Background: Recent data examining urine output (UO) after diuretic challenge found that poor diuretic responsiveness (PDR) is associated with prolonged ventilation. Using data from the Pediatric Cardiac Critical Care Consortium NEPHRON collaborative, we sought to determine whether PDR was associated with prolonged invasive ventilation and non-invasive respiratory support after neonatal cardiac surgery.
Methods: All mechanically ventilated subjects receiving bolus intravenous furosemide followed by 6 hours of UO data within first 24 hours after surgery were included. Primary outcomes included: prolonged invasive mechanical ventilation and prolonged non-invasive mechanical ventilation (i.e. BiPAP, CPAP or high flow nasal cannula). Factors associated with prolonged invasive and non-invasive mechanical ventilation determined observed to expected (O:E) ratios for each subject using a machine-learning algorithm classification and regression tree (CART) analysis, as well as multivariable negative binomial regression. Subjects with O:E ratios greater than the 90th percentile have reached the primary outcomes. Receiver-operative characteristic-area under the curve (ROC-AUC) determined associations between PDR, UO responsiveness and the primary outcomes.
Results: Seven hundred eighty-two subjects were included. For prolonged invasive ventilation, 75 subjects met an O:E ratio >90th percentile (1.4:1), while 80 subjects met an O:E ratio >90th percentile for prolonged respiratory support (1.4:1) (Table 1). In general, PDR was poorly associated with both primary outcomes. For prolonged mechanical ventilation, the highest ROC-AUC was 0.611 for cumulative UO 6 hours (14 ml/k over 6 hours, sensitivity 0.67, specificity 0.53) after furosemide (p=0.0064, 95% confidence interval 0.96-0.99). ROC-AUC for hourly UO were all < 0.6. For prolonged respiratory support, cumulative UO 3-6 hours after furosemide (8.1 ml/kg over 4 hours, sensitivity 0.75, specificity 0.54) showed the highest ROC-AUC (AUC 0.674, p=0.0001, 95% confidence interval 0.91 to 0.97) (Figure 1). No ROC-AUC for O:E >90th%ile for respiratory support for any UO measurement exceeded 0.7.
Conclusions: In a heterogenous multicenter neonatal cohort, PDR is weakly associated with prolonged mechanical ventilation and respiratory support. Broadly incorporating diuretic responsiveness metrics across varying surgical mortality risk cohorts does not precisely predict which patients require prolonged respiratory support. Further subgroup analysis is necessary to determine if PDR may predict poor outcomes in higher risk cardiac surgical cohorts.
