Impact of non-standardized outcome analyses on clinical care of HIV patients: meta-epidemiological study

Lars Hemkens, Katherine Winkel

Status: Abgeschlossen Dauer: Feb. 2016 - Jan. 2017

HIV-infection is treated with highly effective antiretroviral treatment (ART). Healthcare decision makers deem suppression of HIV RNA below a certain blood level highly relevant marker of therapeutic success. However, in clinical trial analysis there is no established standard approach to deal with treatment switches or missing outcome data.This is not just a technical issue of little therapeutic relevance because patients frequently switch or withdraw from a given antiretroviral drug regimen and clinical trials may have missing outcome information for up to 30%, while the virological failure rates are very low (1). Trial analysts frequently counted patients with missing outcome data as treatment failures (missing=failure approach), ignored treatment switches or excluded patients with treatment switches from analysis. This may be further complicated in studies aiming to establish non-inferiority of novel treatments. Overall, all these approaches may lead to biased efficacy assessments and may cause misleading clinical decision making. We aim to conduct a meta-epidemiological study evaluating a large number of recently published trials on ART in HIV using this outcome. Trial characteristics, hypotheses and outcome information will be extracted. Efficacy results will be recalculated using various alternative approaches to deal with missing data. Implications on clinical interpretation of the trial results will be assessed. Finally, we aim to provide healthcare decision makers with an empirical estimate of the impact of non-standardized outcome analyses on clinical care by describing how frequently using a different approach would change the clinical interpretation of trial results.n

Zur Projektübersicht