Improvement of Data Quality and Reduced Time for Data Clearing via Computer-Assisted Plausibility Checks

Abstract

Due to the complexity of studies with peripheral artery disease in the past, data clearing for such studies lasted intolerably long. This could partly be improved by increasing on-site monitoring frequency and duration. In order to improve data quality and reduce time for data clearing further, a computer-assisted plausibility check program was established prior to initiating a multicenter, multinational study concept (PARTNER concept) (1).

In four completed studies (total N = 512 in 36 centers all over Europe) the number of correction lists and inquiries due to open questions after completion of the recruitment period could be remarkably lowered in comparison to previous studies where no computer-assisted plausibility check was employed. The number of missing values was reduced to a minimum (100% of the data with regard to both primary endpoints were available). The number of correction sheets that had to be filled in by the investigator was 3.4 sheets per patient in the study with early plausibility checks, whereas the number was 5.1 sheets per patient in the study where plausibility checks were started later (the average of items to be corrected was 10 in the study with early checks versus 15 per correction sheet with later checks).

A further improvement of data quality was gained by training monitors and investigators on the plausibility program before starting succeeding studies with a similar design. The number of correction lists could be further reduced to two sheets and only an average of three items per sheet per patient. Thus, the time and manpower necessary for correcting data again was remarkably reduced. It has been shown that establishing a protocol to define plausibility prior to starting a clinical trial is a cost-effective way to reduce the time for data clearing of a study and that this should be accompanied by training monitors and investigators.