Study finds significant outcomes differences between administrative and clinical registry data

January 30, 2013 – DCRI-led analysis into the accuracy of administrative datasets for pediatric cardiac programs finds significant differences between administrative and clinical registry data for children undergoing cardiac operations.

A recent DCRI-led analysis into the accuracy of administrative datasets for pediatric cardiac programs found significant differences between administrative and clinical registry data for children undergoing cardiac operations. Disagreement between data sources on case ascertainment (when patients in a data cohort are selected for certain attributes) translates into significant differences in outcomes assessment.

Study author Sara Pasquali, MD, MHS, received assistance in her analysis from the DCRI’s Eric Peterson, MD, MPH; Xia He, MS; and Jennifer Li, MD, MHS. The study was published in the January edition of The Annals of Thoracic Surgery.

The researchers compared data from the Society of Thoracic Surgeons Congenital Heart Surgery database (clinical registry) and the Pediatric Health Information Systems database (administrative database). The 33 hospitals selected for this study entered all patient information into both databases. The overall cohort included more than 59,000 patients, from 0 to 18 years of age, who had undergone a cardiac operation.

In-hospital mortality was used as the primary outcome for comparison of the two databases. Since the patients compared in the databases were the same, disparities in the outcomes related to differences in case ascertainment/coding rather than to the patient population. Although some differences were strategically mitigated by the researchers’ evaluating groups of procedures of similar risk rather than individual procedures, the study results still showed clinically important differences in outcomes between the two databases.

The study authors are careful to point out that administrative datasets contain valuable resource utilization data, medication data, and other information that may not be contained in disease-specific or procedure-specific registries. It is their recommendation that linking clinical and administrative databases would capitalize on the strengths and mitigate some of the weaknesses present in these databases and allow for analyses that would not be possible with either dataset alone.