Comparative Effectiveness Research Using Health Data

Comparative effectiveness research (CER) compares the benefits and harms of alternative interventions for preventing, diagnosing, treating, or monitoring a health condition under real-world conditions. When conducted using routinely collected health data, it draws on electronic records, claims, and registries to estimate how interventions perform in everyday practice rather than only in controlled trials.

Definition

Comparative effectiveness research using health data is the comparison of the real-world benefits and harms of alternative health interventions through analysis of routinely collected data sources such as electronic health records, administrative claims, and registries.

Scope

This topic covers the use of observational and routinely collected health data to compare interventions, the analytic challenges of doing so (notably confounding and data quality), and the relationship between CER and trial-based evidence. It treats CER as a methodological and informatics topic. It describes how comparative evidence is generated and is not a source of treatment recommendations.

Key concepts

• Real-world evidence
• Routinely collected (secondary) data
• Confounding by indication
• Causal inference from observational data
• Pragmatic versus explanatory designs
• Value in health care (outcomes per cost)
• Patient-centered outcomes
• Generalizability and external validity

Mechanisms

CER asks which of two or more options works better for whom and under what circumstances. Randomized trials answer this with high internal validity but often in selected populations; CER complements them by using large bodies of routinely collected data that reflect everyday practice. Because such data are observational, the central methodological problem is confounding, especially confounding by indication, where the reasons a treatment was chosen are also related to outcomes. Analysts therefore rely on causal-inference methods and careful design, and the credibility of conclusions depends heavily on the quality and completeness of the underlying data. CER is frequently framed around value, defined as health outcomes achieved relative to cost.

Clinical relevance

CER informs guidelines, coverage decisions, and shared decision-making by indicating how alternative interventions compare in real-world populations. Reading it critically requires attention to confounding and data quality, since biased comparisons can mislead. This topic explains how comparative evidence is produced; it is not itself prescriptive and does not direct individual treatment choices.

Evidence & guidelines

The U.S. Institute of Medicine's 2009 report defined comparative effectiveness research and set national research priorities, helping to establish CER as a field and motivating investment in patient-centered outcomes research. The report is a foundational policy reference rather than a clinical practice guideline.

History

Interest in comparing interventions under real-world conditions grew as policymakers sought evidence to improve quality and control costs. The 2009 Institute of Medicine report formalized comparative effectiveness research and its priorities, and the subsequent expansion of electronic health records and claims databases made large-scale observational comparison feasible, while also sharpening concerns about confounding and data quality in such analyses.

Debates

Can observational health data substitute for randomized trials in comparing interventions?

Routinely collected data offer scale and real-world relevance but are vulnerable to confounding by indication and quality limitations; debate continues over when observational CER yields trustworthy causal comparisons and when only randomization can resolve the question.

Key figures

• Harold Sox
• Sheldon Greenfield
• Michael E. Porter

Seminal works

• sox-greenfield-2009
• porter-2010

Frequently asked questions

How is comparative effectiveness research different from a randomized controlled trial?

A randomized trial usually tests an intervention against a comparator under controlled conditions in a selected population. Comparative effectiveness research, especially when based on routinely collected data, compares interventions as they are used in everyday practice, trading some internal validity for greater real-world relevance.

Why is confounding such a concern in data-based comparative effectiveness research?

Because treatments in routine care are chosen for reasons related to patient prognosis, the groups being compared often differ systematically. This confounding by indication can make one intervention appear better or worse than it is unless it is carefully addressed by design and analysis.