目的 调查近5年发表的康莱特注射液治疗非小细胞肺癌与肝癌的随机对照试验（即文中提及“随机”二字）的关键特征，并评价安全性信息的报告质量。方法 计算机检索中国知网、万方、PubMed和Embase数据库收集相关研究，检索时限均为建库至2023年4月11日。纳入2018—2022年发表的相关论文，总结其关键特征。使用CONSORT-Harms评价纳入研究安全性信息的报告质量，并基于多重线性回归分析探索报告质量的预测因素。Excel 2019与Stata 17/SE用于统计分析与绘图。P<0.05与95%置信区间（confidence interval，CI）不包括0为差异具有统计学意义。结果 共纳入59篇发表于2018—2022年的相关研究，全部为中文文献，15篇（25.42%）发表于科技核心期刊。调查非小细胞肺癌的共51篇（86.44%），最常见的干预措施为康莱特注射液＋GP方案（吉西他滨＋顺铂，35.59%）；调查肝癌的8篇（13.56%），最常见的干预措施为康莱特注射液＋肝动脉化疗栓塞（6.78%）。34篇（57.63%）研究报告了正确的随机方法，仅1篇（1.69%）提及了盲法。样本量为30～289例，但没有研究报告样本量的计算方法。论文的作者数量为1～8个，参与的机构数量为1～3个。14篇（23.73%）报告接受了基金资助，但没有研究提及利益冲突声明；同样没有研究呈现受试者流程图和使用图形呈现安全性信息。纳入研究的安全性信息报告均存在明显问题，尤其是CONSORT-Harms清单的条目3、4、5、6和9。多重线性回归分析表明论文页数是安全性信息报告质量的预测因素（β=0.65，95% CI（0.15，1.15），P=0.01）。结论 康莱特注射液治疗非小细胞肺癌和肝癌的随机对照试验的方法学和安全性信息报告均存在较大的问题。建议研究者未来参考相关指南规范以严格设计、规范实施和良好报告此类研究，同时学术期刊应减少对论文篇幅的过度限制，并严格把控研究质量。

Objective To investigate the key characteristics of "randomized controlled trials" (i.e., mentioned the "randomized") of Kanglaite Injection (康莱特注射液) for non-small cell lung cancer (NSCLC) and liver cancer published in the past five years, and to evaluate the reporting quality of harms. Methods Relevant studies were collected by systematically search CNKI, WanFang, PubMed, and Embase databases, and the time span was set from the establishment to April 11, 2023. Relevant articles published from 2018 to 2022 were included to summarize their key characteristics, and the CONSORT-Harms was used to evaluate the reporting quality of harms and predictors of the reporting quality were explored based on multiple linear regression analysis. Excel 2019 with Stata 17/SE was used for data analysis and plotting. P < 0.05 and 95% CI excluding 0 were considered statistically significant. Results A total of 59 relevant articles published from 2018 to 2022 were included, all of which were Chinese literature, and 15 (25.42%) were published on journals indexed in CSTPCD. There were 51 articles (86.44%) about NSCLC, and the mostly reported intervention was Kanglaite Injection + GP (gemcitabine + cisplatin) (35.59%); eight articles (13.56%) were focused on liver cancer, and the mostly reported treatment was Kanglaite injection+TACE (transcatheter arterial chemoembolization) (6.78%). The correct randomization method was reported in 34 (57.63%) studies and blind was mentioned in only one (1.69%) study. The sample size ranged from 30 to 289, but no study reported the calculation method. The number of authors ranged from one to eight, and the number of institutions ranged from one to three. Fourteen (23.73%) studies received funding support, but none of the studies mentioned a declaration of conflict of interest. There were also no studies that presented flow diagrams of subjects and used graphics to present harms. There were significant problems regarding the reporting of harms in the included studies, particularly regarding the items 3, 4, 5, 6, and 9 of the CONSORT-Harms items. Multiple linear regression analysis showed that the number of pages was a predictor of the reporting quality of harms (β=0.65, 95% CI (0.15, 1.15), P=0.01). Conclusion There are major problems in the methodology and harms reporting in randomized controlled trials of Kanglaite Injection for NSCLC and liver cancer. It is suggested that researchers should refer to relevant guidelines in the future to design, implement, and report such studies in a strict, standardized, and clear manner. Meanwhile, academic journals should reduce excessive restrictions on the length of articles and strictly control the quality of relevant studies.

R24；R286.91；G350

中国中医科学院科技创新工程项目（CI2021A00702）