Accurate Data Entering a Difficult Task

Accurate patient data reporting is essential for the validity of the scientific conclusions based on the international databases (DB) of CIBMTR and EBMT. A study request from EMBT listing patients with CML, raised suspicion that the list was incomplete and that data had not been correctly reported. To identify if data were missing and if so whether the problem was, a) internal (local DB, data entering etc.), b) external (EBMT/CIBMTR DB), c) random, or d) systematic, a quality control survey was initiated.

Aim: To ensure all CML patients were reported correctly and to identify the cause of possible errors.

Methods: Before the 1^st of July 2013 the previous 20 patients with CML treated with allogeneic stem cell transplantation (SCT) were audited for data accuracy (Nov 5^th 2007 - July 1^st2013). The following data were checked for accuracy: diagnosis, date of diagnoses, molecular markers, cytogenetic, pre transplant treatment, and use of thyrosine kinase inhibitors (TKI). The data accuracy was checked in the EBMT, the CIBMTR and the local DB. Data entering have been performed by the EBMT, the CIBMTR and online by the data managers at our institution.

Results: For the 20 CML patients, the diagnoses, the date of diagnosis and cytogenetics were reported correctly in all DBs. In relation to all other data, no error was identified in the CIBMTR DB. However, two of 20 patients had a coding error in relation to molecular markers in the EBMT DB, but not in the local DB. In relation to treatment, two patients had a coding error in the EBMT DB and one patient in the local DB. In relation to the first course of TKI treatment three patients had a coding error in the EBMT DB and one in the local DB. In relation to the second course of TKI eight patients had a coding error in the local DB and two in the EBMT DB.

Conclusion: All data were entered correctly in the CIBMTR DB, however errors were identified in the EBMT and the local DB. Accordingly, the cause of errors is not related to lack of knowledge of CML biology. The identified errors were not systematically since they did not related to all patients. Therefore it can be assumed that the errors were random data entering errors. When analyzing the data entering process in the three different data bases, the CIBMTR form and DB are identical. However, the MED A form is not identically to the EBMT DB and neither of the forms is identical to the local DB. Therefore interpretation of data during the data entering process may be a source for coding errors. In the future regular audit of data are recommended.