You are here
VALIDATION OF A CHRONIC KIDNEY DISEASE RISK SCORE IN HIV+ PATIENTS IN THE US
Anthony Mills1, Kathy L. Schulman2, Jennifer Fusco2, Ricky Hsu3, Andrew Beyer4, Girish Prajapati4, Karam Mounzer5, Gregory Fusco2
1Southern California Men's Medical Group, Los Angeles, CA, USA,2Epividian, Durham, NC, USA,3AIDS Healthcare Foundation, Los Angeles, CA, USA,4Merck & Co, Inc, Kenilworth, NJ, USA,5Philadelphia FIGHT, Philadephia, PA, USA
The risk of chronic kidney disease (CKD) increases with both HIV infection and aging, substantially complicating clinical decision-making. Our objective was to assess the validity, in an exclusively U.S. based cohort, of an easy-to-calculate CKD risk score, developed using data from the Data collection on Adverse events of Anti-HIV Drugs (D:A:D), a prospective, international, multi-site study.
HIV+ patients with no previous exposure to potentially nephrotoxic antiretroviral agents and ≥3 estimated glomerular filtration rate (eGFR) test results were identified in the OPERA® Observational Database, an aggregation of prospectively collected electronic medical records from HIV caregivers in 79 clinics across 15 states. Patients were followed from the first observed eGFR >60 ml/min/1.73 m2 (2002-2015) until last eGFR test result, the occurrence of the study outcome, lost to follow-up, or study end (31JUL2017). The study outcome was defined as a confirmed (≥2 consecutive results, >90 days apart) decrease in eGFR to < 60 ml/min/1.73m2. Three cohorts were drawn independently using Cockcroft-Gault (CG), MDRD and CKD-EPI eGFR estimation methods. Both full and short D:A:D risk scores were applied. Poisson models estimated incidence as a function of D:A:D risk score. Kaplan Meier survival curves estimated progression at five years. Incidence rate ratio's (IRR), adjusted IRR (aIRR), and Harrell's discrimination statistic were used to assess validity.
After applying study eligibility criteria, there were 19444, 22727 and 22748 patients in the CG, CKD-EPI and MDRD samples, respectively. Median short and full risk scores were -3 in all three OPERA cohorts with very similar IQR. CKD incidence (95% CI) ranged from a low of 7.3 per 1000 person years (6.8, 7.9) in OPERA CG to a high of 11.0 (10.4, 11.6) in OPERA MDRD. While overall incidence was higher than observed in the D:A:D derivation cohort at 6.2 (5.7 - 6.7), IRR's by risk group were similar. Using the full risk score, the aIRR was 1.3 in all three OPERA cohorts, regardless of eGFR method, equivalent to the D:A:D derivation cohort. Harrell's c-statistic ranged from 0.87 to 0.92 in the three OPERA cohorts, comparable to that reported by D:A:D (0.92). Similar findings were observed after applying the D:A:D short risk score [Table].
This study supports the validity of the D:A:D short and full risk scoring method for assessing the probability of CKD in an exclusively U.S. based cohort regardless of eGFR method.