我国急性脑卒中临床试验疗效判断方法分析

Methods of Outcome Assessment in Chinese Acute Stroke Trials

目的 了解国内急性脑卒中临床试验疗效判断方法的现状和存在的问题。 方法 对 2 0 0 3年 3月以前发表的国内 6种神经内科杂志进行人工检索。纳入所有研究急性脑卒中治疗措施、自述为随机或半随机对照试验的人体研究报告。排除关于蛛网膜下腔出血和TIA的试验。对疗效判断指标的类型、随访时间、是否采用盲法判断疗效、是否报告失访、统计分析方法及阳性结果率等方面进行分析。 结果 共入选符合纳入标准的试验 2 10个。使用病理水平指标的试验 119个 (5 7% ) ,病损水平指标 16 1个 (77% ) ,2 5个 (12 % )试验采用残疾 /生活能力水平指标 ,但没有试验将数据变换为二项分类资料进行分析。采用残障 /生存质量水平指标的试验为 0个 ,33个 (16 % )试验在结果中报告了死亡人数但多未将病死率设计为疗效指标。随访时间 3h～ 3年 ,中位时间 17d。仅 4个试验 (2 % )报告了失访率。 97%的试验未描述是否采用盲法评价疗效。 199个 (95 % )试验显示出有统计学意义的阳性结果。 结论 国内急性脑卒中临床试验在采用与患者相关的疗效判断指标、盲法评价疗效、随访时间及报告失访率等方面与国外存在较大差异。

Objective To learn if appropriate methods and clinically relevant outcomes were used by analyzing methods of outcome assessment in Chinese acute stroke trials. Methods Randomised and quasi-randomised controlled trials on acute stroke published before March 2003 in 6 Chinese neurological journals were included. Types of outcome measures, blinding of outcome assessment, duration of follow up, statistical methods used for data analysis and the significance of the results were evaluated. Types of outcomes were classified as death and four levels: ① Pathology. ② Impairment. ③ Disability. ④ Handicap/quality of life. Results Two hundreds and ten trials were included in this analysis. 57% of the trials used outcomes in pathology level, 77% in impairment level, 12% in disability level and none in the quality of life level. No dichotomous data was analyzed for disability measures. Only 16% of the trials reported number of death but few of them designed death as an outcome measure. Duration of follow up ranged from 3h to 3 years (median 17 d, interquartile range 14-30 d). Most trials did not assess outcomes blindly. Results in 95% of the trials were favorable to the tested interventions. Conclusions In Chinese acute stroke trials, outcome measures used were mainly in pathology and impairment levels and very few trials used functional outcome or death. Blinding of outcome assessment was not commonly used. The average duration of follow up was short.