Exploring the methodological challenges of investigating comparison groups with different underlying characteristics:...

Area of Science:

• Pharmacoeconomics and Health Outcomes Research
• Clinical Epidemiology
• Health Services Research

Background:

• Claims data are frequently used for comparative effectiveness research.
• However, differences in patient populations can lead to biased conclusions.
• Confounding variables, such as disease severity and clinical complexity, must be addressed.

Purpose of the Study:

• To identify challenges in drawing accurate conclusions from claims data when comparing patient groups.
• To emphasize the necessity of adjusting for confounding variables using statistical methods.

Main Methods:

• Identified patients aged 60+ initiating amlodipine or felodipine from a managed care claims database.
• Stratified patients by hypertension severity and clinical complexity using a Burden of Illness (BOI) score.
• Analyzed patient demographics, comorbidities, and medication dosage.

Main Results:

• Amlodipine users (n=12,389) were similar in age to felodipine users (n=5,278) but had higher clinical complexity and disease burden.
• Amlodipine users exhibited greater prevalence of high-severity hypertension and ischemic heart disease/angina.
• A statistically significant difference in average daily dose (ADD) was observed between the two groups.

Conclusions:

• Head-to-head comparisons using claims data necessitate rigorous assessment of confounding variables.
• Risk adjustment or stratification is crucial for addressing differing patient characteristics.
• Accurate conclusions depend on accounting for patient heterogeneity in comparative effectiveness studies.