Telephone triage and dispatch of ambulances to patients with suspected and verified acute stroke - a descriptive study

We performed a descriptive retrospective study at Oslo University Hospital Ullevål (OUS), Norway, based on hospital emergency department and EMCC records during a six-month study period from, September 1st, 2019, until February 29th, 2020. OUS serves both as a local- and a comprehensive stroke center for a population of approximately 550 000 inhabitants with residential address in the municipality of Oslo. The EMCC Oslo serves a population of approximately 1.7 million inhabitants in both urban and rural parts of south-eastern Norway, including the Oslo metropolitan area, and handles approximately 250 000 emergency calls annually.

All emergency calls to EMCCs in Norway are answered directly by specially trained EMCC nurses or paramedics using a Criteria-Based Dispatch (CBD) protocol. CBD protocols are decision triage support tools based on a patient’s signs and symptoms and require the dispatcher to have medical skills [27]. The current CBD dispatch protocol in use in Norwegian EMCCs is “The Norwegian Index for Emergency Medical Assistance” (Index) [28]. Index consists of a Start page and 39 different symptom-based criteria cards. The Start page focuses on obtaining critical information on the location of the event, patient state of consciousness, and initial assessment of whether the patient is suffering from an acute life-threatening illness or injury. Except for the Start page and criteria cards 1–3 (suspected cardiac arrests and unconsciousness), the remaining criteria cards each consists of listed criteria with corresponding responses and priorities, additional questions and supplemental information about symptoms and potential causes. Dispatch priorities are coded as red (acute and life-threatening conditions), yellow (urgent and potentially life- threatening conditions) and green (non-urgent conditions). After selecting the appropriate criteria card, the dispatchers assess listed criteria from top (the most acute criteria) to the bottom (non-urgent criteria) until a criteria is met. Stroke related dispatch criteria included in this study were identified from criteria card 27 (“Altered levels of consciousness/paralysis”) and criteria card 39 (“The eyes”) in the 3rd edition of the Norwegian Index (Supplemental material Illustration 1 and 2). Recent reports has shown a large variation in use of Index in assessment of acute stroke in Norwegian EMCCs [29]. EMCC operators have a relatively high self-reported use of Index as a support tool in dispatch decisions, however the study reports of with variations on both individual and EMCC level.

The ambulance service in Norway includes both ground- and boat ambulances staffed by paramedics or emergency medical technicians (EMT). It also includes HEMS and Search and Rescue (SAR) helicopters, staffed by prehospital critical care physicians. If patient symptoms are identified as suspected stroke symptoms by the EMCC dispatcher, the default response will be to dispatch the nearest ground ambulance with a stroke criteria and acute priority (lights and sirens). In remote areas, HEMS or SAR helicopters should be dispatched to reduce prehospital delay [11]. According to national legislation, 90% of all emergency calls to the emergency medical telephone number 113 must be answered within 10 s [30].

Patients with suspected stroke symptoms are routinely assessed by the ambulance service using the “Face– Arm– Speech– Time” test (FAST) [31] or prehospital National Institutes of Health Stroke Scale (NIHSS) due to the Paramedic Norwegian Acute Stroke Prehospital Project (ParaNASPP) [32], which involved five of the ambulance stations in Oslo during the study period. After clinical assessment, all patients with suspected stroke symptoms were consulted directly with the on-call stroke physician at OUS. If not accepted for hospitalization the patients were transported to their local hospital, general practitioner (GP) or out-of-hours GP acute clinic for diagnostic follow up.

Based on EMCC and hospital records we identified all patients who had an ambulance dispatched with one of the stroke criteria after initial EMCC assessments and/or was discharged from OUS with a main diagnosis of acute stroke. From the EMCC database we identified all dispatch records with stroke– and non-stroke criteria among patients discharged with stroke diagnosis. We extracted patient identity (name and Norwegian National Identity Number (11-digits)), dispatch criteria and priority, prehospital time stamps, and patient destination. All duplicates, false registrations, aborted or reallocated dispatches, were excluded to identify unique incidents. From the hospital records we identified all patients discharged with a main diagnosis of acute ischemic stroke (AIS) (I63.0-I63.9), intracerebral hemorrhage (ICH) (I61.0-I62.9), transient ischemic attack (TIA) (G45.8-G45.9), and other ICD-10 non-stroke diagnoses. We also extracted sex, age and referring body. We excluded patients referred from other hospitals, GP or out-of-hours GP acute clinic, patients admitted to local hospitals outside of Oslo, patients with missing EMCC records, and patients under the age of 18 years.

Patients that had an ambulance dispatched with stroke criteria and who were discharged from OUS with a confirmed stroke diagnosis were defined as EMCC true positive. Patients that had an ambulance dispatched with stroke criteria who were not admitted to OUS or discharged with a non-stroke diagnosis, were defined as EMCC false positive. Patients discharged with a stroke diagnosis despite a non-stroke criteria dispatch, were defined as EMCC false negative. Definitions are illustrated in Fig. 1.

To identify subgroups of stroke patients with alternate prehospital pathways, such as patients initially referred to out-of-hours GP acute clinics by the ambulance service and thereafter admitted to OUS within 24 h after the initial emergency call, we matched hospital records with EMCC records, using the patients’ national identity number and incident date and time.

Destination for all patients were identified and categorized as; referred/admitted to OUS, referred to other hospitals, GP or out-of-hours GP acute clinics, or left on scene. Patients referred to OUS by the ambulance services without registered hospital admissions were defined as outpatients.

EMCC stroke sensitivity and positive predictive value (PPV) were calculated according to the definitions of EMCC true positive, false positive and false negative stroke patients. Due to the necessity of identifying and excluding all non-unique dispatches, patients referred from GP, out-of-hours GP or other hospitals, and patients admitted to hospitals outside of Oslo, the total number of true negative stroke patients was not identified. Consequently, EMCC negative predictive value (NPV) could not be calculated.

Prehospital time intervals were calculated based on EMCC electronic time stamps. Due to a lack of access to complete patient hospital records, patient delay (time interval from symptom onset until the initial emergency call to EMCC) was not included. Prehospital timeline and time interval definitions are presented in Fig. 2.

R version 3.6.1 was used to perform statistical calculations. Descriptive statistics are presented in numbers (%) and median with 25- and 75-percentiles. Time intervals are presented as median minutes and seconds (mm:ss) with 25- and 75-percentiles.

Groups were compared using Kruskal-Wallis Test. Wald test was used to calculate 95% confidence intervals for EMCC stroke sensitivity and positive predictive values (PPV).

The study is part of The Dispatch–Norwegian Acute Stroke Prehospital Project (Dispatch NASPP). The study has been performed according to the Helsinki convention [34]. Protocol for the Dispatch NASPP project was approved by the Regional Research Ethics Committee (REC) (ref. no. 2018/1909) and the local data protection officer at OUS (ref. no. 18/25,297).

After initial matching of data records from EMCC, ambulance, and hospital, all person identifiable data were replaced with a case number and the codebook was organized in Medinsight® Release 2.17.4.0. The identified dataset was handled in TSD - Service for Sensitive Data, at the University of Oslo, Norway. All EMCC nurses and paramedics employed in EMCC Oslo have been informed about the study.