Template for documenting and reporting data in physician-staffed pre-hospital services: a consensus-based update

No exact criterion exists concerning selection of participants for a Delphi study.

Many European countries share similarities with regards to infrastructure, socio-political system and health care services, favouring research collaboration [17]. Representatives from European p-EMS were invited to join an expert panel using the same inclusion criteria as the original template:

The experts were identified via the European Prehospital Research Alliance (EUPHOREA) network. The EUPHOREA network consists of representatives from p-EMS throughout central Europe, UK and Scandinavia. Experts were invited via e-mail. Non-responders were reminded via e-mail. For all rounds non-responders were reminded twice per e-mail.

A Delphi technique was applied to achieve a consensus from a panel of selected experts interacting via e-mail. No physical meetings were held. A research coordinator interacted with the participants, administered questionnaires and collected the responses until a consensus was reached. The experts were blinded to each other until an agreement was reached. All responses were anonymized. The Delphi process ran from Feb. 19 to Oct. 1, 2019. The final dataset was approved by all experts.

The experts were asked to propose variables within each of five predefined sections:

There was also an optional sixth section for proposals of variables that did not fit into one of the pre-defined sections.

Each expert suggested 10 variables considered to be most important for routine documentation in p-EMS within each of the five predefined sections.

The results from the first round were structured in a worksheet (Excel for Mac, version 16.31, 2019 Microsoft). Duplicate suggestions were removed before the variables were returned to the experts. Variables from the original template were included if not suggested by the experts. Experts were asked to review and rate all variables from 1 (totally disagree) to 5 (totally agree) based on relevance.

Variables rated ≥4 by more than 70% of the experts were included in the template draft and presented to the experts [18, 19]. In addition, the experts received a number of questions pertaining to the wording of questions, consent to delete some questions because of overlap, relevance of alternatives under a main question, and whether there should be a free-text field for addressing key lessons. Furthermore, they were instructed to provide comments and grade the variables as either compulsory or optional. Later, the experts were asked to suggest the frequency of variable reporting (for each mission, monthly or annually). Variables rated ≥4 by less than 50% of the experts were excluded. Variables rated ≥4 by more than 50% of the experts were summarized and re-rated by the experts. If more than 70% of the experts rated a variable ≥4 in this second round, the variable was included in the final template.

After summarizing the feedback from round III, the list of variables achieving consensus, accompanying comments, and further questions were distributed to the experts. All variables were numbered. This round provided an opportunity for the experts to revise their judgements and combine similar variables.

Feedback from round IV was summarized into a final version of the template and sent to the experts to elicit any objections and/or to give final approval of the template for routine reporting in p-EMS.

The study was drafted according to the Standards for Reporting Qualitative Research (SRQR) [20].

Thirty experts were invited to join the consensus process and 15 agreed to participate. Eleven experts responded in the first Delphi round, ten responded in the second round and nine responded in the last three rounds.

The experts suggested 194 unique variables in the first round (Fig. 1). All variables from the original template were among the suggested variables.

The experts rated the variables suggested in round I from 1 (totally disagree) to 5 (totally agree) based on relevance. A total of 68 main variables (24 fixed system variables, 10 event operational descriptors, 15 patient descriptors, 10 process mapping variables, 9 outcome and quality indicators and no other variables) were rated ≥4 by more than 70% of the experts and included in the preliminary template. Thirty-five main variables and 32 sub-variables were rated < 4 by 50–70% of the experts. Ninety-one variables were rated ≥4 by less than 50% of the experts and were excluded.

The preliminary template was presented to the experts. Additionally, the experts rated the 35 main variables and 32 sub-variables that were initially rated ≥4 by 50–70% once more. Five more main variables and 9 sub-variables were included after this second rating. In total, 73 main variables were included (Fig. 2). The experts agreed that all fixed system variables should be reported annually while all event operational descriptors, patient descriptors, process mapping variables and outcome and quality indicators should be reported after each mission.

The included variables were presented to the experts. After feedback from the experts the wording of variables 1.23.6 and 3.5.6. were changed from “Chest pain, excluding MI” to “Chest pain, MI not confirmed”. Variable 3.8.4. “Systolic blood pressure (SBP) not recordable” and 3.10.4. “SpO2 not recordable” were added. Variables 3.13.1. and 3.13.2. were changed to record the VAS score instead of pain as none, moderate or severe and variable 4.6.17. was changed from “Resuscitative endovascular balloon occlusion of the aorta (REBOA)” to “Endovascular Resuscitation (EVR)”.

The experts approved the final version of the template (Table 1, 2, 3, 4 and 5).

Using Delphi methodology, we have updated a template for standard documentation in p-EMS. The new dataset includes new data variables and the template was expanded from 45 to 73 main variables.

Throughout the world, there are large differences between p-EMS [21‐23], and fixed system variables are important to analyse any influence of system factors and compare systems [11, 24]. The experts suggested reporting all fixed system variables annually. Furthermore, the experts chose to include two variables related to quality. The reason for including these data in this section is that they describe the quality of the system rather than the quality delivered during each mission.

There is no consensus in the literature on how to report mission times [15, 25, 26] and the experts had several suggestions, i.e., exact times (hh:mm), time intervals (dispatch time, on-scene time, etc.) and time reported as year/month/day/hour of event. Response time (time from unit is dispatched to at patient side), on-scene time and transport time (from patient leaving the scene to arrival at the hospital) and time from alarm to arrival at the hospital are all reported in various templates. We argue that by reporting exact times, all desired time intervals can easily be calculated; therefore, exact times should be documented.

The time of the event is usually not possible to accurately identify. In trauma, the time of the event will be distinct, but for other diagnoses a clearly defined start time is often missing. The time when a call is received at the emergency medical communication centre (EMCC) is a distinct time that is easy to document, substituting for the time of the event. This was also emphasized by the experts.

P-EMS differ in service profile, and documenting dispatch type is important for benchmarking. Some services are dispatched to all types of emergency missions, whereas others are dispatched to specific types, e.g., trauma. Some services have an extensive workload due to consultation responsibilities and medical direction for ordinary EMS. This may affect availability if work hours are restricted.

Comorbidity is an important risk adjustment measure, but there is no consensus on comorbidity reporting. The original template for reporting in p-EMS used the American Society of Anesthesiologists Physical Status (ASA-PS) scale in a dichotomized form. However, using full ASA-PS scale has been found to be feasible in p-EMS [27], and it is recommended by the experts.

Reporting the present medical problem is crucial for benchmarking. P-EMS have traditionally reported symptoms, but point-of-care diagnostical options are increasingly available, allowing more precise pre-hospital diagnoses [28‐30].

The experts recommended reporting physiological data at two different time points: at arrival of the p-EMS and at hand-over or the end of patient care. This corresponds with the original template. Reporting data at two different time points allows for monitoring changes in the patient state and may serve as a surrogate measure for p-EMS performance [31]. For SBP and SpO2, the experts also suggest reporting the lowest value measured. Hypotension is an independent predictor of mortality for traumatic brain injury (TBI) patients [32], and reporting the lowest SBP value will capture hypotensive episodes. Further, automated data capture from monitors are increasingly available, enabling continuous measurement of physiological variables. Continuous reporting may capture dynamic changes in patient state, thereby increasing the precision of p-EMS research.

Pain is frequent in the p-EMS patient population, and pain relief is considered good clinical practice [33]. The original template used a three-part scale for reporting pain while the expert group of the revised template suggest reporting pain according to the Visual Analogue Scale (VAS) [34].

The resulting physiological effects of p-EMS treatment and its relation to outcome remains largely unknown in pre-hospital critical care. Such changes in physiology have earlier been difficult to capture but doing so is now more feasible due to technological developments. The experts emphasized this, and as such an expansion of the process mapping section was suggested.

To date, there is no agreement on standard quality indicators in p-EMS but Haugland et al. recently developed a set of quality indicators for p-EMS [35]. Several of these indicators are documented in the revised template but under various sections. Additionally, the experts suggested several other context-specific quality variables related to the individual patient, but these are yet to be validated.

The experts recommend an event-specific long-term outcome measure to be included on a regular basis. The feasibility of capturing this variable as part of a standardized documentation in the p-EMS population remains to be determined.

Several consensus-based templates for reporting in EMS and p-EMS have been created (e.g., trauma, airway handling and cardiac arrest) [14, 15, 26, 36], and studies have proven that data collection according to such templates are feasible [12, 16, 37]. However, to increase the relevance of templates, variables should be coordinated. Of 26 variables in the template on quality indicators in p-EMS [35], five are identical to variables in the current template, six can easily be calculated and three are partially similar. Thus, little extra effort is required to document according to both templates. We believe that the coordination of variables and linking of templates will add value by reducing workload and increasing data capture, thereby facilitating future p-EMS research.

P-EMS are constantly developing, with new diagnostic and therapeutic options available, e.g. pre-hospital blood products, Tranexamic acid, extracorporeal membrane oxygenation (ECMO), thoracotomy and endovascular resuscitation on-scene. To capture these important trends, templates need to be updated regularly. Additionally, the variables shown to be not feasible to document should either be changed or removed. Physiological variables are often reported to be the most often missing variables [38, 39]. In the original template we found the feasibility of collecting physiological data to be good [16], and these variables were not substantially changed in the updated template. Thus, we expect feasibility to be good for physiological variables in the updated template as well.

To be able to compare outcomes, data must be unambiguously defined [26]. A data dictionary with precise definitions will be created for the present template. Furthermore, when implementing the template, it is important to ensure that all requested data are collected. Each service is free to choose whatever supplementary variables it wants, but all core variables should be captured by default, thereby facilitating future research.

Physician-staffed services are more expensive compared to ordinary EMS services making it a limited resource. This emphasize our obligation to use the service for the right patients. Therefore, we continually should strive to identify patients where p-EMS has an additional effect.

To provide a tool for collection of high-quality data is only a first step towards the improvement of p-EMS research. The next step is implementation, which is pivotal for template success. Aiming to increase awareness of the template, we invited experts from all over Europe to participate in its development. We believe this may facilitate implementation. Furthermore, to increase the implementation rate of the template, targeted efforts, such as involvement of stakeholders and highlighting the possibilities which lies within template data research, must be initiated.

Registries (e.g. for trauma and cardiac arrest) have facilitated a large amount of research [14, 40, 41]. In p-EMS there is currently no joint register and each national service manages its own data. Furthermore, data are often registered on paper and later converted to digital format. Automated data capture from monitors and updated digitized data catchment tools could allow for complete template data to be imported directly into a common registry. This would provide a substantial opportunity for joint research. If such a registry could also link template data to outcomes and standardized coding systems for process and outcome issues, we may be able to assess e.g. for which patients p-EMS are useful, which procedures should be performed out-of-hospital and which procedures should not. However, the ethical and legal requirements of data sharing for research purposes (e.g. General Data Protection Regulation (GDPR)) must be taken into account and a substantial work to adhere to the current regulations are needed to succeed.

In the present study, we applied a Delphi method. This approach is in contrast with the Nominal Group Technique (NGT) that was used in the development of the original template. The classic Delphi method applies questionnaires with e-mails whereas the NGT involves a physical meeting with experts to reach a consensus [42]. The methods can also be combined into a modified NGT that starts with a Delphi process and ends with a physical meeting as a final step before consensus. Because this is an update of an existing template, we considered a physical meeting to be unnecessary. Furthermore, we wanted to ensure anonymity of the experts to prevent authors from favouring certain responses.

Reaching agreement is fundamental in Delphi studies, but a commonly accepted definition of consensus is absent [43]. In the present study we defined consensus as variables rated ≥4 (on a scale from 1 to 5) by > 70% of experts. We consider this a transparent and systematic method for reaching a consensus.