Abstract 11328: Risk Analytics and Clinical Decision Support System Role in Identifying Elevated Serum Lactate Level for Pediatric Cardiac Patients

Abstract

Introduction: Pediatric patients with congenital heart defect are predisposed for low cardiac output syndrome (LCOS). Serum lactate is a well-defined marker of LCOS. To improve the quality and safety of care, new tools that are less invasive, more sensitive and specific are needed. Hypothesis: We hypothesize that analytics algorithms can compute risk indices that are associated with elevated lactate. Methods: Quality improvement project in tertiary pediatric cardiac intensive care unit. Data used were collected retrospectively throughout 6/2019-8/2020. The risk analytics algorithm uses a model of patient physiology with vitals and lab data to compute probability density curves for various physiologic variables in real-time. These densities are used to calculate the Hyperlactatemia Index (HLA) and Inadequate Delivery of Oxygen Index (IDO2) [Etiometry Inc., Boston, MA]. HLA and IDO2 represent the integrated probabilities that the current lactate is greater than 4 mmol/L and that the current SvO2 is less than 40% respectively. The minimum data required to calculate HLA and IDO2 is heart rate every 60 seconds, oxygen saturation via pulse oximetry, blood pressures every ten minutes, and at least one lactate lab per day (HLA only). The reported lactates were categorized in four groups. We subdivided the preceding three hours period before the lactate into 30 minutes windows. Subsequently, we calculated the mean IDO2 and HLA and compared the positive predictive value (PPV) of lactate > 4 for both. Results: We included 484 patients in our analysis, 20% of which were neonates. The median patient age was 7.7 months (IQR:2,46). IDO2 and HLA trended higher when lactate levels were elevated, especially in the 30 minutes immediately preceding the lactate check. When the individual probability > 50%, lactate was > 4 in 44% ( + 10%) for HLA and 3% for IDO2. Conclusions: HLA shows a strong association with lactate and is superior to IDO2 across all time points. Performance of the indices was better in the 30min prior to the elevated lactate. Future studies validating this new tool prospectively are critical before its usage in the clinical setting.