Estimation of Input Costs for a Markov Model in a German Health Economic Evaluation of Newer Antidepressants

Abstract

Background: Estimating input costs for Markov models in health economic evaluations requires health state–specific costing. This is a challenge in mental illnesses such as depression, as interventions are not clearly related to health states. We present a hybrid approach to health state–specific cost estimation for a German health economic evaluation of antidepressants. Methods: Costs were determined from the perspective of the community of persons insured by statutory health insurance (“SHI insuree perspective”) and included costs for outpatient care, inpatient care, drugs, and psychotherapy. In an additional step, costs for rehabilitation and productivity losses were calculated from the societal perspective. We collected resource use data in a stepwise hierarchical approach using SHI claims data, where available, followed by data from clinical guidelines and expert surveys. Bottom-up and top-down costing approaches were combined. Results: Depending on the drug strategy and health state, the average input costs varied per patient per 8-week Markov cycle. The highest costs occurred for agomelatine in the health state first-line treatment (FT) (“FT relapse”) with €506 from the SHI insuree perspective and €724 from the societal perspective. From both perspectives, the lowest costs (excluding placebo) were €55 for selective serotonin reuptake inhibitors in the health state “FT remission.” Conclusion: To estimate costs in health economic evaluations of treatments for depression, it can be necessary to link different data sources and costing approaches systematically to meet the requirements of the decision-analytic model. As this can increase complexity, the corresponding calculations should be presented transparently. The approach presented could provide useful input for future models.

Analyzing maternal mortality rate in rural China by Grey-Markov model

“…Markov chain model is widely used in cost-effectiveness analysis. [11–13] This model is a dynamic system which based on the state transition. [14] The system's state is randomized at all time and independent with prior states, this characteristic is called non-aftereffect property or Markov process.…”