Improve Edit Check Specification Through Clinical Data Discrepancy Analysis

During Clinical trials, usually lots of different data modules will be collected, such as Adverse Event, Medical History, Demography, etc. Data discrepancies will be generated when there are differences between data entered and expected data standard. Currently, most discrepancies are produced by edit check specifications, which help to detect and solve the data issue quickly. However, there are still some shortcomings in the edit check specifications, such as duplicate edit check, wrong edit check, etc.Objective:Analyze the data modules with high discrepancy rate; explore the optimization of edit check specifications. Under the premise of ensuring data consistency, integrity and accuracy, reduce the number of unnecessary edit check specification and improve data managers’working efficacy.Method:1. Collect all clinical trial data completed from 2011 to 2013 in a large multinational company. Collect the study number, clinical phase, completed year and therapeutic areas, number of patients and data points, etc.2. Compare the data discrepancy rate under different clinical phase, therapeutic area and completed year.3. Analyze the edit check specifications set up for the data module and corresponding data discrepancies; propose the optimization through below three areas:system optimization, procedure simplification and balance concern between automatic check and manual review.4. For the proposal of each module, pick up some clinical trial with moderate patient number and representative module discrepancy rate, do some retrospective validations. Check whether it can reduce edit check specification number and unnecessary discrepancy after applying the proposal.Result:1. Under different completed year, clinical phase and therapeutic area, data modules with high discrepancy rate are mainly focused on Adverse Event, Concomitant Drug, Dosing and Lab. Range of AE discrepancy rate is from 4%-15%, range of concomitant drug is 7.5%-27.6%, range of dosing is 1.3%-18.3%, range of lab is 0.6% to 23.7%.2. For the edit check specification set up for the four data modules, suggest optimize it through below three aspects:① system optimize the related data, free text and missing data on the case report form, so that the data issue can be reduced during the data entry process, which can reduce the time for data cleaning. ② For some edit check specifications set up for same data point, eliminate the duplicate or similar check to simplify it. ③ It’s not true that more edit check, the higher data quality. In real study, the balance between edit check and manual review should take into consideration. For the lab module, suggest replace the edit check to manual review (Lab review tool).3. For Adverse Event, study A can reduce 22.45% edit check specification by using this proposal. For Concomitant drug module, study B can reduce 26.3% edit check specification by taking this suggestion. For dosing module, study C can reduce 21.1% edit check specification. For lab module, study D can reduce 16.7% edit check specification.Conclusion:1. Data modules with high discrepancy rate are mainly adverse event, concomitant drug, dosing and lab, through which more emphasis will be put on the four modules during the development of clinical data management plan.2. After the optimization, numbers of unnecessary edit check specification will be reduced; also the workload of data managers can be lowered, which ensures the data consistency, accuracy and integrity. It provides valuable experimental evidence and reference for the standardization of edit check specification in clinical data management.