Summary of findings for the main comparison. Pre‐hospital cooling compared to in‐hospital cooling for survival, neuroprotection, and adverse events after out‐of‐hospital cardiac arrest.
| Survival, neurological outcome, and adverse events: pre‐hospital cooling compared to in‐hospital cooling after out‐of‐hospital cardiac arrest | ||||||
| Patient or population: out‐of‐hospital cardiac arrest Settings: emergency medicine and intensive care, worldwide Intervention: pre‐hospital cooling Comparison: in‐hospital cooling | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect (95% CI) | No of participants (studies) | Quality of the evidence (GRADE) | Comments | |
| Assumed risk | Corresponding risk | |||||
| I n‐hospital cooling | Pre‐hospitalcooling | |||||
| Survival and good neurological outcome | Study population | Not pooled | 2369 randomized participants (7 RCTs) (Bernard 2010; Bernard 2012; Castren 2010; Debaty 2014; Kämäräinen 2009; Kim 2007; Kim 2014) |
⊕⊝⊝⊝ VERY LOW 1,2,3,4,5,6,7 | — | |
| Not pooled | Not pooled | |||||
| Not pooled | Not pooled | |||||
| Adverse events ‐ re‐arrest after randomization | Study population | RR 1.23 (1.02 to 1.48) | 1713 participants with available information
(4 studies) (Castren 2010; Kämäräinen 2009; Kim 2007; Kim 2014) |
⊕⊝⊝⊝ VERY LOW 1,2,3,4, 7 | — | |
| 183 per 1000 | 225 per 1000 (187 to 271) | |||||
| Moderate | ||||||
| 186 per 1000 | 229 per 1000 (190 to 276) | |||||
We downgraded the quality of the evidence to 'very low' for the following reasons:
1Inappropriate application of intervention or control, or both.
2Indirectness in the intervention: two studies evaluated intra‐arrest cooling while all others evaluated post‐arrest cooling (Castren 2010; Debaty 2014).
3Indirectness in the intervention: the rate of application of pre‐hospital cooling varied over all studies; up to 50% of all participants did not receive the full intervention (Bernard 2010; Bernard 2012; Kim 2007; Kim 2014); up to 16% of participants did not receive the intervention at all (Bernard 2010; Bernard 2012; Kim 2007; Kim 2014).
4Indirectness in the comparator: the rate of application of in‐hospital cooling varied over all studies; some studies did not provide information (Castren 2010; Kämäräinen 2009; Kim 2007; Kim 2014); in some only some of the participants received in‐hospital cooling (Kämäräinen 2009; Kim 2007; Kim 2014); temperature curves in some studies indicated that a relevant proportion of participants were not cooled according to the then current guidelines (Bernard 2010; Bernard 2012); in some studies the target temperature was at the upper limit of the then current guidelines (Castren 2010).
5Indirectness in the population: one study included only adults with non‐ventricular fibrillation cardiac arrest (Bernard 2010); another only ventricular fibrillation cardiac arrest (Bernard 2012); the others did not make restrictions.
6Imprecision: three studies were feasibility or pilot studies with sample sizes too small to evaluate clinical outcomes (Castren 2010; Kämäräinen 2009; Kim 2007). Due to the above described reasons we refrained from pooling the estimates, therefore we are left with the lower precision of the individual studies.
7Risk of bias within studies: three studies lacked blinding of outcome assessment, which may substantially bias the assessment of neurological outcome (Castren 2010; Kämäräinen 2009; Kim 2007); insufficient administration and continuation of the intervention and comparator/no information on administration and continuation of the intervention and comparator (see above).