RAPID ASSESSMENT OF MODELLING SOFTWARE BY END-USERS: THE EXAMPLE OF GLYCEMIC MANAGEMENT OPTIONS FOR TYPE 2 DIABETES

Purpose:

Modelling softwares to compare effectiveness of interventions for patients with type 2 diabetes mellitus (T2DM) are more available to managers of healthcare organizations than before.  There is a need for operations managers to more thoroughly understand these tools and their appropriate application in decision analysis. 

SMDM and ISPOR-AMCP-NPC have instituted task forces to develop guidelines to assist decision makers in assessing these models. Can such guidelines help managers choose among available models to answer specific policy questions?

We applied one such questionnaire developed by ISPOR-AMCP-NPC to assess two models of T2DM management, comparing the effectiveness of newer medications to a generic option for second line therapy, among those with recently diagnosed T2DM in the U.S. Models selected were ARCHeS from Evidera Inc. and the CDC/RTI model for T2DM.

Method:

First, both models were assessed using the published ISPOR-AMCP-NPC questionnaire. Next, they were populated with aligned inputs and three glycemic control strategies competed. Each strategy used metformin as first-line therapy and third line therapy chosen using ADA guidelines. The strategies differed in second line therapy used: glyburide, exenitide or sitagliptin. Outputs from each model were then compared. 

Result: The ISPOR-AMCP-NPC questionnaire efficiently identified specific domains concerning design (incident versus prevalent disease in population, duration of T2DM), data (medication efficacy), and analysis (sensitivity analysis) where the two models differed and where such differences could be anticipated to impact results. These insights were corroborated during subsequent analyses using the two models. Using a 15 year horizon and 3% discount rate, the ARCHeS model rank-ordered strategies QALY(glyburide) < QALY(exenatide) < QALY(sitagliptin). CDC/RTI model rank-ordered strategies QALY(glyburide) < QALY(sitagliptin) < QALY(exenatide). The cumulative incidence for a cardiovascular event over 15 years was 49.8% (glyburide), 46.9% (exenatide) and 49.2% (sitagliptin) for the ARCHeS model, and 44.7% for the CDC/RTI model (no difference by medication).

Conclusion: Upfront assessment of two T2DM models using the ISPOR-AMCP-NPC questionnaire showed that the questionnaire efficiently identified specific domains where models were not comparable and where model design, data and analysis resulted in predictably different results. The questionnaire enables managers to identify critical domains that could impact the credibility of results for the anticipated decision-making audience.