Association between weight-for-length percentile and ICU length of stay in patients with a single ventricle undergoing bidirectional Glenn repair: A retrospective cohort study.

BACKGROUND: Poor weight gain has been identified as an independent risk factor for increased surgical morbidity and mortality for patients with single-ventricle physiology undergoing staged surgical palliation. Conversely, excessive weight gain has also emerged as an independent risk factor predicting increased morbidity and mortality in a single-center study. Given this novel single-center concept, we investigated the impact of excessive weight on patients with single-ventricle physiology undergoing bidirectional Glenn palliation in a multicenter study model. METHODS: Patients from the Pediatric Heart Network Single Ventricle Reconstruction Trial (n = 387) were analyzed in a retrospective cohort study examining the independent effect of weight percentile on intensive care unit (ICU) length of stay (LOS) and ventilator days. Locally estimated scatterplot smoothing (LOESS) regression was used to plot weight-for-length (WFL) percentiles by ICU LOS and ventilator days. Unadjusted and adjusted ordinal regression was used to model ICU LOS and ventilator days. RESULTS: Scatterplots and LOESS regression curves demonstrated increasing ICU LOS and ventilator days for increasing WFL percentiles. Unadjusted ordinal regression analysis of ICU LOS demonstrated a trend of increasing ICU LOS for increasing WFL percentiles that was not statistically significant (P = 0.11). A similar trend was demonstrated in adjusted ordinal regression that was not statistically significant (P = 0.48). Unadjusted and adjusted ordinal regression analysis of ventilator days did not reach statistical significance (P = 0.07). CONCLUSION: Excessive weight gain has a clinically relevant but not statistically significant association with increased ICU LOS and ventilator days for those patients in the >90th WFL percentile for age.

Duke Scholars

Author Christopher Lindsell Biostatistics & Bioinformatics, Division of Biostatistics