Well-designed data flows

Document and implement well-designed data flows.

The ultimate aim of clinical trials is to generate new knowledge based on accurate trial data. This often requires integration of data from multiple sources, including clinical data, lab data, and health economics data. These data will be generated at different locations and times, as well as different formats that may change over time. This makes robust, flexible and locally-appropriate data management systems and processes important, and well-qualified data management staff essential.

The main STREAM trial database was developed in-house by MRC CTU at UCL, which has a dedicated programming team. This was important for STREAM, as the programmers could make database changes quickly to address protocol amendments, modifications in data formats and new case report forms (CRFs). In addition, the database was built with robust internal checks to ensure validity and reliability of data being entered at site-level. Having a dedicated data management team to develop and document efficient data flows was also essential. For example, STREAM microbiology data transfers were required bi-directionally – from MRC CTU at UCL to ITM (information on randomized patients, follow-up visits, and local cultures) and from ITM to MRC CTU at UCL (DST and sequencing results). These transfers required close coordination and communication between data management teams at MRC CTU at UCL and ITM to ensure data could be linked to trial outcomes and were consistent in content and format.

Ensure the data entry system has capacity to detect errors in real time to minimize the number of queries to resolve later.

Site-level Data Entry Operator, STREAM Trial

STREAM’s data management systems also needed to cater to conditions at trial sites. For example, STREAM used paper, rather than electronic CRFs to collect data because some STREAM sites experienced unreliable internet access and might have struggled with an internet-based e-CRF system. Appropriate staffing at trial sites was also a key component in efficient site-level data management. Site Data Managers not only input data, but also manage and resolve data queries raised at the central level. The STREAM experience indicates that – to fulfill these requirements – the ideal Data Manager will have clinical or clinical trial experience and English language proficiency, as well as data entry skills. This enables them to understand and resolve data queries, interacting with the Sponsor and clinicians, where necessary.

Recommended best practices

Design and manage efficient and accurate data systems and flows through the following measures:

  • Clearly map all data sources prior to trial initiation and documenting data transfers in data transfer agreements

  • Where possible, maintain a dedicated database programming and data management team at Sponsor level

  • Pilot data transfers/flows from different origins to help anticipate and identify issues

  • Ensure sites understand the roles, responsibilities and skills/qualification required for site-level data managers so that appropriately qualified staff are hired

  • Hold regular meetings of data management and clinical teams at trial sites to improve communication and coordination and site-level data management, especially query resolution