Risk-adjusted Phase IV study — Post-marketing Surveillance with Risk Adjustment
Risk-adjusted Phase IV Post-marketing Study · Also known as: risk-adjusted post-marketing surveillance study, adjusted Phase IV trial, risk-stratified post-authorization study, PASS with risk adjustment
A risk-adjusted Phase IV study is an observational or semi-experimental post-marketing study conducted after a drug or device has received regulatory approval. It uses statistical risk-adjustment techniques — such as propensity score matching, inverse probability weighting, or multivariable regression — to control for confounding by indication and baseline patient differences, thereby producing more credible safety, effectiveness, and utilization estimates than unadjusted real-world analyses.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use a risk-adjusted Phase IV study when a drug or device is already marketed and you need real-world evidence on safety, effectiveness, or utilization in populations underrepresented in pre-approval trials — including elderly patients, patients with comorbidities, or those on polypharmacy. It is appropriate when a randomized controlled trial is infeasible, unethical, or too slow, and when a sufficiently large real-world database with relevant covariate information is available. Do not apply this design when the required confounders are not measured in the data source (unmeasured confounding cannot be addressed by propensity scores alone), when the exposure is rare and the database lacks sufficient sample size, or when the causal question requires a prospective randomized answer that regulators will not accept in observational form.
Strengths & limitations
- Enables long-term safety and effectiveness surveillance in diverse real-world patient populations that randomized trials rarely include.
- Risk adjustment markedly reduces confounding by indication, improving causal credibility of observational estimates.
- Leverages large existing databases, allowing study of rare outcomes with sample sizes far exceeding any feasible trial.
- Aligns with regulatory post-authorization safety study (PASS) requirements in the EU and post-marketing commitments in the US, facilitating regulatory acceptance.
- Allows examination of comparative effectiveness across subgroups and real-world dosing patterns.
- Propensity score and regression methods can only adjust for measured confounders — unmeasured confounders (e.g., disease severity indicators not recorded in the database) remain a fundamental threat to validity.
- Data quality is constrained by the source system: coding errors, incomplete records, and misclassification of exposure or outcomes can bias results in unpredictable directions.
- The design does not support strong causal claims on its own; findings must be interpreted alongside biological plausibility and corroborating evidence.
- Complex risk-adjustment pipelines require specialized biostatistical expertise and increase the risk of analytic errors or selective reporting if the protocol is not pre-specified.
Frequently asked
How is a risk-adjusted Phase IV study different from a standard observational cohort study?
Both are observational designs using real-world data, but a risk-adjusted Phase IV study is specifically embedded in the post-marketing regulatory context — it targets a drug or device already on the market — and applies formal, pre-specified risk-adjustment methods (propensity scoring, IPTW, or multivariable adjustment) as a core design element rather than an afterthought. The regulatory framing also imposes protocol pre-registration, defined safety outcomes, and structured reporting to health authorities.
Can propensity score matching completely eliminate confounding?
No. Propensity score methods balance groups only on the confounders that are measured in the dataset. Unmeasured confounders — factors that influence both who receives the treatment and what outcome they experience but are not recorded — remain. Sensitivity analyses such as E-values, instrumental variable approaches, or negative control outcomes can quantify or partially probe unmeasured confounding, but they cannot eliminate it.
What databases are typically used for these studies?
Common data sources include national or regional electronic health record databases (e.g., the UK Clinical Practice Research Datalink), administrative claims databases (e.g., US Medicare/Medicaid or commercial insurer claims), disease registries, and hospital data warehouses. The suitability of a database depends on whether the relevant exposures, confounders, and outcomes are reliably recorded within it.
Is regulatory approval required to conduct a risk-adjusted Phase IV study?
In the EU, post-authorization safety studies (PASS) that are imposed as a condition of authorization must be submitted to and approved by the EMA through the ENCEPP framework. Company-initiated voluntary studies follow good pharmacoepidemiology practice guidelines but may not require prior regulatory approval. In the US, post-marketing commitments under FDA are tracked but voluntary observational studies do not require FDA pre-approval.
When should I prefer an active comparator new-user design?
Whenever feasible. Comparing new users of the drug of interest to new users of an active comparator (rather than non-users or prevalent users) eliminates immortal time bias, avoids the healthy-user effect, and makes the comparison groups more clinically similar at baseline. This design is now considered best practice in pharmacoepidemiology and is recommended in methodological guidelines from the FDA Sentinel program and the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance (ENCePP).
Sources
- Strom, B. L. (Ed.). (2005). Pharmacoepidemiology (4th ed.). John Wiley & Sons. ISBN: 978-0470863107
- Austin, P. C. (2011). An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behavioral Research, 46(3), 399–424. DOI: 10.1080/00273171.2011.568786 ↗
How to cite this page
ScholarGate. (2026, June 3). Risk-adjusted Phase IV Post-marketing Study. ScholarGate. https://scholargate.app/en/epidemiology/risk-adjusted-phase-iv-study
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Cohort StudyEpidemiology↔ compare
- Inverse Probability WeightingCausal inference↔ compare
- Propensity Score MatchingResearch Statistics↔ compare