Using Natural Language Processing to Examine Social Determinants of Health in Prehospital Pediatric Encounters and Associations with EMS Transport Decisions.

BACKGROUND: Social determinants of health (SDOH) influence access to health care and are associated with inequities in patient outcomes, yet few studies have explored SDOH among pediatric EMS patients. The objective of this study was to examine the presence of SDOH in EMS clinician free text notes and quantify the association of SDOH with EMS pediatric transport decisions. METHODS: This was a retrospective analysis of primary 9-1-1 responses for patients ages 0-17 years from the 2019 ESO Data Collaborative research dataset. We excluded cardiac arrests and patients in law enforcement custody. Using natural language processing (NLP) we extracted the following SDOH categories: income insecurity, food insecurity, housing insecurity, insurance insecurity, poor social support, and child protective services. Univariate and multivariable associations between the presence of SDOH in EMS records and EMS transport decisions were assessed using logistic regression. RESULTS: We analyzed 325,847 pediatric EMS encounters, of which 35% resulted in non-transport. The median age was 10 years and 52% were male. Slightly over half (53%) were White, 31% were Black, and 11% were Hispanic. Child protective services (n = 2,620) and housing insecurity (n = 1,136) were the most common SDOH categories found in the EMS free text narratives. In the multivariable model, child protective services involvement (odds ratio (OR)=2.04 [95% confidence interval (CI) 1.84-2.05]), housing insecurity (OR = 1.46 [95% CI 1.26-1.70]), insurance security (OR = 2.44 [95% CI 1.93-3.09]), and poor social support (OR = 10.48 [95% CI 1.42-77.29]) were associated with greater odds of EMS transport. CONCLUSIONS: SDOH documentation in the EMS narrative was rare among pediatric encounters; however, children with documented SDOH were more likely to be transported. Additional exploration of the root causes and outcomes associated with SDOH among children encountered by EMS are warranted.

Identification of Factors Associated with Return of Spontaneous Circulation after Pediatric Out-of-Hospital Cardiac Arrest Using Natural Language Processing.

INTRODUCTION: Prior studies examining prehospital characteristics related to return of spontaneous circulation (ROSC) in pediatric out-of-hospital cardiac arrest (OHCA) are limited to structured data. Natural language processing (NLP) could identify new factors from unstructured data using free-text narratives. The purpose of this study was to use NLP to examine EMS clinician free-text narratives for characteristics associated with prehospital ROSC in pediatric OHCA. METHODS: This was a retrospective analysis of patients ages 0-17 with OHCA in 2019 from the ESO Data Collaborative. We performed an exploratory analysis of EMS narratives using NLP with an a priori token library. We then constructed biostatistical and machine learning models and compared their performance in predicting ROSC. RESULTS: There were 1,726 included EMS encounters for pediatric OHCA; 60% were male patients, and the median age was 1 year (IQR 0-9). Most cardiac arrest events (61.3%) were unwitnessed, 87.3% were identified as having medical causes, and 5.9% had initial shockable rhythms. Prehospital ROSC was achieved in 23.1%. Words most positively correlated with ROSC were "ROSC" (r = 0.42), "pulse" (r = 0.29), "drowning" (r = 0.13), and "PEA" (r = 0.12). Words negatively correlated with ROSC included "asystole" (r = -0.25), "lividity" (r = -0.14), and "cold" (r = -0.14). The terms "asystole," "pulse," "no breathing," "PEA," and "dry" had the greatest difference in frequency of appearance between encounters with and without ROSC (p < 0.05). The best-performing model for predicting prehospital ROSC was logistic regression with random oversampling using free-text data only (area under the receiver operating characteristic curve 0.92). CONCLUSIONS: EMS clinician free-text narratives reveal additional characteristics associated with prehospital ROSC in pediatric OHCA. Incorporating those terms into machine learning models of prehospital ROSC improves predictive ability. Therefore, NLP holds promise as a tool for use in predictive models with the goal to increase evidence-based management of pediatric OHCA.

Applying a set of termination of resuscitation criteria to paediatric out-of-hospital cardiac arrest.

OBJECTIVE: Prehospital Termination of Resuscitation (TOR) protocols for adults can reduce the number of futile transports of patients in cardiac arrest, yet similar protocols are not widely available for paediatric out-of-hospital cardiac arrest (POHCA). The objective of this study was to apply a set of criteria for paediatric TOR (pTOR) from the Maryland Institute for Emergency Medical Services Systems (MIEMSS) to a large national cohort and determine its association with return of spontaneous circulation (ROSC) after POHCA. METHODS: We identified patients ages 0-17 treated by Emergency Medical Services (EMS) with cardiac arrest in 2019 from the ESO dataset and and applied the applicable pTOR certeria for medical or traumatic arrests. We calculated predictive test characteristics for the outcome of prehospital ROSC, stratified by medical and traumatic cause of arrest. RESULTS: We analyzed records for 1595 POHCA patients. Eighty-eight percent (n = 1395) were classified as medical. ROSC rates were 23% among medical POHCA and 27% among traumatic POHCA. The medical criteria correctly classified >99% (322/323) of patients who achieved ROSC as ineligible for TOR. The trauma criteria correctly classified 93% (50/54) of patients with ROSC as ineligible for TOR. Of the five misclassified patients, three were involved in drowning incidents. CONCLUSIONS: The Maryland pTOR criteria identified eligible patients who did not achieve prehospital ROSC, while reliably excluding those who did achieve prehospital ROSC. As most misclassified patients were victims of drowning, we recommend considering the exclusion of drowning patients from future pTOR guidelines. Further studies are needed to evaluate the long-term survival and neurologic outcome of patients misclassified by pTOR criteria.