Knepper David, Fenske Christian, Nadolny Patrick, Bedding Alun, Gribkova Elena, Polzer John, Neumann Jennifer, Wilson Brett, Benedict Joanne, Lawton Andy
1 Business Operations, Allergan, Jersey City, NJ, USA.
2 Clinical Risk Management, Eli Lilly and Company, Indianapolis, IN, USA.
Ther Innov Regul Sci. 2016 Jan;50(1):15-21. doi: 10.1177/2168479015620248.
Data quality issues in clinical trials can be caused by a variety of behaviors including fraud, misconduct, intentional or unintentional noncompliance, and significant carelessness. Regardless of how these behaviors are defined, they may compromise the validity of the study results. Reliable study results and quality data are needed to evaluate products for marketing approval and for decisions that are made on the use of medicine. This article focuses on detecting data quality issues, irrespective of origin or motive. Early detection of data quality issues are important so that corrective actions taken can be implemented during the conduct of the trial, recurrence can be prevented, and data quality can be preserved.
A survey was distributed to TransCelerate member companies to assess current strategies for detecting and mitigating risks involving fraud and misconduct in clinical trials. A review of literature across many industries from 1985 to 2014 was conducted using multiple platforms.
Eighteen TransCelerate member companies anonymously responded to the survey. All of the respondents had one or more existing strategies for fraud and misconduct detection. The literature search identified current practices and methodologies across many industries.
TransCelerate recommends the creation of an integrated, multifaceted approach to proactively detect data quality issues. Detection methods should include a strategy tailored to the characteristics of the study. Some sponsors are taking advantage of more advanced methods and integrated processes and systems to proactively detect and address issues, relying on advances in technology to more efficiently review data in real time. Further research is underway to assess statistical data quality detection methodology in clinical trials.
临床试验中的数据质量问题可能由多种行为导致，包括欺诈、不当行为、有意或无意的不遵守规定以及重大疏忽。无论如何定义这些行为，它们都可能损害研究结果的有效性。需要可靠的研究结果和高质量的数据来评估产品是否可获批上市以及用于有关药物使用的决策。本文重点关注检测数据质量问题，无论其来源或动机如何。早期发现数据质量问题很重要，这样可以在试验进行期间采取纠正措施，防止问题再次出现，并保持数据质量。
向跨加速（TransCelerate）成员公司分发了一份调查问卷，以评估当前检测和降低临床试验中涉及欺诈和不当行为风险的策略。使用多个平台对1985年至2014年多个行业的文献进行了综述。
18家跨加速成员公司匿名回复了调查问卷。所有受访者都有一项或多项现有的欺诈和不当行为检测策略。文献检索确定了多个行业的当前做法和方法。
跨加速建议创建一种综合的、多方面的方法来主动检测数据质量问题。检测方法应包括根据研究特点量身定制的策略。一些申办者正在利用更先进的方法以及集成的流程和系统来主动检测和解决问题，依靠技术进步更高效地实时审查数据。目前正在进行进一步研究，以评估临床试验中的统计数据质量检测方法。
