ADVANCE: a biomedical informatics approach to investigate acute kidney injury in infants

Abstract

Background: Acute kidney injury (AKI) occurs in up to half of infants admitted to the neonatal intensive care unit (NICU) and is associated with increased risks of death and more days of mechanical ventilation, hospitalization, and vasopressor drug support. Our objective was to build a granular relational database to study the impact that AKI has on infants admitted to Level-IV NICUs.

Methods: A relational database was created by linking data from the Children's Hospitals Neonatal Database with AKI-focused data from electronic health records from 9 centers.

Results: The current cohort consists of 24,870 infants with a median (IQR) gestational age of birth of 37 weeks (32 weeks, 39 weeks), and a median birth weight of 2.720 kg (1.750 kg, 3.310 kg). There was a male predominance with 14,214 (57%) males. In all, 2434 (9.8%) of the mothers were of Hispanic ethnicity. The maternal race breakdown of the cohort was as follows: 741 (3.0%) Asian, 5911 (24%) Black, and 14,945 (60%) White. Overall mortality was 5.8%.

Conclusion: The ADVANCE relational database is an innovative research tool to rigorously study the epidemiology of AKI in a large national cohort of infants admitted to Level-IV NICUs involved in the Children's Hospital Neonatal Consortium.

Impact: We used a biomedical informatics approach to build a relational database to study acute kidney injury in infants. We highlight our methodology linking Children's Hospital Neonatal Consortium and electronic health record data from nine neonatal intensive care units. The ADVANCE relational database is a granular and innovative research tool to study risk factors and in-hospital outcomes of acute kidney injury and mortality in a vulnerable patient population.

Fig. 1. Map depicting ADVANCE Sites.

The United States map depicts the locations of the 13 ADVANCE sites, including Arkansas Children’s Hospital, Children’s National Hospital, Atrium Health Wake Forest Baptist Brenner Children’s Hospital, Nationwide Children’s Hospital, UW Health American Family Children’s Hospital, Cincinnati Children’s Hospital Medical Center, Texas Children’s Hospital, Riley Hospital for Children, Children’s Healthcare of Atlanta, Children’s Minnesota, Le Bonheur Children’s Hospital, and St. Louis Children’s Hospital, Children’s of Alabama.

Fig. 2. ADVANCE Data integration strategy.

Electronic health record (EHR) data from each site are de-identified and converted into a relational database linked to CHND data. The core of the relational database structure is a participant table that stores information unique to each participant, accompanied by a study ID. Calculated data elements are stored in additional tables.