Competing risk analysis in a large cardiovascular clinical trial: An APEX substudy. (24th August 2017)

• Record Type: Journal Article
• Authors: Arbetter, Douglas F.; Jain, Purva; Yee, Megan K.; Michalak, Nathan; Hernandez, Adrian F.; Hull, Russell D.; Goldhaber, Samuel Z.; Harrington, Robert A.; Gold, Alex; Cohen, Alexander T.; Gibson, C. Michael
• Is Part Of: Pharmaceutical statistics. Volume 16:Number 6(2017)
• Page Start: 445
• Page End: 450
• Publication Date: 2017-08-24
• Subjects: cardiology -- clinical trials -- competing risks -- proportional hazards -- venous thromboembolism; Pharmacy -- Statistical methods -- Periodicals; Pharmacy -- Statistics -- Periodicals; 615.10727
• DOI: 10.1002/pst.1823
• Languages: English
• ISSNs: 1539-1604

Abstract

Summary: Competing risk methods are time‐to‐event analyses that account for fatal and/or nonfatal events that may potentially alter or prevent a subject from experiencing the primary endpoint. Competing risk methods may provide a more accurate and less biased estimate of the incidence of an outcome but are rarely applied in cardiology trials. APEX investigated the efficacy of extended‐duration betrixaban versus standard‐duration enoxaparin to prevent a composite of symptomatic deep‐vein thrombosis (proximal or distal), nonfatal pulmonary embolism, or venous thromboembolism (VTE)–related death in acute medically ill patients (n = 7513). The aim of the current analysis was to determine the efficacy of betrixaban vs standard‐duration enoxaparin accounting for non‐VTE–related deaths using the Fine and Gray method for competing risks. The proportion of non‐VTE–related death was similar in both the betrixaban (133, 3.6%) and enoxaparin (136, 3.7%) arms, P = .85. Both the traditional Kaplan‐Meier method and the Fine and Gray method accounting for non‐VTE–related death as a competing risk showed equal reduction of VTE events when comparing betrixaban to enoxaparin (HR/SHR = 0.65, 95% 0.42‐0.99, P = 0.046). Due to the similar proportion of non‐VTE–related deaths in both treatment arms and the use of a univariate model, the Fine and Gray method provided identical results to the traditional Cox model. Using the Fine and Gray method in addition to the traditional Cox proportional hazards method can indicate whether the presence of a competing risk, which is dependent of the outcome, altered the risk estimate.