To address questions regarding stroke during HIV infection we developed a stroke adjudication protocol for HIV cohort research that enables comparisons with traditional cohort studies, addresses issues specific to HIV, and allowed us to examine factors associated with stroke in HIV.
CNICS is a U.S multisite clinical cohort of HIV-infected patients receiving longitudinal HIV care. The CNICS stroke protocol was based on a modified traditional protocol (Multi-Ethnic Study of Atherosclerosis). Potential events at 5 CNICS clinical sites through 12/2012 were identified. Case identification criteria included a range of diagnoses and procedure codes such as cerebral angiography. For each potential event, the site assembled de-identified packets with provider notes, imaging results, and ECGs. Antiretroviral medication exposure was redacted to allow blinded review. Using standardized criteria in the setting of ongoing quality control, two neurologists reviewed each packet, followed by a 3rd reviewer when discrepancies occurred, and categorized each stroke as Definite, Probable, or Possible. Stroke types and subtypes and whether the event was related to infection or illicit drug use was determined. Multivariable Cox regression analyses were used to determine traditional and HIV-specific risk factors for stroke comparing patients with and without stroke followed from enrollment at these 5 sites (N=16,924).
Among 500 potential adjudicated events, 175 (35%) had a stroke. Ischemic strokes made up 81% of events, 10% were hemorrhagic, and in 9% the type was unidentifiable. Ischemic stroke subtypes included large vessel atheroembolic (19%); cardioembolic (28%); small vessel (29%); and other/unknown subtypes (23%). Strokes occurred in the setting of illicit drug use in 19% and infection in 20%. The case-fatality rate was 9%. In addition to traditional risk factors, HIV-specific factors (lower CD4 count and higher viral load) were associated with stroke (see Table).
Strokes were predominantly ischemic and were associated not only with traditional risk factors but with lower CD4 count and higher viral load suggesting potential additional benefits of earlier antiretroviral treatment initiation. Standard adjudication protocols facilitate cross-cohort comparisons but require modification to address issues such as identification of strokes related to infection or illicit drug use that is more common in HIV-infected patients.