Related Experiment Videos
Confidence intervals for cost-effectiveness ratios: a comparison of four methods
D Polsky1, H A Glick, R Willke
1Division of General Internal Medicine, University of Pennsylvania, USA.
Health Economics
|May 1, 1997
Summary
The nonparametric bootstrap and Fieller theorem methods accurately compute confidence intervals for cost-effectiveness ratios. These methods improve decision-making for healthcare interventions and policy.
Area of Science:
- Health Economics
- Biostatistics
- Clinical Trial Analysis
Background:
- Cost-effectiveness ratios (CERs) are crucial for healthcare decision-making.
- Accurate confidence intervals (CIs) for CERs are essential for reliable interpretation.
- Existing methods for CI calculation may have limitations.
Purpose of the Study:
- To compare the performance of four methods for calculating CIs for CERs.
- To assess the impact of cost and effect distributions and correlations on CI accuracy.
- To identify the most reliable methods for reporting CIs in cost-effectiveness analyses.
Main Methods:
- Monte Carlo simulation experiment.
- Evaluation of four CI methods: box, Taylor series, nonparametric bootstrap, and Fieller theorem.
- Analysis under varying cost/effect distributions (normal, log-normal) and correlations (-0.50 to 0.50).
Main Results:
- Nonparametric bootstrap and Fieller theorem methods demonstrated superior accuracy in coverage probability.
- Taylor series method showed asymmetric underestimation of interval upper limits.
- Nonparametric bootstrap and Fieller theorem methods provided more dependable accuracy than box or Taylor series methods.
Conclusions:
- Nonparametric bootstrap and Fieller theorem methods are recommended for calculating CIs for CERs.
- Accurate CIs enhance the identification of cost-effective healthcare interventions.
- Routine use of reliable CI methods supports informed clinical and policy judgments.