Predicting Post-Fontan Length of Stay: The Limits of Measured Variables

Abstract

Post-operative length of stay (LOS) is an important metric for both healthcare providers and patients and their families. Predicting LOS is a challenge as it is sensitive to multitudinous patient and system factors. All subjects undergoing a Fontan from 1996-2016 who survived to hospital discharge were included. Details about the pre-operative status, operative conduct, and post-operative course of each patient were obtained. The association between patient characteristics and post-Fontan LOS were determined using stepwise multivariable regression models. Of 320 subjects who underwent a Fontan, 314 (98.1%) survived to hospital discharge. Median age at Fontan was 3.3years (IQR 2.8, 4.0) and the most common underlying diagnosis was hypoplastic left heart syndrome (106, 33.8%). Median post Fontan LOS was 11days (IQR 8, 17). Univariable risk factors for longer LOS included number of previous surgeries, post-Glenn LOS, cardiopulmonary bypass time, post-operative chylothorax, and failure to extubate in the operating room (all p < 0.05). In multivariable models, number of previous operations, extubation in the operating room, and postoperative complications predicted LOS (R2 = 0.5185 for full model). The proportion of patients discharged on week days (14.7-18.8% per day) was significantly higher than the proportion discharged on weekend days (5.1-9.9% per weekend day). Pre-operative variables have limited use in predicting post-Fontan length of stay. The most important predictors of post-operative LOS are extubation in the operating room and the occurrence of post-operative complications. However, a significant proportion of variability in LOS was not explained by available measurable variables.