Abstract
Drug development can take up to 15 years, costing as much as $11 billion USD, and relies heavily on high-quality data. The goal of this investigation of contract research organizations (CROs) was to assess the impact of data management activities (such as curation, quality assessment and integration) on model-informed drug development (MIDD) deliverables. A survey was sent to a diverse sample of CROs, to evaluate their baseline experience with assessing the data quality of sponsor-provided data and the time required to create analysis-ready datasets. It was distributed to 44 colleagues from 32 companies offering pharmacometrics services, including data management. The survey included 11 questions; 9 were multiple choice and 2 open-ended. Responses were gathered anonymously to ensure confidentiality and intellectual property protection and later shared with all participants. Of the 17 survey respondents, most develop data specifications and create analysis-ready datasets. The majority (65%) said the data they received from sponsors was rarely (< 10%) immediately usable due to improper formatting and quality issues like missing data and inconsistencies. Over 50% cited lack of definition/specifications as the primary reason. Assuming an average programming cost of $250/hour, cleaning client data takes CROs 3 to 24 h, costing between $750 and $6000 per dataset. Significant time is spent on rectifying poor-quality data. Automated data quality assessments can improve efficiency checks, though automation alone cannot resolve all quality issues. Better communication, collaboration, and systematic approaches to address data quality issues involving automation and AI are essential to further improve data quality.