PURPOSE In this study, we compared the assessments of progression-free survival (PFS) carried out by the local investigator or by a blinded independent central review in the framework of phase III registration randomized controlled trials (RCT) in oncology. METHODS We carried out a search in the clinicatrials.gov database, looking at the RCTs reporting the results of both independently assessed and investigator-assessed PFS. The hazard ratios (HRs) of investigator-assessed PFS and independently assessed PFS were recorded, and a discrepancy index was obtained by calculating the ratio of their respective HRs. Moreover, we investigated possible factors of discrepancy by analyzing the trials in different groups (by year, by tumor type, by drug type, by study design). RESULTS We analyzed 28 RCTs meeting the search criteria. The estimated mean discrepancy index was 0.98 (confidence interval 0.927-1.032 (n = 32)). Subgroup analysis showed that the confidence intervals in all cases included the value 1, except in the subgroup of studies started in the period 2003-2006. CONCLUSION In phase III oncology trials, we found no significant differences between the hazard ratios estimated by local investigators and those estimated by blinded independent central reviews. A relatively higher variability, in terms of large CI, was found in trials with biological agents.

中文翻译：

---

本地研究者对无进展生存期的评估与三期肿瘤学试验中的独立独立中央评价之间的比较。

目的在这项研究中，我们比较了在肿瘤学的III期注册随机对照试验（RCT）框架下，由当地研究人员或由盲人的独立中央评价对无进展生存期（PFS）进行的评估。方法我们在Clinicalatrials.gov数据库中进行了搜索，查看了报告独立评估和研究者评估的PFS结果的RCT。记录研究者评估的PFS和独立评估的PFS的危险比（HRs），并通过计算它们各自的HRs的比率获得差异指数。此外，我们通过分析不同组中的试验来调查差异的可能因素（按年份，肿瘤类型，药物类型，研究设计）。结果我们分析了符合搜索标准的28个RCT。估计的平均差异指数为0.98（置信区间0.927-1.032（n = 32））。亚组分析显示，所有情况下的置信区间都包含值1，但从2003-2006年开始的研究亚组除外。结论在第三阶段的肿瘤学试验中，我们发现当地研究人员估计的危险比与盲目独立的中央评价所估计的危险比之间没有显着差异。在大的CI方面，在生物制剂试验中发现了相对较高的变异性。我们发现，当地调查人员估计的危险比与盲目独立的中央评估所估计的危险比之间没有显着差异。在大的CI方面，在生物制剂试验中发现了相对较高的变异性。我们发现，当地调查人员估计的危险比与盲目独立的中央评估所估计的危险比之间没有显着差异。在大的CI方面，在生物制剂试验中发现了相对较高的变异性。