Method On Analysis And Classification Of Frequently Occurred Data Issues In Data Management Of Clinical Trials

OBJECTIVETo reveal the status of the data quality of clinical trials so as to provide a reference to data manager on their data managements works. To summarize the classification of clinical data issues by investigating the amount and class of source clinical data. Finally, to explore the best practice to improve the clinical data management based on experience of historical.METHODIn this study, we collect data of9phase IV clinical trials. And detect data issues according to the number of variables involved and the modules they belong to.RESULTMissing data is the most common type of data issues. Among all the Case Report Form modules, medical history (25.00%) and drug information (24.32%) are the main sources of data issues. Abnormal data is often found in quantitative data with units and data with two options like demography and vitals. The type of logistic issues is more complex.Common logistic issues can be divided into follows:derived data related issues, description related data issues, status related data issues, common sense related data issues and time related data issues. Derived data related issues only occur in demography information with frequency of38which is closely related to the calculation of age in demographic module. Description related data issues occurred in adverse events, physical examination, and summary information of the frequency, respectively346,32, and20. Status related data issues occur in the adverse event, medical history, concomitant medications and summary information. Common sense related data issues occur in vital signs module. Issues occurred in adverse events are complex. The proportion (6.96%) of description related data issues occurred in adverse events is higher than the status related data issues (1.31%) and time related data issues (1.17%). The status related data issues occurred in medical history module have the proportion as high as10.90%. In addition, data about date and time are more likely to be problematic and their type of issues is more complex.Provide a reference for improving the efficiency of data edit checks and SAS programming by reviewing historical studies. CONCLUSIONIn the data management process, we should detect data according to their attributes and the modules they belong to. Attention should be paid to medical history data, test drug information data and adverse event data since they contribute large proportion to the data issues in Phase IV studies.