Krepper Daniela, Giesinger Johannes Maria, Dirven Linda, Efficace Fabio, Martini Caroline, Thurner Anna Margarete Maria, Al-Naesan Imad, Gross Franziska, Sztankay Monika Judith
Department of Psychiatry, Psychotherapy Psychosomatics and Medical Psychology, University Hospital of Psychiatry II, Medical University of Innsbruck, Innsbruck, Austria.
Department of Psychiatry, Psychotherapy Psychosomatics and Medical Psychology, University Hospital of Psychiatry II, Medical University of Innsbruck, Innsbruck, Austria.
J Clin Epidemiol. 2023 Oct;162:1-9. doi: 10.1016/j.jclinepi.2023.07.012. Epub 2023 Jul 29.
This review addresses the common problem of missing patient-reported outcome (PRO) data in clinical trials by assessing the current practice of their statistical handling as reported in publications of randomized controlled trials (RCTs) in patients with breast cancer.
We searched PubMed to identify RCTs evaluating biomedical treatments in breast cancer patients with at least one PRO endpoint published between January 2019 and February 2022. Two reviewers independently assessed the eligibility of the publications for this scoping review and extracted prespecified information on missing PRO data and related statistical practices.
Of 1,598 publications identified, 118 trials met the inclusion criteria. Eighty-eight (74.6%) trials reported the extent of missing data, with 11 (9.3%) not containing any missing PRO data. Twenty-one (19.6%) trials explicitly stated the statistical approach for handling missing data, with a preference for single imputation over multiple imputation approaches (57.2%/19.0%). Only six (5.6%) trials reported a sensitivity analysis to examine the extent to the results being affected by changes in assumptions made about missing PRO data.
International efforts to raise awareness of the importance of accurately reporting state-of-the-art handling of missing PRO data are not yet fully reflected in the current literature of breast cancer RCTs.
本综述通过评估随机对照试验（RCT）在乳腺癌患者中的出版物中报告的统计处理的当前实践，来解决临床试验中患者报告结局（PRO）数据缺失这一常见问题。
我们检索了PubMed，以识别在2019年1月至2022年2月期间发表的、评估至少有一个PRO终点的乳腺癌患者生物医学治疗的RCT。两名评审员独立评估这些出版物是否符合本范围综述的纳入标准，并提取有关PRO数据缺失及相关统计实践的预先指定信息。
在识别出的1598篇出版物中，118项试验符合纳入标准。88项（74.6%）试验报告了数据缺失的程度，其中11项（9.3%）不包含任何PRO数据缺失。21项（19.6%）试验明确说明了处理缺失数据的统计方法，更倾向于单一插补而非多重插补方法（57.2%/19.0%）。只有6项（5.6%）试验报告了敏感性分析，以检查结果受关于PRO数据缺失所做假设变化影响的程度。
提高对准确报告PRO数据缺失的最新处理方法重要性的认识的国际努力，尚未在当前乳腺癌RCT文献中得到充分体现。
