A pilot study on the quality of data management in a cancer clinical trial.

Twelve institutional data managers were asked to independently code the data from a patient chart of one patient in an ovarian cancer trial. They abstracted data from the medical record and filled out three types of trial forms (on-study, chemotherapy, and summary forms). The analysis of the processed data revealed that the median rate of errors was 13% for the 12 data managers. The error rate differed among the types of trial forms. The factors causing these errors were mistakes in interpretation, documentation, and coding. The level of experience of the data managers proved to be an important factor. It became clear that the documentation in the medical record was inadequate. We conclude that data managers as well as physicians involved in cancer clinical trials need specific training and that the quality of data management in cancer clinical trials is an important issue for further investigation.