• Punkaj Gupta, Jeffrey M Gossett, and Mallikarjuna Rettiganti.
• Section of Cardiac Critical Care, Methodist Children's Hospital, San Antonio, TX.
• Crit. Care Med. 2018 Dec 1; 46 (12): e1112-e1120.

ObjectivesTo use two national databases to quantify the pace and magnitude of improvement in hospital performance (as measured by mortality) across hospitals caring for critically ill children in the United States.DesignWe used empirical Bayes shrinkage estimators to obtain shrinkage estimators of observed/expected mortality ratios for each hospital assuming a Gamma Poisson posterior distribution. Revised mortality rates for each hospital were obtained from the shrunken incidence ratios.SettingPediatric Health Information System participating hospital and Kids' Inpatient Database participating hospital.PatientsPatients less than or equal to 18 years old who received invasive mechanical ventilation during their hospital stay at a Pediatric Health Information System participating hospital (2005-2015) or a Kids' Inpatient Database participating hospital (1997-2012) were included.InterventionsNone.Measurements And Main ResultsA total of 486,838 patients from 48 Pediatric Health Information System hospitals and 798,840 patients from 947 Kids' Inpatient Database hospitals were included. For the Pediatric Health Information System hospitals, the median shrunken adjusted mortality decreased from 11.66% in 2005 to 7.11% in 2015; for the Kids' Inpatient Database hospitals, it decreased from 5.79% in 1997 to 3.86% in 2012. By 2015, more than 95% of the Pediatric Health Information System hospitals performed better than or as well as the best 25% of the hospitals in 2005. By 2012, 33.7% of Kids' Inpatient Database hospitals performed better than or as well as the best 25% of the hospitals in 1997.ConclusionsThis study provides insight into the magnitude of improvement in patient mortality in hospitals caring for critically ill children in the United States. This study quantifies hospital performance in pediatric critical care over time, and it provides benchmarks against which individual hospitals can assess their own performance. In future pediatric epidemiologic studies, we should identify outcomes other than mortality to quantify improvement in hospital performance.

25 keywords...

hide…