Prehospital traumatic cardiac arrest: a systematic review and meta-analysis

The methods in this review are described based on the preferred reporting items for systematic reviews and meta-analyses (Prisma) Checklist [7] and the Prisma-S extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews [8]. The search was developed in Embase.com, optimized for sensitivity and then translated to other databases following the method as described before [9]. The search was carried out in the databases Embase.com, Medline ALL via Ovid, Web of Science Core Collection (Science Citation Index Expanded; Social Sciences Citation Index; Arts & Humanities Citation Index; Conference Proceedings Citation Index-Science; Conference Proceedings Citation Index-Social Science & Humanities and Emerging Sources Citation Index and the Cochrane Central Register of Controlled Trials via Wiley). Additionally, a search was performed in Google Scholar where the 200 top relevant references were downloaded. After the original search was performed in March 2019, the search was updated twice (last update; July 13, 2020). The references were imported into EndNote and duplicates were removed as described previously [10, 11].

The search strategies for Embase and Medline used relevant thesaurus terms from Emtree and Medical Subject Headings (MeSH) respectively. In all databases terms were searched in titles and abstracts of references. The search contained terms for (1) prehospital cardiac arrest or prehospital Advanced Life Support, and (2) injuries and trauma. Terms were combined with Boolean operators AND and OR and proximity operators were used to combined terms into phrases. The full search strategies of all databases are available in the appendix. The searches in Embase and Web of Science were limited to exclude conference papers. The reference lists of retrieved non-included relevant review articles and of the included references have been scanned for relevant references missed by the search. No authors or subject experts were contacted, and unindexed journals in the field were not browsed.

Studies meeting the following criteria were included: (1) articles written in English, French, German, or Dutch, and (2) studies with reported outcomes of interest for prehospital TCA. Exclusion criteria were: (1) (> 10%) pediatric patients, (2) military report or combat patients, (3) studies evaluating only a specific treatment (e.g., thoracotomy or REBOA), (4) letters to the editor, expert opinions, systematic reviews, and meta-analyses, (5) animal studies, (6) articles for which the full text was not available to the researchers, and (7) studies published before 1995. Two authors independently (NJV and MGVV) screened the titles and abstracts for relevance and then extracted and selected relevant full text records, where possible. Subsequently, any leftover duplicates were removed, and the full texts of the selected articles were assessed. Discrepancies were resolved through discussion at each stage.

Studies were assessed on quality using the Newcastle–Ottawa Quality Assessment Scale (NOS) [12]. No stars were awarded for comparability since the current literature is divided whether there are prognostic factors that need to be adjusted for. Follow up was rated sufficient if patients were followed up for at least 14 days or were followed up until discharge from hospital. The loss to follow up cut-off point was set on 2.5%. In the Newcastle–Ottawa Quality Assessment Scale, studies are awarded stars based on their quality. The maximum of achievable stars was seven, and studies with less than six stars were excluded.

The risk of bias was assessed using the RevMan Risk of Bias Tool, were all included studies were scored for different types of bias. A study was excluded if it scored either a high risk of bias in one or more categories or if it scored an unknown risk of bias in two or more categories.

Data were extracted by two researchers independently (NJV and MGVV) using a data collection sheet. Any discrepancy was resolved by discussion. The following variables were extracted: author name, year of publication, journal name, country of patient inclusion, start and end date of patient inclusion, study design, study database, database registry type (Studies including prehospital deaths and studies excluding prehospital deaths; Studies from prehospital registries and studies from hospital registries where prehospital care providers are not allowed to declare a patient dead on scene and thus every patient is transported to a hospital were marked as “study including prehospital deaths”), organization of prehospital EMS system (physician or no physician), number of patients included, patient characteristics (sex, age), trauma type (penetrating, blunt, road traffic accident, fall from height), arrest characteristics (witnessed, unwitnessed, bystander CPR), first monitored rhythm (shockable, non-shockable), prehospital interventions (intubation, administration of epinephrine), survival rates, and long-term neurological outcome. Favorable neurological outcome was defined as a Cerebral Performance Category (CPC) I or II, or as a Glasgow Outcome Scale (GOS) 4 or 5 [13, 14]. The database registry types were divided into two predefined categories: (1) studies using databases including patients declared dead on-scene, (2) studies using databases excluding patients declared dead on-scene. Studies from EMS systems where patients can only be declared dead by an ER-physician and patients are thus always transported to a hospital were included in the first category. Similarly, the EMS systems were divided in two predefined categories: (1) studies from countries or regions with a physician-based EMS service and (2) studies from countries or regions without the availability of a HEMS / EMS physician.

Data regarding registry type and TCA survival, and registry type and neurological outcome were pooled using MedCalc Statistical Software version 18.2.1 (MedCalc Software bvba, Ostend, Belgium; http://​www.​medcalc.​org). Likewise, the data regarding organization of EMS system and TCA survival, and organization of EMS system and neurological outcome are analyzed using MedCalc. Data regarding prognostic factors for TCA survival were analyzed using Review Manager statistical software (RevMan 5.4.1, The Nordic Cochrane Collaboration, Copenhagen, Denmark).

Heterogeneity was assessed using a Q-test and I² statistic. In MedCalc, the Q-test was used as a binary test to investigate the presence of statistically significant heterogeneity, and the I² statistic as a quantitative measure of heterogeneity. In RevMan, the Chi² test was used as a binary test to investigate the presence of statistically significant heterogeneity, and again the I² statistic as a quantitative measure of heterogeneity. In the I² statistic, the limit for the quantitative measure of heterogeneity was set at 40%, with values above 40% set as any significant level of heterogeneity. Mantel–Haenszel models were applied in accordance with the heterogeneity of the data; if the result of the I² statistic was below 40%, fixed effect models were used and if the result of the I² statistic was above 40%, random effect models were used.

Data are reported as pooled estimate or risk ratio with corresponding 95% confidence interval (95% CI), as applicable. Forrest plots and funnel plots are shown.

RevMan funnel plots are evaluated using Egger’s regression test to investigate if there is a significant amount of publication bias within the analysis, Meta-Essentials version 1.50 in Microsoft Excel was used to perform this analysis [15] (Microsoft Excel 2016, Microsoft Corporation, Redmond, United States of America, http://​office.​microsoft.​com/​excel).

A power analysis on the prognostic factors analysis was performed using the Shinyapps Meta Power Calculator (Shinyapps, RStudio, Boston, United States of America, http://​jtiebel.​shinyapps.​io/​MetaPowerCalcula​tor).

As systematic reviews and meta-analysis are exempt of IRB approval in the Netherlands, no medical ethical approval was needed to conduct this study.

We found 2957 articles, of which 2865 were excluded based on title and/or abstract. The remaining 92 articles were then screened on full text, if available. Fifty-six articles were excluded for various reasons (Fig. 1), resulting in 36 articles being included [1‐3, 5, 6, 16‐47]. All studies scored 6 or 7 points on the Newcastle–Ottawa Quality Assessment Scale and therefore no study was excluded for quality reasons (Table 1). Likewise, no studies were excluded after the risk of bias evaluation, since no study scored either a high risk of bias in one or more categories, or an unknown bias risk in two or more categories. The risk of bias evaluation is displayed as a RevMan Risk of Bias Summary Tool and funnel plots (supplemental Figs. 8–14).

In total, 51.722 patients were involved in this systematic review and meta-analysis.

Most included studies (n = 21) were published between 2016 and 2019 [27‐47], and most of the 36 primary studies were carried out in either Europe (n = 14) [2, 3, 6, 19‐21, 28, 32, 33, 35, 36, 40, 44, 46] or Asia (n = 13) [22, 25, 26, 30, 31, 34, 37‐39, 41, 43, 45, 47]. Almost all included studies were retrospective cohort studies (n = 32) [1‐3, 16‐18, 20‐38, 40‐44, 46, 47] and included both penetrating and blunt trauma cases (n = 30) [1‐3, 16, 17, 19‐21, 23‐25, 27, 28, 30‐36, 40‐47]. More detailed characteristics of the included primary studies are shown in Table 2.

The overall pooled mortality rate for TCA was 96.2% (95% CI 95.0–97.2) (Fig. 2). Within the 36 included studies, 13 studies reported neurological outcome as defined in the methods section [1, 2, 19‐22, 25, 26, 30, 31, 33, 37, 43]. Of all TCA survivors, a (pooled) favorable neurological outcome was observed in 43.5% of the TCA survival patients (95% CI 32.3–55.0).

To investigate TCA mortality for different database registry types, pooled mortality rates were calculated separately for 27 studies including prehospital deaths [1, 3, 6, 17, 21, 22, 24‐27, 29‐32, 34‐41, 43‐47] and for nine studies excluding prehospital deaths (Fig. 3) [2, 16, 18‐20, 23, 28, 33, 42]. The pooled mortality rates were 97.2% (95% CI 96.3–98.0) and 92.3% (95% CI 85.7–96.9) for studies including patients declared dead on-scene and for studies excluding patients declared dead on-scene, respectively.

Likewise, neurological outcome was investigated for different database types by calculating the pooled proportion of surviving patients with a favorable neurological outcome for nine studies [21, 22, 25, 26, 29‐31, 37, 43] including and for four studies excluding prehospital deaths separately (Fig. 4) [2, 19, 20, 33]. A favorable neurological outcome was seen in 35.8% (95% CI 29.8–42.2) of surviving patients in studies including prehospital deaths and in 49.5% (95% CI 23.3–75.9) of surviving patients in studies excluding prehospital deaths.

To investigate mortality differences between EMS systems with and without a physician on scene, mortality rates were calculated separately for studies from countries or regions with and countries or regions without a physician-based service (Fig. 5). The pooled mortality rates were 93.9% (95% CI 89.3–97.2) in 10 studies where a physician was available at the prehospital scene [2, 3, 6, 28, 33, 36, 37, 40, 44, 46] and 97.6% (95% CI 96.8–98.4) in 17 studies where no physician was available at the prehospital scene [1, 6, 17‐19, 24‐27, 29, 35, 37‐39, 41, 43, 45].

Similarly, the proportion of surviving patients with a favorable neurologic status was calculated for studies from countries or regions with and countries or regions without a physician-based service. For three studies from a country or region with a physician-based EMS system [2, 33, 37], the pooled proportion of patients with a favorable neurologic outcome was 57.0% (95% CI 32.8–79.6). For six studies from a country or region without a physician-based EMS system, the pooled proportion of patients with a favorable neurologic outcome was 38.0% (95% CI 26.4–50.3) (Fig. 6) [1, 19, 25, 26, 37, 43].

Risk ratios of possible prognostic factors were calculated to investigate the association between pre- and intra-arrest factors and TCA mortality. Risk ratios were calculated for studies including prehospital deaths and for studies excluding prehospital deaths separately (Fig. 7) (Table 3; supplement). In studies including prehospital deaths, only the first monitored ECG-rhythm was associated with mortality (RR 1.12; 95% CI 1.03–1.21; p = 0.006). No risk factors were identified in studies excluding prehospital deaths. The evaluation of the funnel plots using Egger’s regression test showed that only in the analysis of the prognostic factor ‘sex’ in the studies excluding prehospital deaths, there might have been a significant amount of publication bias. The power analysis showed that all risk ratio analysis had sufficient power to support our conclusions; all prognostic factor analysis had a power of 1.00.

In conclusion, prehospital TCA is associated with a high mortality rate, with approximately one in twenty patients surviving to discharge. When interpreting results from studies on this subject, factors such as the in- or exclusion of patients that have deceased on-scene and the type of prehospital EMS system (physician-based) should be considered. Apart from first monitored ECG rhythm, this study found no other prognostic factors available to differentiate between survivors and non-survivors.