Using a treatment transition model to Evaluate the Effects of Neglecting Hba1c Drift In oral anti-diabetic Drugs for Type 2 Diabetes


      HbA1c drift is the gradual loss of treatment efficacy typically for oral anti-diabetic drugs. Modeling studies including the effects of HbA1c drift in oral anti-diabetic drugs are rare in the literature. Clinical outcomes can be distorted if the drifts are not considered. Our study compared the simulated clinical outcomes of type 2 diabetic mellitus (T2DM) treatments with and without drift.


      Lifetime clinical outcomes of treatments were estimated using a Monte Carlo microsimulation model. The simulated cohort was newly diagnosed age 45 T2DM patients with an initial 8.5% HbA1c level. Demographic and clinical characteristics were sourced from NHANES data. Two scenarios were simulated. Both scenarios were aligned with Vijan et al., 2014. The first scenario compared the benefit of metformin with and without drift over a stable HbA1c of 8.5%. The second compared metformin with drift to basal-bolus insulin. Drift data were extrapolated from clinical trials.


      Metformin treatment without drift gained 3.02 life years (LY), while with drift the LY gain was only 0.08. For 100,000 simulated patients, metformin without drift prevents 1,960 myocardial infarctions (MIs), 320 strokes, 10,610 cases of blindness, 27,390 lower extremity amputations (LEAs), and 19,010 end stage renal disease cases requiring dialysis (ESRDs). When HbA1c drift was considered, the benefits of metformin treatment shrank to no additional strokes, prevention of 500 cases of blindness, 700 LEAs, and 410 ESRDs, with an increase of 100 MIs. The direct comparison of LY of metformin with drift and insulin treatment showed that the insulin gained 2.51 LY over metformin. Compared with metformin treatment with drift, insulin treatment prevented more complication events including 2,110 MIs, -140 strokes, 8,340 cases of blindness, 20,450 LEAs, and 16,390 ESRDs.


      HbA1c drift in oral anti-diabetic drugs has substantial effects on clinical outcomes and should be incorporated into diabetes health outcome models.