A consensus-based template for documenting and reporting in physician-staffed pre-hospital services

Background Physician-staffed pre-hospital units are employed in many Western emergency medical services (EMS) systems. Although these services usually integrate well within their EMS, little is known about the quality of care delivered, the precision of dispatch, and whether the services deliver a higher quality of care to pre-hospital patients. There is no common data set collected to document the activity of physician pre-hospital activity which makes shared research efforts difficult. The aim of this study was to develop a core data set for routine documentation and reporting in physician-staffed pre-hospital services in Europe. Methods Using predefined criteria, we recruited sixteen European experts in the field of pre-hospital care. These experts were guided through a four-step modified nominal group technique. The process was carried out using both e-mail-based communication and a plenary meeting in Stavanger, Norway. Results The core data set was divided into 5 sections: "fixed system variables", "event operational descriptors", " patient descriptors", "process mapping", and "outcome measures and quality indicators". After the initial round, a total of 361 variables were proposed by the experts. Subsequent rounds reduced the number of core variables to 45. These constituted the final core data set. Emphasis was placed on the standardisation of reporting time variables, chief complaints and diagnostic and therapeutic procedures. Conclusions Using a modified nominal group technique, we have established a core data set for documenting and reporting in physician-staffed pre-hospital services. We believe that this template could facilitate future studies within the field and facilitate standardised reporting and future shared research efforts in advanced pre-hospital care.


Background
Pre-hospital emergency medicine aims to provide early, high-quality, goal-directed treatment. The time to "restoration of adequate flow and physiology" is recognised as crucial in several conditions, including cardiac arrest, cardiac ischaemia, severe trauma, sepsis, respiratory distress, stroke and obstetric emergencies [1]. As our knowledge of acute illness and injury evolves, rapid diagnostics and treatment are becoming more important.
It has been argued that an appropriately trained physician is able to provide superior treatment compared to other emergency medical care providers, and, based on this assumption, having a physician as an integrated part of pre-hospital emergency medical care will positively affect patient outcomes [2][3][4][5]. These services typically consist of rapid and flexible means of transportation. In larger cities, rapid response cars are often available, and helicopters are common in rural areas. These means of transportation [6,7] facilitate rapid access to the patient and can reduce the time to definitive care at appropriate medical facilities.
The role of the physician-staffed pre-hospital services is not solely medical treatment. The physician often takes an active part in patient and resource triage and acts as a medical advisor/supervisor for ambulance crews.
In Scandinavia, physician-staffed pre-hospital services are well-established and commonly manned with anaesthesiologists to deliver critical emergency care. These services are publicly funded and operate 24 hours a day. The means of transportation (helicopter, rapid response car, ambulance) are provided according to logistical needs. The majority of services are well integrated within their local emergency medical systems [6].
Physician-staffed pre-hospital services demand more resources and are more expensive than standard paramedic-staffed services. The evidence base for which patient groups benefit from physician-manned services and which system factors are related to a favourable patient outcome is currently weak [8].
More research is required to evaluate the effect and efficiency of such services [9]. The well-organised and relatively homogeneous systems of Scandinavia are suited for collaborative research projects. Unfortunately, proper documentation and reporting of medical and operational activity is uncommon and the data that is recorded lacks common variables and definitions. This lack of systematisation of documentation within and between countries is nearly generic in an international perspective and makes it difficult to compare services and to conduct systematic searches in the published literature and multicentre-based prospective or registrybased research [10][11][12][13].
In several emergency medical fields, shared research efforts based on common reporting templates have proven useful [14][15][16][17]. The Utstein Template for documenting and reporting in cardiac arrest is probably the most widely used template and has facilitated a large amount of research [18][19][20][21]. The rationale is to document and report data prospectively with a predefined template developed by experts [22][23][24]. These templates gather data that can be easily and routinely reported, making multicentre-based research feasible [25,26]). Considerably less agreement currently exists on the reporting of the effect and efficiency of emergency medical systems. There is no international agreement on the reporting of fundamental variables, such as outcome variables [10]. Research addressing structure, process and outcome quality indicators, and factors associated with increased value is lacking [27,28].
Templates for documenting and reporting data may be developed using multidisciplinary expert panel consensus methods [27,29,30]. Expert panel assessments allow a combination of evidence-based knowledge, personal experience, and general insight in the characteristics of the patient cohort assessed or problem addressed [31], and has been the method of choice in the development of previous Utstein templates [24,[32][33][34][35][36][37] The aim of this project was to develop a common Utstein-like template for documenting and reporting on pre-hospital emergency medical services manned with trained pre-hospital physicians.

Methods
The current work is the third part of the ScanDoc project, a Scandinavian consortium working for better research infrastructure in advanced pre-hospital care.
We used a modified nominal group technique [31], and defined "physician-staffed pre-hospital unit" as a dedicated unit staffed with a physician with competency in critical emergency care above that of an ordinary oncall general practitioner.

The experts
The ScanDoc representatives assisted in proposing experts from the Scandinavian countries. Several countries in central Europe operate pre-hospital physician staffed services similar to those in Scandinavia, and, therefore, it was decided to invite representatives from other countries in Europe. As the original purpose of the ScanDoc-project was to standardise reporting from the Scandinavian countries, the project group decided to weight the panel composition toward Scandinavia, resulting in three representatives from Scandinavia per one from another European country. Based on the following inclusion criteria, panel members were recruited from Sweden, Finland, Denmark, Norway, Italy, Germany, Austria, Switzerland, Holland and the United Kingdom.
1. Relevant clinical experience working in a physician pre-hospital service to ensure personal insight into the operative and medical characteristics of advanced prehospital care.
2. Scientific and/or substantial leadership responsibilities in pre-hospital care to ensure competency in research methods and governance of pre-hospital emergency systems.
3. Ability to communicate in English. The experts were identified using Medline and Google Scholar searches and via the professional network of the project group. Each expert was invited into the consensus process by e-mail or telephone when feasible. Non-responders were reminded three times by email, followed by direct contact from one of the members of the project group. The selected expert panel participated throughout the complete process, and no additional experts were included after initiating the process.

The modified nominal group process
The consensus process was arranged in a four-step sequence (details of each stage are described below). Stages 1, 2 and 4 were carried out by email. During stage 3, the panel met in Stavanger, Norway for a 2-day consensus meeting.
Keeping proposals anonymous is of major methodological importance to reduce the influence of "loud-speaking" experts and to facilitate the influence of their "silent-speaking" peers [31,38]. The expert panel members were aware of its composition, but anonymity related to proposals was secured until stage 3. Stage 1. The expert panel received a predesigned worksheet by email. Each expert was asked to propose the ten most important variables to be routinely documented for shared research and benchmarking within each of five predefined sections: 1. Fixed system variables Variables relating to system characteristics concerning how the service is organised, the operational capacities of the service and its integration with the EMS with which it operate.

Event operational descriptors
Variables documenting the context of a mission (dispatch) or episodes of use (for services with advisory functions).

Patient descriptors
Variables documenting information related to the patient's profile, e.g., age, gender, co-morbidity and type of medical complaint.

Process mapping
Variables recording what happened to the patient and how the episodes of care proceeded. 5. Outcome and quality indicators Variables relating to patient and/or mission outcomes, as well as measures of quality.
An optional sixth section for proposals of variables that did not fit into one of the predefined sections was provided. The experts were informed that a subsidiary aim of the process was to establish a core data set that was easy to routinely collect and did not require excessive database alterations.
The project organisers recognised the challenge that several reports on how to document and report data in various parts of emergency care have already been published and implemented [24,32,37,39]. These templates (e.g., Utstein for cardiac arrest, the Utstein template for major trauma and the Utstein template for the reporting of advanced airway management) contain some common variables with slightly different definitions. The expert panel was supplied these published templates and asked to make the new variables compatible with existing template variables, if feasible.
The proposed variables were returned to the project group by email and systemised. Different variables with identical meaning were combined carefully so as not to interfere with the expert panel's proposals. No single proposed variables were deleted. The variables within each section were ranked according to how many times the variables had been proposed by the different expert panel members.
Stage 2. The revised worksheet containing aggregated results from stage 1 was sent to the expert panel. The panel were then requested to rank the ten most important variables in each section from 10 (most important) to 1 (least important). The variables with no ranking were then removed from the list. The results from this ranking provided the basis for the consensus meeting.
Stage 3. The expert panel gathered in Stavanger, Norway and, during a 2-day meeting, agreed upon a core data set. At this meeting, moderators (DL and HML)) led the experts through the results from stage 2. The experts were divided into two groups and discussed portions of the preliminary dataset. The experts subsequently presented their discussions in plenum, and variables were discussed, debated and agreed upon. The variables were decided upon on day 1 and were defined and categorised on day 2. Some variables were not finally defined during the meeting, and the expert panel approved the project group to propose definitions for the remaining variables before stage 4. Stage 4. Based on step 3, the final data set, including definitions, was prepared by the project group and submitted to the expert panel for final approval. At this point, no additional variables were accepted, but minor changes related to answer categories and definitions were allowed.
Consensus was defined as agreement on the proposed variables at the consensus meeting among the attending experts. Furthermore, we informed the experts during stage 4 that no response was interpreted as agreement to the final core data set.

Results
Of the 17 invited experts, 16 expressed willingness to participate. Fifteen experts returned their proposals during stage one and two, and ten experts attended the consensus meeting. The final core data set was sent to all 15 experts.
In stage one, a total of 361 unique variables were proposed by the experts (additional file 1). After the aggregation of variables with similar content by the project group, 162 different variables were returned to the experts in stage 2 (additional file 2). During stage 2, 98 variables were rated and included in the preliminary data set discussed at the consensus meeting (additional file 3).
The final core data set consisted of 45 variables (Table 1,2,3,4 and 5) of which seven consisted of the same variable (physiological data) measured at two different time points (first and last values).
After stage 4, six experts returned minor comments related to a small number of the established definitions. A few minor changes were made accordingly. Moreover, some issues raised in this stage may be important for future revision of the data set. A revision of the template is planned in approximately two years.     The "fixed system variables" section contains nine variables, all of which relate to the context of care (Table 1). In the "event operational descriptors" section, seven variables were included derived from four time variables (Table 2). In the third section, "patient descriptors", nineteen variables were decided upon (Table 3). Eight variables were included to describe the diagnostic and therapeutic interventions within the episode of care in the "process mapping" section (Table 4). Finally, 2 variables were included in the "quality Indicator and mission outcome" section ( Table 5).

Background information and fixed system variables
The expert panel recommended that a description of the system under investigation be reported. These data will typically be "fixed", as they are not patient specific. In cooperative research, these variables can significantly influence comparisons between systems [40][41][42]. The expert panel divided this section into two parts; one related to the service's context (funding, integration within local EMS, medical capacities, training programs required) and the other including more specific variables, such as medical specialty required for staffing physician, population and service area covered, mode of transportation, operating hours and dispatch system. Strictly defining these variables makes it more straightforward to analyse the influence of system factors.

Event operational descriptors Mission times
Most physician-manned pre-hospital services document time variables, but there is no international consensus of the definition of these variables. Since emergency care is often time-critical, standardising how time is documented seems paramount. The expert panel agreed upon documenting time variables as exact times (hh:mm) and not as intervals. Most data registration tools will easily calculate the desired intervals, and it was thought that exact times would provide greater flexibility in statistical handling than intervals.

Call receipt at Emergency Medical Communication Centre (EMCC)
Response time is commonly documented, and frequently defined as the time from alarm to crew arrival at the scene. This time interval indicates the system's ability to respond to an emergency in a timely manner and is used as a quality indicator in the European Union [43]. The expert panel realised that the receipt of the emergency call at the EMCC is more difficult to collect than the actual alarm time to the attending crew. Nevertheless, the time of receipt of the emergency call from the scene is closer to the actual out-of-hospital time from the patient's perspective, and the expert panel agreed to include this time as the "start of event" indicator.

Type of dispatch
This indicator variable is important for benchmarking purposes [44], and will not only describe the crude mission profile of the service, but will also be useful for the stratification of mission type data extraction. Some physician-staffed pre-hospital services are dispatched to undifferentiated emergency calls but many are targeted to particular types or severity of incident [6,7]. A substantial workload is imposed onto some services, because of consultation responsibilities for lower-level services (paramedics, dispatch centre) [6]. The expert panel emphasised that this function should also be documented.

Patient descriptors Co-morbidity
Robust risk-adjustment measures are fundamental for any effort to compare the characteristics of care against hard outcomes [45]. Clearly, the same treatment for the same medical problem can yield different outcomes in groups of patients with different levels of co-morbidity [46]. The expert panel included the well-established and recommended American Society of Anesthesiologists Physical Status Classification System (ASA-PS) [47], albeit in a dichotomised form, as ASA-PS = 1 or ASA-PS>1, as an indicator of co-morbidity. It was considered that while it would be ideal to use the full ASA-scale, obtaining the complete medical history from seriously ill patients is likely to be impossible. It is emphasised that the patient's morbidity prior to the acute medical incident is to be recorded.

Medical problem
It is crucial for benchmarking to have key variables to serve as inclusion criteria. Because an accurate diagnosis , the expert panel recommended inclusion of a complaint-based indicator for the medical reason for dispatch to better reflect the real-world setting.

First and last values
As monitoring technology evolves, the amount of vital patient data is almost unlimited, and monitoring strategy is often case-specific. The expert panel recommended documenting these vital data: Glasgow Coma Scale [49], heart rate, systolic blood pressure, cardiac rhythm, oxygen saturation (SpO2) and respiratory rate. Typically, such measurements are recorded in a continuous pattern but reporting a complete dynamic profile of the patient's physiology during the process of care will be impractical. The expert panel recommended to report the first and last of these measurements. This approach will form the basis for later analyses of therapeutic and diagnostic interventions related to the patient's clinical condition within the episode of care [50].

Cardiac rhythm
Advanced anti-arrhythmic treatment is usually confined to physician-staffed pre-hospital services. Several experts noted that recording the cardiac rhythm is often unreliable and not always indicated. Nevertheless, interpretation of cardiac rhythm and treatment of malignant arrhythmias is clearly an advanced intervention. This intervention should be documented, and a four-point scale was accordingly proposed. This scale is inspired from the Mainz Emergency Evaluation Score and separates the cardiac rhythm into normal (sinus rhythm), pathological (atrial fibrillation, AV-block grade 1-2), pathological with circulatory compromise, and cardiac arrest [51]. The expert panel believes that this classification can be applicable using a 3-channel electrocardiogram monitor.

Pain
Adequate analgesia is regarded as good clinical practice [52], is one major task for pre-hospital services, and should be documented accordingly [53]. There is currently no consensus on which method for documenting pain is most applicable in a pre-hospital setting [54]. Several of the experts argued that the visual analogue scale (VAS) should be used because it is the most established scale in pain assessment. Nevertheless, use of the VAS requires a conscious and cooperative patient, which is not always available in emergency care. A reasonable approach is to document a more coarse scale assessed by the physician. The experts recommend including a three-level pain scale was chosen, consisting of the following classifications: no pain, moderate pain and severe pain.

Process mapping
Diagnostic procedures were included in an effort to document the specially trained physician's competency in diagnostics. The expert panel agreed that this role in diagnostics will probably be more important in the future. Therapeutic procedures were systemised according to the traditional ABCD algorithm, and the procedures used in treatment of each were recommended for reporting accordingly.

Medications
The expert panel agreed upon dividing the medication variable into two types. The first type describes whether any medication was administered, whereas the second documents which type of medication was administered. Emphasis is placed on the type of medication that usually is administered only by physicians. These medications include potent analgesics and sedatives, neuromuscular blocking agents, vasopressors, intravenous antibiotics and anti-arrhythmics.

Quality indicator and outcome measure
The expert panel suggested the outcome measure should be a description of mission outcome rather than a specific patient medical outcome. Although the recording of long-term patient outcome is ideal it was recognised that this is not achievable in many systems because services often transport patients to different locations and medical facilities.
The expert panel did not agree upon a universal quality indicator during the consensus process. Nevertheless, there seems to be a common opinion that such quality indicators should be developed, and that improvement or deterioration in the clinical condition during the episode of care can serve as a relevant and patient-centred quality indicator. How such a quality indicator should be developed was outside the scope of this consensus process. Several proposed variables can prove useful for selection criteria in future benchmarking and research: variable 11, "type of dispatch", will indicate the main operational indication for dispatch, whereas variable 17, "medical problem", will further indicate the medical basis for dispatch. When establishing categories within these variables, a compromise between precision and user-friendliness must be achieved; too many categories will be accompanied by difficulties in definitions, whereas too few will result in loss of precision. Therefore, an extended version of the "first-hour quintet" is proposed which [43] will include > 90% of the medical complaints met in daily service for most services (unpublished data from ScanDoc phase 1b).

Primary findings
Using experts within the field and the modified nominal group technique, we defined a core data set to standardise documenting and reporting in physician-staffed pre-hospital care.
If implemented, we believe that shared research efforts within the field will be enhanced.
The fairly strict definitions and categories within each variable will form the basis for robust documentation and facilitate the extraction of comparable data. The established variables are recommended to reflect ordinary pre-hospital care but also seek to document variables on the specific functions of specially-trained physicians and their exclusive diagnostic and interventional capacities. In addition to being experts in securing vital functions (airway, breathing and circulation), these physicians can apply more aggressive pain control, interpret electrocardiograms and perform advanced diagnostics [8,[55][56][57][58]. The expert panel agreed on the importance of robust documentation of these advanced interventions, especially those related to the ongoing debate on whether advanced pre-hospital care improves patient outcome.
Shared research efforts require unique patient group identifiers. In the current core data set, several variables can act as such identifiers. Traditionally, pre-hospital treatment strategies have been symptom-based. A diagnostic uncertainty is generally present, and interventions are applied according to the physician's preliminary diagnosis. In the dispatch phase, this diagnostic uncertainty is even more present. Nevertheless, as more advanced diagnostic equipment, such as ultrasound and near patient biochemical tests, are commonly available, diagnostic precision may improve. However since this level of diagnostic equipment is not always accessible, a focus on chief complaints remains fundamental.

Implementation of core data
The expert panel emphasised that it should be possible to use the core data set without replacing the current infrastructure of documentation. This project does not intend to construct a unique documentation software, as this software could interfere with already developed and wellfunctioning local systems. Therefore, we believe the main benefit of this data set is to standardise current reporting and create increased understanding of how variables should be documented and defined in the future.

Limitations
Using a consensus methodology to develop a core data set for physician-manned pre-hospital services presents methodological challenges exist [38]. First, there is a risk of selection bias when appointing experts. We believe that the multi-national selection of experts with predefined selection criteria reduced the risk of establishing a biased expert panel. Second, a well-known risk related to the project group's influence on the expert panel is a provision of biased literature before the first round [31]. The project group did provide the expert panel members with a selection of background literature because previously developed and implemented templates were too important to be neglected. Nevertheless, the possibility of this "literature-based" bias seems minimal, given the huge number of variables proposed in the first round.
The specific operational and medical characteristics of the EMS connected to individual experts might have influenced the priorities and proposals during the process. This problem was evident during the consensus meeting, but the panel agreed that the proposed core data set was feasible for most mission types. It is underlined that the current data set represents a minimum, and supplementary variables usually are documented according to service preference.

Summary and conclusions
Using a modified nominal group technique, we have established an Utstein-like template for documenting and reporting in physician-staffed pre-hospital services. The core data set consists of 45 variables grouped in five different categories. We recommend that future studies in the field report their data according to the template. We believe the template can facilitate future shared research efforts in advanced pre-hospital care.

Additional material
Additional file 1: Results Stage 1. All proposals from experts after completion of stage 1 Additional file 2: Results Stage 2. All rankings from experts on proposals from stage 1 Additional file 3: Proposals before consensus meeting. All proposals organized by the sum of ranking points from stage 2.