3123 — The Effect of Hypoglycemic Medications on Patients with Type 2 Diabetes
Kheirbek RE, Washington DC VA/GW University; Alemi F, BayPines VA/Georgetown University;
This paper reports the pharmacoepidemiology of hypoglycemic medications. After FDA approves a medication, there is limited data on impact of medications on patients. Since pre-release data are typically gathered on patients that do not have various complications, post release evaluation of medications are important. This study provides data on the effect of hypoglycemic agents on veterans with diabetes, after adjustments have been made for patients’ severity of illness and comorbidities
Diabetic patients were selected based on (1) at least two diabetic inpatient or outpatient visits in last 5 years, and (2) at least 180 days between first and last visit. For each patient, we examined exposure to various hypoglycemic medications within 365 days of their last visit. Severity of patients’ illness was based on Alemi and Walter’s diagnoses-based composite severity index. 17,773 patients met the study criteria. Stepwise logistic regression was used to first explain mortality by the patients’ severity of illness and demographics; subsequently the effect of different oral hypoglycemic and insulin medications was measured.
Patients’ severity of illness correctly classified mortality for 89.6% of the patients and was highly statistically significant (Chi-square = 142.95, p <.0001). Being married (odds = 0.8, 95% confidence interval, C.I., from 0.72 to 0.90), younger (odds = 0.97, 95% C.I. from 0.97 to .98), and white (odds = 0.63, 95% C.I. from 0.55 to 0.71) decreased risk of mortality. Being Hispanic increased risk of mortality when not adjusted for severity but decreased the risk of mortality when adjusted for severity (odds = 0.86, C.I. of 0.76 to 0.97). After adjustments were made for composite severity of illness and patient demographics, the remaining variance in mortality was explained by exposure to five medications: cholorpropamide (odds = 1.003, 95% Confidence Interval, C.I. from 1.000 to 1.006), glipizide (odds = 1.014, 95% C.I. from 1.003 to 1.024), glyburide (odds = 1.034, 95% C.I. from 1.018 to 1.049), insulin (odds = 1.086, 95% C.I. from 1.044 to 1.130) and rosiglitazone (odds = 1.182, 95% C.I. from 1.115 to 1.254). None of the other medications (metformin, acarbose, glimepiride, pioglitazone, tolazamide, repaglinide, troglitazone, and TDPP4) were associated with excess mortality beyond what could be expected from the patients’ severity of illness or demographic characteristics. The paper reports the sensitivity of our conclusions to various assumptions including the examination of 76 classes of non-diabetic medications as possible causes of increased mortality. A simulation study showed that for patients within the same severity subgroup, risk of mortality could be reduced by 10.55%, if optimal combination of diabetic medications were used.
Our study found that some medications increased severity adjusted mortality in patients with type 2 diabetes. These data suggest that it is important and possible to create a decision aid that could predict patients’ response to medications. Such an aid could rely on patients-like-me algorithms to reduce all-cause mortality risks among diabetes patients
Medications Increase Severity Adjusted Risk of Mortality. Response to oral glycemic medications can be accurately predicted for 89.9% of the patients. If patients are given optimal combination of medications and if these optimal combinations work as anticipated, then 30-day survival rate could increase by 10%.