Success and complications by team composition for prehospital paediatric intubation: a systematic review and meta-analysis

Background Clinical team composition for prehospital paediatric intubation may affect success and complication rates. We performed a systematic review and meta-analysis to determine the success and complication rates by type of clinical team. Methods We searched MEDLINE, EMBASE, and CINAHL for interventional and observational studies describing prehospital intubation attempts in children with overall success, first-pass success, and complication rates. Eligible studies, data extraction, and assessment of risk of bias were assessed independently by two reviewers. We performed a random-effects meta-analysis of proportions. Results Forty studies (1989 to 2019) described three types of clinical teams: non-physician teams with no relaxants (22 studies, n = 7602), non-physician teams with relaxants (12 studies, n = 2185), and physician teams with relaxants (12 studies, n = 1780). Twenty-two (n = 3747) and 18 (n = 7820) studies were at low and moderate risk of bias, respectively. Non-physician teams without relaxants had lower overall intubation success rate (72%, 95% CI 67–76%) than non-physician teams with relaxants (95%, 95% CI 93–98%) and physician teams (99%, 95% CI 97–100%). Physician teams had higher first-pass success rate (91%, 95% CI 86–95%) than non-physicians with (75%, 95% CI 69–81%) and without (55%, 95% CI 48–63%) relaxants. Overall airway complication rate was lower in physician teams (10%, 95% CI 3–22%) than non-physicians with (30%, 95% CI 23–38%) and without (39%, 95% CI 28–51%) relaxants. Conclusion Physician teams had higher rates of intubation success and lower rates of overall airway complications than other team types. Physician prehospital teams should be utilised wherever practicable for critically ill children requiring prehospital intubation.


Background
Airway management is a critical component of prehospital care for severely ill and injured children. Airway management is arguably even more important in children than in adults, as cardiac arrest is more likely to be hypoxic in origin and therefore amenable to airway and ventilation intervention. As hypoxia correction is a time critical intervention, an emergency medical service (EMS) system must be able to provide airway management as early as possible, preferably at the incident scene.
Intubation is generally considered to be the gold standard for airway management in the critically ill and injured. Children however typically comprise only about 5% of total EMS cases [1][2][3], and those requiring intubation vary from 0.1% of all EMS responses [3,4] to approximately 5% of paediatric cases when advanced intervention teams are selectively utilised [1,2]. Success rates are also reported to be lower in children and the complication rate higher [5,6]. Traditionally, ground EMS systems have intubated children without muscle relaxants, but many systems are introducing relaxants into their clinical protocols with the expectation that overall success rates would improve and that intubation could be offered for a wider range of pathologies. There are also recent reports that physician staffed helicopter EMS (PS-HEMS) may produce particularly high procedural success with low complication rates [7][8][9][10].
The purpose of this study was to systematically review the available literature and perform a meta-analysis to determine whether there exists an association between type of prehospital team and intubation success and complication rates.  We hypothesised that utilisation of muscle relaxants by non-physician teams would improve procedural success in prehospital paediatric intubation over teams without relaxant access and that the greater experience and training of physician teams might produce further performance gains above those associated with relaxant access for non-physician teams.

Methods
The systematic review was conducted and reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) Guidelines [11].

Data sources and literature search strategy
We created search strategies for the concepts of 'intubation', 'prehospital', and 'paediatric' using a combination of standardised terms and keywords drawn from indices, thesauri, and on-topic articles (Supplementary eAppendix) in consultation with a medical librarian. The electronic databases Ovid MEDLINE, EMBASE, and CINAHL were searched from database inception to November 11, 2019. Additionally, we conducted a manual search of reference lists of included and other relevant articles. All articles were reviewed for inclusion by two independent reviewers (AG and NB). Any discrepancies were resolved by consensus with a third reviewer (AW).

Study selection
Interventional and/or observational studies were eligible for inclusion if they reported data on the success, firstpass success, and/or complication rates of prehospital paediatric intubation attempts. Studies that did not separately report the number of patients in whom intubation was attempted were excluded from analysis, as were abstract-only and grey literature reports. There was no language restriction.

Data extraction
Successful intubation, first-pass success rate, and complication rates were extracted from the included articles by two independent investigators (AG and NB). Where there was discrepancy, a third author (AL) adjudicated. We also extracted data about authors, publication year, study location, setting, professional background of team members, availability of muscle relaxants, and participant characteristics by inclusion age and intubation indication. For each study, the team composition (exposure variable) was classified into three groups: non-physicians with no relaxants, non-physicians with relaxants, and physicians with relaxants. We made no contact with authors for missing data as many studies were old.

Outcomes
The primary outcomes were the proportions of overall intubation and first look success rates. Secondary outcomes included the rate of intubation complications, specifically unrecognised oesophageal and endobronchial intubation, three or more attempts at intubation, hypoxia, or aspiration.

Assessment of study quality
The criteria used by Fouche and colleagues [12] were used to evaluate study quality. The checklist consists of 8 items that assess external and internal validity through 4 domains: selection bias, non-response bias, measurement bias, and bias related to the analysis, with each item graded as low or high [12]. The overall risk of bias for the study was rated 'low' if 7 or more domains were rated low, 'moderate' if 4 to 6 domains were rated low, and 'high' if 1 to 3 domains were rated low [12]. Each included study was assessed by AL and reviewed by AG.

Data synthesis and statistical analysis
We used the macro 'metaprop_one' in STATA 16.0 (Sta-taCorp, College Station, TX) to pool proportions using the Freeman-Tukey double arcsine option to ensure that the confidence intervals around the estimates did not fall outside 0 or 1 with stable variances [13]. We used a logistic-normal random-effects model [13] and assessed the heterogeneity as low, moderate, and high using I 2 values of 25%, 50%, and 75% [14]. We performed subgroup analyses by team composition a priori to explain heterogeneity and conducted a sensitivity analysis on low risk of bias trials to estimate the robustness of primary outcome results. Meta-regression with robust variance estimates (to take into account within-study correlation between different team types) was used to explore differences in the primary outcomes by team composition subgroups over time (year of publication) [15]. As there were large variations in the clinical population (mixed, trauma, head injury, and arrested) studied, subgroup meta-analyses by team composition were also performed for the overall intubation rate, first-pass success rate, and overall airway complication rate. We did not assess publication bias with a funnel plot as it has been shown to be problematic in meta-analysis of proportions [16].

Quality assessment
None of the 40 studies were rated as high risk of bias. Eighteen studies (n = 7820) were rated moderate risk of  (Fig. 3). Non-response bias (item 4) more than 20% was the most common type of bias affecting external validity (Fig. 3).

Adverse events
Most adverse events by team type showed large intragroup and inter-group heterogeneity, supporting the need to report individual team pooled estimates ( Table 2). Sixteen studies, involving 1975 children, examined the overall intubation complication rate. The overall airway complication rate was lower in physician teams than in non-physicians with and without relaxants ( Table 2). Physician teams were not associated with the occurrence of oesophageal intubations, aspirations, or the need for three or more multiple intubation attempts (P = 1.00, P = 0.27, P = 0.24, respectively).
Of the 40 studies included in this systematic review, only six [7,10,30,37,45,46] (651 children) reported hypoxia after intubation. Much of the meta-analysis result was influenced by Martinon and colleagues' study [37] (n = 296) that examined the effect of national guidelines on prehospital intubation in severely head-injured children. A post hoc sensitivity analysis, excluding Martinon and colleagues' study [37], showed that there was a significant inter-group heterogeneity (P < 0.001), with a pooled estimate for physician team decreasing to 3% (95% CI 0-10%).

Discussion
To our knowledge, this is the first meta-analysis to compare prehospital intubation success and complication rates of different teams of intubator providers specifically in children. The success and complication rates for physician teams are better than nonphysician teams either with or without muscle relaxants. Although reported clinical populations varied between studies, the success and complications rates followed the same pattern when population subgroup meta-analyses were performed. The overall success and first-pass success estimates were robust in the sensitivity analyses. Even after adjusting for the year of publication in the meta-regressions, team composition differences in the overall success estimate remained significant. The overall quality of evidence was graded as moderate to high after assessing for the presence of selection and non-response bias, measurement bias, and bias related to data analysis [12].
Two previous meta-analyses [12,49] examining the success and complication rates by physician and nonphysician teams regardless of patient age where both team types utilised relaxants demonstrated higher overall and first-pass success for physician teams compared with non-physician teams. Our review indicates that this is also observed in the paediatric subgroup. A possible contributor to higher success rates by physician teams is in-hospital exposure to NR not reported as less than 3 groups, no intra-group heterogeneity reported NR not reported as less than 3 groups, no intra-group heterogeneity reported; NA not applicable *Martinon et al. [37] before and after national guidelines paediatric intubation compensating for the rare requirement for this procedure in prehospital practice. All of the identified physician team studies utilised HEMS for at least some responses, and it may be that there is an additive effect from HEMS increasing team experience by allowing small numbers of clinicians to cover a larger population thereby concentrating exposure. As non-physician teams utilising relaxants have higher success rates when transported by HEMS compared with ground transport lends additional support to this theory.
Successful use of a clinical bundle to avoid periintubation hypoxia by a non-physician team utilising relaxants in non-arrested adults has been reported [50]. The bundle mandated intubation attempts be abandoned in favour of mask ventilation and urgent transport when pre-oxygenation failed to achieve a SpO 2 of at least 94%. The complete bundle reduced peri-intubation hypoxia rates from 44.2 to 3.5% and suggests that avoiding prehospital intubation in hypoxic patients may minimise risk for teams with lower experience levels. This approach however also potentially denies intubation to patients with critical hypoxia who are arguably the most likely to benefit from early intubation. A focus on oxygenation rather than procedural success is suggested for future studies given this is the primary aim of all airway management. It is noteworthy that in our systematic review, only six studies could be identified that reported hypoxia as an outcome from 40 studies that met the inclusion criteria.
Caution is also needed in interpreting our metaanalysis subgroup analysis results as these are observational in nature. However, we believe that the results of the within-study comparisons of different team composition performances in four studies [8,28,32,38], together with insights from our recent study [7], are credible and supportive of higher overall intubation success, first-pass success, and lower complication rates associated with physician teams. Our results were also robust when sensitivity analysis and meta-regressions were performed. The definition of paediatric age group varied between studies ranging from < 13 to < 19 years. Inclusion of a large proportion of teenage patients in a sample is unlikely to reflect the specific issues of paediatric airway management as the greatest difficulty and complication rates occur in smaller children.
Differences in airway training between studies and between team types are a possible explanation for the observed performance differences. As a major difference between physicians and non-physicians is the training programmes to which they have been exposed, it is intuitive to suggest that further training of non-physician teams may decrease or eliminate the observed differences. Reporting of airway training was too heterogeneous to support an analysis however. Some studies provided no description of training [19,21,24,26,39,44,45], and some reported pooled data from multiple agencies [3,28,32], whilst others described the studied teams simply as Advanced Life Support and/or Paediatric Advanced Life Support certified [5,20,[33][34][35].
It is possible that there is variability between team types in willingness to report complications. Studies have demonstrated under-reporting of prehospital intubation complications by non-physician personnel [51] and physician teams [7] when documentation is compared with electronic monitor data. Similarly, under-reporting has been documented in the emergency department setting when video recordings of the resuscitation are reviewed [52]. We are not aware of any studies that compare the rates of underreporting between team types however. Underreporting is also likely to be affected by factors such as organisational and national cultures which may confound any difference by team type as well as the status of legal protection for disclosure of complications in the reporting jurisdiction. Ideally, future studies should report complications based on monitor data and/or video review.

Conclusions
Our systematic review supports higher overall success and first-pass success with lower complication rates by teams incorporating physicians when intubating children in the prehospital environment. The results of the metaanalysis suggest that this applies regardless of nonphysician team utilisation of neuromuscular blockade. Physician prehospital teams should be utilised wherever practicable for critically ill children requiring prehospital intubation.