Resultado económico y Financiero

I employed a modified version of the evidence evaluation process used by the International Liaison Committee on Resuscitation (ILCOR) to create the foundation for their 2015 resuscitation guidelines (Morrison et al., 2015). This modified approach consisted of:

1. PICOS question development

2. Iterative search strategy development

3. Article selection by two independent reviewers

4. Evidence review with focus on the risk of confounding and bias 5. Discussion and interpretation of findings

4.2.1 Inclusion and exclusion criteria

As discussed in Chapter 1, OHCAs occurring either due to trauma or in children tend to be caused by different pathophysiological processes and where therefore excluded from this review. Also in Chapter 1, I describe the different outcomes used to measure the quality of prehospital care for patients with OHCA. As return of spontaneous circulation (ROSC) is not a patient-focused outcome, I excluded studies which only reported rates of ROSC following OHCA. In regards to study designs, I excluded publications which described outcomes following prehospital critical care for OHCA without comparison to ALS care (e.g. case series or prehospital chart reviews). Box 4.1 summarises the inclusion and exclusion criteria, according to the PICOS structure. The PICO acronym is used frequently to ask precise clinical questions in evidence-based medicine (Hecht, Buhse and Meyer, 2016). For systematic reviews, an S is frequently added, representing the term study design(s) (Liberati et al., 2009). Box 4.2 summarises the interventions used to define prehospital critical care for this review; see Chapter 1 for a more detailed discussion of these interventions.

Box 4.1 Inclusion criteria according to the PICOS system

Patients All cases of non-traumatic out-of-hospital cardiac arrest in adults (age 18 or older)

Intervention Prehospital critical care by any provider group (paramedics or physicians) with interventional capacity beyond ALS algorithms and dedicated dispatch to critically ill patients*

Comparator ALS by any prehospital provider

Outcomes Any patient-focused outcome such as short or long-term survival or quality of life; ROSC alone was not considered a patient- focused outcome

Study designs Any comparative design such as randomized trials, but also observational studies with a comparative element

*See Box 4.2 for detailed list of potential critical care interventions ALS: Advanced Life Support, ROSC: Return of spontaneous circulation

Box 4.2 List of prehospital critical care interventions for out-of-hospital cardiac arrest, based on previous research (von Vopelius-Feldt and Benger, 2014a)

Prehospital anaesthesia

Rapid sequence induction of anaesthesia (RSI) Sedation or paralysis

Advanced cardiovascular management Central intravenous access

Cardiac ultrasound Thrombolysis

Non-ALS intravenous drugs (for example Magnesium, Calcium, Sodium Bicarbonate) Use of vasopressors after return of spontaneous circulation

Special circumstances Thoracostomy

Peri-mortem hysterotomy ALS: Advanced Life Support

4.2.2 Search strategy

After initial consultation with a University of Bristol librarian, I created reproducible search strings (see Table 4.1) that were customised for searches in each of the following electronic databases: PubMed; EmBASE; CINAHL Plus and AMED (via EBSCO); Cochrane Database of Systematic Reviews; DARE; Cochrane Central Register of Controlled Trials; NHS Economic Evaluation Database; NIHR Health Technology Assessment Database; Google Scholar;

ClinicalTrials.gov. The wide spread of databases was chosen to optimise capture of grey literature, such as conference abstracts or government reports, as well as reliably identify higher level evidence such as randomised trials or existing systematic reviews.

Each search string was designed to meet the inclusion and exclusion criteria but also to balance sensitivity and specificity within each database. Therefore, the search strings for databases that returned only a few results on initial test searches consist of very broad categories, whereas searches within databases with large numbers of results (for example PubMed) were more restrictive. I tested for sensitivity of the more restrictive search strings by checking if known key publications (von Vopelius-Feldt, Coulter and Benger, 2015; Olasveengen et al., 2009; Mitchell et al., 1997) were identified by the search.

I excluded research published prior to 1990 as it was deemed very unlikely that this would be relevant to modern EMS practice. ALS guidelines are updated every 5 years and have changed significantly since pre-1990 (Link et al., 2015; Chamberlain, 1989). Likewise, advances in technology and evidence-based medicine have changed practice and standards of prehospital care to a level that is not comparable to pre-1990 (Spaite et al., 2014; Helm et

al., 1991).

Despite the fact that the review aimed to evaluate prehospital critical care by any provider, the search strategy reflects the fact that prehospital critical care is often provided by physicians or helicopter medical services (HEMS). Search terms such as doctor or helicopter were always used to broaden the search (OR syntax), rather than restricting it; see Table 4.1. Also included in the results were all cited and citing articles of publications which were retrieved for full text analysis during the review process. In addition, I used social media requests (Twitter and Research Gate) to identify further grey literature. The final searches for each database were undertaken between April and June 2016. See Table 4.1 for a comprehensive overview of search strings, databases and search results.

Table 4.1 Literature search strategies and results

Database Search strategy Results

Pubmed (prehospital emergency care[MeSH Terms]) AND cardiac arrest, out of hospital[MeSH Terms]

622 Pubmed (pre-hospital OR prehospital OR EMS OR

"emergency medical services" OR ambulance) AND (arrest OR sudden death) AND ("critical care" OR "intensive care" OR doctor OR physician OR HEMS OR helicopter)

1377

EmBASE (pre-hospital or prehospital or EMS or "emergency medical services") and (arrest or "sudden death") and ("critical care" or "intensive care" or doctor or physician or HEMS or helicopter)

963

CINAHL Plus and AMED (via EBSCO)

( pre-hospital or prehospital or EMS or 'emergency medical services' ) AND ( arrest or 'sudden death' ) AND ( critical care or intensive care or doctor or physician ) 359 Cochrane Database of Systematic Reviews cardiac arrest 24 Database of Abstracts of Review of Effects (DARE) cardiac arrest 70 Cochrane Central Register of Controlled Trials

(pre-hospital OR prehospital OR EMS OR emergency medical services) AND (arrest OR sudden death) AND (critical care OR intensive care OR doctor OR physician) 97 NHS Economic Evaluation Database cardiac arrest 42 NIHR Health Technology Assessment Database

[MeSH] out-of-hospital cardiac arrest 2

Google Scholar (limited to 500 most relevant)

prehospital critical care out-of-hospital cardiac arrest

500

Clinicaltrials.gov (prehospital OR pre-hospital OR EMS OR emergency medical services) AND (cardiac arrest OR sudden death)

133

Cited and citing articles of full text publications reviewed

365

Social media 0

4.2.3 Article selection

Article selection followed a three-step approach. First, two researchers (myself and co- supervisor JBR) independently reviewed all study titles and removed all publications which were obviously not related to the study question as well as duplicate results. Next, we independently reviewed the abstracts of all remaining publications, removing those that did not fulfil the inclusion criteria outlined in Box 4.1. Finally, both researchers independently reviewed the full text of all remaining publications to assess for inclusion in the final analysis. If there were discrepancies in the researchers’ opinions during step one or two, the publication in question was moved forward to the next step. If there were discrepancies in step three, consensus was sought between the two researchers. If no consensus was achieved, a third researcher (supervisor JB) would have been asked to review the publication, however, this final step was not required for this review. The final full narrative analysis of all included manuscripts was undertaken by myself.

4.2.4 Analysis and presentation of results

The narrative analysis was undertaken in a three-step approach for each publication. First, I read each manuscript multiple times, to fully immerse myself in the research. I then extracted key aspects of each study into an evidence table, using the Strengthening The Reporting of Observational Studies in Epidemiology (STROBE) checklist (von Elm et al., 2014) as guidance; see Table 4.2. Finally, I examined the publications for risk of bias and confounding, paying particular attention to sample size and population, differences in treatments or follow-up and methods of adjustments. Potential sources of bias or confounding for each study are presented in Section 4.4, together with descriptions of prehospital critical care and ALS care for each study. Care was taken to provide only a description of publications at this stage, rather than interpretation.

Finally, the overall reliability of the evidence, drawing on the findings of the evidence review, is discussed in Section 4.5. It also provides further interpretation of the systematic review findings by referencing other research and providing a wider context of the findings.

Overall, the systematic review was structured and presented according to the PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions (Liberati et al., 2009). In keeping with good research practice, I prospectively registered the review with the International Prospective Register of Systematic Reviews (PROSPERO), registration number CRD42016039995.