Adherence to the ABCDE approach in relation to the method of instruction: a randomized controlled simulation study

Nurses, nurse practitioners (NP), physician assistants (PA), pediatric residents, neonatal fellows, and neonatologists employed in the neonatal intensive care (NICU) and high care units of the Radboud medical center participated. In as much as NP, PA, and pediatric residents performed the same work on the ward, they were collectively referred to as neonatal ward clinicians. Participants were automatically recruited for this study, since they had to take part in periodic neonatal advanced life support (NALS) simulation training in this center. The only additional eligibility criterion was the availability to participate in the training program during the study period. The exclusion criterion was refusal to give informed consent to use the video recordings of the simulation scenarios for research purposes. This did not occur; written informed consent was obtained from all participants. The eventual number of participants depended on the number of trainings given within the time frame of the study and the attendance rate of the healthcare professionals. Background characteristics were collected with a short questionnaire (Additional file 1).

Simulations of an existing training program were used, since scheduling separate simulations exclusively for this study was not possible. The latter would interfere too much with the usual activities of the simulation center, and it was expected that a limited number of healthcare professionals would be able to attend these additional, study-related simulations. On training days, once or twice a month, the instruction method for that day’s team of eight nurses, one neonatal ward clinician, and one neonatal fellow/neonatologist was randomly allocated using sequentially numbered opaque sealed envelopes (SNOSE). Randomization was performed by a person not involved in the study, with an allocation ratio of 1:1. The intervention group was trained using VBI, the control group was trained using a CL. Each training day lasted eight hours. Except for the instruction method, training days were identical in both groups. Participants were asked not to inform future participants about the contents and proceedings of the training.

The instructional video (15 min) showed a NICU nurse and senior pediatric resident consecutively performing perfect demonstrations of the ABCDE approach during a NALS simulation scenario. The video was interspersed with brief screenshots with text emphasizing the tasks performed for the various domains of the algorithm. The video was specifically made for this study by a neonatologist with experience in simulation-based training (MB). The lecture (20–30 min) concerned a PowerPoint presentation of 30 slides (Microsoft Office PowerPoint 2007), in which the application of the ABCDE approach during NALS was explained step-by-step. The lecture was given by an experienced simulation specialist, the video was shown without additional comments. Questions of the participants were only briefly answered to elucidate the content of the lecture or video, without digressions. Both lecture and video were presented once, and they were only available on training days, not online or elsewhere. MB meticulously checked that the video and lecture contained the same teaching content.

After instruction of the ABCDE approach, participants received a 15 min orientation to the simulation environment, equipment, and manikin by the simulation operator. The simulation room closely mimicked the clinical work environment, including a patient bed, monitor, ventilator, T-piece oxygen delivery device, suction, intravenous setup, defibrillator, and crash cart with basically the same equipment and medications as available on the ward (Table 1). All equipment had standardized locations within the simulation room and was carefully demonstrated. The phone (to call for back-up assistance) and intercom (to communicate with the simulation operator) were introduced to the participants. All features, possibilities (e.g. vascular access, umbilical cord catheterization, chest drain placement, tracheal intubation, pupil reflexes, and clonic seizures), and impossibilities (e.g. changes in skin color and temperature, occurrence of skin abnormalities, chest retractions, and capillary refill time) of the manikin were clarified. An originally low-fidelity manikin (Newborn Anne, Laerdal Benelux, Amersfoort, the Netherlands) was used, which had been equipped with various features (mechanics to enable active breathing; internal speakers for crying, grunting, breath, and heart sounds) and recording capabilities (magnetic switch to measure compression depth; pressure sensor to assess hand placement, compression rate, and recoil; and flow sensor to quantify tidal volumes) by a technical simulation expert (Tim Antonius), transforming it into a high-fidelity manikin. Tidal volumes, airway pressures, and chest compression characteristics had been calibrated beforehand to ensure reliable mechanical measurements.

Subsequently, the team participated in four to five NALS scenarios. Each scenario was run by two different nurses, the same ward clinician, and the same fellow/neonatologist. The remaining participants (6 nurses) observed the performance in the adjacent briefing/debriefing room through a live audiovisual connection. Scenarios were controlled by a simulation operator from behind a one-way mirror. All scenarios started with the two nurses. This was usual practice in simulation training in this center. It mirrored clinical reality, in which a nurse instantaneously asked a colleague for help when a patient showed aberrant vital signs, so they performed the initial assessment together. When the situation deteriorated, the nurses summoned the ward clinician, who almost invariably called for back-up assistance by the fellow/neonatologist. Scenarios lasted approximately 15 min, during which all three profession categories had to perform a systematic ABCDE approach. The nurses performed the first ABCDE assessment when the newborn started to deteriorate. The ward clinician and fellow/neonatologist performed the second and third ABCDE assessments, respectively, upon their arrival at the scene. During assessment, the principle of a ‘sterile cockpit’ was applied. Notes, pocket cards, or other adjuncts were not permitted, because the aim was to investigate the effect of the instruction method on adherence in isolation. Otherwise, it would have been very difficult to control for differences in the use of these adjuncts. Each scenario was immediately followed by a non-scripted, video-assisted, operator-led debriefing (30 min) in the debriefing room with all participants, including the observing nurses [14]. Formative feedback was provided, not only on the ABCDE assessment, but also on other aspects, such as skill performance, clinical reasoning, differential diagnosis, and crew resource management principles. Debriefers were unaware of group assignment.

Various patient scenarios were used (e.g. apnea, arrhythmia, sepsis, seizures, and metabolic derangements). Scenarios involving resuscitation at birth were also part of the existing training program. The medical staff of the department considered it undesirable to discontinue these birthing scenarios for the duration of the study (> 6 months). Consequently, these scenarios did take place, but were excluded from the study, for they require another algorithm. The scenarios were semi-structured, with scenario progression according to prespecified triggering events. All scenarios were videotaped using three high-definition cameras: one on the manikin’s side, one affixed to the overhead radiant warmer, and one ceiling-mounted overview camera. These views were combined with the vitals from the monitor into one 4-screen window for assessment.

A suitable assessment tool to score adherence to the ABCDE algorithm could not be found. The assessment form developed by Drost-de Klerck et al. was also not compatible with the scoring procedure, for it was based on the slightly different ABCDE structure of the (adult) Advanced Life Support [15]. Therefore, a novel assessment instrument with a conventional, trichotomous scoring system was developed by MB and MH and applied in this study. All 24 items of the ABCDE approach, as presented in the Dutch Advanced Pediatric Life Support (APLS) course manual [4], were incorporated and categorized under the appropriate domains (A, B, C, D, and E) (Additional file 2). Note that the Dutch NALS course manual was not issued yet at the time of this study. For all items, 2, 1, and 0 points were awarded for adequate and timely performance, for incomplete or out-of-sequence performance, and for inadequate and out-of-sequence performance, respectively. Detailed scoring instructions were formulated. Each participant could receive a maximum score of 48 points. Items were classified as ‘not assessable’ whenever participants could not finish their ABCDE approach, because they rapidly summoned extra help. Such a situation was strictly defined in the scoring instructions. The final score for individuals’ adherence to the ABCDE algorithm was expressed as a percentage score and calculated by dividing the number of awarded points by the number of points for all assessable items, multiplied by 100%.

Prior to the actual study, the assessment tool was tested for intra-observer reliability. Ten previously videotaped simulation scenarios were assessed twice by the same observer (ML) with an interval of 2 weeks to prevent recall of ratings [16]. The intraclass correlation coefficient (ICC) was 0.87 (95% CI 0.74–0.94), with an overall difference of 1.4% (p = 0.39), which signifies an almost perfect agreement. Assessment of inter-observer reliability was not needed, for all videos were rated by the same observer (ML). Face validity was established, since consensus on the assessment instrument was reached among 6 experts in the fields of neonatal and pediatric life support. Content validity was ensured, because all items were derived from the Dutch APLS course manual.

All videos were assessed several weeks later by one researcher (ML), who was specifically trained in video-based scoring of the ABCDE approach in NALS scenarios through elaborate instruction by the designers of the assessment instrument and by means of proof scoring videos from earlier training sessions. ML was blinded to the received instruction method; she was not present during the scenarios. Only primary ABCDE assessments were scored, not the re-assessments after interventions. In as much as the two nurses worked closely together, their performance was combined and they were scored as one healthcare professional.

A sample size calculation was performed to estimate the number of participants needed to achieve a power of 80% with a statistical significance of 0.05 (two-tailed) for the primary outcome (between-group difference in adherence). Twenty-one participants per group were required for an expected realistic and clinically relevant difference of 50% of the percentage score in the intervention group compared to the control group. The standard deviation used in the sample size calculation (16%) was obtained by assessing 31 previously videotaped ABCDE approaches performed during earlier NALS simulation scenarios.

The main strengths of this study were the randomization of participants to prevent selection bias, blinding of the video assessor to the intervention, and the fact that the participants were not fully aware of the specific aim of this study, since they gave informed consent for research purposes in general. An assessment tool with face and content validity and high intra-observer reliability was used, substantiating the validity and reliability of the results. Sufficient power for the primary outcome was reached. The conclusion that VBI is superior to CL in terms of adherence to the ABCDE algorithm was not only corroborated by the overall between-group difference, but also by the results on the individual domains.

In addition to the considerations mentioned above regarding generalizability of this study and the imperfections of the ABCDE algorithm in general, some limitations arose while conducting this study. Most limitations were related to the use of an existing training program. The two nurses were assessed as one despite possibly different background characteristics. The training sessions took place during regular day shifts. As a result, fellows/neonatologists were sometimes unable to attend the simulations, since urgent matters demanded their presence on the ward. This explained the limited inclusion of these healthcare professionals and precluded conclusions about their adherence. It was not possible to compare both groups at baseline regarding their adherence to the ABCDE algorithm. However, considering their similar background characteristics, the two groups were probably not importantly divergent in this regard. Since participants were randomly assigned to either the intervention or control group, possible confounding factors were probably divided equally over both groups and therefore did not substantially influence the primary outcome of this study. In as much as physicians (i.e. ward clinicians and fellows/neonatologists) participated in multiple scenarios, they may have benefitted from a learning curve. This learning curve did probably not affect the difference in adherence between the VBI and CL groups, because, if it occurred, it did so in both groups. However, it may have influenced the comparison between nurses and physicians, because nurses only performed one scenario and could therefore benefit less from a learning curve. Support for this may be found in comparing nurses and physicians only in their first scenario, since this comparison does not show a significant difference. However, these results could not be reliably interpreted, because the study was underpowered for such a subanalysis.

Further studies into this subject matter are highly recommended. Some follow-up studies are already planned by our research group, including evaluation of adherence to the ABCDE approach in the clinical situation. Additional research is needed to find out which (bundle of) interventions, other than VBI, can be employed to further improve adherence to the ABCDE approach. Possible interventions include: alternative learning strategies, self-efficacy training, video review sessions, use of pocket cards, decision support tools, and/or augmented reality devices [6]. By conducting more research on the ABCDE approach, this algorithm may become evidence-based instead of consensus-based.