Impact of Confounding on Cost, Survival, and Length-of-Stay Outcomes for Neonates with Hypoplastic Left Heart Syndrome Undergoing Stage 1 Palliation Surgery.

The objective of this analysis was to update trends in LOS and costs by survivorship and ECMO use among neonates with hypoplastic left heart syndrome (HLHS) undergoing stage 1 palliation surgery using 2016 data from the Healthcare Cost and Utilization Project Kids' Inpatient Database. We identified neonates ≤ 28 days old with HLHS undergoing Stage 1 surgery, defined as a Norwood procedure with modified Blalock-Taussig (BT) shunt, Sano modification, or both. Multivariable regression with year random effects was used to compare LOS and costs by hospital region, case volume, survivorship, and ECMO vs. no ECMO. An E-value analysis, an approach for conducting sensitivity analysis for unmeasured confounding, was performed to determine if unmeasured confounding contributed to the observed effects. Significant differences in total costs, LOS, and mortality were noted by hospital region, ECMO use, and sub-analyses of case volume. However, other than ECMO use and mortality, the maximum E-value confidence interval bound was 1.71, suggesting that these differences would disappear with an unmeasured confounder 1.71 times more associated with both the outcome and exposure (e.g., socioeconomic factors, environment, etc.) Our findings confirm previous literature demonstrating significant resource utilization among Norwood patients, particularly those undergoing ECMO use. Based on our E-value analysis, differences by hospital region and case volume can be explained by moderate unobserved confounding, rather than a reflection of the quality of care provided. Future analyses on surgical quality must account for unobserved factors to provide meaningful information for quality improvement.