目的 基于诊断准确性研究报告规范2015（STARD2015）评价国内诊断准确性研究质量。 方法 检索中国知网、万方数据知识服务平台的中文核心期刊，收集2017—2022年发表的诊断准确性研究文献，主要检索词为诊断试验、敏感度、特异度、受试者工作特征（ROC）曲线等。根据纳排标准筛选文献，2位评价者独立评价纳入文献的质量，计算各篇文献STARD2015报告符合率及STARD2015各条目符合率，根据文献STARD2015报告符合率将文献报告质量分为低中高3个水平，统计各水平文献比例及各年份处于中、高水平文献比例。根据所报告条目的内容是否完善将其分为规范和未规范报告，统计未规范报告条目文献比例。采用组内相关系数（ICC）分析2名评价者评估文献一致性；趋势卡方检验分析各年份中、高水平文献比例变化趋势；各年份文献STARD2015报告符合率多组间比较采用单因素方差分析。 结果 共纳入2017—2022年诊断准确性研究文献6 771篇，STARD2015报告符合率为39.56%±4.90%，最低为17.65%，最高为64.71%（报告项目数范围6~22项），93.53%（6 333/6 771）文献处于中水平。不同年份文献的STARD2015报告符合率差异有统计学意义（F=25.023，P<0.01），2021年文献报告符合率高于其他年份（P均<0.01）；各年份报告处于STARD2015中、高水平文献比例呈上升趋势（χ 2=14.099，P<0.01）。STARD2015中各条目报告情况差异较大，条目符合率0~100%。在报告条目的文献中，10.34%（569/5 503）的文献对条目6、4.15%（277/6 677）的文献对条目8、21.84%（1 447/6 626）的文献对条目20、66.67%（24/36）的文献对条目22、26.03%（877/3 369）的文献对条目26未规范报告。 结论 国内2017—2022年发表的诊断准确性研究文献对STARD2015的总体报告质量呈中等水平，各条目符合率差异较大，国内研究人员对STARD2015认识不足，需加强STARD2015推广。 Objective The quality of domestic diagnostic accuracy research was evaluated to explore the quality level of domestic diagnostic accuracy research based on Standards for Reporting of Diagnostic Accuracy 2015 (STARD2015). Methods The Chinese core journals of CNKI and Wanfang Database were searched, and research literature on diagnostic accuracy published from 2017 to 2022 were collected. The main search terms are diagnostic test, sensitivity, specificity, receiver operating characteristic (ROC) curve, etc. The literature was selected according to the inclusion criteria, and the quality of the included literature was evaluated independently by two reviewers, and the conformity rate of the STARD2015 report and the STARD2015 article was calculated. The report quality of the literature was divided into three levels (low, medium and high) according to the conformity rate of the STARD2015 report. The proportion of literature at each level and the proportion of literature at medium and high level in each year were counted. According to quality analysis on the contents of the included articles, articles were divided into standardized reports and unstandardized reports. Intragroup correlation coefficient (ICC) was used to analyze the consistency of two reviewers. The trend Chi-square test was used to analyze the trend of the proportion of medium and high level literature in each year. One-way analysis of variance was used to compare the coincidence rates evaluated by STARD2015 for each year. Results A total of 6 771 studies on diagnostic accuracy published from 2017 to 2022 were included. The compliance rate sccording to STARD2015 was 39.56%±4.90%, and the reported compliance rate ranged from 17.65% to 64.71% (the number of reported items ranged from 6 to 22), and 93.53% (6 333/6 771) literatures were in the middle level. Compliance rate of STARD2015 reports varied significantly among different years (F=25.023, P<0.01), and the compliance rate of 2021 was significantly higher than that of other years (P<0.01). The proportion of medium and high level literatures according to STARD2015 showed an increasing trend (χ 2=14.099, P<0.01). The reporting situation of each item varied significantly, and the conformity rate of items raned from 0 to 100%. According to report item, non-standard report rate was 10.34% (569/5 503) for item 6, 4.15% (277/6 677) for item 8, 21.84% (1 447/6 626) for item 20, 66.67% (24/36) for item 22, and 26.03% (877/3 369) for item 26. Conclusions The overall report quality of published domestic literature on diagnostic accuracy from 2017 to 2022 is at a medium level according to STARD2015, and the reports conformity rate of each item vary significantly, indicating significant knowledge gap on STARD2015 among domestic researchers. The promotion of STARD2015 needs to be strengthened.
Quality evaluation on diagnostic accuracy research in China from 2017 to 2022 based on STARD2015
Objective The quality of domestic diagnostic accuracy research was evaluated to explore the quality level of domestic diagnostic accuracy research based on Standards for Reporting of Diagnostic Accuracy 2015 (STARD2015). Methods The Chinese core journals of CNKI and Wanfang Database were searched, and research literature on diagnostic accuracy published from 2017 to 2022 were collected. The main search terms are diagnostic test, sensitivity, specificity, receiver operating characteristic (ROC) curve, etc. The literature was selected according to the inclusion criteria, and the quality of the included literature was evaluated independently by two reviewers, and the conformity rate of the STARD2015 report and the STARD2015 article was calculated. The report quality of the literature was divided into three levels (low, medium and high) according to the conformity rate of the STARD2015 report. The proportion of literature at each level and the proportion of literature at medium and high level in each year were counted. According to quality analysis on the contents of the included articles, articles were divided into standardized reports and unstandardized reports. Intragroup correlation coefficient (ICC) was used to analyze the consistency of two reviewers. The trend Chi-square test was used to analyze the trend of the proportion of medium and high level literature in each year. One-way analysis of variance was used to compare the coincidence rates evaluated by STARD2015 for each year. Results A total of 6 771 studies on diagnostic accuracy published from 2017 to 2022 were included. The compliance rate sccording to STARD2015 was 39.56%±4.90%, and the reported compliance rate ranged from 17.65% to 64.71% (the number of reported items ranged from 6 to 22), and 93.53% (6 333/6 771) literatures were in the middle level. Compliance rate of STARD2015 reports varied significantly among different years (F=25.023, P<0.01), and the compliance rate of 2021 was significantly higher than that of other years (P<0.01). The proportion of medium and high level literatures according to STARD2015 showed an increasing trend (χ 2=14.099, P<0.01). The reporting situation of each item varied significantly, and the conformity rate of items raned from 0 to 100%. According to report item, non-standard report rate was 10.34% (569/5 503) for item 6, 4.15% (277/6 677) for item 8, 21.84% (1 447/6 626) for item 20, 66.67% (24/36) for item 22, and 26.03% (877/3 369) for item 26. Conclusions The overall report quality of published domestic literature on diagnostic accuracy from 2017 to 2022 is at a medium level according to STARD2015, and the reports conformity rate of each item vary significantly, indicating significant knowledge gap on STARD2015 among domestic researchers. The promotion of STARD2015 needs to be strengthened.
万方数据
