School of Epidemiology & Public Health, University of Ottawa, 600, Peter Morand Crescent, Ottawa, Canada K1G 5Z3; Clinical Epidemiology Program, Ottawa Hospital Research Institute, 501, Smyth Road, Ottawa, Canada K1H 8L6.
Department of Medicine, Université de Montréal, 2900, Boulevard Edouard-Montpetit, Montréal, Canada H3T 1J4; Centre de recherche du Centre Hospitalier Universitaire de Montréal, 900, Saint Denis Street, Montréal, Canada H2X 0A9.
J Clin Epidemiol. 2018 Jun;98:9-15. doi: 10.1016/j.jclinepi.2018.02.001. Epub 2018 Feb 9.
The aim of the study was to determine the impact of using the win ratio approach and investigate whether this approach alters the interpretations or conclusions of a randomized trial in kidney transplantation.
We present an application of the win ratio approach in a kidney transplant trial that assessed the clinical effectiveness of ramipril treatment vs. placebo. The primary composite outcome included the time to death, kidney transplant failure, or doubling of serum creatinine. We compare the win ratio to a conventional hazard ratio (HR) from the original trial. A win ratio with a lower 95% confidence limit greater than 1 indicates a positive treatment effect with statistical significance.
For the primary composite end point, ramipril treatment resulted in a win ratio of 1.21 (95% confidence interval [CI], 0.55-2.59) vs. a HR of 0.76 (95% CI, 0.38-1.51). With extended follow-up (mean 48 months), ramipril was associated with a win ratio of 1.02 (95% CI, 0.54-1.83) vs. a HR of 0.96 (95% CI, 0.55-1.65).
The win ratio approach produced results similar to the original time-to-event analysis. Using this approach would not alter the conclusion of the original trial and lessens concerns associated with composite components of unequal clinical importance.
本研究旨在探讨胜率比法的影响，并研究该方法是否改变了肾移植随机试验的解释或结论。
我们介绍了胜率比法在一项评估雷米普利治疗与安慰剂在肾移植中的临床效果的肾移植试验中的应用。主要复合结局包括死亡、肾移植失败或血清肌酐加倍的时间。我们将胜率比与原始试验中的传统风险比（HR）进行比较。胜率比的置信区间下限大于 1 时，表示治疗效果具有统计学意义。
对于主要复合终点，雷米普利治疗的胜率比为 1.21（95%置信区间[CI]，0.55-2.59），而 HR 为 0.76（95% CI，0.38-1.51）。随访时间延长（平均 48 个月）后，雷米普利的胜率比为 1.02（95% CI，0.54-1.83），而 HR 为 0.96（95% CI，0.55-1.65）。
胜率比法的结果与原始时间事件分析相似。使用该方法不会改变原始试验的结论，并减轻了对临床重要性不等的复合成分的担忧。
