Rates and Correlates of Incident Type 2 Diabetes Mellitus Among Persons Living With HIV-1 Infection
Yuanfan Ye, Sadeep Shrestha, Greer Burkholder, Anju Bansal, Nathaniel Erdmann, Howard W. Wiener, Jianming Tang
Abstract
The prevalence of various comorbidities continue to rise in aging persons living with HIV-1 infection (PLWH), and our study here aimed to assess the rates and correlates of incident type 2 diabetes mellitus (T2DM) in PLWH from a retrospective, southeastern U.S. cohort. Based on electronic health records, we examined patient demographics, body mass index (BMI), HIV-1-related outcomes, hepatitis C virus co-infection, common comorbidities (e.g. shingles and asthma), usage of protease inhibitors, and usage of statins as potential correlates for T2DM occurrence. Among 3,975 PLWH with ≥12 months of follow-up between January 1999 and March 2018, the overall rate of incident T2DM was 135 per 10,000 person-years, almost 2-fold higher than the rate reported for the general U.S. population. In multivariable models (354 T2DM patients and 3,617 control subjects), sex, BMI, nadir CD4 + T-cell count, HIV-1 viral load (VL) and duration of statin use were independent correlates of incident T2DM (adjusted P <0.05 for all), with clear consistency in several sensitivity analyses. The strongest associations (adjusted odds ratio/OR >2.0 and P <0.0001) were noted for: i) statin use for ≥6 months (OR = 10.2), ii) BMI ≥30 kg/m 2 (OR = 3.4), and iii) plasma VL ≥200 copies/ml (OR = 2.2). Their collective predictive value was substantial: the C-statistic for area under the receiver operating characteristics curve was 0.87 (95% CI = 0.84-0.91), showing close similarity between two major racial groups (C-statistic = 0.87 for African Americans and 0.91 for European Americans). Overall, these findings not only establish a promising algorithm for predicting incident T2DM in PLWH but also suggest that patients who are obese and use statins should require special consideration for T2DM diagnosis and prevention.