Nonparametric inference for median costs with censored data

Area of Science:

• Health Economics
• Biostatistics
• Health Policy

Background:

• Health care cost estimations are vital for treatment evaluation and disease expenditure assessment.
• Current focus is on mean cost, but median cost is also important for payers and consumers.
• Right censoring in prospective studies presents challenges for accurate cost data analysis.

Purpose of the Study:

• To propose methods for estimating the median cost and its confidence interval (CI) with right-censored data.
• To extend these methods for estimating the ratio and difference of two median costs and their CIs.
• To provide tools for analyzing informatively censored cost data.

Main Methods:

• Development of novel statistical methods for median cost estimation under right censoring.
• Application of these methods to estimate ratios and differences of median costs.
• Validation through simulation studies and real-world data analysis.

Main Results:

• The proposed methods effectively estimate median costs and their CIs in the presence of right censoring.
• The methods are robust and perform well in simulation and real data analyses.
• The study provides a framework for analyzing other quantiles and informatively censored data.

Conclusions:

• New statistical methods are presented for median health care cost estimation with censored data.
• These methods address the challenges of induced informative censoring in cost data.
• The findings support more accurate economic evaluations and health policy decision-making.