A dynamic model for estimating changes in health status and costs

Area of Science:

• Health Economics
• Biostatistics
• Cancer Research

Background:

• Accurately assessing total treatment costs is complex due to dynamic patient health status and incomplete data.
• Existing methods often struggle to incorporate patient-specific factors influencing both cost and health outcomes.
• Cancer treatment costs require robust evaluation methods that account for disease progression and healthcare utilization.

Purpose of the Study:

• To develop and demonstrate an innovative joint regression model for assessing total treatment costs over a finite period.
• To incorporate patient health status dynamics, medical care use, and patient characteristics into cost estimations.
• To estimate net present values (NPVs) of expenditures as a function of patient variables in cancer patients.

Main Methods:

• Utilized a Markov model to estimate health state transition probabilities and the impact of patient variables.
• Employed a mixed-effects model for sojourn costs, treating transition times as random effects and patient variables as fixed effects.
• Combined these models to estimate NPVs of expenditures, accommodating data complexities like censoring, heteroscedasticity, and skewness.

Main Results:

• Applied the joint regression model to a dataset of 624 incident cancer cases, using Medicare claim files for 2-year cost data.
• Estimated NPVs for charges incurred over 2 years, stratified by cancer site and stage.
• Demonstrated the model's flexibility in assessing patient characteristic influence on both cost and health outcomes.

Conclusions:

• The developed joint regression model offers a flexible and robust approach to estimating total treatment costs.
• This method effectively integrates patient health dynamics, healthcare utilization, and individual characteristics.
• The findings highlight the importance of patient-specific factors in understanding and managing cancer treatment expenditures.