Differential case reporting in a national clinical quality database: an analysis of the impact on comparison of outcome between departments.

INTRODUCTION: Selection biases due to difference in reporting may cause spurious findings. The purpose of this study was to illustrate the effect of case incompleteness on the differences in 180-day survival rate when comparing departments.
METHODS: Completeness was estimated as the proportion of patients reported to the Danish Lymphoma Database compared with the National Patient Registry. The effect of differential reporting between departments was investigated using plots were the proportion of patients who survived 180 days in the individual departments were compared with the national average as well as examining the association between the department and 180-day survival.
RESULTS: In total, 8015 patients were registered in the National Patient Registry with a lymphoma diagnosis. Of these, 1824 patients were not reported to the Danish Lymphoma database equaling a rate of completion of 77.2%. There were large differences between departments with regard to patient completeness (range, 52.4%-91.6%). The survival of the patients reported to the database was significantly better than those not reported and the differential reporting affected the department ranking.
CONCLUSION: Differential case reporting may cause spurious findings. To ensure valid conclusions validation against a gold standard and continuous feedback to data providers is important.