Variation in antidiabetic medication intensity among Medicare beneficiaries with diabetes mellitus

Recent guidelines for treating older patients with diabetes mellitus (DM) and significant disease burden place less emphasis on glycemic control and stress the potential harms that may arise from adherence to strict regimens with antidiabetic medications. However, there are few empirical benchmarks against which clinicians can compare their prescribing practices for patients who have DM and varying levels of comorbidity.

The current study had 2 goals: (1) to provide national estimates showing how the intensity of antidiabetic medication regimens for Medicare beneficiaries with DM varies by level of medical spending (a proxy for overall disease burden); and (2) to identify potential predictive factors associated with the observed differences.

This study analyzed 2002 Medicare Current Beneficiary Survey (MCBS) data to benchmark intensity of antidiabetic medication regimens for Medicare beneficiaries with DM arrayed by decile of cumulative medical care spending. The study involved 3 steps: (1) stratification of the study population into 10 mutually exclusive deciles by cumulative all-source annual medical spending; (2) assessment of the unconditional association between decile assignment and intensity of antidiabetic medication use; and (3) identification of mediating factors that differentially explain medication intensity across the spectrum of disease burden. We evaluated 3 outcomes: (1) prevalence of any antidiabetic agent in 2002; (2) annual utilization rates for 5 different classes of oral hypoglycemic agents (sulfonylurea, metformin, thiazolidinedione, alpha-glucosidase inhibitors, and meglitinides) plus insulins; and (3) counts of annual prescription fills.

The final study sample comprised 1956 Medicare beneficiaries representing 23.1% of the MCBS sample after exclusions. We found a pronounced inverted U-shaped pattern in intensity of antidiabetic treatment. Compared with individuals in the group with the highest prevalence of antidiabetic use (decile 7), the unadjusted treatment odds ratios were 0.40 in decile 1 (95% CI, 0.26-0.60) and 0.54 in decile 10 (95% CI, 0.36-0.81). We found similar patterns in the complexity of drug regimens and numbers of antidiabetic prescriptions filled among users. Controlling for disease severity and other factors eliminated the inverted U-shaped pattern among higher cost beneficiaries but not for those in the lower spending deciles.

This national study found that high-cost Medicare beneficiaries with DM received substantially less intensive antidiabetic regimens compared with those incurring more modest medical expenditures in 2002. Longitudinal analysis is necessary to determine whether this finding indicates suboptimal therapy or has a more benign explanation. However, the magnitude of the association warrants the attention of clinicians who treat elderly and disabled diabetic patients with high disease burden.