for observational research, drawing on the HARPER template and international best practice.

• the HARPER protocol template — structuring research questions, study design, data sources, population definitions and planned analyses
• defining the study population — inclusion and exclusion criteria, prevalent versus incident users and the consequences of imprecise definitions
• specifying outcomes and follow-up — locking endpoint definitions before data access and defining follow-up periods that satisfy HTA assessors
• pre-registration and transparency — where to register protocols, why it is increasingly expected and how it protects against selective reporting

Module 5: Bias, Confounding and Threats to Validity

Every observational study is vulnerable to biases that randomisation would have eliminated. This module covers the threats to validity that arise most frequently in RWE research and explains why ignoring them produces results that payers will not believe.

• confounding by indication — why patients receiving a new intervention may differ systematically from comparator groups and why naive comparisons mislead
• selection bias and immortal time bias — two of the most common and damaging biases in retrospective studies
• information bias and misclassification — inaccurate coding, inconsistent endpoint definitions and missing data
• why these biases go unaddressed — the pattern of ignoring confounding, comparing non-comparable groups and presenting unadjusted results as causal

Module 6: Statistical Methods for Causal Inference from Observational Data

When randomisation is absent, statistical methods must make treatment groups comparable. This module covers the core analytical techniques at a level that enables participants to commission, conduct or critically appraise analyses of RWE.

• propensity score methods — matching, stratification, weighting and covariate adjustment, when each is appropriate and what to look for when evaluating quality
• multivariable regression — assumptions, limitations when confounding is severe and warning signs of inadequate adjustment
• instrumental variable analysis and advanced methods — a commissioner-level introduction to techniques addressing unmeasured confounding
• survival analysis — Kaplan-Meier curves, Cox proportional hazards models, competing risks and time-to-event methods for long-term outcomes

Module 7: Reporting Standards and Quality Assurance

An observational study is only as credible as its reporting is transparent. This module covers the reporting frameworks and quality appraisal tools that journals and HTA bodies use to evaluate RWE — and how to build compliance in from the outset.

• STROBE and RECORD — reporting guidelines for observational studies and routinely collected health data
• ROBINS-I — the risk of bias tool for non-randomised studies and how earlier design choices determine the assessment received
• GRADE for observational evidence — how certainty is rated, why observational studies start at low certainty and what that means for payer decision-making
• internal quality assurance — review and sign-off processes that catch errors, ensure protocol adherence and produce a defensible audit trail

Module 8: From Real-World Evidence to HTA Submission

Generating credible real-world evidence is only valuable if it reaches the right decision-makers in the right format. This final module covers how RWE findings are used within HTA submissions, value dossiers and procurement cases.

• when HTA bodies accept RWE as primary evidence — circumstances under which NICE, G-BA and other agencies consider observational data as the principal basis for appraisal
• integrating RWE into economic models — how real-world resource utilisation and outcomes data feed into budget impact and cost-effectiveness analyses
• RWE in clinical evaluation reports and procurement — supporting regulatory compliance and value-based procurement
• building a long-term RWE strategy — planning a programme of observational studies that accumulates evidence and supports re-evaluation and geographic expansion