THE APPLICATION OF CONTINUOUS-TIME SEMI-MARKOV MODELS IN HEALTH ECONOMIC DECISION MAKING: A CASE STUDY IN HEART FAILURE DISEASE MANAGEMENT
Candidate for the Lee B. Lusted Student Prize Competition
To conduct a model-based economic evaluation of three disease management programs (DMPs) for heart failure (HF).
Method:
The interventions evaluated in this study were taken from the COACH, which is a multi-center randomized controlled trial in which 1023 HF patients were randomly assigned to three different DMPs: usual care consisting of routine management by a cardiologist and two intervention groups consisting of additional basic and intensive support by a nurse specialized in HF management. To extrapolate the findings from COACH beyond the study’s 18 month follow-up, we developed a continuous-time semi-Markov model with three health states: discharged alive (state 1), HF-related hospital readmission (state 2), and dead (state 3). Parametric survival models with the treatment indicator included as a covariate were fitted to estimate the distributions of the elapsed time in state 1 and state 2 before making a transition to a new state. Logistic regression models were used to estimate the probability of the next state visited conditional on the time spent in the present state before making this transition.
The long-term cost-effectiveness of the thee different DMPs were expressed in terms of costs per life year gained. The cost attached to state 1 were set to €3.40/d for care-as-usual, €3.99/d for basic support, and €5.21/d for intensive support. The cost attached to state 2 was set to €769/d for all DMP groups. Monte Carlo simulation was applied to evaluate the model and generate the cost and effectiveness outcomes. The time horizon for this analysis was set to 10 years.
Result:
The average survival time was found to be 4.43 years for care-as-usual, 4.90 years for basic support, and 5.50 years for intensive support with €8511, €9294, and €13453 being the corresponding average costs. This resulted in an incremental cost-effectiveness ratio (ICER) of €1666 between basic support and care-as-usual and an ICER of €6932 between intensive support and basic support.
Conclusion:
Out of the three DMPs considered, an intensive form of nurse-led HF disease management yielded the most favorable cost-effectiveness ratio over a 10-year time horizon.