Use of Regression Models in Cost-effectiveness Analysis

Michigan State University, East Lansing MI

Investigators

Linked publications & trials

Abstract

[unreadable] DESCRIPTION (PROVIDED BY APPLICANT): In an era of economic constraints, methods for assessing costs and outcomes of health care programs, and comparing costs with outcomes of competing interventions have come to the forefront in the pursuit of optimizing health benefits from a specified budget, or in finding the lowest cost strategy for a specified health effect. The objective of this research is to develop, test, and apply innovative statistical methods for analyses of health care costs, utilization, and outcomes with the goal of informing decision-making in the allocation of health care resources. We propose a unified stochastic framework in which costs of an intervention are incurred dynamically through resource use as a patient's health history unfolds over time. Our models recognize limitations in data that are typically present in epidemiologic and clinical studies. We incorporate observable characteristics of patients, such as demographics and comorbid conditions, and account for unmeasured variables that might influence both cost and health outcomes. We estimate summary measures commonly used in economic evaluations (e.g., life expectancy, quality-adjusted life years, net present value, net health benefit, and cost-effectiveness ratios) and derive the basis for statistical inference on these measures. We then test the performance and sensitivity of our procedures with both real and simulated data. [unreadable] [unreadable] We propose applications of our methods using national state databases and ongoing clinical studies. 1) Using the Nationwide Inpatient Sample (NIS), we will estimate inpatient costs in relation to comorbidity, patient demographics, and clinical attributes: (a) for patients with acute myocardial infarction (AMI) undergoing cardiac procedures (coronary artery bypass surgery, percutaneous coronary intervention); (b) for women with breast cancer, undergoing mastectomy or lumpectomy; (c) for patients with colorectal cancer undergoing colectomy. 2) Using the Michigan Medicaid and Medicare claims database for 1996 to 2000, we examine the impact of treatments, cancer stage at diagnosis, patient demographics, and comorbid illnesses on cost and survival in patients with breast, colorectal, lung, and prostate cancer. 3) In a trial of a nurse-managed protocol emphasizing strong patient-provider relationships in high-utilizing patients with no evidence of organic disease, we will estimate costs and cost-effectiveness in relation to improvements in quality of life (QOL), mental and physical health functioning and patient satisfaction. 4) In a trial of an intervention in women undergoing surgery for breast cancer, we assess costs, health care utilization, QOL, and return to presurgery physical and mental health functioning. 5) In the Heart After Hospital Recovery Planner study of patients after AMI, we examine the impact on QOL of an intervention aimed at efficient use of resources in health education and management. [unreadable] [unreadable] This application contributes to an international research effort to develop rigorous methods for analyses of health care costs and outcomes, and, applied to clinical and epidemiologic studies, it directly translates research into practice. [unreadable] [unreadable] [unreadable]