临床意义与统计学意义结合的临床试验结果分类方法及评价研究

Integration of clinical significance and statistical significance on clinical study results categorization:a Meta-epidemiology study

目的统计学意义在临床试验结果解读中发挥重要作用,但在获得有统计学意义结果时,临床意义的评估常被忽略,本研究尝试提出一种简单且明确的试验结果新分类方法,重点针对已获得统计学阳性的试验,评估其结果是否具有临床意义。方法以2019年美国心脏病学会(ACC)和欧洲心脏病学会(ESC)大会公布结果的临床试验为研究对象,通过荟萃流行病学方法,提取研究水平的特征变量。主要评价指标包括试验设计阶段假设的目标效应值及试验结果所得的观察效应值,基于两者间的差异,对已获得统计学阳性的试验进行细分,识别出临床意义可能不充分的研究;并基于检验把握度的理论提出阈值,作为对该类问题进行识别的基础。结果最终纳入分析的12项临床试验多发表于顶尖专业期刊、具有较高的研究设计与报告质量;观察效应值与目标效应值间有一定相关性(r=0.892),但基于两者间的差异,并聚焦获得了有统计学意义结果的7项试验,其中被分类为临床意义不充分的共2项,按ACC和ESC公布的试验进行划分,有统计学意义却可能无临床意义的研究各1项(1/3和1/4)。结论已获得统计学阳性结果的试验,仍有必要对其临床意义的充分性进行评估,本研究提出一种新的将临床意义与统计学意义结合的分类标准及在其基础上对临床试验结果可靠性进行评估的方法,辅助研究者识别因临床意义不充分导致的潜在风险,为临床研究结果的合理解读提供一定的参考和帮助。

Objective Statistical significance plays an important role in the interpretation of clinical trial results.However,on the basis of obtaining statistical significance,the assessment of clinical significance is often neglected.This study attempted to propose a simple and unambiguous new classification method for study results,focusing on studies with statistical positive findings to evaluate whether the results have clinical significance.Methods Our study subjects were the clinical studies in 2019 ACC and ESC annual meetings.Meta-epidemiology methods were used to extract the characteristic variable from each study.The primary evaluation indicators included target effect-size and observed effect-size.Based on the difference between the two indicators,the studies that had statistical significance were subdivided to identify studies with possible insufficient clinical significance;Furthermore,the theoretical threshold based on power analysis was proposed,which was used as the basis for the interpretation of study results.Results There were 12 clinical studies included in the final analysis.All of them were published on top journals.Those studies had relative high quality on both study design and reporting.The correlation coefficient between the observed and target effect-size was 0.892.Among the 7 studies with statistical significance,two of them were classified as insufficient clinical significance.The counts was 1(1/3)and 1(1/4)for the studies reported in ACC and ESC respectively.Conclusions The achievement of clinical significance is critical even in the study with positive results.This paper proposes a new classification standard that combines clinical significance with statistical significance and further suggests a method to evaluate the reliability of clinical study results in order to assist researchers in identifying potential risks caused by insufficient clinical significance,and provide some reference and help for the reasonable interpretation of clinical study results.