A systematic review and meta-analysis comparing mortality in pre-hospital tracheal intubation to emergency department intubation in trauma patients

Pre-hospital airway management is an important area for research in pre-hospital critical care [1]. Tracheal intubation (TI) with a correctly positioned cuffed tracheal tube is considered the gold standard for securing an airway [2‐5]. Pre-hospital intubation (PHI) of trauma patients is performed in many advanced emergency medical systems (EMS). Alternatively, conservative airway measures may be used before hospital admission, with TI performed in the emergency department (ED) [6]. Outside the operating theatre and in out-of-hospital settings, TI is challenging, with relatively high complication rates and limited resources for managing complications [3, 7‐11]. The reported success rates for PHI vary, but the best-performing systems show success rates similar to those of in-hospital emergency TI [12‐16]. For patients not in cardiac arrest, emergency department intubation (EDI) is normally performed as rapid sequence induction intubation (RSI), which includes the use of a rapid-onset neuromuscular blocking agent before TI, whereas PHI is done both with and without drugs [17].

The indications, techniques and providers used for the procedure vary widely, and interpretations of the current evidence of the effects of PHI on patient outcome differ considerably [18]. Although several guidelines suggest that TI should be considered for all trauma patients with a Glasgow coma scale (GCS) score of 8 or below, the evidence supporting the use of a particular GCS score as a threshold for intubation is poor [2, 4, 5]. A 2009 Cochrane review of all types of emergency TI included three studies that fulfilled the Cochrane criteria and in which the majority of patients experienced out-of-hospital cardiac arrest. The authors’ conclusion regarding the subgroup of trauma patients in this analysis was that the current evidence base provided no imperative to expand the practice of pre-hospital intubation in urban systems [19]. This systematic review was performed to compare the mortality rates of adult trauma patients undergoing PHI to those undergoing EDI.

The search identified 3211 unique references through the search process described in Fig. 1. After the initial screening of titles and abstracts of all records, 64 studies were examined in full text by both authors responsible for the selection process. Of the 64 studies, 42 were excluded because PHI or EDI was not confirmed for all patients. Twenty-two studies met our inclusion criteria and compared mortality rates of patients who underwent PHI with patients who underwent EDI (Table 1) [16, 27‐47]. One study was considered to have poor methodological quality and was excluded. Seven studies that met the inclusion criteria reported data from the same health registries during the same period; of these, the three that best agreed with our defined aims were included in the meta-analysis, the others were excluded (Table 3). Two studies investigated different subgroups from a large trauma registry and both were included in the meta-analyses. One randomized controlled trial (RCT) and sixteen observational studies were included in the mortality meta-analysis. Five of the seventeen studies examined pre-hospital RSI. One RSI study and six of the twelve studies involving no RSI or some RSI provided adjusted data in their analyses (Table 2). Data from the primary analysis were used to perform separate subgroup analyses of four studies with no significant differences between groups in the ISS and four studies with a verified similar pre-hospital GCS score <9 in both groups (Table 3).

Of the 21 studies that met all the eligibility criteria, twelve concluded that PHI was associated with a worse outcome than EDI, seven found no differences in mortality between the groups, one found a lower mortality rate when PHI performed by aeromedical crews was compared with EDI provided after ground transport, and for one study, a mortality analysis was beyond the scope of the article.

The aim of this review was to compare the mortality rates in adult trauma patients intubated before and immediately after hospital arrival. Despite differences between studies, our forest plots quite consistently showed a higher mortality rate for PHI than EDI. When all available data, both adjusted and unadjusted, were considered, no studies identified a positive effect on the mortality rate when PHI was compared to EDI. Eight separate analyses of subgroups were made, five of these found a significantly higher mortality rate in the PHI-patients; Crude mortality rate in both RSI (five studies) and non RSI (12 studies) studies, non RSI-studies after adjusting for injury severity (six studies), studies with no significant differences in ISS (four studies), and non RSI-studies with patients with a similar GCS (two studies). Three subgroup analyses did not identify a significant higher mortality rate after PHI; RSI-studies after adjusting for injury severity, based on one study, RSI-patients with a similar GCS, based on two studies, and studies from a European-organized EMS, based on two studies. However, there are some major objections towards doing a meta-analysis on this material: most importantly a high risk of selection bias and a high level of heterogeneity in the included studies.

The effect of selection bias in observational studies in this material should not be underestimated, as sicker patients are more prone to undergo more aggressive airway procedures. The fact that the only RCT included was also the only study with a non-significant trend towards a better mortality rate in the PHI group underlines this [29, 50]. We tried to weaken the impact of selection bias in this systematic review by only including studies with a high level of indication for intubation, reflected in all patients either undergoing PHI or EDI. Except for 2 studies, the articles examined in this review included only patients who had circulation at hospital admission, and patients who died shortly after hospital admission were excluded from the analysis in 11 of the 21 studies. In most of our included studies, the ISS in PHI patients was higher than in EDI patients. The lack of physiological parameters has been raised as an objection to the validity of the ISS when comparing patients, and a significantly higher mortality rate in the PHI group was shown in the four studies in which there were no differences between groups in the ISS [51]. The association between PHI and a higher mortality rate was similar for unadjusted and adjusted numbers, with an unadjusted OR of 2.54 (2.05, 3.15) and an adjusted OR of 2.59 (1.97, 3.39). The fact that the adjustments had little impact on the results is an interesting finding, which may imply that correcting for other factors associated with injury severity should be considered.

The other major factor in this meta-analysis is the high level of heterogeneity between the studies. Tracheal intubation (TI) is a complex intervention, patient populations are heterogeneous and there are major differences in staffing and EMS infrastructure. Only approximately 10% of the PHI patients in this meta-analysis had full access to RSI drugs; this reflects the clinical reality, but weakens the direct comparison of PHI to EDI. The subgroup analyses of studies where all PHI patients had access to RSI showed a less negative trend than for the studies in which RSI was not available for all, which suggests that access to pre-hospital RSI is of importance. One common objection to the comparison of PHI and EDI is that personnel outside the hospital, in general, receive less training in TI than their counterparts in the ED, which may lead to a prolonged performance time with increased exposure to hypoxia and possibly a higher rate of complications and failed intubation [12, 52‐54]. Most studies in our analysis were from an American-organized EMS, in which paramedics perform most PHI; this differs from parts of Europe, where emergency physicians and anaesthesiologists perform most PHI (Table 1) [6]. Our subgroup analysis from a European-organized EMS was based on two studies and did not show a significant difference in mortality rates between PHI and EDI. A recent meta-analysis that examined success rates for PHI found a significantly higher median physician success rate of 98.8%, compared to a non-physician success rate of 91.7% (p = 0.003) [55]. The reported differences in success rates between PHI and EDI seem to be relatively low compared with the differences in mortality rates in our included studies, indicating that the differences in success rates alone may be insufficient to explain the observed differences in mortality rate. Success rate is, however, a very crude parameter with only two possible outcomes, and detailed information on time spent on the procedure, number of attempts before successful intubation, and adverse events that may influence patient status were not supplied in most studies. We aimed to examine subgroups of studies in which PHI was performed by personnel with the same level of expertise as those performing EDI, but the only study in which all patients were treated by physicians did not show any deviation from the other studies [43].

The high heterogeneity in this review is reflected in mortality rates of 7.7–93.5% for PHI and 6.25–66.5% for EDI, which gives an I ² value of 91% in the crude data analysis (Fig. 2) and 86% in the adjusted OR analysis (Fig. 3). Any precise effect estimates or numbers needed to treat drawn from these heterogeneous data are necessarily invalid. One might argue that a meta-analysis of this material can be misleading and vague, but the high level of consistency present across a wide range of studies is still interesting. Despite the importance of selection bias and heterogeneity, to completely reject all negative results on grounds of methodology is not something that should be done without serious consideration, and a thorough investigation into other possible causes for differences in mortality rates seems to be strongly indicated.

Adverse events associated with TI are related not only to difficulty in inserting the airway but also the physiological consequences of the actual intubation and positive-pressure ventilation. The pre-hospital environment can be hostile, with few viable ways to treat complications. When muscle relaxants are administered, patients who previously had intact airway reflexes may face a greater risk of aspiration and hypoxia if difficulties occur. One study found transient hypoxia in more than half of the patients undergoing PHI RSI, which is significantly higher than the respective incidence for trauma intubations in the ED [52]. PHI may predispose to tension pneumothorax, and both the condition itself and therapeutic thoracotomy, if performed, have a relatively high morbidity rate [56]. Cardiovascular collapse is a known complication of TI in this patient group, and some centres deliberately postpone in-hospital TI in patients in shock until after initial stabilization [57, 58]. The only RCT included in our review identified a significantly higher occurrence of pre-hospital cardiac arrest after PHI; this may be related to Wang et al's finding of a  highly significant higher mortality rate after pre-hospital advanced airway management in patients with haemorrhagic shock, but no significantly higher mortality in patients without shock [59]. The studies in this review did not provide sufficiently detailed information to perform a separate analysis of patients in shock; this is a very important subgroup to investigate in future research into pre-hospital airway management.

None of the studies in this meta-analysis identified a significant positive effect on the mortality rate after PHI, but to interpret this as evidence that PHI is generally unfavourable does not seem to be valid. Many authors advocate the use of PHI, and the rationale for securing a seriously compromised airway as soon as possible seems reasonable, as the compromised patients are the same patients with the same problems, earlier in their pathway of care [18, 60]. It is unlikely that any pre-hospital services will achieve the level of care and equipment provided by a full in-hospital trauma team, which means that the rationale for PHI is that early protection and control of the airway outweighs the increased risks associated with performing the procedure in a less favourable setting. Regardless of the weaknesses concerning low-quality evidence, the consistent finding of worse outcomes after PHI compared with EDI should raise some questions. Variable effects in subgroups of patients have led to recommendations for a tailored approach to interventions in other fields of emergency care, and this may also be valid for pre-hospital airway management [61, 62].

This systematic review quite consistently shows higher mortality rates when patients undergoing PHI are compared to patients intubated in the ED. However, reducing the analysis of a complex intervention to a dichotomous first-past-the-post approach discounts the comprehensive nature of the intervention. The association between PHI and a higher mortality rate does not necessarily contradict the importance of the intervention, but it does call for a thorough investigation by clinicians and researchers into possible causes for this finding. Further comparisons of widely defined patient and personnel groups are not likely to provide results that differ extensively from earlier reports; future research should include well-conducted subgroup analyses to investigate in which situations PHI may improve the outcome.