PM1050: Hospital Variation in Mortality for High Risk Infant Diagnoses

Background: A large proportion of perioperative mortality after pediatric surgery occurs among infants with specific high-risk diagnoses. It is unclear if there is hospital-level variation in mortality for these high-risk pathologies. In addition, if variation exists, the magnitude of perioperative mortality variation across hospitals for these diagnoses has yet to be characterized.

Methods: The Pediatric Health Information System® database (2012–2020) was used to identify patients who underwent an operation for one of eight high-risk infant diagnoses: necrotizing enterocolitis (NEC), isolated intestinal perforation (IIP), meconium peritonitis, intestinal atresia, volvulus, tracheoesophageal fistula (TEF), gastroschisis, congenital diaphragmatic hernia (CDH). Reliability-adjusted inpatient mortality rates were calculated controlling for complex chronic conditions (defined with Feudtner ICD-9/10 codes) and use of extracorporeal membrane oxygenation. Hospitals were stratified into quintiles of adjusted inpatient mortality rates (very low mortality [i.e., best performance in quintile 1, Q1] to very high-mortality [i.e., worst performance in quintile 5, Q5]) and diagnosis-specific inpatient mortality rates were calculated across hospitals. Nonparametric test of trend was used to assess mortality differences across mortality quintiles.

Results: In total, 20,430 infants were identified across 48 academic, pediatric hospitals. All hospitals had level III or IV neonatal intensive care units during the study period. The most common diagnoses in the overall cohort were gastroschisis (36.5%), volvulus (23.7%), atresia (15.8%), and NEC (15.7%). The overall inpatient mortality rate was 8.2% [95% CI 7.8–8.6]. The diagnosis with the highest mortality rate was NEC (27.5%) followed by IIP (20.0%), meconium peritonitis (18.9%), CDH (7.2%), TEF (7.1%), volvulus (3.7%), gastroschisis (3.5%), and atresia (3.4%). Adjusted hospital inpatient mortality rates (Figure 1) ranged from 2.7% [0.0–6.4] to 13.2% [9.8–16.6]. Nine hospitals had significantly lower adjusted inpatient mortality compared to the other hospitals in the cohort while five hospitals had significantly higher rates. The absolute difference in mortality rates comparing the very low and high mortality hospitals (Figure 2) was 6.0% (Q1, 5.1; Q5, 11.1; p< 0.01). There was at least a 2-fold difference in mortality between very low and high mortality hospitals for 6 of the 8 diagnoses: CDH (Q1, 5.4%; Q5, 14.8%; p< 0.01), IIP (Q1, 11.1%; Q5, 28.7%; p< 0.01), TEF (Q1, 5.1%; Q5, 12.6%; p< 0.01), meconium peritonitis (Q1, 11.4%; Q5, 25.2%; p< 0.01), atresia (Q1, 2.2%; Q5, 4.8%; p< 0.01), and NEC (Q1, 16.9%; Q5, 35.9%; p< 0.01).

Conclusion: For high-risk infant diagnoses, there is significant variation in inpatient mortality rates across academic, pediatric hospitals. For some diagnoses, there was >2-fold variation in inpatient mortality comparing very low and very high mortality hospitals. Further work is needed to understand the reasons for such dramatic mortality variation and whether there are specific care processes at low mortality hospitals that can be identified and implemented more broadly at other hospitals.