Quality Assessment in Pediatric Cardiology Using Network Technologies

Background: Multicenter consortia, real-world databases, networks, and voluntary registries have been established with the purpose of research in a larger scale and generalizability of findings. The MIRACUM (Medical Informatics in Research and Care in University Medicine) consortium has the goal to set up a nationwide infrastructure for the secondary use and sharing of health care data to improve medical care and research. With this new network technologies and data storage, we aim to implement a robust algorithm to process information on clinical outcome and quality assessment for pediatric cardiovascular patients treated in our hospital.

Methods: Patients who had undergone cardiac surgery in our department in the years 2011 to 2020 were included in the analysis when aged <18 years. CHD was classified in four cardiac diagnosis groups (uncomplicated biventricular, complex biventricular, univentricular group I and II) defined by both OPS and ICD codes. We evaluated preoperative, demographic (i.e., weight, age, presence of chromosomal abnormalities, or concomitant malformations), peri-procedural (i.e., aortic cross clamp and bypass time), and postsurgical risk factors such as inflammatory and renal markers for the primary outcome mortality.

Results: A total of 1,774 patients with 2,227 hospitalization encounters were included. Heart disease groups comprised uncomplicated biventricular with 1,335 (59.95%), complex biventricular with 603 (27.08%), univentricular stage I with 98 (4.40%), and univentricular stage II and III with 191 (8.58%) surgeries. Most significant risk factors for increased mortality were weight <2500 g (HR 12.16; p < 0.001), creatinine-ratio (HR 3.03; p < 0.001) postsurgical leukocyte count <4000/µL (HR 3.2; p < 0.05) and concomitant malformations as well as chromosomal abnormalities. Lower values of the ratio aortic clamp time/bypass time were strongly associated with reduced mortality (HR 0.04, p < 0.01).

Conclusion: Using data integration, we were able to get data assessment on a larger scale. Our analysis confirmed known non modifiable risk factors for increased mortality such as weight and chromosomal abnormalities. Interestingly, laboratory parameters such as leukocytopenia or creatinine-ratio showed a high predictive value. We understand our analysis as preparatory ground work for pooling data across different centers in a privacy-preserving manner by utilizing established infrastructure of the German Medical Informatics Initiative.

Publikationsverlauf

Artikel online veröffentlicht:
13. Februar 2024