Telemedical support for prehospital emergency medical service in severe emergencies: an open-label randomised non-inferiority clinical trial

The institutional ethics committee of the University Hospital Aachen approved this trial on 25 November 2015 (approval number: EK 170/15). The ethics committee waived the need for informed consent for the initial randomisation and EMS-treatment, and it was only required for the follow-up procedures.

This open-label, randomised, controlled, parallel-group, non-inferiority trial with two interim analyses and no trial adaptation was conducted within the ground-based emergency medical district of Aachen. All resulting EMS files and hospital information system-based outcome data of admitted patients were analysed within the scope of data quality assurance for outcome measures. Patients admitted to five hospitals (one university hospital and four primary care hospitals) were followed. The latest version of the published study protocol [17] and amended trial statistical analysis plan (TSAP) are presented in Additional file 1 and Additional file 2, respectively. The TSAP was finalised before database export and analysis. The trial was registered with ClinicalTrials.gov (NCT02617875) on 30 November 2015 and is reported in accordance with the CONSORT statement for non-inferiority trials [18]. Additional methods are presented in Additional file 3.

The data safety monitoring committee reviewed non-blinded safety data. The study director at the University Hospital RWTH Aachen was responsible for study conduction.

We included all routine patients aged ≥ 18 years with severe (such as acute coronary syndrome, stroke or hypertensive emergency) emergencies. Acute life-threatening emergencies (such as polytrauma, respiratory insufficiency, apnoea or cardio circulatory arrest) [17] were excluded as for these patients an EMS physician on scene cannot be denied because of potentially required manual skills (such as intubation, chest drainage or resuscitation). Enrolment occurred independent of the final hospital destination or decision for conveyance to the emergency department.

Technical details are described elsewhere [19]. In brief, the tele-EMS transfers all vital, audio, and video data from the emergency scene to a teleconsultation centre equipped with an EMS physician and supported by software (TM-Documentation, Umlaut Telehealthcare GmbH, Aachen, Germany) using the latest checklists and guidelines.

All paramedics on a telemedically-equipped ambulance received a one-day training focused on indication for teleconsultation and telemedical equipment. Every conventional EMS physician is at least in the 4th year of residency in anaesthesiology and passed an examination in emergency medicine. In addition, every tele-EMS physician is a board-certified anaesthetist and completed at least 500 on-site emergency missions combined with further training in telemedicine.

Paramedics are obliged to call the tele-EMS physician 24/7, if drugs or invasive procedures are to be carried out according to an SOP, as they are not allowed to carry out them alone according to German law [13, 19, 20]. If the tele-EMS physician is not accessible, a physician on site is mandatory by law in such cases.

During the emergency call, the dispatchers in the central dispatching centre of the fire department in Aachen typed the suspected symptoms into the dispatching software (Cobra 4, ISE GmbH, Aachen, Germany). An algorithm in the dispatching software initiated randomisation automatically if essential operational conditions were met, e.g. availability of an EMS physician in both study arms and the entered symptom matched a severe condition according to the EMS physician indication catalogue of the German federal medical association. Patients were allocated to receive either tele-EMS (tele-group) or conventional physician-based EMS (control group) using complete randomisation with a 1:1 allocation ratio. Before implementation of the tele-EMS physician for each of the patients, an EMS-physician was alerted. Since implementation, a parallel alert can only be dispensed because the paramedics can consult the tele-EMS physician by pushing a button. The dispatcher assigned the patients to their respective groups as randomised. A manual conversion of the initial assignment in terms of the tele-group was required, if in the further course of the phone call the dispatcher assumed a life-threatening emergency. Further, the initially dispatched team was always allowed to pass the mission to the other group on both sides based on the situational assessment.

Paramedics were initially dispatched in a ground-based ambulance alone and were authorised to decide based on their situational assessment and specifications in standard operating procedures (SOPs), whether they needed tele-EMS physician involvement for the respective patient. However, if the patient unexpectedly presented on scene with an acute life-threatening condition, meeting the exclusion criteria, they were obliged to alert an EMS-physician and inform the tele-EMS physician. Whether they have appropriately followed this SOP was not directly assessed in this study, but indirectly by the occurrence of AEs.

Paramedics in a ground-based ambulance and the conventional EMS physician in a separate vehicle were dispatched simultaneously in a so-called rendezvous system and treated the patients on scene as in the clinical routine. If paramedics arrived first on scene, they were also authorised to cancel the involvement of a conventional EMS physician if not needed.

A standardised paramedic-EMS file was completed by paramedics, and if involved, also a physician-EMS file by physicians in both groups.

Blinding was not possible because of the obvious treatment procedure. A follow-up was carried out on days 30 and 90 after the intervention.

The primary outcome with respect to the non-inferiority of the tele-group was the frequency of four predefined intervention-related AEs during EMS treatment with suspected causality to the group assignment, as one would expect benefit or minimising risk having an EMS physician on scene.

The predefined AEs, which were not the original reason for the emergency call, were chosen by the authors because they represent the most common AEs associated with drug therapy in the EMS and comprised the following three new immediate AEs: (i) allergic reaction (documented in the free text or indicated by the use of anti-allergic drugs after any treatment by the EMS team); (ii) blood pressure decrease (necessity of vasopressors); and (iii) respiratory insufficiency (decrease in peripheral oxygen saturation < 90%, or need for assistive, manual or controlled ventilation). Further, it comprised one new AE with a time lag of 24 h, in order to capture also delayed complications possibly resulting from not guideline-conform treatment of the EMS: (iv) cardiac arrest within 24 h of the intervention. The term “intervention-related” covered beside wrong treatments (e.g. iatrogenic allergic reaction because of inadequate survey of medical history; iatrogenic blood pressure drop, apnoea or respiratory insufficiency after wrong drug dosage or selection; or circulatory arrest after wrong drug dosage, or wrong hospital referral) also the cases where the respective EMS did not provide any treatment. An independent clinical endpoint committee (CEC) judged whether the reported AEs were suspected to have occurred because of group assignment based on blinded data (see also Additional file 3).

Secondary outcomes regarding the superiority hypotheses for the tele-group were analysed by means of the EMS files and comprised the guideline-conform medical history survey, treatment and documentation quality, AEs independent of the EMS care used, and the necessity of a physician for the EMS operation. For further details on outcome measures, see Additional file 1 and Additional file 2. The proportion of fulfilled SAMPLE acronym [21] items (consisting of the survey of S = Symptoms, A = Allergies, M = Medication, P = Past pertinent medical history, L = Last oral intake, and E = Events leading to present Illness/Injury) indicated the quality of the medical history survey, which was assessed by the proportion of documented patient vital data in accordance with the respective guidelines [22]. Treatment quality was evaluated by the frequency of adherence to six predefined clinically relevant components (correct hospital choice, electrocardiography (ECG) use, intravenous access placement, non-invasive blood pressure, oxygen saturation, and blood glucose level). The secondary outcomes evaluated by means of automatically generated reports of the dispatching software comprised the physician engagement time and further operating times, including cancellation of the EMS physician or conversions between the groups. The secondary outcomes death within 24-h, 30-days, 90-days and hospitalisation; discharge destination; intensive care unit and hospital length of stay were analysed using the EMS files, hospital database, and follow-up interviews.

Data are presented as means and standard deviations (SDs) for continuous variables, and as frequencies and percentages for categorical variables. Because of the lack of evidence for EMS-dependent AEs, our sample size calculation was based on analysis of our own data with an assumed AE rate of 2% for the conventional physician-based EMS. We assumed a non-inferiority margin of 1.5% and a power of 80%, and we allocated an overall significance level of 5% to K = 3, as two interim analyses for non-inferiority were planned [23]. (Additional file 3). All AE cases were included in the interim analyses of the primary endpoint (Additional file 2).

The final analysis of all randomised cases was performed at a significance level of 2.25%. Considering an assumed dropout rate of 10% evenly distributed over the stages and groups, the calculated sample size using Addplan 6.0 [24] was 3344 patients or 1672 per group. Statistical analyses were performed using SAS (version 9.4; SAS Institute Inc.). To evaluate the non-inferiority of the tele-group regarding the rate of causal AEs, the respective 99.97% and 99.29% confidence intervals (CIs) for the rate difference between the groups were calculated (proc FREQ in SAS). For the final evaluation of the primary endpoint, the 97.75% CI of rate differences (proc FREQ in SAS) was calculated using the Newcombe hybrid score method, as only one causal AE was observed in the tele-group.

To evaluate all dichotomous secondary endpoints, we used two-sided chi-square tests (proc FREQ). The exact Fisher test was applied for values < 5, and no further evaluation was performed in the case of 0 events in both groups. All continuous endpoints were evaluated with two-sided t-tests (proc TTEST), and the bootstrap statement with 10 000 bootstrap samples was used to calculate the bootstrap t intervals for skewed data.

Except for the primary endpoint, p-values < 5% were considered statistically significant. Deviations from the original TSAP are described in the Additional file 3.

From 13 August 2018 to 5 September 2019, 3531 of 25,900 out-of-hospital emergency patients were enrolled and randomised into either the control group (n = 1767) or tele-group (n = 1764) (Fig. 1). A manual conversion of the assigned treatment group was performed in 119 cases by the dispatching personnel or the initially dispatched team (n = 99 were switched from tele-EMS to conventional physician-based EMS treatment; n = 20 vice versa). The paramedics cancelled the involvement of the respective EMS-physician after arrival at the emergency scene because of their own situational judgment in 108 of 1676 cases (6.4%) and 893 of 1544 cases (57.8%) in the control and tele-groups, respectively. Thus, 1568 and 651 patients received treatment by an EMS-physician in the control and tele-groups, respectively. A total of 3220 patients were analysed for the primary endpoint (Fig. 1).

Tables 1 and 2 present the data for the entire population and the cases with physician contact. The baseline characteristics were similar in both groups (Tables 1; Additional file 3: Table S1). Patients had a mean age of 61.3 years and 53.8% were female.

The statistical analysis using the Newcombe hybrid score method revealed that the non-inferiority margin of − 0.015 was not covered by the 97.75% CI of − 0.0046 to 0.0025. Thus, the non-inferiority hypothesis could be confirmed (Table 2), indicating that tele-EMS is non-inferior to the conventional physician-based EMS. Only one of the intervention-related AEs had suspected causality to the group assignment. This AE occurred in the tele-group (a patient with cancellation due to tele-EMS involvement and treatment by paramedics only). A hypertensive emergency was not medically treated in the prehospital setting as the guidelines would have demanded, and the patient had a cardiac arrest within 24 h of the emergency call due to intracranial bleeding.

The secondary outcomes are presented in Tables 2 and 3 and Additional file 3: Tables S2–S4.