Simulation modeling to evaluate costbenefit of multi-level screening strategies involving behavioral components to improve compliance: The example of diabetic retinopathy

Recent health care reform brings forth the importance of preventive strategies such as screening. However, concerns about the growing population and rising healthcare spending necessitate health plans and public health policymakers to consider and determine cost-beneficial population-based screening strategies. Screening strategies often vary on compliance and, thus, result in suboptimal cost-benefit for the population. To achieve maximum cost benefit, especially for screenings that often suffer low compliance, such as Diabetic Retinopathy screening, policymakers need to consider multi-level (e.g. both patient- and provider-level) interventions to improve screening compliance. To determine which of such interventions are cost-beneficial, many different screening strategies need to be considered. Yet, it is costly, impractical, and time-consuming to do clinical trials to test all the different screening strategies. Simulation models provide a more cost- and time-effective way to help determine cost-beneficial screening strategies.;Simulation models have been used extensively to address cost-benefit of screening. However, there are shortcomings to existing models. First, they do not have a structure that enables the evaluation of strategies that include policies affecting compliance, even though screening compliance is often low. Thus, strategies that include policies targeting the improvement of compliance may be necessary to achieve maximum cost-benefit. Additionally, even policies not specifically targeting compliance may affect compliance, and models evaluating such policies without considering their impact on compliance will under- or over-estimate the policy impact. Second, current simulation models do not have a structure that can evaluate multi-level strategies (e.g. those targeting patient, providers, and clinics) even though they are more likely to have sustained or powerful effect than those targeting only the individual-level. This research develops a generic conceptual model for screening services that addresses the two shortcomings and then constructs the model for the case of Diabetic Retinopathy screening to illustrate the model.;The first shortcoming is addressed by including compliance as a mediating variable in the model, rather than a fixed input variable as in current methods. Compliance is influenced by patient characteristics (demographics, disease severity, self-care, health belief) and screening strategy used, and in turn compliance influences disease progression and healthcare utilization and, thus, cost and benefit of the screening strategy. The second shortcoming is addressed by using hierarchical simulation with nested design where policy effects are manifested not universally but through a hierarchical structure. This enables evaluation of the impact of policies at a higher aggregate level (e.g. policies that target providers) on individual (patient) outcomes. The multi-level design has the benefit of taking into account behavior at several levels and enabling the incorporation of economy of scale (e.g. how clinic size may affect cost-benefit of certain screening strategies) in the model. The developed model is then compared with models without compliance as a mediating variable and models without the hierarchical structure to illustrate its advantages in terms of policy impact assessment. In addition, the models were used to evaluate cost-benefit of different screening strategies using multiple perspectives. As the results indicate differing decisions made by policymakers viewing through these different perspectives, a case for context-based decision making is made.