Does Disease Management for Diabetes Reduce Medical Expenditures? Evidence Using a Three-Period Difference-in-Difference Analysis

Abstract

Disease management (DM) is one of the most popular concepts among policymakers in OECD countries. Governments of these countries — including Germany, the United Kingdom, and the United States — not only implement pilot projects but also allocate substantial budgets toward DM programs. Nevertheless, evaluations about DM program on cost are still controversial.

This study provides new evidence regarding DM program on cost. We focus on the DM program to prevent from renal dialysis for type 2 diabetic nephropathy patients (in the stage three and more). The DM program was implemented by a public health insurer (Kure City in Hiroshima Prefecture in Japan). We collate a unique database by linking medical claims data for the 2010–2012 fiscal years with intervention records by the DM program on both 2010 and 2011 fiscal years. We obtained unbalanced panel data for monthly medical costs in the form of several subtotals; for example, inpatient costs, outpatient costs, drug costs, and dialysis costs. There were 172 diabetic patients in our treatment group and 258 diabetic patients in our control group.

To tackle with selection bias by non-randomized allocation of participants of the DM program, we adopted Fixed Effect Model of panel data analysis. We inserted individual constant term to absorb unobserved time irrelevant heterogeneity (i.e. motivation to control chronic illness) in the model. We also employ “Charlson Comorbidity Index” as explanatory variable in the model to control complexity of diabetic patient based on number of complications. We expanded Difference in Difference analysis from traditional two periods model to newly three periods model. Then, we can distinguish “rectification effect” on medical cost by case management in the DM program from “health improvement effect” on medical cost by preventing undergo artificial dialysis thanks to the DM program.

Results indicate that dialysis cost was 51.33% lower in treatment group comparing with control group. The rectification effect was 33.63% and larger than the health improvement effect which was 25.84%. In case of all medical claim cost excluding the dialysis cost, there was only rectification effect and it was 8.52% lower than that of control group.

The study concludes that the DM program focus on serious diabetic nephropathy patients would have significant cost containment effect on medical cost.

Keywords: Disease Management, Diabetes, DID, Charlson Comorbidity Index Score

JEL Classification: I18, I12, C23, H51