MOPE070 - Poster Exhibition
Re-assessing the cardiovascular risk of abacavir in the Swiss HIV Cohort Study (SHCS) using a flexible marginal structural model
J. Young1, Y. Xiao2, M. Abrahamowicz3, E.E.M. Moodie3, M.B. Klein4, P. Vernazza5, E. Bernasconi6, A. Cusini7, R. Weber8, H.C. Bucher1,9
1University Hospital Basel, Basel Institute for Clinical Epidemiology & Biostatistics, Basel, Switzerland, 2Institut National d'Excellence en Santé et en Services Sociaux, Montreal, Canada, 3McGill University, Department of Epidemiology, Biostatistics, and Occupational Health, Montreal, Canada, 4McGill University Health Centre, Division of Infectious Diseases and Immunodeficiency Service, Montreal, Canada, 5Cantonal Hospital St Gallen, Division of Infectious Diseases and Hospital Epidemiology, St. Gallen, Switzerland, 6Regional Hospital of Lugano, Lugano, Switzerland, 7University Hospital Bern and University of Bern, Division of Infectious Diseases, Bern, Switzerland, 8University of Zurich, Division of Infectious Diseases and Hospital Epidemiology, Zurich, Switzerland, 9University Hospital Basel, Division of Infectious Diseases and Hospital Epidemiology, Basel, Switzerland
Background: Recent but not cumulative exposure to abacavir has been shown to increase the risk of cardiovascular disease in some studies. Attempts to find a mechanism that explains this increase in risk have proved inconclusive.
Methods: We assessed the risk of a composite cardiovascular endpoint - a myocardial infarction, cardiovascular related death or invasive cardiovascular procedure - in SHCS patients exposed to abacavir. We fitted a new Cox model that estimates a flexible curve representing the effect of cumulative exposure while accounting for time dependent confounding from cardiovascular risk factors.
Results: Between April 2000 and October 2012, 11,625 patients were assessed for cardiovascular risk, 4474 patients were exposed to abacavir, and 350 patients experienced the composite endpoint. In a conventional Cox model, hazard ratios (95% confidence interval) for the effect of abacavir on the composite endpoint were: 1.04 (0.99-1.09) for cumulative exposure (per year) and 1.52 (1.13-2.04) for any use within the last 6 months. The equivalent estimates reported by the D:A:D study were 1.03 (0.96-1.10) and 1.63 (1.30-2.04) respectively. A new Cox model with a curve based on a one knot spline constrained to equal zero at both ends of a period of influence (Figure) provided a better fit to these data than other curve functions. The constraint at the beginning of the period of influence implies a lag before exposure has an effect. We therefore fitted an alternative conventional Cox model for the effect of abacavir: 1.04 (0.99-1.09) for cumulative exposure (per year), 3.36 (2.04-5.53) for any use within the last 1 to 6 months, and 0.42 (0.25-0.69) for current use.
Conclusion: Current exposure to abacavir does not appear to present a risk; rather exposure in the previous 5 to 20 months has the most impact on cardiovascular risk, suggesting a mechanism other than acute inflammation is responsible.
Download the e-Poster
Back to the Programme-at-a-Glance