Topics to Cover in a Data Management Plan

1. CRF design 2. Study setup 3. Tracking CRF data (CRF workflow) 4. Entering data 5. Cleaning data 6. Managing lab data 7. SAE handling 8. Coding reported terms 9. Creating reports and transferring data 10. Closing studies

Figure 1.1 Elements of the data management process that should be addressed in a data management plan.

Revising the DMP How big a change must it be to appear as a revision in the DMP? Is normal change control sufficient or does it warrant attention in the plan? If a document referred to in the DMP is updated, does its version number have to be updated in the DMP? If an appendix is updated, does the DMP, as a whole, require a revision with a new version number? Who needs to sign off on revisions? Using plans with CROs An experienced data manager from the sponsor company should expect to review in detail, and sign off on, the CROs DMP. The CRO should explain what the process will e for revising the DMP during the course of the study. It is the sponsors responsibility to allocate resources to get the initial DMP and all revisions reviewed and signed in a reasonable period of time. Quality assurance and DMPs Quality assurance (QA) is the prevention, detection, and correction of errors or problems. In biopharmaceutical firms, QA is closely tied to regulatory compliance because good practice must be closely tied to following regulations. SOPs for DMPs and study files Using data management plans The work to be done and responsibilities are clearly stated at the start of the study so that everyone knows what is expected. The expected documents are listed at the start of the study so they can be produced during the course of, rather than after, the conduct of the study. The document helps everyone fulfill regulatory requirements. Data management tasks become more visible to other groups when the DMP is made available to the project team. The DMP provides continuity of process and a history of a project. This is particularly useful for long-term studies and growing data management groups. CRF design considerations At some companies, the first draft of the CRF is created by a clinical research associate; at others, data managers prepare the initial version. Even if data managers are not responsible for the creation of the CRF, they should be closely involved in its review and approval as part of a cross-functional team. A cross-functional team is the only way to design a

CRF that will collect the necessary data in a way that is clear and easy for the investigator, efficient for data management processing, and appropriate for analysis.

Data cleaning issues Discrepancies in data is time-consuming to identify, track, and resolve especially when they involve a query to a site. Therefore, working on a CRF design that reduces the need for checks on fields that are administrative only (that is, they do not actually add to the quality of the data) has a lot of value to data management. Asking the investigator to provide duplicate or repeat information in different locations is one such source of low-value queries. Other low-value queries will be generated if the CRF designer does not make adequate provision for blank or unavailable responses or permits ambiguous responses. Duplicate data CRFs frequently include questions that are intentionally used as cross-checks to other fields. This can be a very good policy when the data is actually different. For example, cleaning programs can check for logical consistency by comparing the sex of the patient to the patients pregnancy status. Problems do arise, however, when cleaning programs cross-check values that are actually duplicates of the same data. For example, a CRF should not ask for both the patients age and birth date.