Related Experiment Videos
Estimating medical costs from incomplete follow-up data
D Y Lin1, E J Feuer, R Etzioni
1Department of Biostatistics, University of Washington, Seattle 98195, USA.
Biometrics
|June 1, 1997
Summary
Estimating total treatment costs with censored survival data is challenging. This study introduces novel methods to accurately calculate average costs, minimizing bias from unknown survival times.
Area of Science:
- Biostatistics
- Health Economics
- Survival Analysis
Background:
- Estimating average total treatment costs is complex due to censored survival data.
- Standard survival analysis methods are not directly applicable to cost estimation with censored data.
- Naive cost averages can be severely biased.
Purpose of the Study:
- To develop and validate unbiased estimators for average total treatment costs in the presence of censored survival times.
- To address the limitations of existing methods in health economic evaluations.
Main Methods:
- Partitioning the time period into small intervals.
- Utilizing the Kaplan-Meier estimator for survival probabilities within intervals.
- Estimating interval-specific costs conditional on survival.
- Developing two novel cost estimation approaches based on interval analysis.
Main Results:
- The proposed estimators are consistent when censoring occurs at interval boundaries.
- The estimators are asymptotically normal with estimable variances.
- Numerical studies indicate small biases and adequate asymptotic approximations for practical use, even with interior censoring.
Conclusions:
- The developed methods provide a robust approach to estimating average total treatment costs with censored data.
- These estimators offer improved accuracy in health economic evaluations, particularly in oncology.
- The study provides a practical framework applicable to real-world healthcare cost analyses, as demonstrated by an ovarian cancer case study.