We performed a quantitative systematic review following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and according to Cochrane methodology.[6,7] A PRISMA checklist is available as a supplement (Table S1).
Systematic Search and Study Selection
Two authors (MG, AP) searched three electronic databases (PubMed, CENTRAL, EMBASE) from inception to 15 February 2019 using a variety of high-sensitivity and low-specificity search strategies. Details of the search strategy for PubMed are available as a supplement (Methods S1). There was no language restriction. Reference lists of retrieved articles were checked for further potentially relevant publications (backward snowballing). Retrieved articles were screened by three authors (CC, AP, MG). Discrepancies and queries for inclusion were resolved through consensus. If agreement could not be reached, discrepancies were discussed with the fourth author (MRT).
Inclusion and Non-inclusion Criteria
We included full reports of RCTs performed in adults (≥18 years) undergoing surgery that compared any experimental regimen of succinylcholine with the standard regimen, 1.0 mg kg− 1, for RSI. We included studies that tested a 'true' or a 'modified' RSI procedure. A true procedure involves intravenous induction with a hypnotic, intravenous administration of succinylcholine immediately (i.e. without any delay) after loss of consciousness, an apnoea period of no more than 60 s followed by orotracheal intubation. A modified RSI procedure is different in that the delay between loss of consciousness and the administration of the neuromuscular blocking agent is longer and allows, for instance, the additional recording of electromyographic baseline measures. During this time period, the patient is usually ventilated and oxygenated through a facemask. However, as with true RSI, after administration of the neuromuscular blocking agent, the apnoea period before orotracheal intubation is lasting no longer than 60 s.
Eligible trials had to report on intubating conditions using a validated score that evaluated ease of laryngoscopy, vocal cord position and movement, airway reaction, and movement of limbs. We did not consider studies including obese patients for two reasons. Firstly, obesity is an independent risk factor of difficult laryngoscopy and tracheal intubation.[10–12] Secondly, the best succinylcholine regimen in obese patients remains controversial.[13,14] Data from non-randomized trials, paediatric studies, abstracts, and trials that lacked a succinylcholine 1.0 mg kg− 1 group were also not taken into account.
Two authors (MG, CC) read the full-text articles, extracted independently all relevant information and entered the data into a predefined electronic form. Discrepancies were resolved by discussion with a third author (MRT).
According to good clinical research practice in pharmacodynamic studies of neuromuscular blocking agents, excellent, good or unacceptable intubating conditions may be distinguished. Excellent intubating conditions are present when all variables of the intubating score (ease of laryngoscopy, vocal cord position and movement, airway reaction, movement of limbs) are rated as excellent. Unacceptable intubating conditions are present when at least one variable of the intubating score is rated as poor. We chose the incidence of excellent intubating conditions (evaluated 50 to 60 s after the administration of succinylcholine) as the primary outcome as we regarded this outcome as the clinically most relevant in the context of RSI. The incidence of unacceptable intubating conditions was regarded as a secondary outcome. Good intubating conditions were not further analysed as we did not expect these data to inform decision-making. A further secondary outcome was apnoea time. Two definitions of apnoea time were used in the retrieved trials. First, apnoea time was defined as the time in minutes from succinylcholine administration to the occurrence of the first visible diaphragmatic contractions that coincided with movements of the reservoir bag. It was shown that this definition correlated with the incidence of haemoglobin desaturation defined by an oxygen saturation less than 80%. Second, in some trials, apnoea time was defined as obvious recognizable end-tidal CO2 waveforms appearing on the monitor.
Risk of Bias in Individual Studies
Quality of data reporting was assessed by two authors (MG, CC) and independently checked by another author (AP) using the Cochrane Collaboration method and a modified 4-item, 7-point Oxford scale taking into account the method of randomization, concealment of treatment allocation, degree of blinding, and reporting of drop-outs, as previously described. In the case of divergence of opinion, consensus was reached by discussion with the fourth author (MRT).
Many comparisons contained zero cells, which made the calculation of risk ratios impossible. In order not to lose potentially relevant information, we decided to calculate absolute risk differences (ARD) with 95% confidence intervals (CI) for dichotomous data. When the 95% CI around the ARD did not cross 0, the result was considered statistically significant (p value equal or inferior to 0.05). We also computed numbers needed to treat (NNT) with 95% CI as the inverse of the ARD point estimates and the lower and upper limits of their 95% CI. The NNT was the estimated number of patients who needed to be treated with the experimental regimen for one additional patient to have one more outcome compared with the comparator. A positive ARD suggested that an outcome was improved with an experimental regimen compared with the standard regimen and was consequently translated into a positive NNT. A negative ARD suggested that an outcome was worsened with an experimental regimen compared with the standard regimen and was consequently translated into a negative NNT (which may then be interpreted as a "number needed to harm"). An ARD of 0, indicating no difference between the experimental and the standard regimen, was translated into an NNT of infinity (∞). For continuous outcomes, we computed mean differences (MD) with 95% CI. We used a random-effects model throughout (Mantel-Haenszel method). Between studies heterogeneity was quantified using the I2 statistics. We performed sensitivity analyses computing relative instead of absolute risk differences and using a fixed-effect instead of a random effects model. Statistical analyses were performed with Review Manager (RevMan [Computer program], Version 5.3; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark, 2014) and Microsoft Excel 2010 (for Mac).
BMC Anesthesiol. 2020;20(54) © 2020 BioMed Central, Ltd.