Imputing missing patient-level data and propensity score matching in cost-effectiveness analysis in Crohn's disease

Area of Science:

• Health Economics
• Biostatistics
• Pharmacoeconomics

Background:

• The impact of data imputation and propensity score analysis on cost-effectiveness analysis (CEA) remains unclear.
• Accurate economic evaluations are crucial for healthcare decision-making, particularly for novel treatments.

Purpose of the Study:

• To compare different methods of grouping data after imputation and before propensity score calculation in economic evaluations.
• To assess the effect of these methods on the incremental cost-effectiveness ratio (ICER).

Main Methods:

• Utilized patient-level data from 573 children with Crohn's disease in a microsimulation model.
• Compared two approaches for propensity score matching after multiple imputation: 'Within' (separate analysis per dataset) and 'Across' (averaged propensity scores).
• Calculated the incremental cost per remission week gained for early anti-tumor necrosis factor-α treatment versus standard care.

Main Results:

• The incremental cost per remission week gained varied from CAD$2,236 to CAD$12,464 (mean CAD$4,266) using the 'Within' approach.
• The 'Across' approach yielded an incremental cost of CAD$4,679 per remission week gained.
• The 'Across' method resulted in multiple sets of health state transition probabilities.

Conclusions:

• Imputing missing data and using propensity score analysis introduces methodological uncertainty into cost-effectiveness analyses.
• The 'Across' approach appears less complex and may reduce bias and variance in economic evaluations.
• Further research is needed to fully understand the implications of these methods on ICER.