Introduction to Economic Evaluation of Health Promotion Interventions Assignment

Evaluating Cost-effectiveness using Markov cohort models: The HPV case

The goal of this assignment is to provide students with applied practical experience using Markov cohort models to evaluate the cost-effectiveness of alternative decisions, in this case universal HPV vaccination. The Markov cohort model included in this homework assignment is not intended to provide a realistic evaluation of the likely cost-effectiveness of each alternative given the limited nature of Markov cohort models and the artificial nature of a number of assumptions used to populate this model.

Given the complexity of HPV infection and cervical cancer epidemiology and transitions between eating states, decision analysts have determined that microsimulation models will likely provide a better analytic strategy for comparing population health interventions. However, for many of the PAHO interventions, modified cohort models without dynamic transition are still used to aid in transparency of decision-making.  Be sure to review units 7.3 and 7.4 in the asynchronous materials before starting this assignment. This content is also covered well starting on page 331 in the Drummond textbook, continuing through page 335.

Students will be provided with a diagram of the state transitions included in the model as well as the probabilities for transferring between states for each of the one-year model cycles. In contrast to the decision tree example from the previous homework assignment, students will be expected to use discounting to provide undiscounted and discounted health outcomes and cost for each of the alternative strategies in this homework. We will not be including half cycle correction in the homework as this is overly complicates the core concepts that we are trying to communicate here.

The cohort model described in this homework assignment will be run starting at age 11 (t=0) for 79 cycles or until the entire cohort has died. Assume that the discount rate is 3%. Note that a number of the transition probabilities and costs for this model change depending on the cycle. For example, the risk of transitioning from "healthy" to "local cancer" does not begin until the transition from cycle 38 to cycle 39 and the background mortality from healthy and local cancer does not begin until the same time. The costs of treating cancer are time limited and are described in more detail below.

State Transition Diagram

Deliverables

• Completed Markov cohort model in Excel with calculated cost-effectiveness ratio in terms of cost per DALY averted both in discounted and undiscounted terms
• 250 word commentary on the limitations of using the Markov cohort model in this case as well as the broader limitations of this modeling approach.