The Acidemia Index: A Near Real-Time Risk Analytics Algorithm is Associated With Measured Acidemia in Pediatric Critical Care Patients, a Multi-Center Validation Study.

OBJECTIVES

Acidemia frequently evolves in pediatric critical care patients, especially with congenital heart defects. Worsening acidemia secondary to inadequate systemic oxygen delivery can be detrimental to patients' outcomes and the ability to predict it has the potential to prompt early interventions to improve the clinical state. We aimed to evaluate the association of a novel near real-time predictive analytics algorithm with acidemia (ACD) (arterial pH < 7.25) in pediatric patients admitted to a critical care unit.

STUDY

DESIGN

Retrospective observational study in nine tertiary institutions in the United States.

SETTING

Majority of patients were admitted to the cardiac ICU. Using Etiometry platform data (Etiometry, Boston, MA), acidemia (ACD) index was validated.

PATIENTS: Patients 12 years old or younger were admitted to an ICU between February 1, 2018, and November 31, 2020.

INTERVENTION: A total of 24,431 arterial blood pH measurements from 1858 patients were included in the validation dataset. The ACD index was calculated using a physiologic algorithm that incorporates patients' variables including laboratory and clinical data. Based on the previous assessment of the physiologic state of the patient, the physiologic algorithm interprets the new data in a real-time manner using Bayes' theorem.

MEASUREMENT AND MAIN

RESULTS

Based on a complete dataset, the area under the receiver operating characteristic curve of the ACD index was 0.93. As the index value increased, the likelihood of having acidemia increased (p < 0.01). The relative risk of having acidemia when the ACD index is less than 1 was 0.11 (95% CI, 0.07-0.15), and the relative risk of not having acidemia when the ACD index was greater than 99 was 0.38 (95% CI, 0.32-0.46).

CONCLUSIONS

In this large pediatric cohort, higher ACD index values were associated with a higher likelihood of having acidemia. Consequently, this novel index has the potential to identify severe changes in clinical status. Prospective analysis of the ACD index is important to understand its utility in the management of pediatric critical illness.