Soran A, Nesbitt L, Mamounas E P, Lembersky B, Bryant J, Anderson S, Brown A, Passarello M
NSABP Biostatistical Center, Pittsburgh, PA 15213, USA.
Clin Trials. 2006;3(5):478-85. doi: 10.1177/1740774506070747.
There is a need for data quality assurance procedures in phase III cancer trials. At the National Surgical Adjuvant Breast and Bowel Project (NSABP) 'real-time' systems have been developed for quality assurance and study monitoring: (1) manual review and triage of data forms by data managers at the time of submission; (2) computerized edit checking of all submitted data forms; (3) systematic review of eligibility, treatment compliance and toxicity in the first 100 patients of a new protocol; (4) prospective centralized medical review of all reported serious adverse events, treatment failures, second primary cancers and deaths; (5) quarterly review and approval of study summary data files by project statistician; and (6) on-site auditing.
To assess the utility of an additional final comprehensive review of all patient records to confirm eligibility, disease status and vital status prior to manuscript submission.
Four phase III NSABP studies, which had been monitored using the triage-based quality assurance program described above, were selected for analysis (n = 7972). Charts for 5965 patients were identified that had not been previously medically reviewed for protocol events of recurrence, second primary cancer or death. Submitted source documents and data forms of these 5965 NSABP patient records underwent medical review to verify patient eligibility, disease status and vital status.
This final comprehensive review found no additional treatment failures or deaths, identified seven additional cases of ineligibility, was time-intensive requiring enormous use of expensive resources, and was therefore judged not to add significantly to the integrity of the database.
Our findings are influenced by the procedures the NSABP employs for quality assurance and study monitoring for Phase III clinical trials and may have limited generalizability to other settings.
In the presence of multiple quality assurance and data monitoring systems, the rare discrepancies found between the data forms and source documentation does not support the routine use of a final comprehensive chart review for phase III trials at the NSABP Biostatistical Center.
三期癌症试验需要数据质量保证程序。在国家外科辅助乳腺和肠道项目（NSABP）中，已开发出用于质量保证和研究监测的“实时”系统：（1）数据管理人员在提交数据表格时进行人工审核和分类；（2）对所有提交的数据表格进行计算机化编辑检查；（3）对新方案的前100名患者的入选资格、治疗依从性和毒性进行系统审查；（4）对所有报告的严重不良事件、治疗失败、第二原发性癌症和死亡进行前瞻性集中医学审查；（5）项目统计学家每季度对研究汇总数据文件进行审查和批准；以及（6）现场审计。
评估在提交手稿前对所有患者记录进行额外的最终全面审查以确认入选资格、疾病状态和生命状态的效用。
选择四项使用上述基于分类的质量保证计划进行监测的NSABP三期研究进行分析（n = 7972）。确定了5965名患者的病历，这些病历此前未针对复发、第二原发性癌症或死亡等方案事件进行医学审查。对这5965份NSABP患者记录提交的原始文件和数据表格进行医学审查，以核实患者的入选资格、疾病状态和生命状态。
这项最终全面审查未发现额外的治疗失败或死亡病例，发现了另外7例不符合入选资格的病例，耗时较长且需要大量使用昂贵资源，因此被判定对数据库的完整性没有显著增加。
我们的研究结果受到NSABP用于三期临床试验质量保证和研究监测的程序的影响，可能对其他环境的普遍适用性有限。
在存在多个质量保证和数据监测系统的情况下，数据表格和原始文件之间发现的罕见差异不支持在NSABP生物统计中心对三期试验常规使用最终全面图表审查。
