An analysis of the descriptors of acute myocardial infarction used by South Africans when calling for an ambulance from a private emergency call centre

The variables predictive of ambulance non-conveyance of patients in the Western Cape, South Africa

Introduction

Emergency medical service (EMS) resources are limited and should be reserved for incidents of appropriate acuity. Over-triage in dispatching of EMS resources is a global problem. Analysing patients that are not transported to hospital is valuable in contributing to decision-making models/algorithms to better inform dispatching of resources. The aim is to determine variables associated with patients receiving an emergency response but result in non-conveyance to hospital.

Methods

A retrospective cross-sectional study was performed on data for the period October 2018 to September 2019. EMS records were reviewed for instances where a patient received an emergency response but the patient was not transported to hospital. Data were subjected to univariate and multivariate regression analysis to determine variables predictive of non-transport to hospital.

Results

A total of 245 954 responses were analysed, 240 730 (97.88 %) were patients that were transported to hospital and 5 224 (2.12 %) were not transported. Of all patients that received an emergency response, 203 450 (82.72 %) patients did not receive any medical interventions. Notable variables predictive of non-transport were green (OR 4.33 (95 % CI: 3.55-5.28; p<0.01)) and yellow on-scene (OR 1.95 (95 % CI: 1.60-2.37; p<0.01).Incident types most predictive of non-transport were electrocutions (OR 4.55 (95 % CI: 1.36-15.23; p=0.014)), diabetes (OR 2.978 (95 % CI: 2.10-3.68; p<0.01)), motor vehicle accidents (OR 1.92 (95 % CI: 1.51-2.43; p<0.01)), and unresponsive patients (OR 1.98 (95 % CI: 1.54-2.55; p<0.01)). The highest treatment predictors for non-transport of patients were nebulisation (OR 1.45 (95 % CI: 1.21-1.74; p<0.01)) and the administration of glucose (OR 4.47 (95 % CI: 3.11-6.41; p<0.01)).

Conclusion

This study provided factors that predict ambulance non-conveyance to hospital. The prediction of patients not transported to hospital may aid in the development of dispatch algorithms that reduce over-triage of patients, on-scene discharge protocols, and treat and refer guidelines in EMS.

The lay descriptors of out-of-hospital cardiac arrest in the Western Cape province, South Africa

Introduction

Out-of-Hospital Cardiac Arrest (OHCA) is a time-sensitive emergency requiring prompt identification and emergency care to reduce morbidity and mortality. The first step in managing OHCA is rapid identification by the emergency dispatch centre. Identification of these patients remains challenging in South Africa due to multiple languages and widely differing levels of education. This study aimed to identify the key descriptors (words and phrases) of OHCA used by callers in the Western Cape when contacting the provincial Emergency Medical Services' emergency call centre.

Methodology

Computer-aided dispatch data with a corresponding “patient unresponsive” incident type were drawn for a 12-month period (January–December 2018). Corresponding patient care records were extracted to verify OHCA. The original voice recordings between the caller and emergency call taker at the time of the emergency were extracted and transcribed verbatim. Transcriptions were subjected to inductive, qualitative content analysis to the manifest level. Descriptors of OHCA in isiXhosa, English and Afrikaans calls were coded, categorised, and quantified.

Results

A total of 729 confirmed OHCA cases were identified, of which 38 (5.2%) Afrikaans, 24 (3.3%) isiXhosa and a random sample of 50 (6.8%) English calls were transcribed. Following content analysis, five distinct categories were identified. The most prevalent categories were descriptors related to ill health (medical history and suspected diagnosis; 35.5%), level of consciousness (unresponsive; 18.6%) and cardiac activity (pulselessness and suspected death; 17.2%).

Conclusion

The vast majority of callers within the Western Cape province of South Africa use consistent descriptors across languages when requesting an ambulance for OHCA. Future studies should focus on the development and validation of OHCA recognition algorithms, based on these findings.

The Feasibility of Using Machine Learning to Classify Calls to South African Emergency Dispatch Centres According to Prehospital Diagnosis, by Utilising Caller Descriptions of the Incident

This paper presents the application of machine learning for classifying time-critical conditions namely sepsis, myocardial infarction and cardiac arrest, based off transcriptions of emergency calls from emergency services dispatch centers in South Africa. In this study we present results from the application of four multi-class classification algorithms: Support Vector Machine (SVM), Logistic Regression, Random Forest and K-Nearest Neighbor (kNN). The application of machine learning for classifying time-critical diseases may allow for earlier identification, adequate telephonic triage, and quicker response times of the appropriate cadre of emergency care personnel. The data set consisted of an original data set of 93 examples which was further expanded through the use of data augmentation. Two feature extraction techniques were investigated namely; TF-IDF and handcrafted features. The results were further improved using hyper-parameter tuning and feature selection. In our work, within the limitations of a limited data set, classification results yielded an accuracy of up to 100% when training with 10-fold cross validation, and 95% accuracy when predicted on unseen data. The results are encouraging and show that automated diagnosis based on emergency dispatch centre transcriptions is feasible. When implemented in real time, this can have multiple utilities, e.g. enabling the call-takers to take the right action with the right priority.

The Development of Consensus-Based Descriptors for Low-Acuity Emergency Medical Services Cases for the South African Setting

INTRODUCTION

Emergency Medical Services (EMS) are designed to respond to and manage patients experiencing life-threatening emergencies; however, not all emergency calls are necessarily emergent and of high acuity. Emergency responses to low-acuity patients affect not only EMS, but other areas of the health care system. However, definitions of low-acuity calls are vague and subjective; therefore, it was necessary to provide a clear description of the low-acuity patient in EMS.

AIM

The goal of this study was to develop descriptors for "low-acuity EMS patients" through expert consensus within the EMS environment.

METHODS

A Modified Delphi survey was used to develop call-out categories and descriptors of low acuity through expert opinion of practitioners within EMS. Purposive, snowball sampling was used to recruit 60 participants, of which 29 completed all three rounds. An online survey tool was used and offered both binary and free-text options to participants. Consensus of 75% was accepted on the binary options while free text offered further proposals for consideration during the survey.

RESULTS

On completion of round two, consensus was obtained on 45% (70/155) of the descriptors, and a further 30% (46/155) consensus was obtained in round three. Experts felt that respiratory distress, unconsciousness, chest pain, and severe hemorrhage cannot be considered low acuity. For other emergency response categories, specific descriptors were offered to denote a case as low acuity.

CONCLUSION

Descriptors of low acuity in EMS are provided in both medical and trauma cases. These descriptors may not only assist in the reduction of unnecessary response and transport of patients, but also assist in identifying the most appropriate response of EMS resources to call-outs. Further development and validation are required of these descriptors in order to improve accuracy and effectiveness within the EMS dispatch environment.