IMPROVING AN ADHERENT PATIENT ANALYSIS OF THE DIABETES PREVENTION PROGRAM AND OUTCOMES STUDY
Author(s)
Gabriel Gbidye, MS, RN1, Brian Rittenhouse, PhD2;
1MCPHS University, Boston, Boston, MA, USA, 2MCPHS University, Boston, MA, USA
1MCPHS University, Boston, Boston, MA, USA, 2MCPHS University, Boston, MA, USA
OBJECTIVES:
An adherent-patient subgroup (APS) analysis of the 10-year Diabetes Prevention Program (DPP) and Outcomes Study (OS) was published in 2013, comparing placebo, metformin, and lifestyle modification (lifestyle), and concluding that both lifestyle and MET were cost-effective. Four problems limit the interpretation of that analysis. First, it used unconventional QALY calculations. Second, after year 3, placebo and metformin participants were permitted to enter lifestyle, contaminating comparisons. Third, the analysis assumed that all placebo participants were adherent. Fourth, incremental cost-effectiveness ratios (ICERs) were not calculated appropriately.
METHODS: We reanalyzed the APS using only the uncontaminated years 1 to 3 and conventional QALY calculation. Placebo outcomes were modified using published DPP adherence prevalence and a 9 percent risk reduction in diabetes for adherent versus non-adherent participants. Revised placebo outcomes used published diabetes-status-based cost and utility information.
RESULTS:
Our 3 year reanalysis using conventional QALYs changed APS placebo costs and QALYs relative to the original full-adherence placebo assumption. Incorporating observed adherence prevalence (77 percent) and diabetes-risk differences in adherent and non-adherent patients reduced APS placebo costs by approximately 67 dollars and slightly reduced APS placebo QALYs by approximately 0.00023. Proper incremental comparisons showed that metformin was extendedly dominated and that the only APS comparison needed was lifestyle versus placebo, with an ICER of approximately 45,000 dollars per QALY, showing lifestyle to be cost-effective, which is the same conclusion as in the original analysis had ICERs been calculated correctly. The calculations also permit cost-effectiveness analysis of the previously unmodeled non-adherent placebo subgroup, for which placebo was the cost-effective treatment.
CONCLUSIONS:
These corrections yield a more accurate APS result and enable estimation of the non-adherent sample results. A remaining question is how such subgroup results can inform decisions when adherence is not known in advance but must be estimated with potential error.
An adherent-patient subgroup (APS) analysis of the 10-year Diabetes Prevention Program (DPP) and Outcomes Study (OS) was published in 2013, comparing placebo, metformin, and lifestyle modification (lifestyle), and concluding that both lifestyle and MET were cost-effective. Four problems limit the interpretation of that analysis. First, it used unconventional QALY calculations. Second, after year 3, placebo and metformin participants were permitted to enter lifestyle, contaminating comparisons. Third, the analysis assumed that all placebo participants were adherent. Fourth, incremental cost-effectiveness ratios (ICERs) were not calculated appropriately.
METHODS: We reanalyzed the APS using only the uncontaminated years 1 to 3 and conventional QALY calculation. Placebo outcomes were modified using published DPP adherence prevalence and a 9 percent risk reduction in diabetes for adherent versus non-adherent participants. Revised placebo outcomes used published diabetes-status-based cost and utility information.
RESULTS:
Our 3 year reanalysis using conventional QALYs changed APS placebo costs and QALYs relative to the original full-adherence placebo assumption. Incorporating observed adherence prevalence (77 percent) and diabetes-risk differences in adherent and non-adherent patients reduced APS placebo costs by approximately 67 dollars and slightly reduced APS placebo QALYs by approximately 0.00023. Proper incremental comparisons showed that metformin was extendedly dominated and that the only APS comparison needed was lifestyle versus placebo, with an ICER of approximately 45,000 dollars per QALY, showing lifestyle to be cost-effective, which is the same conclusion as in the original analysis had ICERs been calculated correctly. The calculations also permit cost-effectiveness analysis of the previously unmodeled non-adherent placebo subgroup, for which placebo was the cost-effective treatment.
CONCLUSIONS:
These corrections yield a more accurate APS result and enable estimation of the non-adherent sample results. A remaining question is how such subgroup results can inform decisions when adherence is not known in advance but must be estimated with potential error.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
EE112
Topic
Economic Evaluation
Disease
SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity)