Persons infected with HIV have a higher risk of cardiovascular disease (CVD) after adjustment for traditional risk factors. Despite this increased risk, HIV is not accounted for in traditional CVD risk calculations or cholesterol guidelines.
We assessed 10-year CVD events in veterans infected with HIV using the Veterans Affairs (VA) Clinical Case Registry (CCR) from 2001-2010. Baseline (1998-2000) laboratory, comorbidity, and medication data were used to determine patient risk scores according to both the Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) model and the American College of Cardiology/American Heart Association 2013 guidelines using the Pooled Cohort Equations (PCE). Veterans with prior history of CVD, low density lipid-cholesterol (LDL-C) <70 or >190, diabetes with LDL-C>70, receiving statins, and all women were excluded. Events were defined per respective risk model (myocardial infarction [MI], bypass/angioplasty, stroke, carotid artery endarterectomy or death from coronary heart disease for D:A:D; acute coronary syndrome, MI, stable/unstable angina, revascularization, stroke, transient ischemic attack, or peripheral arterial disease for PCE). Kaplan-Meier analyses were used to compare PCE and D:A:D risk models. We also developed our own model specific to the HIV population using proportional hazards modelling of CCR data and PCE event definitions.
In 3171 male veterans infected with HIV, observed ten-year events numbered 1165 (36.7%) by PCE criteria and 1088 (34.3%) using D:A:D criteria. As shown in the figure (by quintiles of risk score), the D:A:D model performed better than the PCE model for risk of outcome. In our new model, Hepatitis C (HCV) coinfection was associated with 50% increased hazard (adjusted HR 1.495, CI 1.275-1.752) of PCE event. HIV viral load (aHR 1.062, CI 1.033-1.092) was significantly associated with risk of outcome while CD4 count and CD4 nadir were not. Traditional risk factors were also incorporated into the model, with older age and systolic blood pressure demonstrating significant association with increased hazard of outcome.
There was a high rate of ten-year observed CVD events in HIV-infected veterans. The D:A:D model had better discrimination than the PCE for risk of outcome. Our new model additionally takes into account viral load and HCV-coinfection, which were important risk factors for PCE events.