Impact of telemedicine on neonatal resuscitation in the emergency department: a simulation-based randomised trial

Abstract

Background

The delivery and initial resuscitation of a newborn infant are required but rarely practised skills in emergency medicine. Deliveries in the emergency department are high-risk events and deviations from best practices are associated with poor outcomes.

Introduction

Telemedicine can provide emergency medicine providers real-time access to a Neonatal Resuscitation Program (NRP)-trained paediatric specialist. We hypothesised that adherence to NRP guidelines would be higher for participants with access to a remotely located NRP-trained paediatric specialist via telemedicine compared with participants without access.

Materials and methods

Prospective single-centre randomised trial. Emergency Medicine residents were randomised into a telemedicine or standard care group. The participants resuscitated a simulated, apnoeic and bradycardic neonate. In the telemedicine group a remote paediatric specialist participated in the resuscitation. Simulations were video recorded and assessed for adherence to guidelines using four critical actions. The secondary outcome of task load was measured through participants’ completion of the NASA Task Load Index (NASA-TLX) and reviewers completed a detailed NRP checklist.

Results

Twelve participants were included. The use of telemedicine was associated with significantly improved adherence to three of the four critical actions reflecting NRP guidelines as well as a significant improvement in the overall score (p<0.001). On the NASA-TLX, no significant difference was seen in overall subjective workload assessment, but of the subscore components, frustration was statistically significantly greater in the control group (p<0.001).

Conclusions

In this study, telemedicine improved adherence to NRP guidelines. Future work is needed to replicate these findings in the clinical environment.

Introduction

Delivery of an infant in the emergency department (ED) is a required but rarely practised skill for emergency physicians.¹ About 10% of neonates require some form of immediate resuscitation to promote breathing at birth, ranging from simple drying and stimulating to positive pressure ventilation (PPV) and endotracheal intubation.² This number may be even higher in the ED and prehospital settings due to increased incidence of precipitous delivery when outside the controlled delivery room environment, and lower likelihood of adequate prenatal care.³ Emergency Medicine training includes basic paediatric and neonatal resuscitation; however, these skills are not often called on and Emergency Medicine residents are not required by the Accreditation Council for Graduate Medical Education to be Neonatal Resuscitation Program (NRP) trained. Without regular refreshers or frequent use, these skills degrade rapidly over time.^{4 5}

Children under 1 year of age account for over 4 million ED visits annually in the USA and more than 27 million visits come from those under 15 years old.⁶ The Nationwide Emergency Department Sample estimates a total of 11 084 deliveries in EDs in the USA per year. Only 8.5% of EDs have 24 hours/day, in-house availability of a paediatrician.⁷ Obstetric services that are hospital based have decreased in the USA and it is unknown if this has been associated with changes in birth location and outcomes.⁸

While it is not possible to staff every ED with a paediatric specialist, modern telecommunications technology can be used to provide round-the-clock, real-time, remote access to paediatric subspecialists. Telemedicine offers substantial opportunities to expand access to clinical experts in outlying areas. This technology has increasingly been used to provide trauma and stroke consultation and expertise to regional hospitals, and has been proposed as a method to staff and monitor intensive care units (ICU) in hospitals lacking a full-time intensivist.^{9 10} High levels of satisfaction have been reported by patients and providers related to a paediatric critical care telemedicine programme for rural and community EDs in California.¹¹ There is a need to describe the impact of paediatric telemedicine on patient outcomes and processes of care. This is challenging due to the variation in patient conditions and providers. A controlled study leveraging simulation would allow for a detailed evaluation of the impact of telemedicine on patient care.

In this study, we aimed to evaluate the impact of telemedicine on the quality of a simulated resuscitation of a neonate in the ED. Our primary outcome was adherence to four predefined ‘critical actions’ stemming from the NRP guidelines.¹² Secondary outcomes included the NASA Task Load Index (NASA-TLX) and a more extensive 40-item NRP checklist.¹³

Materials and methods

Study setting and participants

This single-centre, prospective, simulation-based, randomised trial took place at a simulation centre of a large US academic medical institution in July 2017. Participants were recruited from senior Emergency Medicine resident physicians in their third or fourth year of training for participation in a scripted simulation scenario involving an apnoeic and bradycardic neonate (online supplementary appendix 1). Twelve participants were randomised to one of two groups: telemedicine (intervention) or standard care (control). No formal sample size was calculated, as we considered this an initial pilot study. A convenience sample was deemed adequate, which included recruitment of all Emergency Medicine residents rotating in the Pediatric ED during this study period.

Simulation and telemedicine design

A simulation scenario of a newborn presenting with apnoea and bradycardia immediately after delivery. The scenario was standardised and a confederate nurse was present in the room in all cases. The scripted nurse actor was instructed not to offer any suggestions, but to request clarification when needed. A SimNewB manikin (Laerdal Medical, Stavanger, Norway) was used to simulate the neonate in distress. Vital signs were displayed in real time on a monitor. The room was stocked with a limited supply of paediatric resuscitative equipment including laryngoscopes, an infant warmer, endotracheal tubes, bulb suction, wall suction, a bag valve mask, towels, medications, a Broselow Pediatric Emergency Tape (Armstrong Medical Industries, Lincolnshire, IL) and a printed copy of the NRP algorithm (posted on the warmer) (online  supplementary appendix 2).

In the telemedicine group, a single paediatric trained provider was available to the participant to assist with the management. Telemedicine contact was provided via a bidirectional audio feed, with additional visual information provided to the paediatric emergency medicine (PEM) specialist via video and a one-way window. Participants in the control group did not receive help from the specialist, but had access to identical equipment, personnel and resources.

Data collection and analysis

All scenarios were video recorded in their entirety (SimCapture, B-line Medical, Washington, DC). Three physician raters independently evaluated the recorded sessions using the four-item critical action checklist and the detailed 40-item checklist based on NRP guidelines (table 1). This checklist contained both independent and time-sensitive variables. Raters were calibrated and trained on the checklist during a session where one video rating was demonstrated, then raters had the opportunity to do a rating themselves and then they had the opportunity to ask questions about the rating system. Any disagreements were resolved by consensus in a post hoc meeting. Additional data were collected on specific tasks and choice of equipment.

Table 1.

Checklist items

	Group		P values
	Control (n=6)	Intervention (n=6)
Time-sensitive items
PPV started
Initiated at 60–90 s	3 (60%)	2 (33%)	0.242
Sustained for 30 s prior to intubation	3 (60%)	5 (83%)	0.444
Auscultated lungs	2 (33%)	5 (83%)	0.117
30 s ventilation via endotracheal tube	5 (100%)	6 (100%)	1
Initiated chest compressions 30 s after intubation	0 (0%)	6 (100%)	0.002
Other items
Activated code or called NICU for help	0 (0)	6 (100)	0.002
Team roles assigned	2 (33)	5 (83)	0.242
Turned warmer on	3 (50)	6 (100)	0.182
Dried baby	1 (17)	4 (67)	0.242
Removed wet sheets	3 (50)	5 (83)	0.545
Stimulated baby	4 (67)	6 (100)	0.455
Cleared airway with bulb suction	3 (50)	5 (83)	0.545
Auscultated lungs and voice result	1 (17)	6 (100)	0.015
Auscultated heart and voice result	1 (17)	5 (83)	0.013
Applied pulse oximeter on right hand (correct)	2 (33)	5 (83)	0.043
Applied pulse oximeter on left hand (incorrect)	2 (33)	0 (0)	0.455
Infant face mask used	2 (33)	0 (0)	0.455
Neonatal mask used	3 (50)	6 (100)	0.182
Oxygen source attached (21%)	0 (0)	6 (100)	0.002
Oxygen source attached (other %)	5 (83)	0 (0)	0.015
Placed face mask on face with top of bridge of nose/bottom below mouth	5 (83)	6 (100)	1
Delivered breath with resultant chest rise that appeared effective	5 (83)	6 (100)	1
Used MRSOPA if trouble M=adjust mask, R=reposition head to open airway, S=suction, O=open mouth, P=optimize pressure of delivered breaths, A=artificial airway	6 (100)	6 (100)	1
Auscultated heart	0 (0)	3 (50)	0.182
Prepared for intubation	5 (83)	6 (100)	1
CO2 detector used	1 (17)	4 (67)	0.242
Correct tube size (3.5) verbalised	1 (17)	6 (100)	0.015
Correct blade size (0) verbalised	2 (33)	6 (100)	0.061
Insertion depth verbalised	2 (33)	6 (100)	0.061
Secured tube	1 (17)	6 (100)	0.015
Oxygen increased to 100% after intubation	2 (33)	2 (33)	1
Demonstrated thumb method or two-finger method	6 (100)	6 (100)	1
Force generated to lower one-third of sternum	6 (100)	6 (100)	1
Compressions administered in 3:1 ratio with ventilation	0 (0)	6 (100)	0.002
Asked for intravenous access	6 (100)	5 (83)	1
Need for epinephrine verbalised	5 (83)	6 (100)	1
0.1–0.3 mL/kg of 1:10 000 (0.01 mg/kg intravenous, intraosseous)	5 (83)	6 (100)	1
Asked for ventilator	1 (17)	6 (100)	0.015
Vent settings (PIP 20–25, PEEP 5–6, rate of 40–60 is correct)	0 (0)	6 (100)	0.002

Values in bold reflect statistical significance (p<0.05).

NICU, neonatal intensive care unit; PEEP, positive end expiratory pressure; PIP, peak inspiratory pressure; PPV, positive pressure ventilation.

The primary outcome was adherence to four critical actions: (1) use of warming/drying/stimulation; (2) use of PPV in lieu of immediate chest compressions; (3) choice of 21% oxygen; and (4) initiation of chest compressions only after intubation. These four actions were determined in conjunction with a board-certified neonatologist with expertise in NRP and over 10 years in experience. The 40-item complete NRP checklist was assessed as a secondary outcome (table 1). Most newborns presenting with apnoea will respond to a combination of basic stimulation and suctioning, in conjunction with avoidance of hypothermia (ie, warm/dry/stimulate).¹⁴ Current guidelines strongly advocate for optimised ventilation as the primary resuscitative strategy in the apnoeic and bradycardic newborn not responsive to stimulation and suctioning within 30 s.^14–16 These guidelines also recommend against exposing the neonate to high oxygen levels during initial resuscitation given the absence of evidence of benefit and potential harm from oxygen toxicity.

Task load was assessed via a paper-based NASA-TLX¹⁵ given to participants after the scenario. The TLX is a multidimensional scale developed to estimate the workload of individuals while, or immediately after, completing a task.¹⁷ The TLX and variations thereof have been used to evaluate workload in over 550 studies spanning numerous fields, especially aviation and healthcare.¹⁸

We used a modified form of the NASA-TLX referred to as the raw TLX (rTLX) where the weighting process of items was eliminated. It is simpler to apply as the ratings are averaged or added to create an estimate of overall workload. In the 29 studies in which rTLX was compared with the original version, it was found to be either more sensitive,¹⁹ less sensitive²⁰ or equally sensitive.¹⁸ The rTLX features seven domains, each evaluated on a 20-point Likert scale. The domains represented in the rTLX are mental demand, physical demand, temporal demand, performance, effort and frustration (online supplementary appendix 3 and  4).

All data were manually entered into Microsoft Excel V.14.0 (Microsoft, Redmond, WA) and transferred into SPSS (V.22.0, IBM) for performance of statistical analyses. All data were examined for missing values and for normality and homogeneity in each analysis. We compared differences in demographics by group to which participants were randomised. In the bivariate analyses, we conducted Pearson’s χ² or Fisher’s exact tests for categorical data as appropriate, and independent t-tests for normal continuous data.

Results

Twelve emergency medicine residents participated in the study. Demographics were similar between groups (table 2). Only one resident reported previous NRP certification. Although many residents had previous simulated telemedicine experience, only one had used telemedicine in a real clinical scenario. For three of four critical actions, those in the intervention group had significantly better adherence compared with the control group (table 3). No members of the control group optimised ventilation, as measured by administering PPV with an appropriate oxygen concentration, and intubating if necessary prior to initiating chest compressions. Only one member of the control group attempted to warm, dry and stimulate the infant prior to more aggressive resuscitative steps. In the control group, three of six participants (60%) sustained PPV for at least 30 s compared with five of six participants (83%) in the intervention group. All members of the intervention group met all four critical actions.

Table 2.

Demographics

Demographic survey question	Control n=6 (%)	Intervention n=6 (%)
PGY3	5 (83)	3 (5)
PGY4	1 (17)	3 (50)
How many simulations would you estimate that you have participated in?
6–10 times	1 (17)	0 (0)
>15 times	5 (83)	6 (100)
How many times would you estimate that you have intubated a simulated adult or child patient?
1–5 times	1 (17)	0 (0)
6–10 times	1 (17)	1 (17)
10–15 times	3 (50)	1 (17)
>15 times	1 (17)	4 (67)
How many times would you estimate that you have intubated a REAL patient regardless of the age?
>15 times	6 (100)	6 (100)
How many times would you estimate that you have intubated a neonate (less than 28 days old) in a REAL patient?
Never	4 (67)	5 (83)
1–5 times	2 (33)	1 (17)
How many times would you estimate that you have intubated a simulated neonate (up to the size of a SimBaby)?
1–5 times	6 (100)	3 (50)
6–10 times	0 (0)	1 (17)
10–15 times	0 (0)	2 (33)
How many times have you used telemedicine in a SIMULATED patient?
Never	3 (50)	4 (67)
1–5 times	3 (50)	2 (33)
How many times have you used telemedicine in a REAL patient?
Never	6 (100)	5 (83)
1–5 times	0 (0)	1 (17)
Have you had previous exposure to neonatal resuscitation training?
Never	3 (50)	4 (67)
1–5 times	3 (50)	2 (33)
Mean confidence caring for a baby (SEM)	2 (0.4)	2.5 (0.5)

PGY, postgraduate year.

Table 3.

Critical actions

	Group		P values
	Control n=6 (%)	Intervention n=6 (%)
At least 25 seconds of warming/drying/stimulating performed initially	1 (17)	6 (100)	0.015
Non-invasive positive pressure ventilation attempted prior to intubation	5 (83)	6 (100)	1
Initial oxygen setting of 21%	0 (0)	6 (100)	0.002
Chest compressions initiated only after intubation	0 (0)	6 (100)	0.002

Values in bold reflect statistical significance (p<0.05).

On the 40-item checklist, participants in the control group were statistically significantly less likely to use the correct ratio of breaths to chest compressions, or to complete secondary tasks such as calling for a ventilator, securing the tube or verbalising appropriate ventilator settings. There was a statistically significant improvement in the overall checklist score (control: mean 51.3 (SD=11.1), intervention: 81.6 (SD=6.4), p<0.001). On the NASA-TLX, no significant difference was seen in overall subjective workload assessment, but of the subscore components, frustration was statistically significantly greater in the control group (p<0.001) (table 4)

Table 4.

Teamwork and task load

	Arm		P values
	Control	Telemedicine
Workload: Mean NASA-rTLX team score (SEM), max=100	64 (4)	51 (5)	0.056
Mental demand (max=20) (SEM)	16 (1)	14 (1)	0.061
Physical demand (max=20) (SEM)	5 (2)	6 (1)	0.667
Temporal demand (max=20) (SEM)	14 (1)	14 (2)	0.972
Performance (max=20) (SEM)	10 (1)	5 (1)	0.051
Effort (max=20) (SEM)	15 (1)	12 (1)	0.114
Frustration (max=20) (SEM)	14 (1)	8 (2)	0.05
Anxiety (max=20) (SEM)	16 (1)	12 (1)	0.055

Values in bold reflect statistical significance (p<0.05).

rTLX, raw TLX.

Discussion

This study examined the impact of telemedicine on neonatal resuscitation by ED providers in a simulated setting. Telemedicine was found to significantly improve adherence to critical actions and a significantly higher score on a comprehensive neonatal resuscitation checklist.

Previous work specifically on telemedicine for neonatal resuscitation has shown decreased time to adequate ventilation in simulated patients, increased access to specialists, decreased unnecessary transfers and decreased costs.^21–23 Other interventions using telemedicine in the prenatal and perinatal periods showed an association with decreased infant mortality as well as decreased rates of delivery of very low birthweight infants in hospitals lacking neonatal intensive care unit and neonatal experts.²⁴

Fewer studies have been done using telemedicine for paediatric emergencies. Observational studies have suggested improved quality of care in the ED or ICU for critically ill or injured children at rural hospitals lacking paediatric subspecialty care with the use of telemedicine consultations.^{24 25} Other studies have evaluated adherence to clinical resuscitation guidelines, as well as the use in paediatric critical results with varying results in clinical outcomes, but overall positive acceptability.^23–29

This study illustrated a knowledge gap in this cohort of Emergency Medicine trainees. Critically, participants in the control group did not adhere to NRP guidelines. Anecdotally, we observed that participants often applied Pediatric Advanced Life Support (PALS) algorithms instead. This may be due to participants’ increased familiarity with these algorithms (only one participant had previous NRP certification), as well as their greater exposure to paediatric rather than neonatal resuscitation. The NRP guidelines represent a shift from standard PALS and adult resuscitative guidelines in their prioritisation of ventilation and airway control. While most participants did attempt PPV, on review of recorded sessions, many in the control group gave only limited ‘rescue breaths’ rather than a sustained trial.

Participants in the control group missed simple steps on the checklist such as securing the endotracheal tube and requesting a ventilator. It is unlikely that these participants lack knowledge of these needed steps, but rather, they may have been overlooked in the setting of increased frustration and workload. Cognitive unburdening may be achieved through telemedicine. Increased cognitive workload may account for some of the difference between groups. The NASA-TLX scores for overall workload were borderline statistically significant (p=0.056), likely due to the fact that we did not power the study for this detecting workload differences between groups.

Although beyond the scope of this research study, previous studies have found that simulation-based training promotes better retention of neonatal resuscitation skills than traditional didactic training.²³ Moreover, neonatal resuscitation can be used as a prototype for high-risk, low-frequency events that can fall within the scope of practice for an emergency physician. Given the clear guidelines provided by the NRP, simulation can serve as a method to evaluate telemedicine-based interventions.

Limitations

Despite encouraging results regarding the utility of telemedicine, this study had several key limitations that may preclude broader extrapolation. The number of participants was small and the study was underpowered to show a significant difference in the NASA-TLX. Only one PEM specialist was available to serve as a telemedicine leader, and the individual was not blinded to the clinical scenario. While this reflects similar limitations in previous studies, an ideal investigation would involve multiple subspecialists without foreknowledge to minimise potential bias.²¹

Another limitation is the fact that only one group had access to an NRP-trained specialist in any form. The resulting differences between the groups are surely due to this additional expertise. Nevertheless, the fact remains that telemedicine was shown to be an effective mechanism by which to provide this expertise.

The telemedicine provider was not a neonatologist but a PEM fellow who was part of the study team. This introduced a potential source of bias.

Conclusions

Participants in our simulated neonatal resuscitation scenario receiving telemedicine support adhered significantly better to NRP guidelines. Telemedicine has the potential to provide substantial decision-making support in high-risk, low-frequency events by allowing real-time access to a specialist. Further study is needed to determine both the ideal strategy for teleleadership and the best scenarios in which it should be deployed.