Efficacy and safety of sugammadex versus neostigmine in reversing neuromuscular blockade in adults

Summary of findings for the main comparison. Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg

Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg
Patient or population: adult patients, ASA I to IV, who received non‐depolarizing NMBAs Setting: elective in‐patient or day‐case surgical procedures performed at centres across Europe and Asia Intervention: sugammadex 2.0 mg/kg Comparison: neostigmine 0.05 mg/kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
	Neostigmine 0.05 mg/kg	Sugammadex 2.0 mg/kg
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Mean recovery time from T2 to TOFR > 0.9 was 12.87 minutes	Mean recovery time from T2 to TOFR > 0.9 was 1.96 minutes Mean recovery time from T2 to TOFR > 0.9 in the sugammadex group was10.22 minutes faster (8.48 to 11.96 minutes faster) than neostigmine	‐	835 (10 studies)	⊕⊕⊕⊝^c Moderate	TSA alfa‐boundary adjusted MD is ‐10.22 (95% CI ‐12.11 to ‐8.33; diversity (D²) = 87%, I² = 84%, random‐effects model, 80% power, alpha 0.05). Cumulative Z‐curve crosses the monitoring boundary (Figure 1)
Recovery time^a from post‐tetanic count (PTC) 1 to 5 to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Outcome not clinically relevant for this comparison
Risks of adverse events and serious adverse events^b, bradycardia, PONV, and signs of residual neuromuscular blockade	Outcome not analysed for this comparison
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High quality:We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in post‐anaesthetic care unit, or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies)

Figure 1

TSA of all trials comparing sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg; recovery time from T2 to TOFR > 0.9 minutes. With a required information size of 106, firm evidence in place favours sugammadex in a random‐effects model, with an alfa‐boundary adjusted MD of ‐10.22 (95% CI ‐12.11 to ‐8.33; diversity (D²) = 87%, I² = 84%, random‐effects model). The cumulative Z‐curve crosses the monitoring boundary constructed for the required information size with 80% power and alpha of 0.05. However, none of the included trials had low risk of bias, and because TSA is ideally designed for trials with low risk of bias and cannot be adjusted for risk of bias, the precision of our findings has to be downgraded. Furthermore, the degree of diversity and heterogeneity is high, which once again raises questions about the reliability of the calculated required information size.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Summary of findings 2. Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg

Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg
Patient or population: adult patients, ASA I to IV, who received non‐depolarizing NMBAs Setting: elective in‐patient or day‐case surgical procedures performed in Italy and USA Intervention: sugammadex 4.0 mg/kg Comparison: neostigmine 0.07 mg/kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
	Neostigmine 0.07 mg/kg	Sugammadex 4.0 mg/kg
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Outcome not clinically relevant for this comparison.
Recovery time^a from post‐tetanic count (PTC 1 to 5) to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Mean recovery time from PTC 1 to 5 to TOFR > 0.9 was 48.8 minutes	Mean recovery time from PTC 1 to 5 to TOFR > 0.9 was 2.9 minutes Mean recovery time from PTC 1 to 5 to TOFR > 0.9 in the sugammadex group was 45.78 minutes faster (52.15 to 39.41 minutes faster) than in the neostigmine group	‐	114 (2 studies)	⊕⊕⊝⊝^c Low
Risk of adverse events and serious adverse events^b, bradycardia, PONV, and signs of residual neuromuscular blockade	Outcome not analysed for this comparison
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in the post‐anaesthetic care unit, or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of participants, n = 114)

See Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification; and Characteristics of ongoing studies.

Summary of findings 3. Sugammadex (any dose) vs neostigmine (any dose)

Sugammadex (any dose) compared to Neostigmine (any dose)
Patient or population: Adult patients, ASA I‐IV, who received non‐depolarizing NMBAs Setting: Elective in‐patient or day‐case surgical procedures performed in centres across Europe, USA and Asia Intervention: Sugammadex (any dose) Comparison: Neostigmine (any dose)
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
Outcomes	Risk with neostigmine (any dose)	Risk with sugammadex (any dose)	Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Outcome not clinically relevant for this comparison
Recovery time^a from post‐tetanic count (PTC) 1 to 5 to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Outcome not clinically relevant for this comparison
Risk of composite adverse events^b	283 per 1000	159 per 1000 (137 to 204)	RR 0.60 (0.49 to 0.74)	2298 (28 studies)	⊕⊕⊕⊝^c Moderate	TSA with continuity adjustment for zero event trials (0.001 in each arm); alfa‐boundary adjusted RR 0.62 (95% CI 0.51 to 0.74; diversity (D²) = 34%, I² = 14%, random‐effects model; 80% power, 0.05 alpha; Figure 3)
Bradycardia	84 per 1000	13 per 1000 (6 to 28)	RR 0.16 (0.07 to 0.34)	1218 (11 studies)	⊕⊕⊕⊝^d Moderate
PONV	131 per 1000	68 per 1000 (33 to 115)	RR 0.52 (0.28 to 0.97)	389 (6 studies)	⊕⊕⊝⊝^e Low
Overall signs of postoperative residual paralysis	131 per 1000	52 per 1000 (37 to 75)	RR 0.40 (0.28 to 0.57)	1474 (15 studies)	⊕⊕⊕⊝^f Moderate	TSA with continuity adjustment for zero event trials (0.001 in each arm): alfa‐boundary adjusted RR 0.4 (95% CI 0.27 to 0.59; diversity (D²) = 0%, I² = 0%, random‐effects model, 80% power, 0.05 alpha, Figure 4). Cumulative Z‐curve crosses the monitoring boundary constructed for a required information size of 424 participants indicating firm evidence in favour of sugammadex
Risk of serious adverse events^b	10 per 1000	6 per 1000 (1 to 23)	RR 0.54 (0.13 to 2.25)	959 (10 studies)	⊕⊕⊝⊝^g Low	TSA with continuity adjustment for zero event trials (0.001 in each arm): alfa‐boundary adjusted RR 0.35 (95% CI 0.00 to 3190; diversity (D²) = 0%, I² = 0%, random‐effects model, 80% power, alpha 0.05), Cumulative Z‐curve does not cross the monitoring boundary constructed for a required information size of 8189 participants with 11.71% of the required information size included
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio;
GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in the post‐anaesthetic care unit or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^dDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^eDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of participants‐ n = 389 ‐ and wide confidence interval (CI) ‐ 0.28 to 0.97) ^fDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^gDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of events ‐ 10/1000 in the neostigmine group vs 6/1000 in the sugammadex group ‐ and wide confidence interval (CI) ‐ 0.13 to 2.25)

Figure 3

TSA of dichotomous data on drug‐related risk of adverse events; sugammadex (any dose) vs neostigmine (any dose). This analyses includes continuity adjustment for zero event trials (0.001 in each arm) resulting in an alfa‐boundary adjusted RR of 0.62 (95% CI 0.51 to 0.74; diversity (D²) = 34%, I² = 14%, random‐effects model), with a control event proportion of 27.97%. With the required information size of 502, analyses indicated firm evidence favouring sugammadex with 2298 participants included corresponding to a relative risk reduction (RRR) of 38% with 80% power and alpha of 0.05. Despite the fact that the cumulative Z‐curve does not cross the monitoring boundary directly, it is hard to imagine future trials radically changing the overall picture of this analysis. However, none of the included trials were at low risk of bias, and this does downgrade the reliability of our finding.

Figure 4

TSA of dichotomous data on risk of signs of residual neuromuscular blockade; sugammadex (any dose) vs neostigmine (any dose). With continuity adjustment for zero event trials (0.001 in each arm), TSA resulted in an alfa‐boundary adjusted RR of 0.4 (95% CI 0.27 to 0.59; diversity (D²) = 0%, I² = 0%, random‐effects model, with 80% power and alpha of 0.05), with a control event proportion of 13.08%. Cumulative Z‐curve crosses the monitoring boundary constructed for a required information size of 424 participants, indicating firm evidence in favour of sugammadex. However, none of the included trials had low risk of bias, and this equally diminishes the reliability and precision of our estimates.

Background

After several discussions with the editorial team, a decision was reached to split the original review (Abrishami 2009) into two reviews based on the very extensive number of publications (> 70) identified by the updated search along with various comparators, interventions, and outcome measures.

Description of the condition

Neuromuscular blockade

Neuromuscular blocking agents (NMBAs) are drugs that induce skeletal muscle relaxation primarily by causing a decreased response to the neurotransmitter acetylcholine (ACh) at the neuromuscular junction of skeletal muscle. At that site, ACh normally produces electrical depolarization of the postjunctional membrane of the motor end‐plate, which leads to conduction of muscle action potential and subsequently induces skeletal muscle contraction. Neuromuscular agents are classified as depolarizing or nondepolarizing (PubChem 2016). Non‐depolarizing NMBAs may be further subdivided into aminosteroidal and curariform types of agents.

Use of NMBAs during surgery facilitates tracheal intubation, protects patients from vocal cord injury, and improves surgical conditions by suppressing voluntary or reflex skeletal muscle movements (Bowman 2006; Keating 2016). Following surgery, relaxation is no longer needed, it is important that effects of the NMBA can be quickly and effectively terminated. Postoperative residual neuromuscular blockade and resulting muscle weakness caused by non‐depolarizing NMBAs have been shown to be associated with increased mortality and morbidity (Pedersen 1994; Shorten 1993). Residual neuromuscular blockade may result in pulmonary complications, for example, laboured breathing, low oxygen levels in the blood, lung infection, and entry of gastric contents into the lungs (Berg 1997; Bevan 1996; Eriksson 1993; Eriksson 1997; Murphy 2006; Murphy 2008; Sundman 2000). It can also lead to a postoperative decrease in muscle strength with associated complications, such as visual difficulties and delayed recovery and discharge time (Murphy 2011). Postoperative residual blockade frequently occurs after routine anaesthesia (Viby‐Mogensen 1979). Its incidence varies among trials depending on the type of NMBA used. Some studies have demonstrated a lower incidence of residual block following short‐acting or intermediate‐acting NMBAs in comparison with long‐acting agents (Bevan 1988; Brull 1991). However, postoperative residual neuromuscular blockade may still occur in the short‐acting or intermediate‐acting NMBA group, with incidence ranging from 16% to 60% (Appelbaum 2003; Baillard 2005; Bevan 1996; Debaene 2003; Fawcett 1995; Hayes 2001; Kim 2002; Maybauer 2007; McCaul 2002).

Monitoring of neuromuscular blockade

The degree of neuromuscular blockade is monitored by assessment of various patterns of electrical stimulation. The train‐of‐four (TOF) twitch stimulation was developed as a clinical tool that could be used to assess neuromuscular block in the anaesthetized patient (Ali 1970). This strategy involves stimulating the ulnar nerve with four supramaximal 200 microsecond stimuli separated by 0.5 seconds. This approach is repeated every 10 seconds. Twitches on a TOF pattern fade as relaxation increases. This enables the observer to compare T1 (first twitch of the TOF) versus T0 (control), as well as T4 (fourth twitch of the TOF) versus T1. This T1/T4 ratio is known as the TOF ratio (TOFR). Satisfactory recovery from neuromuscular block and clinical absence of residual curarization have not occurred until the TOFR is > 0.9 (Viby‐Mogensen 2000), contrary to TOFR > 0.7, as previously suggested (Ali 1971). During profound non‐depolarizing neuromuscular block, no response to TOF twitch stimulation may occur. In such circumstances, a post‐tetanic count (PTC) may be useful (Viby‐Mogensen 1981). If a 5 second tetanic stimulus at 50 Hz is administered, after no twitch response has been elicited, followed 3 seconds later by additional single twitches at 1 Hz, response to single twitch stimulation may occur. Although this pattern will not be seen during very profound block, a response will be seen in the early stages of recovery, before the TOF reappears. The number of post‐tetanic twitches is an indication of when the first twitch of the TOF will reappear.

The muscle response to peripheral nerve stimulation can be assessed by visual and tactile methods and by electromyography, acceleromyography, and mechanomyography. Visual observation and palpation of the contracting muscle group are the easiest but least accurate methods of assessing neuromuscular block. Acceleromyography was introduced for clinical use in 1988 (Jensen 1988; Viby‐Mogensen 1988). This technique measures acceleration of a distal digit, which is directly proportionate to the force of muscle contraction and therefore is inversely proportionate to the degree of neuromuscular block.

The monitor consists of an acceleration transducer (i.e. a piezo‐electric ceramic wafer with an electrode on each side) and a stimulation and computing unit. The transducer can be fastened to the thumb, and when the finger is moved in response to nerve stimulation, a voltage difference develops between the two electrodes. The voltage then is measured and is registered in the computing unit.

Description of the intervention

Reversal of neuromuscular blockade

The most commonly used NMBA reversal agents are neostigmine and edrophonium, both of which are cholinesterase inhibitors. They antagonize both aminosteroidal and curariform types of non‐depolarizing NMBAs by inhibiting the breakdown of ACh in the neuromuscular junction (NMJ), causing, ACh to bind the receptor and depolarize the muscle fibre and allowing greater transmission of nerve impulses. These medications, however, require that a muscarinic antagonist (e.g. glycopyrrolate, atropine) be used to compensate for their cholinergic side effects such as bradycardia, hypotension, bronchoconstriction, and postoperative nausea and vomiting (Tramer 1999). Adverse effects associated with the use of muscarinic antagonists include tachycardia, dry mouth, and urinary retention (Mirakhur 1985).

In contrast to cholinesterase inhibitors, the NMBA reversal agent sugammadex does not interfere with acetylcholinesterase receptor systems; therefore, it does not produce the muscarinic side effects associated with other reversal medications for NMBAs. Sugammadex is a synthetically modified ɣ‐cyclodextrin, a chemical structure with a hydrophilic exterior and a hydrophobic core. It was specifically designed to reverse rocuronium‐induced paralysis by encapsulating rocuronium; however, its inner cavity is large enough to encapsulate other aminosteroidal NMBAs such as vecuronium and, to a much lesser degree, pancuronium (Golembiewski 2016; Naguib 2009). Sugammadex does not bind nor does it reverse the neuromuscular blocking effects of curariform NMBAs. Upon binding, it creates a complex formation between the molecule and the aminosteroidal NMBA, which results in more rapid reversal of the neuromuscular blockade than is achieved by anticholinesterase drugs (Park 2015). Sugammadex does not bind to plasma proteins and is not metabolized. It is excreted unchanged in the urine by the kidneys. Renal clearance of sugammadex is rapid ‐ most of the dose (70%) is excreted within six hours (Golembiewski 2016).

How the intervention might work

The positively charged quaternary nitrogen of the aminosteroidal NMBA forms electrostatic bonds with negatively charged interior groups of sugammadex to encapsulate rocuronium and vecuronium (Golembiewski 2016). Sugammadex forms a stable, inactive 1:1 complex with rocuronium or vecuronium; this reduces the amount of free NMBA that is available to bind to nicotinic acetylcholine receptors at the neuromuscular junction, resulting in reversal of neuromuscular blockade (Keating 2016). Once the NMBA is removed from its site of action and is rendered inactive (by encapsulation within the sugammadex molecule in the plasma), neuromuscular transmission and muscle function are restored. By reversing aminosteroid‐induced neuromuscular blockade, one can avoid the associated risks caused by residual block, can shorten time in the operating room, and can improve the patient's quality of recovery and discharge time (Arbous 2005).

Why it is important to do this review

Residual neuromuscular block is a common complication in the post‐anaesthesia care unit, with approximately 40% of patients exhibiting a TOFR < 0.9 (Murphy 2010). The clinical safety and efficacy of sugammadex in reversing rocuronium‐induced neuromuscular blockade have been studied in several randomized controlled trials (RCTs) that compared this medication versus placebo or conventional reversal agents (de Boer 2007; Gijsenbergh 2005; Sacan 2007; Sorgenfrei 2006; Sparr 2007). The aim of our review was to update the best available evidence on this topic and to assess the efficacy and safety of sugammadex and neostigmine in reversal of neuromuscular blockade. We aimed to systematically review RCTs conducted to examine sugammadex and neostigmine administration.

Objectives

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs irrespective of publication status, date of publication, blinding status, outcomes published, or language. We contacted trial investigators and study authors to ask for relevant data. We included unpublished trials only if trial data and methodological descriptions were provided in written form or could be retrieved from the trial authors. We excluded observational studies. We did not include studies using a non‐standard design, such as cross‐over trials and cluster‐randomized trials.

Types of participants

We included adults (> 18 years of age) classified as American Society of Anesthesiologists (ASA) I to IV who had received non‐depolarizing NMBAs for an elective in‐patient or day‐case surgical procedure, and who consented to be included in the study. We did not include paediatric participants, healthy volunteers, or participants not undergoing surgical procedures.

Types of interventions

We included all trials comparing sugammadex versus neostigmine in adults receiving non‐depolarizing NMBAs. We included any dose of sugammadex and neostigmine and any time point of administration of study drug.

We excluded trials that compared sugammadex and neostigmine versus only placebo or no intervention.

Types of outcome measures

Primary outcomes

Recovery time from second twitch (T2) to TOFR > 0.9
Recovery time from post‐tetanic count (PTC) 1 to 5 to TOFR > 0.9

For our first primary outcome "Recovery time from T2 to TOFR > 0.9", we compared sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg. For our second primary outcome "Recovery time from PTC 1 to 5 to TOFR > 0.9", we compared sugammadex 4 mg/kg versus neostigmine 0.07 mg/kg. In all studies, the TOF‐watch assessor used acceleromyography to measure recovery time in minutes from administration of the study drug to TOFR > 0.9 at the same monitoring site (ulnar nerve and adductor pollicis muscle).

Secondary outcomes

Risk of adverse events
Risk of serious adverse events

Study authors defined and safety outcome assessors observed and assessed adverse events and serious adverse events in the operating theatre, in the post‐anaesthetic care unit, or up to seven days after surgery, depending on each study. Furthermore, this review regarded as adverse events overall clinical signs of postoperative residual paralysis reported by trial authors. We measured risk of adverse events as the number of adverse events per all participants and/or the number of participants experiencing one or more adverse events per all participants. We included in risk assessments only adverse events that were possibly, probably, or definitely related to study drug. We included in the analysis adverse events and serious adverse events observed following any administered dose of sugammadex and neostigmine and at any time point of study drug administration. Additionally, for the purposes of this review, we presented adverse events as specific adverse events as well as composite adverse events, defined as the combination of all adverse events.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2016, Issue 4); MEDLINE (WebSPIRS Ovid SP, 1950 to 2 May 2016); and Embase (WebSPIRS Ovid SP, 1980 to 2 May 2016). We applied no language restrictions. We did a top‐up search in May 2017. For specific information regarding our search strategies and results, please see Appendix 1, Appendix 2, and Appendix 3.

Searching other resources

We searched for ongoing clinical trials and unpublished trials at the following Internet sites.

We handsearched the reference lists of reviews, randomized and non‐randomized trials, and editorials for additional trials. We contacted the main authors of trials in this field to ask about missed, unreported, and ongoing trials. We applied no language restrictions to eligible reports.

We conducted the latest search on 2 May 2016, along with a top‐up search in May 2017.

Data collection and analysis

Two review authors (AMH, PD) independently screened and classified all citations as potential primary studies, review articles, or other; independently examined all potentially eligible primary trials and decided on their inclusion in the review; and furthermore independently extracted data from each trial and evaluated data on methods and outcomes in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We (AMH, PD) resolved disagreements by discussion and by consultation with the last review author (AA).

Selection of studies

We assessed articles identified via the described searches and excluded obviously irrelevant reports. Two review authors (AMH, PD) independently examined articles and screened titles and abstracts to identify eligible trials. We completed this process without blinding to study authors, institutions, journals of publication, or results. We resolved disagreements by reaching consensus among two review authors (AMH, PD) and by consultation with the last review author (AA). We listed all excluded trials along with reasons for their exclusion in the Characteristics of excluded studies table.

Data extraction and management

We independently extracted and collected data from each trial without blinding to study authors, source institutions, or publication sources of trials. We resolved disagreements by discussion and approached all first authors of included trials for additional information on risks of bias. For more detailed information, please see Contributions of authors.

Assessment of risk of bias in included studies

We evaluated the validity and design characteristics of each trial.

We evaluated trials for major potential sources of bias (random sequence generation, allocation concealment, blinding of participants, blinding of personnel, blinding of primary outcome assessor, blinding of secondary outcome assessor, incomplete outcome data, selective reporting, funding bias and other bias; see Appendix 4). We assessed each trial quality factor separately and defined trials as having low risk of bias only if they adequately fulfilled all of the criteria described below.

Measures of treatment effect

For our primary outcome (recovery time to TOFR > 0.9), we used mean differences (MDs) with 95% confidence intervals (CIs) because data were continuous and were measured in the same way by all trials. For our secondary outcomes (risks of adverse events and serious adverse events), we calculated risk ratios (RRs) with 95% CIs for dichotomous data (binary outcomes), which were measured in the same way between trials. We also presented data for primary and secondary outcomes as relative differences. (See Data collection and analysis section.)

Unit of analysis issues

Trials with multiple intervention groups

In accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), we combined data for secondary outcomes extracted from trials with two or more groups receiving different doses of sugammadex or neostigmine. We excluded trials that compared only different doses of sugammadex or different doses of neostigmine, as well as trials without a control group.

Cross‐over trials

We planned to exclude cross‐over trials from our meta‐analyses because of potential risk for “carry‐over” of treatment effect. However, we identified no cross‐over trials through our search.

Dealing with missing data

We contacted the authors of trials with missing data to retrieve relevant information. For all included trials, we noted levels of attrition and any exclusions. In cases of missing data, we chose 'complete‐case analysis’ for our primary outcomes, which excludes from the analysis all participants for whom the outcome is missing.

Selective outcome reporting, which occurs when non‐significant results are selectively withheld from publication (Chan 2004), is defined as selection, on the basis of trial results, of a subset of the original variables recorded for inclusion in publication of trials (Hutton 2000). The most important types of selective outcome reporting include selective omission of outcomes from reports; selective choice of data for an outcome; selective reporting of different analyses using the same data; selective reporting of subsets of the data; and selective underreporting of data (Higgins 2011).

Assessment of heterogeneity

We explored heterogeneity using the I² statistic and the Chi² test. An I² statistic above 50% represents substantial heterogeneity (Higgins 2011). In cases of substantial heterogeneity, we tried to determine the cause of heterogeneity by performing relevant subgroup and sensitivity analyses (excluding potential outliers to see visual impact of the overall value of the I² statistic on forest plots). We used the Chi² test to provide an indication of heterogeneity between trials, with a P value ≤ 0.1 considered significant. However, in cases of presumed substantial clinical heterogeneity within an analysis, we planned to use the random‐effects model independent of I² value.

Assessment of reporting biases

We included both published and unpublished studies during the selection process. We attempted to source published protocols for each of our included studies by using clinical trials registers. We compared published protocols versus published study results to assess the risk of selective reporting bias. Two review authors (AMH and PD) resolved disagreements by discussion and by consultation with the last review author (AA). As we included a sufficient number of studies (greater than 10), we assessed reporting biases (such as publication bias) by using funnel plots. We used the asymmetry of the funnel plot to assess risk of publication and other reporting bias (Higgins 2011). An asymmetrical funnel plot may indicate publication of only positive results (Egger 1997).

Data synthesis

Data analysis

We used Review Manager software (RevMan 5.3.5) and calculated MDs with 95% CIs for continuous outcomes, and RRs with 95% CIs for dichotomous variables. We used the Chi² test to obtain an indication of heterogeneity between trials, with P ≤ 0.1 considered significant. We quantified the degree of heterogeneity observed in the results by using the I² statistic, which can be interpreted as the proportion of total variation observed between trials that is attributable to differences between trials rather than to sampling error (Higgins 2011). I² > 75% is considered as very heterogeneous. However, we chose a random‐effects model for all of our analyses because clinical heterogeneity was a considerable issue beside the inter‐study heterogeneity expressed by the I² statistic. Thus, we saw little rationale to carry out comparative analyses examining the impact of the choice between using a fixed‐effect versus a random‐effects model.

Trial sequential analysis

Risk of type 1 errors in meta‐analyses due to sparse data and repeated significance testing following updates with new trials remains a serious concern (Brok 2009; Thorlund 2009; Wetterslev 2008; Wetterslev 2009). As a result, spurious P values due to systematic errors from trials with high risk of bias, outcome reporting bias, publication bias, early stopping for benefit, and small trial bias may result in false conclusions. In a single trial, interim analysis increases the risk of type 1 errors. To avoid type 1 errors, group sequential monitoring boundaries (Lan 1983) are used to decide whether a trial could be terminated early because of a sufficiently small P value, with the cumulative Z‐curve crossing the monitoring boundary.

Sequential monitoring boundaries can be applied equally to meta‐analyses and are labelled 'trial sequential monitoring boundaries’. In 'trial sequential analysis’ (TSA), the addition of each new trial to a cumulative meta‐analysis is viewed as an interim meta‐analysis, which provides useful information on the need for additional trials (Wetterslev 2008).

It is appropriate and wise to adjust new meta‐analyses for multiple testing on accumulating data to control overall type 1 error risk in cumulative meta‐analysis (Pogue 1997; Pogue 1998; Thorlund 2009; Wetterslev 2009).

When TSA is performed, the cumulative Z‐curve crossing the boundary indicates that a sufficient level of evidence has been reached; as a consequence, one may conclude that no additional trials may be needed. However, evidence is insufficient to allow a conclusion if the Z‐curve does not cross the boundary or does not surpass the required information size.

To construct trial sequential monitoring boundaries (TSMBs), one needs a required information size, which is calculated as the least number of participants required in a well‐powered single trial with low risk of bias (Brok 2009; Pogue 1998; Wetterslev 2008).

In this updated review, we adjusted the required information size for heterogeneity by using the diversity adjustment factor (Wetterslev 2009). We applied TSA, as it prevents an increase in the risk of type 1 errors (20%). If the actual accrued information size was too small, we provided the required information size in the light of actual diversity (Wetterslev 2009).

Subgroup analysis and investigation of heterogeneity

We conducted the following subgroup analyses.

Sugammadex 2.0 mg/kg versus neostigmine 0.05 mg/kg: recovery time from T2 to TOFR > 0.9
1. Total intravenous anaesthesia (TIVA) versus volatile anaesthetics
Sugammadex, any dose, versus neostigmine, any dose: adverse events
1. Composite adverse events: different dosages of sugammadex versus neostigmine
2. Composite adverse events: TIVA versus volatile anaesthetics
3. Bradycardia: atropine versus glycopyrrolate
4. Postoperative nausea and vomiting (PONV): TIVA versus volatile anaesthetics

If analyses of various subgroups were significant, we planned to perform a test of interaction (Altman 2003). We considered P values < 0.05 as indicating significant interaction between treatments and subgroup categories. However, because subgroup analyses showed no significant differences, we performed no tests of interaction.

Sensitivity analysis

We conducted the following sensitivity analyses.

Sugammadex 2.0 mg/kg versus neostigmine 0.05 mg/kg, recovery time from T2 to TOFR > 0.9, excluding meeting abstracts
Sugammadex, any dose, versus neostigmine, any dose, composite adverse events, excluding meeting abstracts

Summary of findings table and GRADE

We used the principles of the GRADE approach to perform an overall assessment of evidence related to all of our outcomes. We constructed a 'Summary of findings' table using GradePro software. As outcomes of clinical interest, we chose to present recovery time from T2 to TOFR > 0.9 (moderate block); recovery time from PTC 1 to 5 to TOFR > 0.9 (deep block); risks of adverse events, serious adverse events, bradycardia, and PONV; and signs of residual neuromuscular blockade (see summary of findings Table for the main comparison; summary of findings Table 2; and summary of findings Table 3).

Results

Description of studies

Results of the search

In May 2016, through electronic searches and searches of the references of potentially relevant articles, we identified 2502 publications. We excluded 2431 publications, as they were duplicates (n = 675), measured clearly irrelevant outcomes, or were not RCTs. We retrieved a total of 72 relevant publications for further assessment. Of these, 14 were ongoing trials, one trial was awaiting classification, and 16 were excluded with reasons. We reran the search in May 2017 and identified 513 citations (503 by searching databases and 10 by searching clinical trials). Upon reading titles/excluding duplicates, we found 11 studies of interest; of these, two are awaiting classification, six are ongoing, and three were excluded with explanation. In total, 41 RCTs (N = 4206) met our inclusion criteria. Of these, 31 trials (N = 2559) were eligible for meta‐analyses, 20 are ongoing, and three are awaiting classification. We have provided search results in a flow chart in Figure 5.

Figure 5

Study flow diagram.

Included studies

We included 41 trials (4206 participants) in our review.

Publication type

Of the 41 included trials, 29 (71%) were published as full‐text papers (Adamus 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Castro 2014; Cheong 2015; Flockton 2008; Gaszynski 2011; Geldner 2012; Hakimoglu 2016; Illman 2011; Isik 2016; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Koyuncu 2015; Lemmens 2010; Martini 2014; Mekawy 2012; Pongracz 2013; Rahe‐Meyer 2014; Sabo 2011; Schaller 2010; Tas 2015; Woo 2013; Wu 2014; Yagan 2015). Twelve (29%) of the 41 trials were available only as meeting abstracts (Balaka 2011; Foletto 2014; Georgiou 2013; Grintescu 2009; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Raziel 2013; Riga 2014; Sherman 2014; Sustic 2012). All of the included trials were published in English, with the exception of one article that was published in Turkish (Koc 2015). We contacted all 41 trial authors for missing information; 12 (29%) replied and provided supplementary data.

Participants and settings

We reported full details of participants and settings in the Characteristics of included studies section.

Of the 41 included studies, 30 were single‐centre studies conducted in 15 countries: Turkey (seven studies: Hakimoglu 2016, Isik 2016, Kizilay 2016, Koc 2015, Koyuncu 2015, Tas 2015, Yagan 2015), Croatia (five studies: Kogler 2012, Kvolik 2012a, Kvolik 2012b, Kvolik 2013, Sustic 2012), Greece (three studies: Balaka 2011, Georgiou 2013, Riga 2014), Germany (two studies: Kaufhold 2016, Schaller 2010), Israel (two studies: Raziel 2013, Sherman 2014), Italy (two studies: Carron 2013, Foletto 2014) and one study each in Egypt (Mekawy 2012), Hungary (Pongracz 2013), Netherlands (Martini 2014), Czech Republic (Adamus 2011), Portugal (Castro 2014), Poland (Gaszynski 2011), Romania (Grintescu 2009), Korea (Cheong 2015), and USA (Brueckmann 2015). Eleven were multiple‐centre studies: 22 European centres in Rahe‐Meyer 2014, 13 European centres in Blobner 2010 and Khuenl‐Brady 2010, 10 European centres in Geldner 2012, nine US centres in Jones 2008 and Lemmens 2010, eight European centres in Flockton 2008, seven Korean centres in Woo 2013, six Chinese plus four European centres in Wu 2014, two Finnish centres in Illman 2011, and an unspecified number of US centres in Sabo 2011.

The sample size of included trials ranged from 22 to 1198 adults (aged > 18 years) with ASA status I to IV. Among studies reporting ASA status, the distribution of participants across groups was as follows: ASA I: 1003 participants (32%); ASA II: 1772 participants (56%); ASA III: 331 participants (11%); and ASA IV: 31 participants (1%).

Five trials included only morbidly obese (MOB) participants (Carron 2013; Castro 2014; Foletto 2014; Gaszynski 2011; Raziel 2013), and one trial focused on super‐obese (SO) patients (Georgiou 2013). One trial included participants classified as New York Heart Association (NYHA) II to III (Kizilay 2016), and one trial investigated participants with myasthenia gravis (Balaka 2011).

Participants underwent diverse elective surgical procedures under general anaesthesia: extreme lateral interbody fusion (Adamus 2011); trans‐sternal thymectomy (Balaka 2011); laparoscopic or open abdominal surgery (Brueckmann 2015); laparoscopic removal of adjustable gastric banding (Carron 2013); laparoscopic bariatric surgery (Castro 2014); laparoscopic sleeve gastrectomy (Foletto 2014; Raziel 2013; Sherman 2014); elective bariatric surgery (Gaszynski 2011); laparoscopic cholecystectomy or appendectomy (Geldner 2012); laparoscopic cholecystectomy (Grintescu 2009; Sustic 2012); open bariatric surgery (Georgiou 2013); arthroscopic surgery (Hakimoglu 2016); non‐cardiac surgery (Kizilay 2016); interventional bronchoscopy (Kogler 2012); extremity surgery (Koyuncu 2015); thyroidectomy (Kvolik 2012a; Kvolik 2012b); thyroidectomy or breast cancer surgery (Kvolik 2013); laparoscopic prostatectomy or nephrectomy (Martini 2014); endoscopic sinus surgery with or without septoplasty (Mekawy 2012); hip or knee joint replacement or hip fracture surgery (Rahe‐Meyer 2014); open abdominal and urogenital surgery (Sabo 2011); and septoplasty (Tas 2015).

Four studies combined participants who underwent diverse elective surgical procedures (Blobner 2010; Cheong 2015; Lemmens 2010; Woo 2013). Twelve studies provided no data on the type of elective surgical procedure performed (Flockton 2008; Illman 2011; Isik 2016; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Koc 2015; Pongracz 2013; Riga 2014; Schaller 2010; Wu 2014; Yagan 2015).

Investigators maintained anaesthesia with opioid most often in combination with volatile anaesthetics, specifically with sevoflurane in 15 trials (Adamus 2011; Blobner 2010; Cheong 2015; Grintescu 2009; Jones 2008; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Lemmens 2010; Pongracz 2013; Riga 2014; Sabo 2011; Tas 2015; Woo 2013; Yagan 2015); desflurane in six trials (Carron 2013; Castro 2014; Gaszynski 2011; Hakimoglu 2016; Isik 2016; Koyuncu 2015); isoflurane in one trial (Mekawy 2012); and sevoflurane or desflurane in one trial (Illman 2011). Twelve trials used propofol for maintenance (Flockton 2008; Foletto 2014; Geldner 2012; Georgiou 2013; Kaufhold 2016; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Martini 2014; Schaller 2010; Wu 2014); and two trials used any anaesthetic, according to usual practice (Brueckmann 2015; Rahe‐Meyer 2014). Four trials provided no information on anaesthesia maintenance (Balaka 2011; Raziel 2013; Sherman 2014; Sustic 2012).

Most trials used rocuronium as a non‐depolarizing neuromuscular blocking‐agent (NMBA). However, Lemmens 2010 used vecuronium; Rahe‐Meyer 2014 used rocuronium or vecuronium, according to usual practice at the site; Flockton 2008 compared sugammadex following rocuronium versus neostigmine following cisatracurium; and Martini 2014 compared atracurium for induction and mivacurium for maintenance versus rocuronium for both induction and maintenance. Two studies provided no information on the NMBA agent used (Castro 2014; Sherman 2014).

Interventions

We summarized the interventions reported in included studies under Characteristics of included studies.

All studies compared sugammadex and neostigmine, but investigators administered these drugs in different doses: Adamus 2011 and Sustic 2012 compared sugammadex 2 mg/kg versus neostigmine 0.04 mg/kg; and 15 trials compared sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg (Blobner 2010; Castro 2014; Cheong 2015; Flockton 2008; Foletto 2014, Grintescu 2009, Illman 2011; Kvolik 2012a, Kvolik 2012b, Khuenl‐Brady 2010; Koc 2015; Tas 2015; Woo 2013; Wu 2014; Yagan 2015). Two trials compared sugammadex 2 mg/kg versus neostigmine 0.07 mg/kg (Kogler 2012; Koyuncu 2015).

Three studies compared sugammadex 2 mg/kg versus neostigmine 2.5 mg (Balaka 2011; Raziel 2013; Sherman 2014). Kizilay 2016 compared sugammadex 3 mg/kg versus neostigmine 0.03 mg/kg, Isik 2016 compared sugammadex 4 mg/kg versus neostigmine 0.04 mg/kg. Four trials compared sugammadex 4 mg/kg versus neostigmine 0.05 mg/kg (Geldner 2012; Hakimoglu 2016; Mekawy 2012; Sabo 2011). Three trials compared sugammadex 4 mg/kg versus neostigmine 0.07 mg/kg (Carron 2013; Jones 2008; Lemmens 2010). Rahe‐Meyer 2014 compared sugammadex 4 mg/kg versus usual care (neostigmine with glycopyrrolate or atropine, no dose specified, or placebo/spontaneous recovery). Martini 2014 compared sugammadex 4 mg/kg versus neostigmine 1 to 2 mg, and Riga 2014 did not specify dose for sugammadex or neostigmine. Four trials compared several different doses of sugammadex versus several different doses of neostigmine (Brueckmann 2015; Kaufhold 2016; Pongracz 2013; Schaller 2010). Georgiou 2013 compared sugammadex 2 mg/kg ideal body weight versus sugammadex 2 mg/kg corrected body weight versus neostigmine 50 µg/kg ideal body weight versus neostigmine 50 µg/kg corrected body weight, Carron 2013 compared sugammadex 4 mg/kg total body weight versus neostigmine 70 μg/kg lean body weight, and Gaszynski 2011 compared sugammadex 2 mg/kg corrected body weight versus neostigmine 50 µg/kg corrected body weight.

Outcomes

Of the 41 RCTs that met our inclusion criteria, 12 trials (n = 949) were eligible for meta‐analysis of the primary outcome (recovery time > TOFR 0.9) (Blobner 2010; Carron 2013; Cheong 2015; Foletto 2014; Gaszynski 2011; Georgiou 2013; Grintescu 2009; Illman 2011; Jones 2008; Koc 2015; Woo 2013; Wu 2014).

Of the 41 trials, 28 (N = 2298) were eligible for meta‐analysis of secondary outcomes (adverse events and serious adverse events): Adamus 2011; Balaka 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Castro 2014; Cheong 2015; Flockton 2008; Gaszynski 2011; Geldner 2012; Hakimoglu 2016; Illman 2011; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Lemmens 2010; Mekawy 2012; Pongracz 2013; Sabo 2011; Schaller 2010; Woo 2013; Wu 2014; Yagan 2015).

Ten RCTs (N = 1647) were ineligible for meta‐analysis (Isik 2016; Kvolik 2012a; Kvolik 2013; Martini 2014; Rahe‐Meyer 2014; Raziel 2013; Riga 2014; Sherman 2014; Sustic 2012; Tas 2015) for the reasons provided in Table 1 (table of studies ineligible for meta‐analysis).

Table 1. Table of studies ineligible for meta‐analysis

Study ID	Reasons for ineligibility	Comparisons	Conclusions
Isik 2016	Primary endpoint: acute effects of sugammadex and neostigmine on renal function	Sugammadex 4 mg /kg at reappearance of PTC 1 to 2 or T2 vs neostigmine 40 µg/kg + atropine 10 µg/kg at reappearance of T2	We believe that the use of more specific and sensitive new‐generation markers such as Cystatin C to evaluate kidney function will provide better understanding and interpretation of our results. Sugammadex has more tolerable effects on kidney function than does neostigmine. However, when compared with preoperative values, negative alteration of postoperative values can be seen. Neostigmine and sugammadex do not cause renal failure but may affect kidney function
Kvolik 2012a	TOFR recovery data available only as mean, no data on standard deviation, study author has not replied	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg	Recovery of cough reflexes was faster and respiration more efficient in patients receiving sugammadex. Safe extubation was determined by age, TOFR recovery, and effects of other anaesthetics
Kvolik 2013	TOFR recovery data available only as mean, no data on standard deviation, study author has not replied	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg + atropine 25 µg/kg	An increase in BIS Index registered after reversal of rocuronium effects was faster during the recovery period in patients who were given sugammadex as compared with neostigmine. Although rapid increase in BIS Indices was registered in sugammadex group, more sensitive measurements are needed to confirm clinical value of this observation
Martini 2014	Primary endpoint: influence of depth of the NMB on SRS (surgical rating score)	Neostigmine 1 to 2 mg + atropine 0.5 to 1 mg (for reversal of moderate NMB) vs sugammadex 4 mg/kg (for reversal of deep NMB)	Application of 5‐point SRS showed that deep NMB results in improved quality of surgical conditions compared with moderate block in retroperitoneal laparoscopy, without compromise to patients’ perioperative and postoperative cardiorespiratory conditions
Rahe‐Meyer 2014	Comparison: sugammadex 4 mg/kg vs usual care (neostigmine with glycopyrrolate or atropine, or placebo/spontaneous recovery). Study author has not replied with separate data on neostigmine with glycopyrrolate or atropine or placebo/spontaneous recovery.	Sugammadex 4 mg/kg vs usual care (neostigmine with glycopyrrolate or atropine, or placebo/spontaneous recovery)	Sugammadex produced limited, transient (< 1 hour) increases in activated partial thromboplastin time and prothrombin time but was not associated with increased risk of bleeding vs usual care
Raziel 2013	No useable data available for quantitative meta‐analysis on recovery time or risk of adverse events	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg + atropine 10 µg/kg	Sugammadex facilitates reversal of neuromuscular blockade after bariatric surgery, depending on the depth of neuromuscular blockade induced
Riga 2014	Primary outcome: cognitive function assessed by change in Mini‐Mental State Evaluation test (MMSE), Clock Drawing Test, and Isaacs Set Test, performed preoperatively, 1 hour postoperatively, and at discharge (1 to 15 days postoperatively)	Sugammadex vs neostigmine/atropine	No significant difference was observed regarding cognitive function after neostigmine/atropine combination or sugammadex was received for reversal of rocuronium‐induced neuromuscular blockade for elective surgery
Sherman 2014	Primary outcome: postoperative complications, data not available in useful format	Sugammadex 2 mg/kg vs neostigmine 2.5 mg/kg	Use of sugammadex (compared with neostigmine) as reversal agent following laparoscopic sleeve gastrectomy; surgery was associated with higher postoperative oxygen saturation despite lower TOF count before administration of reversal agent. Lack of differences in other measured variables may stem from the small size of patient groups studied
Sustic 2012	Outcome: gastric emptying evaluated by paracetamol absorption test	Sugammadex 2 mg/kg vs neostigmine 40 µg/kg + atropine group 15 µg/kg	Although study results show a tendency toward faster gastric emptying in sugammadex group, this difference is not significant in most, possibly owing to small sample size in this study
Tas 2015	Aim: to evaluate effects of sugammadex on postoperative nausea‐vomiting, pain, coagulation parameters, and quantity of postoperative bleeding. Data not available in useful format	Neostigmine 0.05 mg/kg + atropine 0.02 mg/kg vs sugammadex 2 mg/kg	Sugammadex was associated with greater postoperative bleeding than neostigmine in septoplasty patients. For surgical procedures with high risk of bleeding, the safety of sugammadex needs to be verified

Acronyms:

BIS ‐ Bispectral Index

MMSE ‐ Mini‐Mental State Examination

NMB ‐ neuromuscular blockade

T2 ‐ second twitch in train‐of‐four stimulation

TOFR ‐ train‐of‐four ratio

PTC ‐ post‐tetanic count

SRS ‐ surgical rating score

See Characteristics of included studies for further information on the included studies.

Excluded studies

Among 83 identified relevant trials, we excluded 19 publications (Aho 2012; Baysal 2013; Dahaba 2012; Gaona 2012; Ghoneim 2015; Harazim 2014; Kakinuma 2013; Kara 2014; Kzlay 2013; Nagy 2014; Ozgun 2014; Pecek 2013; Sacan 2007; Schepens 2015; Stourac 2016; Veiga Ruiz 2011; Nagashima 2016; Nemes 2016; NCT03111121).

We have explained reasons for exclusion of each trial in the Characteristics of excluded studies table.

Ongoing studies

We identified 20 ongoing and unpublished trials by searching www.controlled‐trials.com, clinicaltrials.gov, and www.centerwatch.com. The following five trials have been completed but to the best of our knowledge, no data from these trials have yet been published: NCT01539044; NCT01748643; NCT02160223; NCT02330172; NCT02414880). Six trials are currently recruiting participants (NCT02256280; NCT02361060; NCT02454504; NCT02666014; NCT02698969; NCT02860507). Six trials are classified as ongoing (NCT02909439; NCT02697929; NCT03108989; NCT03116997; NCT02939430; NCT03144453) and three trials are not yet open for recruiting participants (NCT02648503; NCT02845375; NCT02861131).

See Characteristics of ongoing studies for details.

Studies awaiting classification

We reran the search in May 2017 and found three trials (NCT02243943; Kim 2016; Sen 2016) that published data after we had completed our main search in May 2016; we will include these trials in the next updated version of this review.

Risk of bias in included studies

We assessed the risk of bias of included studies using the 'Risk of bias' tool developed by Cochrane. The first review author (AMH) and the second review author (PD) independently assessed risk of bias for each study and resolved disagreements by discussion or by consultation with the last review author (AA). We have presented the various bias domains in Figure 2 ‐ Risk of bias graph ‐ and Figure 6 ‐ Risk of bias summary

Figure 6

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Random sequence generation (selection bias)

Twenty‐seven trials (66%) reported adequate generation of random sequence that was computer‐based (Adamus 2011; Brueckmann 2015; Carron 2013; Hakimoglu 2016; Illman 2011; Isik 2016; Jones 2008; Kaufhold 2016; Martini 2014; Mekawy 2012; Pongracz 2013; Raziel 2013; Riga 2014; Schaller 2010; Sustic 2012; Tas 2015; Yagan 2015); or was performed by using a central randomization system (Blobner 2010; Flockton 2008; Geldner 2012; Khuenl‐Brady 2010; Koyuncu 2015; Lemmens 2010; Rahe‐Meyer 2014; Sabo 2011; Woo 2013; Wu 2014).

Furthermore, one trial (2%) reported randomization by lots (Kizilay 2016). Thirteen trials (32%) did not report sufficient information for assessment of risk of bias(Balaka 2011; Castro 2014; Cheong 2015; Foletto 2014; Gaszynski 2011; Georgiou 2013; Grintescu 2009; Koc 2015; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Sherman 2014).

Allocation concealment (selection bias)

Eighteen trials (44%) reported adequate allocation concealment performed by using sequentially numbered opaque sealed envelopes (SNORES) (Adamus 2011; Carron 2013; Isik 2016; Jones 2008; Martini 2014; Tas 2015; Yagan 2015); or secondary to a central randomization system (Blobner 2010; Flockton 2008; Geldner 2012; Khuenl‐Brady 2010; Koyuncu 2015; Lemmens 2010; Rahe‐Meyer 2014; Raziel 2013; Sabo 2011; Woo 2013; Wu 2014).

One trial (2%) reported using no allocation concealment (Kizilay 2016). Twenty‐two trials (54%) did not describe their method of allocation concealment (Balaka 2011; Brueckmann 2015; Castro 2014; Cheong 2015; Foletto 2014; Gaszynski 2011; Georgiou 2013; Grintescu 2009; Hakimoglu 2016; Illman 2011; Kaufhold 2016; Koc 2015; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Pongracz 2013; Riga 2014; Schaller 2010; Sherman 2014; Sustic 2012).

Blinding

Blinding of participants (performance bias)

Fourteen trials (34%) adequately blinded participants and therefore had low risk of performance bias (Adamus 2011; Brueckmann 2015; Geldner 2012; Georgiou 2013; Illman 2011; Kizilay 2016; Martini 2014; Pongracz 2013; Rahe‐Meyer 2014; Raziel 2013; Riga 2014; Schaller 2010; Woo 2013; Wu 2014).

Eight trials (20%) did not adequately blind participants and therefore had high risk of performance bias; two of these specifically reported that participants were not blinded (Sustic 2012; Yagan 2015), and six were marked as “open‐label” trials (Blobner 2010; Flockton 2008; Grintescu 2009; Jones 2008; Khuenl‐Brady 2010; Lemmens 2010).

The remaining 19 trials (46%) did not provide sufficient data on participant blinding and we assigned risk of performance bias as unclear(Balaka 2011; Carron 2013; Castro 2014; Cheong 2015; Foletto 2014; Gaszynski 2011; Hakimoglu 2016; Isik 2016; Kaufhold 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Sabo 2011; Sherman 2014; Tas 2015).

Blinding of personnel (performance bias)

Seven trials (17%) reported adequate blinding of the anaesthesiologist and therefore had low risk of performance bias (Cheong 2015; Illman 2011; Kaufhold 2016; Mekawy 2012; Pongracz 2013; Rahe‐Meyer 2014; Schaller 2010).

Seventeen trials (41%) did not report adequate blinding of anaesthesiologists and therefore had high risk of performance bias; 11 of these specifically reported that the anaesthesiologist was not blinded: (Adamus 2011; Brueckmann 2015; Kizilay 2016; Martini 2014; Raziel 2013; Riga 2014; Sabo 2011; Sustic 2012; Woo 2013; Wu 2014; Yagan 2015), and six trials were marked as “open‐label” trials (Blobner 2010; Flockton 2008; Grintescu 2009; Jones 2008; Khuenl‐Brady 2010; Lemmens 2010).

The remaining 17 trials (41%) did not provide sufficient data on anaesthesiologist blinding and therefore had unclear risk of performance bias (Balaka 2011; Carron 2013; Castro 2014; Foletto 2014; Gaszynski 2011; Geldner 2012; Georgiou 2013; Hakimoglu 2016; Isik 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Sherman 2014; Tas 2015).

Blinding of TOF‐watch assessment (detection bias)

Two trials (5%) specifically reported that the anaesthesiologist was also the TOF‐watch assessor: (Adamus 2011; Illman 2011). Four trials (10%) reported adequate blinding of the TOF‐watch assessor and therefore had low risk of performance bias (Brueckmann 2015; Illman 2011; Martini 2014; Schaller 2010).

Twelve trials (29%) did not provide adequate blinding of the TOF‐watch assessor and therefore had high risk of detection bias; six of these trials specifically reported that the anaesthesiologist was not blinded (Adamus 2011; Kizilay 2016; Raziel 2013; Woo 2013; Wu 2014; Yagan 2015), and six trials were marked as “open‐label” trials (Blobner 2010; Flockton 2008; Grintescu 2009; Jones 2008; Khuenl‐Brady 2010; Lemmens 2010).

For two trials (5%), risk of bias assessment was of no relevance, as trial authors presented no TOF‐watch data (Rahe‐Meyer 2014; Sustic 2012).

The remaining 23 trials (56%) did not provide sufficient data on TOF‐watch assessor blinding and had unclear risk of detection bias (Balaka 2011; Carron 2013; Castro 2014; Cheong 2015; Foletto 2014; Gaszynski 2011; Geldner 2012; Georgiou 2013; Hakimoglu 2016; Isik 2016; Kaufhold 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Pongracz 2013; Riga 2014; Sabo 2011; Sherman 2014; Tas 2015).

Blinding of safety assessment (detection bias)

Twenty trials (49%) reported adequate blinding of the safety assessor and therefore had low risk of detection bias (Blobner 2010; Brueckmann 2015; Carron 2013; Flockton 2008; Geldner 2012; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Lemmens 2010; Martini 2014; Rahe‐Meyer 2014; Raziel 2013; Riga 2014; Sabo 2011; Schaller 2010; Sustic 2012; Tas 2015; Woo 2013; Wu 2014; Yagan 2015).

Two trials (5%) did not adequately blind the safety assessor and therefore had high risk of detection bias; one of these specifically reported that the safety assessor was not blinded (Kizilay 2016), and the other trial was marked as an “open‐label” study (Grintescu 2009).

The remaining 19 trials (46%) did not provide sufficient data on safety assessor blinding and had unclear risk of detection bias (Adamus 2011; Balaka 2011; Castro 2014; Cheong 2015; Foletto 2014; Gaszynski 2011; Georgiou 2013; Hakimoglu 2016; Illman 2011; Isik 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Pongracz 2013; Sherman 2014).

Incomplete outcome data

The following 28 trials (68%) had low risk of attrition bias as either all participants were accounted for, or missing outcome data were properly balanced among groups: Adamus 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Castro 2014; Cheong 2015; Flockton 2008; Gaszynski 2011; Geldner 2012; Hakimoglu 2016; Illman 2011; Isik 2016; Jones 2008; Kaufhold 2016; Kizilay 2016; Koc 2015; Koyuncu 2015; Martini 2014; Mekawy 2012; Pongracz 2013; Rahe‐Meyer 2014; Raziel 2013; Riga 2014; Sabo 2011; Tas 2015; Woo 2013; Wu 2014; Yagan 2015.

For three trials (7%), missing outcome data were not balanced across intervention groups (Khuenl‐Brady 2010; Lemmens 2010; Schaller 2010); these studies therefore had high risk of attrition bias.

The remaining 10 trials (24%) did not provide sufficient data on incomplete outcomes and had unclear risk of attrition bias (Balaka 2011; Foletto 2014; Georgiou 2013; Grintescu 2009; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Sherman 2014; Sustic 2012).

Selective reporting

Twenty trials (49%) had low risk of reporting bias, as they were registered online: 16 on clinicaltrials.gov (Blobner 2010 – NCT00451217; Brueckmann 2015 – NCT01479764; Flockton 2008 ‐ NTC00451100; Geldner 2012 – NCT00724932; Georgiou 2013 ‐ NCT01629394; Jones 2008 ‐ NCT00473694; Khuenl‐Brady 2010 – NCT00451217; Lemmens 2010 – NCT00473694; Martini 2014 – NCT 01631149; Rahe‐Meyer 2014 – NCT01422304; Raziel 2013 – NCT01631396; Riga 2014 – NCT02419352; Schaller 2010 – NCT00895609; Woo 2013 – NCT01050543; Wu 2014 – NCT00825812; Yagan 2015 – NCT02215382); one on SYNABA – The Polish Clinical Trials authorization (Gaszynski 2011 – 252922); one on ANZCTR ‐ Australian New Zealand Clinical Trials Registry (Hakimoglu 2016 ‐ ACTRN12614000651684); and finally two on Eudra‐CT (Illman 2011 ‐ 2009‐013537‐22; Pongracz 2013 ‐ 2011‐001683‐22).

The remaining 20 trials (49%) were not registered online, but it is clear that the published article or meeting abstract includes all expected outcomes (Adamus 2011; Balaka 2011; Carron 2013; Castro 2014; Cheong 2015; Grintescu 2009; Isik 2016; Kaufhold 2016; Kizilay 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Sabo 2011; Sherman 2014; Sustic 2012; Tas 2015). Therefore, these trials had low risk of reporting bias.

One trial (2%) did not provide sufficient information for assessment of risk of bias and had unclear risk of reporting bias (Foletto 2014).

Other potential sources of bias

Funding bias

Merck, Sharp and Dohme or Schering‐Plough provided financial support for 11 trials (27%), indicating high risk of funding bias (Blobner 2010; Geldner 2012; Illman 2011; Jones 2008; Khuenl‐Brady 2010; Lemmens 2010; Martini 2014; Rahe‐Meyer 2014; Sabo 2011; Woo 2013; Wu 2014). Authors of the following trials were former employees, current employees, or members of advisory boards of Merck, Sharp and Dohme/Schering‐Plough, or had received honoraria for lectures, consultancy, or advisory board membership, or travel grants from Merck, Sharp and Dohme/Schering‐Plough: Adamus 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Flockton 2008; Gaszynski 2011; Geldner 2012; Illman 2011; Kaufhold 2016; Khuenl‐Brady 2010; Koyuncu 2015; Lemmens 2010; Martini 2014; Rahe‐Meyer 2014,Schaller 2010; Woo 2013; Wu 2014). These studies had high risk of funding bias.

We could not assess funding risk of bias for the following 14 trials (34%) owing to insufficient information: Balaka 2011; Castro 2014; Foletto 2014; Grintescu 2009; Hakimoglu 2016; Koc 2015; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Pongracz 2013; Sherman 2014; Sustic 2012; these studies had unclear risk of funding bias.

Eight trials (20%) had low risk of funding bias, as they were funded by departmental sources (Georgiou 2013; Isik 2016; Kaufhold 2016; Koyuncu 2015; Raziel 2013; Riga 2014; Schaller 2010; Tas 2015). Trial authors funded two trials (5%) (Kizilay 2016; Yagan 2015), and in two cases (5%), study authors received research grants (Gaszynski 2011; Polish Government grant; and Cheong 2015; Inje University research grant).

Other bias

Twenty‐one trials (51%) had low risk of other bias, as they reported specific information on sample size calculation (Adamus 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Cheong 2015; Flockton 2008; Geldner 2012; Hakimoglu 2016; Illman 2011; Isik 2016; Jones 2008; Kaufhold 2016; Koyuncu 2015; Lemmens 2010; Martini 2014; Pongracz 2013; Rahe‐Meyer 2014; Sabo 2011; Woo 2013; Wu 2014; Yagan 2015).

Of these 21 trials, 12 (29%) were powered to address this review’s primary outcome (Adamus 2011; Blobner 2010; Carron 2013; Cheong 2015; Flockton 2008; Illman 2011; Jones 2008; Lemmens 2010; Pongracz 2013; Sabo 2011; Woo 2013; Wu 2014), and seven trials (17%) were powered to address this review’s secondary outcome (Brueckmann 2015; Geldner 2012; Hakimoglu 2016; Isik 2016; Koyuncu 2015; Rahe‐Meyer 2014; Yagan 2015). Twenty trials (49%) did not provide information on sample size calculation (Balaka 2011; Castro 2014; Foletto 2014; Gaszynski 2011; Georgiou 2013; Grintescu 2009; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Kogler 2012; Kvolik 2012a; Kvolik 2012b; Kvolik 2013; Mekawy 2012; Raziel 2013; Riga 2014; Schaller 2010; Sherman 2014; Sustic 2012; Tas 2015).

Treatment groups were generally comparable with respect to baseline characteristics, except Cheong 2015, which described significant differences in body weight between groups that might have influenced the dosage of administered drugs; and Flockton 2008, which reported a higher proportion of women, higher mean age, and a higher percentage of ASA II to III participants in the sugammadex group. Furthemore, Lemmens 2010 discontinued one intervention group owing to a marked difference in efficacy between groups after interim analysis. Therefore, these trials had high risk of other bias.

All trials used the same method (acceleromyography) and at the same monitor site (ulnar nerve, adductor pollicis muscle). We analysed quality variables of neuromuscular recording methods among full‐text trials have provided a summary in Table 2 ‐ Quality variables of neuromuscular monitoring methods among included trials.

See: Summary of findings for the main comparison Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg; Summary of findings 2 Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg; Summary of findings 3 Sugammadex (any dose) vs neostigmine (any dose)

Table 2. Quality variables of neuromuscular monitoring methods among included trials

Study ID	Method of recording	Monitor site	Arm fixation	Supramaximal stimulation	Temperature maintained and recorded	Initial signal stabilization	Twich height calibration	Preload used
Adamus 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Yes	Not mentioned
Blobner 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Yes	Not mentioned
Brueckmann 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Yes	Not mentioned
Carron 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Not mentioned	Not mentioned	Yes	Yes	No
Castro 2014	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Cheong 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Not mentioned	Not mentioned	Not mentioned
Flockton 2008	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Gaszynski 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Not mentioned	Not mentioned
Geldner 2012	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Yes	Yes	Not mentioned
Hakimoglu 2016	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Illman 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes		Yes	No
Isik 2016	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Not mentioned	Yes	Not mentioned	Not mentioned	Not mentioned
Jones 2008	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Kaufhold 2016	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Not mentioned
Khuenl‐Brady 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Yes	Yes	Not mentioned
Kizilay 2016	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Koc 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Not mentioned	Yes	Not mentioned
Koyuncu 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	No	Yes	No	No	Yes	Not mentioned
Lemmens 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Martini 2014	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Not mentioned	Yes	Yes	Yes
Mekawy 2012	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Yes	Not mentioned
Pongracz 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Yes
Sabo 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Schaller 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Not mentioned
Tas 2015	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Woo 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes *	No *	Yes *	Yes	Yes	No *
Wu 2014	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Not mentioned	Yes	Yes	Not mentioned
Yagan 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned

Studies with only abstracts were not included in this table because they did not document information regarding neuromuscular monitoring

List of abbreviations:

N. ulnaris ‐ ulnar nerve

M. adductor pollicis ‐ adductor pollicis muscle

Effects of interventions

See summary of findings Table for the main comparison; summary of findings Table 2; and summary of findings Table 3.

Comparison 1. Sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg for rocuronium reversal

1.1 Primary outcome 1: recovery time from T2 to TOFR > 0.9

Ten trials were included in this category (Blobner 2010; Cheong 2015; Foletto 2014; Gaszynski 2011; Georgiou 2013; Grintescu 2009; Illman 2011; Koc 2015; Woo 2013; Wu 2014).

All trials used rocuronium for intubation and maintenance. The intubating dose of rocuronium was 0.6 mg/kg in five trials (Blobner 2010; Cheong 2015; Koc 2015; Woo 2013; Wu 2014), 0.6 to 1 mg/kg in Illman 2011, and 1 mg/kg in Gaszynski 2011. The maintenance dose of rocuronium was 0.1 to 0.2 mg/kg in four trials (Blobner 2010; Koc 2015; Woo 2013; Wu 2014), 0.06 mg/kg corrected body weight (CBW) with maximum two additional doses in Gaszynski 2011, and 5 to 10 mg in two trials (Cheong 2015; Illman 2011). No information on rocuronium dosage was available for three trials (Foletto 2014; Georgiou 2013; Grintescu 2009).

Meta‐analysis of results showed that sugammadex 2 mg/kg reversed neuromuscular blockade from T2 to TOFR > 0.9 in 1.96 minutes, and neostigmine 0.05 mg/kg reversed neuromuscular blockade from T2 to TOFR > 0.9 in 12.87 minutes. Therefore, sugammadex 2 mg/kg was on average 10.22 minutes (6.6 times) faster than neostigmine 0.05 mg/kg in reversing neuromuscular blockade at T2 reappearance (MD 10.22 minutes, 95% CI 8.48 to 11.96; I² = 84%; 10 studies; n = 835; random‐effects model; Analysis 1.1; GRADE quality of evidence: moderate; summary of findings Table for the main comparison). We downgraded the GRADE quality of evidence by one owing to high risk of bias.

The following trials used NMBAs other than rocuronium and therefore were not included in the meta‐analysis.

Flockton 2008 compared rocuronium‐sugammadex 2 mg/kg versus cisatracurium‐neostigmine 0.05 mg/kg and found that reversal with sugammadex was 4.7 times faster than with neostigmine (geometric mean recovery time of 1.9 vs 9.0; P < 0.0001).

Khuenl‐Brady 2010 investigated the effect of sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg in reversing vecuronium‐induced neuromuscular blockade (induction 0.1 mg/kg, maintenance 0.03 to 0.03 mg/kg) and described that the geometric mean time of recovery to TOFR > 0.9 was significantly faster with sugammadex than with neostigmine (2.7 minutes, 95% CI 2.2 to 3‐3 vs 17.9, 95% CI 13.1 to 24.3, respectively; P < 0.0001; n = 93).

Other trials did not provide enough information or compared doses of sugammadex and neostigmine other than those previously mentioned and as such could not be included in the meta‐analysis: Kvolik 2012a compared sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg and reported T2 to TOFR > 0.9 recovery time of 2.5 minutes versus 8.5 minutes, respectively (P = 0.045, n = 38), but these data could not be included in the meta‐analysis, as standard deviation (SD) data were not reported in the paper and could not be obtained. Mekawy 2012 examined recovery time from T2 to TOFR > 0.9 comparing sugammadex 4 mg/kg (n = 20) versus neostigmine 0.05 mg/kg plus atropine 0.02 mg/kg (n = 20) and reported that mean reversal time (SD) was 2.47 (0.51) versus 24.21 (4.7) minutes, respectively.

Subgroup analysis

1.2 TIVA versus volatile anaesthetics

Seven trials maintained anaesthesia with volatile anaesthetic (Blobner 2010; Cheong 2015; Gaszynski 2011; Grintescu 2009; Illman 2011; Koc 2015; Woo 2013), and three trials used TIVA for maintenance (Foletto 2014; Georgiou 2013; Wu 2014). Subgroup analysis of results showed no significant subgroup differences in recovery time to TOFR > 0.9 (Analysis 1.2).

Sensitivity analysis

1.3. Excluding meeting abstracts

Sensitivity analysis that excluded data from meeting abstracts (MD 9.27 minutes, 95% CI 7.40 to 11.14; I² = 82%; n = 767; random‐effects model; Analysis 1.3) did not change overall results regarding significance.

Primary outcome 2: recovery time from PTC 1 to 5 to TOFR > 0.9

This outcome is not clinically relevant as dosages of sugammadex 2 mg/kg and neostigmine 0.05 mg/kg are too low to reverse the deep rocuronium‐induced neuromuscular blockade seen at PTC 1 to 5.

Secondary outcomes: risk of adverse events and risk of serious adverse events

We have described these outcomes in detail under Comparison 3 (Analysis 3.2).

Comparison 2. Sugammadex 4 mg/kg versus neostigmine 0.07 mg/kg for rocuronium reversal

Primary outcome 1. Recovery time from T2 to TOFR > 0.9

This outcome is not clinically relevant as dosages of sugammadex 4 mg/kg and neostigmine 0.07 mg/kg are too high to reverse the moderate rocuronium‐induced neuromuscular blockade seen at T2.

2.1 Primary outcome 2: recovery time from PTC 1 to 5 to TOFR > 0.9

We combined two trials in this category (Carron 2013; Jones 2008). Both trials used rocuronium 0.6 mg/kg as a single intubating dose and rocuronium 0.15 mg/kg for maintenance. Carron 2013 combined neostigmine with atropine 0.01 mg/kg, and Jones 2008 combined neostigmine with glycopyrrolate 0.014 mg/kg. Carron 2013 administered sugammadex or neostigmine at reappearance of PTC 1 to 5, and Jones 2008 at reappearance of PTC 1 to 2. Carron 2013 included morbidly obese female participants. Carron 2013 maintained anaesthesia with desflurane, and Jones 2008 with sevoflurane.

Meta‐analysis of trial results showed that sugammadex 4 mg/kg reversed neuromuscular blockade from PTC 1 to 5 to TOFR > 0.9 in 2.9 minutes, and neostigmine 0.07 mg/kg reversed neuromuscular blockade from PTC 1 to 5 to TOFR > 0.9 in 48.8 minutes. Sugammadex 4 mg/kg was therefore on average 45.78 minutes (16.8 times) faster than neostigmine 0.07 mg/kg in reversing neuromuscular blockade at reappearance of PTC 1 to 5 (MD 45.78 minutes, 95% CI 39.41 to 52.15; I² = 0%; two studies; n = 114; random‐effects model; Analysis 2.1; GRADE quality of evidence: low; summary of findings Table 2). We downgraded GRADE quality of evidence two levels owing to high risk of bias and imprecision.

The following trials used NMBAs other than rocuronium, gave a dose of neostigmine different from the one described above, or had missing SD values and were not included in the meta‐analysis. Lemmens 2010 investigated the effect of sugammadex 4 mg/kg versus neostigmine 0.07 mg/kg in reversing vecuronium‐induced neuromuscular blockade (induction 0.1 mg/kg, maintenance 0.015 mg/kg) and described that the geometric mean time of recovery to TOFR > 0.9 was 15‐fold faster with sugammadex than with neostigmine (4.5 vs 66.2 minutes, respectively; P < 0.0001; n = 83). Geldner 2012 reported that participants receiving sugammadex 4 mg/kg administered at PTC 1 to 2 recovered 3.4 times faster than those given neostigmine 0.05 mg/kg plus atropine 0.01 mg/kg (geometric mean recovery time of 2.4 (2.1 to 2.7) vs 8.4 (7.2 to 9.8) minutes, respectively; P < 0.0001). Kogler 2012, reported that median recovery time from PTC 1 to 2 to TOFR > 0.9 after sugammadex 2 mg/kg was 1.1 minutes versus 10.13 minutes for neostigmine 0.07 mg/kg (P < 0.001; n = 31; no SD value reported).

Secondary outcomes: risk of adverse events and risk of serious adverse events

We have described these outcomes in detail under Comparison 3 (Analysis 3.2).

Other recovery times

Some trials measured recovery times other than those described in the comparisons above. Only single trials measured these data; therefore, we could not include them in the meta‐analysis, but we can describe the qualitative data as follows.

Balaka 2011 reported mean recovery time from TOFR of 50% to > 90% as 9.7 minutes after administration of neostigmine 2.5 mg and 2.8 minutes after administration of sugammadex 4 mg/kg (P < 0.05; n = 40). Yagan 2015 compared sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg administered at T4/T1 20% and found that extubation time (defined as time to TOFR > 0.9) was seven minutes in the neostigmine group and two minutes in the sugammadex group (P > 0.05; n = 36). Martini 2014 compared moderate NMB (T1 to 2) induced by atracurium/mivacurium reversed by neostigmine 1 to 2 mg plus atropine 0.5 to 1 mg (n = 12) versus deep NMB (PTC 1 to 2) induced by high‐dose rocuronium and reversed by sugammadex 4 mg/kg (n = 12). Recovery times to TOFR > 0.9 expressed as mean (SD) were, respectively, 10.9 (4.9) versus 5.1 (2.4) (P < 0.01). Pongracz 2013 investigated adequate doses for reversal of reappearance of four twitches of TOF and discovered that sugammadex 1 mg/kg, unlike neostigmine, rapidly and effectively reverses rocuronium‐induced block that has recovered spontaneously to threshold TOF count four. Furthermore, sugammadex 0.5 mg/kg reverses a similar block within eight minutes. Sabo 2011 compared sugammadex 4.0 mg/kg versus neostigmine 0.05 mg/kg plus glycopyrrolate 0.01 mg/kg administered when the TOF‐blinded anaesthesiologist considered the patient ready for reversal of NMB. The anaesthesiologist could ask the TOF‐watch operator whether the patient had recovered to at least 1 to 2 PTC before administering the reversal agent. This trial demonstrated significantly faster recovery to TOFR > 0.9 ratio within two minutes (95% CI 1.8 to 2.5) in the sugammadex group versus eight minutes (95% CI 3.8 to 16.5 minutes) in the neostigmine group. Schaller 2010 investigated the efficacy of sugammadex (0.0625, 0.125, 0.25, 0.5, or 1.0 mg/kg), neostigmine (5, 8, 15, 25, or 40 µg/kg), and saline, and by using a bi‐exponential model and regression analysis concluded that sugammadex 0.22 mg/kg and neostigmine 34 µg/kg effectively and comparably reverse a rocuronium‐induced shallow residual neuromuscular block at TOFR = 0.5 (n = 99). Kaufhold 2016 investigated several different doses of sugammadex or neostigmine as well as placebo administered at TOFR ≥ 0.2 and found that residual neuromuscular block of TOFR = 0.2 cannot be reversed reliably with neostigmine within 10 minutes. However, substantially lower doses of sugammadex than the approved dose of 2.0 mg/kg may be sufficient to reverse residual rocuronium‐induced neuromuscular block at recovery of TOFR ≥ 0.2. Koyuncu 2015 looked at the effects of sugammadex 2 mg/kg (n = 50) versus neostigmine 70 µg/kg + atropine 0.4 mg per 1 mg neostigmine administered when four twitches of TOF were visible with fade and found that sugammadex speeds recovery of neuromuscular strength but only slightly (P > 0.01; n = 100).

Comparison 3. Sugammadex (any dose) versus neostigmine (any dose)

Primary outcome 1: recovery time from T2 to TOFR > 0.9

This outcome was not clinically relevant as doses for sugammadex and neostigmine used are specific to the depth of the neuromuscular blockade.

Primary outcome 2: recovery time from PTC 1 to 5 to TOFR > 0.9

This outcome was not clinically relevant as doses for sugammadex and neostigmine used are specific to the depth of the neuromuscular blockade.

3.1. Secondary outcomes: risks of adverse events and serious adverse events

The following 28 trials investigated adverse events possibly, probably, or definitely related to study drug: Adamus 2011; Balaka 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Castro 2014; Cheong 2015; Flockton 2008; Gaszynski 2011; Geldner 2012; Hakimoglu 2016; Illman 2011; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Kogler 2012; Koyuncu 2015; Kvolik 2012a; Lemmens 2010; Mekawy 2012; Pongracz 2013; Sabo 2011; Schaller 2010; Woo 2013; Wu 2014; Yagan 2015.

Meta‐analysis of trial results showed significantly fewer adverse events in the sugammadex group than in the neostigmine group (RR 0.60, 95% CI 0.49 to 0.74; I² = 40%; 28 studies, n = 2298; random‐effects model; Analysis 3.1; GRADE quality of data: moderate; summary of findings Table 3; quality of evidence downgraded one level owing to high risk of bias). Specifically, the risk of composite adverse events was 283/1000 in the neostigmine group and 159/1000 in the sugammadex group. With number needed to treat for an additional beneficial outcome (NNTB) of eight to avoid an adverse event, sugammadex appears to have a stronger safety profile than neostigmine. Furthermore, data show significantly fewer participants with one or more adverse events (RR 0.62, 95% CI 0.48 to 0.81; I² = 0%; n = 1766; random‐effects model; Analysis 3.5; GRADE quality of data: moderate; summary of findings Table 3) in the sugammadex group than in the neostigmine group.

Data on specific adverse events show significantly less risk of the following adverse events in the sugammadex group than in the neostigmine group: bradycardia (RR 0.16, 95% CI 0.07 to 0.34; I² = 0%; n = 1218; random‐effects model; Analysis 3.6; NNTB 14; GRADE quality of data: moderate; summary of findings Table 3; downgraded one level owing to high risk of bias), PONV (RR 0.52, 95% CI 0.28 to 0.97; I² = 0%; n = 389; random‐effects model; Analysis 3.7; NNTB 16; GRADE quality of data: low; summary of findings Table 3; downgraded two levels owing to high risk of bias and imprecision), desaturation (RR 0.23, 95% CI 0.06 to 0.83; I² = 0%; n = 134; random‐effects model; Analysis 3.8), need for transitory oxygen supplementation (RR 0.24, 95% CI 0.09 to 0.66; I² = 0%; n = 76; random‐effects model; Analysis 3.10), and procedural complications (RR 0.12, 95% CI 0.02 to 0.97; n = 168; I² = 0%; random‐effects model; Analysis 3.9). Also, significantly fewer participants were unable to perform 5 seconds of sustained head‐lift at extubation (RR 0.34, 95% CI 0.15 to 0.78; I² = 0%; n = 395; random‐effects model; Analysis 3.11) in the sugammadex group than in the neostigmine group.

Data show no significant differences between sugammadex and neostigmine with regard to nausea (RR 0.83, 95% CI 0.44 to 1.56; I² = 0%; n = 719; Analysis 3.13), vomiting (RR 2.05, 95% CI 0.50 to 8.48; I² = 0%; n = 297; Analysis 3.14), postprocedural nausea (RR 1.39, 95% CI 0.27 to 7.12; I² = 0%; n = 168; Analysis 3.15), headache (RR 1.02, 95% CI 0.48 to 2.18; I² = 0%; n = 388; Analysis 3.16), hypertension (RR 1.45, 95% CI 0.23 to 9.05; I² = 0%; n = 287; Analysis 3.17), hypotension (RR 1.23, 95% CI 0.38 to 3.96; I² = 0%; n = 465; Analysis 3.18), cough (RR 1.42, 95% CI 0.42 to 4.81; I² = 65%; n = 200; Analysis 3.19), dry mouth (RR 0.44, 95% CI 0.10 to 1.87; I² = 17%; n = 289; Analysis 3.20), dizziness (RR 0.98, 95% CI 0.10 to 9.23; I² = 0%; n = 168; Analysis 3.21), tachycardia (RR 0.44, 95% CI 0.09 to 2.22; I² = 0%; n = 338; Analysis 3.22), pruritus (RR 1.62, 95% CI 0.20 to 12.88; I² = 0%; n = 175; Analysis 3.23), pyrexia (RR 1.43, 95% CI 0.23 to 8.91; I² = 0%; n = 264; Analysis 3.24), shivering (RR 0.75, 95% CI 0.40 to 1.43; I² = 0%; n = 190; Analysis 3.25), chills (RR 4.04, 95% CI 0.46 to 35.85; I² = 0%; n = 166; Analysis 3.26), rash (RR 0.83, 95% CI 0.17 to 3.96; I² = 0%; n = 701; Analysis 3.27), supraventricular extrasystoles (RR 0.32, 95% CI 0.03 to 3.05; I² = 0%; n = 189; Analysis 3.28), laryngospasm (RR 0.34, 95% CI 0.07 to 1.65; I² = 0%; n = 100; Analysis 3.29), increased upper airway secretion (RR 0.37, 95% CI 0.09 to 1.59; I² = 0%; n = 442; Analysis 3.30), procedural complications (RR 0.12, 95% CI 0.02 to 0.97; I² = 0%; n = 168; Analysis 3.9), procedural hypertension (RR 1.65, 95% CI 0.33 to 8.21; I² = 0%; n = 267; Analysis 3.31), procedural hypotension (RR 0.49, 95% CI 0.02 to 14.15; I² = 60%; n = 391; Analysis 3.32), abdominal pain (RR 0.98, 95% CI 0.10 to 9.27; I² = 0%; n = 196; Analysis 3.33). Furthermore, data show no significant differences in reported clinical signs of residual NMB (RR 1.0; n = 646; Analysis 3.34), inadequate reversal of NMB (RR 0.11, 95% CI 0.01 to 2.02; n = 368; Analysis 3.35), and recurrence of NMB (RR 0.74, 95% CI 0.05 to 10.74; I² = 33; n = 1289; Analysis 3.36). Clinical tests revealed no significant differences in the number of participants reporting general muscle weakness at extubation (RR 0.61, 95% CI 0.31 to 1.18; I² = 0%; n = 288; Analysis 3.12), at PACU discharge (RR 0.49, 95% CI 0.12 to1.90; I² = 0%; n = 410; Analysis 3.37), or in the number of participants unable to perform five seconds of sustained head‐lift at PACU discharge (RR 1.0; n = 399; Analysis 3.38).

A single trial observed some drug‐related adverse events; therefore, we could not include them in a meta‐analysis of specific adverse events, but we used the data to calculate overall risk of adverse events. The following isolated adverse events were observed in the sugammadex group: three cases of breath‐hold (10%) in Hakimoglu 2016, two cases of strange taste in the mouth (6%) in Gaszynski 2011, two cases of increased beta‐N‐acetyl‐D‐glucosaminidase (6%) in Flockton 2008, two cases of bronchospasm (4%) in Koyuncu 2015, and one case of each of the following: severe abdominal pain (2%), pharyngolaryngeal pain (2%), diarrhoea (2%), and tinnitus (2%) in Blobner 2010; decreased hematocrit (1%) and procedural haemorrhage (1%) in Brueckmann 2015; tremor (3%) and altered facial sensation (3%) in Flockton 2008; postprocedural hypertension (3%), paraesthesia (3%), and increased blood creatinine phosphokinase (3%) in Jones 2008; retching (2%), airway complication to anaesthesia (2%), and hot flush (2%) in Khuenl‐Brady 2010; procedural pain (2%) in Sabo 2011; leukocytosis (2%) in Lemmens 2010; mild hypoventilation (1%) in Wu 2014; and finally one case of intraoperative movement (2%) in Schaller 2010.

In the neostigmine group, the following isolated drug‐related adverse events were reported: four cases of breath‐hold (13%) in Hakimoglu 2016; two cases of albumin present in the urine (4%) in Blobner 2010; two cases of leukocytosis (5%) in Lemmens 2010; and one case of each of the following: involuntary muscle contractions (2%), visual accommodation disorder (2%), increased urine beta‐2 microglobulin (2%), severe bradycardia (2%), and productive cough (2%) in Blobner 2010; respiratory distress (1%) and delayed recovery from anaesthesia (1%) in Brueckmann 2015; hyperhidrosis (3%), decreased blood protein (3%), restlessness (3%), chest discomfort (3%), incision site complication (3%), and postprocedural complication (3%) in Jones 2008; ventricular extrasystoles (2%), sleep disorder (2%), and increased gamma‐glutamyltransferase (2%) in Khuenl‐Brady 2010; anxiety (3%), depression (3%), and fatigue (3%) in Lemmens 2010; dyspepsia (2%) and somnolence (2%) in Sabo 2011; severe muscle weakness (1%) in Wu 2014; and finally one case of intraoperative movement (2%) in Schaller 2010.

We have described in Table 3 each observed adverse event possibly, probably, or definitely related to sugammadex or neostigmine. This table also presents risk of adverse events in descending order, as well as the number of studies observing each adverse event.

Table 3. Table of adverse events

	Sugammadex			Neostigmine
Specific adverse events	Number of AEs	Number of participants	Risk of AEs, %	Number of AEs	Number of participants	Risk of AEs, %	RR (95% CI)	Number of studies	Total number of participants
Cough	20	100	20,0	14	100	14,0	1.42 (0.42‐4.81)	3	200
Shivering	13	91	14,3	19	99	19,2	0.75 (0.40‐1.43)	3	190
Desaturation	2	63	3,2	11	71	15,5	0.23 (0.06‐0.83)	2	134
General muscle weakness after extubation	13	142	9,2	22	146	15,1	0.61 (0.31‐1.18)	4	288
Breath‐hold	3	30	10,0	4	30	13,3	‐	1	60
PONV	13	206	6,3	24	183	13,1	0.52 (0.28‐0.97)	6	389
Laryngospasm	2	50	4,0	6	50	12,0	0.34 (0.07‐1.65)	2	100
Not able to perform 5 second head‐lift after extubation	7	193	3,6	23	202	11,4	0.34 (0.15‐0.78)	6	395
Bradycardia	4	621	0,6	50	597	8,4	0.16 (0.07‐0.34)	11	1218
Procedural complications	0	85	0,0	7	83	8,4	0.12 (0.02‐0.97)	2	168
Postprocedural nausea	8	128	6,3	5	122	4,1	1.34 (0.47‐3.81)	3	250
Dry mouth	3	146	2,1	9	143	6,3	0.44 (0.10‐1.87)	3	289
Headache	12	195	6,2	11	193	5,7	1.02 (0.48‐2.18)	4	388
Increased beta‐N‐acetyl‐D‐glucosaminidase	2	34	5,9	0	39	0,0	‐	1	73
Strange taste in mouth	2	35	5,7	0	35	0,0	‐	1	70
Nausea	17	364	4,7	20	355	5,6	0.83 (0.44‐1.56)	9	719
Leukocytosis	1	46	2,2	2	36	5,6	‐	1	82
Albumin present in the urine	0	48	0,0	2	48	4,2	‐	1	96
Vomiting	6	149	4,0	2	148	1,4	2.05 (0.50‐8.48)	4	297
Bronchospasm	2	50	4,0	1	50	2,0	‐	1	100
Chills	3	82	3,7	0	84	0,0	4.04 (0.46‐35.85)	2	166
General muscle weakness at PACU discharge	3	208	1,4	6	202	3,0	0.49 (0.12‐1.90)	5	410
Procedural hypertension	4	133	3,0	2	134	1,5	1.65 (0.33‐8.21)	3	267
Tremor	1	34	2,9	0	39	0,0	‐	1	73
Altered facial sensation	1	34	2,9	0	39	0,0	‐	1	73
Postprocedural hypertension	1	37	2,7	0	38	0,0	‐	1	75
Paraesthesia	1	37	2,7	0	38	0,0	‐	1	75
Increased blood PK	1	37	2,7	0	38	0,0	‐	1	75
Increased upper airway secretions	2	223	0,9	6	219	2,7	0.37 (0.09‐1.59)	2	442
Hyperhidrosis	0	37	0	1	38	2,6	‐	1	75
Decreased blood protein	0	37	0	1	38	2,6	‐	1	75
Restlessness	0	37	0	1	38	2,6	‐	1	75
Chest discomfort	0	37	0	1	38	2,6	‐	1	75
Incision site complication	0	37	0	1	38	2,6	‐	1	75
Procedural hypotension	1	200	0,5	5	191	2,6	0.49 (0.02‐14.1)	2	391
Postprocedural complication	0	37	0,0	1	38	2,6	‐	1	75
Tachycardia	1	165	0,6	4	173	2,3	0.44 (0.09‐2.22)	3	338
Pruritus	2	87	2,3	1	88	1,1	1.62 (0.20‐12.88)	2	175
Intraoperative movement	1	43	2,3	1	51	2,0	‐	1	94
Anxiety	0	46	0	1	46	2,2	‐	1	92
Depression	0	46	0	1	46	2,2	‐	1	92
Fatigue	0	46	0	1	46	2,2	‐	1	92
Hypotension	5	227	2,2	5	238	2,1	1.23 (0.38‐3.96)	4	465
Supraventricular extrasystoles	0	96	0,0	2	93	2,2	0.32 (0.03‐3.05)	2	189
Clinical signs of inadequate reversal of NMB	0	188	0,0	4	180	2,2	0.11 (0.01‐2.02)	4	368
Leukocytosis	1	46	2,2	0	36	0,0	‐	1	82
Ventricular extrasystoles	0	48	0,0	1	45	2,2	‐	1	93
Sleep disorder	0	48	0,0	1	45	2,2	‐	1	93
Increased gamma‐glutamyl‐transferase	0	48	0,0	1	45	2,2	‐	1	93
Retching	1	48	2,1	0	45	0,0	‐	1	93
Airway complication to anaesthesia	1	48	2,1	0	45	0,0	‐	1	93
Hot flush	1	48	2,1	0	45	0,0	‐	1	93
Abdominal pain	1	48	2,1	0	48	0,0	‐	1	96
Severe abdominal pain	1	48	2,1	0	48	0,0	‐	1	96
Pharyngolaryngeal pain	1	48	2,1	0	48	0,0	‐	1	96
Diarrhoea	1	48	2,1	0	48	0,0	‐	1	96
Tinnitus	1	48	2,1	0	48	0,0	‐	1	96
Involontary muscle contractions	0	48	0,0	1	48	2,1	‐	1	96
Visual accomodation disorder	0	48	0,0	1	48	2,1	‐	1	96
Increased B₂‐microglobulin	0	48	0,0	1	48	2,1	‐	1	96
Severe bradycardia	0	48	0,0	1	48	2,1	‐	1	96
Productive cough	0	48	0,0	1	48	2,1	‐	1	96
Pyrexia	2	133	1,5	1	131	0,8	1.43 (0.23‐8.91)	3	264
Hypertension	2	143	1,4	1	144	0,7	1.45 (0.23‐9.05)	3	287
Decreased hematocrit	1	74	1,4	0	77	0,0	‐	1	151
Procedural haemorrhage	1	74	1,4	0	77	0,0	‐	1	151
Delayed recovery from anaesthesia	0	74	0,0	1	77	1,3	‐	1	151
Respiratory distress	0	74	0,0	1	77	1,3	‐	1	151
Dizziness	1	85	1,2	1	83	1,2	0.98 (0.10‐9.23)	2	168
Abdominal pain	1	99	1,0	1	97	1,0	0.98 (0.10‐9.27)	2	196
Rash	2	355	0,6	3	346	0,9	0.83 (0.17‐3.96)	5	701
Severe muscle weakness	0	149	0,0	1	142	0,7	‐	1	291
Mild hypoventilation	1	149	0,7	0	142	0,0	‐	1	291
Clinical signs of recurrence of residual NMB	1	674	0,1	2	615	0,3	0.74 (0.05‐10.7)	13	1289
Clinical signs of residual NMB	0	341	0,0	0	305	0,0	‐	7	646
Not able to perform 5 second head‐lift at PACU discharge	0	205	0,0	0	194	0,0	‐	5	399
Redness at injection site	0	50	0,0	0	50	0,0	‐	1	100
Hypersensitivity	0	60	0,0	0	30	0,0	‐	1	90

Table of reported adverse events possibly, probably, or definitely related to sugammadex or neostigmine, listed in descending order according to risk of adverse events. Furthermore, the number of studies observing for each adverse event is presented

List of abbreviations:

NMB ‐ neuromuscular blockade

PACU ‐ post‐anaesthesia care unit

The largest trial in this review (Rahe‐Meyer 2014) randomized 1198 participants and reported that 64 out of 596 participants (10.7%) in the sugammadex group and 72 out of 588 (12.2 %) in the usual care group had at least one drug‐related adverse event. Unfortunately, we could not include these data in our meta‐analysis, as the "usual care" group combined participants who received either neostigmine or placebo, and we were not able to obtain data from the neostigmine group.

Subgroup analysis of composite adverse events

3.2 Different dosages of sugammadex and neostigmine

Different trials used different dosages of sugammadex and neostigmine.

Adamus 2011 compared sugammadex 2 mg/kg versus neostigmine 0.04 mg/kg. Twelve trials compared sugammadex 2 mg/kg versus neostigmine 0.05 mg/kg (Blobner 2010; Castro 2014; Cheong 2015; Flockton 2008; Gaszynski 2011; Illman 2011; Khuenl‐Brady 2010; Koc 2015; Kvolik 2012b; Woo 2013; Wu 2014; Yagan 2015). Two trials compared sugammadex 2 mg/kg versus neostigmine 0.07 mg/kg (Kogler 2012; Koyuncu 2015). Balaka 2011 compared sugammadex 2 mg/kg versus neostigmine 2.5 mg. Kizilay 2016 compared sugammadex 3 mg/kg versus neostigmine 0.03 mg/kg. Four trials compared sugammadex 4 mg/kg versus neostigmine 0.05 mg/kg (Geldner 2012; Hakimoglu 2016; Mekawy 2012; Sabo 2011). Three trials compared sugammadex 4 mg/kg versus neostigmine 0.07 mg/kg (Carron 2013; Jones 2008; Lemmens 2010). Four trials compared several different doses of sugammadex versus several different doses of neostigmine (Brueckmann 2015; Kaufhold 2016; Pongracz 2013; Schaller 2010). Subgroup analysis of data showed no significant subgroup differences in RR for composite adverse events (Analysis 3.2).

3.3. TIVA versus volatile anaesthetics

Twenty trials maintained anaesthesia with volatile anaesthetic (Adamus 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Castro 2014; Cheong 2015; Gaszynski 2011; Hakimoglu 2016; Illman 2011; Jones 2008; Khuenl‐Brady 2010; Kizilay 2016; Koc 2015; Koyuncu 2015; Lemmens 2010; Mekawy 2012; Pongracz 2013; Sabo 2011; Woo 2013; Yagan 2015). Seven trials used TIVA for maintenance (Flockton 2008; Geldner 2012; Kaufhold 2016; Kogler 2012; Kvolik 2012b; Schaller 2010; Wu 2014). One trial provided insufficient information (Balaka 2011). Subgroup analysis of trial results showed no significant subgroup differences in RR for composite adverse events (Analysis 3.3).

Sensitivity analysis of composite adverse events

3.4 Excluding meeting abstracts

Sensitivity analysis excluding data from meeting abstracts (RR 0.60, 95% CI 0.49 to 0.74; I² = 35%; n = 2091; random‐effects model; Analysis 3.4) did not change overall results regarding significance.

Subgroup analysis of bradycardia

3.7 Atropine versus glycopyrrolate

All trials reporting bradycardia combined neostigmine with an antimuscarinic drug. Six trials used atropine (Carron 2013; Gaszynski 2011; Geldner 2012; Koc 2015; Koyuncu 2015; Wu 2014). Five trials used glycopyrrolate (Blobner 2010; Brueckmann 2015; Cheong 2015; Schaller 2010; Woo 2013). Subgroup analysis of trial results showed no significant subgroup differences in RR for bradycardia (Analysis 3.6).

Subgroup analysis of PONV

3.9 TIVA versus volatile anaesthetics

Five trials maintained anaesthesia with volatile anaesthetic (Adamus 2011; Castro 2014; Cheong 2015; Hakimoglu 2016; Yagan 2015), One trial used TIVA for maintenance (Schaller 2010). Subgroup analysis of trial results showed no significant subgroup differences in RR for PONV (Analysis 3.7).

Qualitative data

Investigators reported effects of sugammadex and neostigmine on the following parameters in data format that was ineligible for meta‐analysis.

Intraocular pressure (IOP)

Hakimoglu 2016 described that post‐extubation intraocular pressures (IOPs) were similar between sugammadex and neostigmine groups (P > 0.05; n = 60); Yagan 2015 reported lower end‐extubation IOPs when sugammadex 2 mg/kg was used in comparison with neostigmine 0.05 mg/kg ‐ atropine 0.02 mg/kg (P < 0.05; n = 36), suggesting that sugammadex may be a better option for reversal of neuromuscular blockade in conditions for which an increase in IOP is not desired, such as glaucoma and penetrating eye injury.

Haemodynamic effects

Kizilay 2016 (n = 90) examined the haemodynamic effects of sugammadex and neostigmine in cardiac participants undergoing non‐cardiac surgery. Investigators found that the sugammadex group had lower systolic, diastolic, and mean blood pressures and heart rate when compared with the neostigmine group (P < 0.05). They reported no significant differences between and within groups in terms of QTc interval values. Study authors suggest that sugammadex might be preferred to neostigmine‐atropine combination for reversal of rocuronium‐induced neuromuscular blockade in cardiac patients undergoing non‐cardiac surgery,

Bleeding events

The largest trial in this review(Rahe‐Meyer 2014; n = 1198) included participants undergoing hip/knee surgery or hip fracture surgery and compared sugammadex 4 mg/kg versus usual care (neostigmine or spontaneous recovery). Investigators reported bleeding events within 24 hours in 17 (2.9%) sugammadex and 24 (4.1%) usual care participants (RR 0.70, 95% CI 0.38 to 1.29). Compared with usual care, increases of 5.5% in activated partial thromboplastin time (aPTT; P < 0.001) and 3.0% in prothrombin time (P < 0.001) from baseline occurred with sugammadex 10 minutes after administration and resolved within 60 minutes. Data show no significant differences between sugammadex and usual care for other blood loss measures (transfusion, 24‐hour drain volume, drop in haemoglobin, and anaemia) or for risk of venous thromboembolism, and trials reported no cases of anaphylaxis. Sugammadex induced limited (< 8% at 10 minutes) and transient (< 1 hour) increases in aPTT and prothrombin time but was not associated with increased risk of bleeding or increased severity of bleeding. A much smaller trial (Tas 2015; n = 50) investigated effects of sugammadex and neostigmine on postoperative coagulation parameters and bleeding after seroplasty with sugammadex, increasing postoperative bleeding measured by nasal tip dressings (4.1 ± 2.7 mL in the sugammadex group vs 2.5 ± 2.7 mL in the neostigmine group; P = 0.013) without significantly affecting prothrombin time (PT) (P = 0.953), aPTT values (P = 0.734), or international normalized ratio (INR) values (P = 0.612).

Mekawy 2012 reported no differences in intraoperative blood loss between sugammadex 4 mg/kg (n = 20) and neostigmine 0.05 mg/kg plus atropine 0.02 mg/kg groups (104.6 ± 13.2 vs 111.2 ± 9.8 mL, respectively; P = 0.060)

Renal function

Isik 2016 (n = 50) investigated effects of neostigmine and sugammadex on kidney function and found that both drugs may affect kidney function but sugammadex has more tolerable effects than neostigmine.

Gastric emptying

Sustic 2012 measured gastric emptying by using the paracetamol absorption test. Values of plasma paracetamol concentration (PPC) immediately after arrival of participants in the recovery room (T0) were significantly higher between the sugammadex 2 mg/kg group (1.2 ± 0.9) and the neostigmine 0.04 mg/kg/atropine 0.015 mg/kg group (0.4 ± 0.4) (P < 0.01). Values of PPC at 15, 30, 60, 120, and 150 minutes were higher without reaching statistical difference: T15, 2.1 ± 1.5 vs 1.5 ± 1.4; T30, 3.7 ± 3.8 vs 2.9 ± 2.2; T60, 4.2 ± 2.8 vs 3.5 ± 2.7; T120, 5.0 ± 3.4 vs 4.6 ± 3.6; and T150, 5.9 ± 3.4 vs 4.9 ± 3.2.

Values for PPC at 90 minutes were minimally higher in the neostigmine‐atropine group: time 90, 4.6 ± 3.4 vs 4.7 ± 3.4 (P = NS). Study authors concluded that although results show a tendency toward faster gastric emptying in the sugammadex group, this difference did not reach statistical difference, possibly owing to the small sample size of the study .

Thyroid function

Kvolik 2012a (n = 24) investigated effects on thyroid function and observed a significant increase in T4 levels compared with baseline one hour after anaesthesia (from 13.3 to 17.5 in the neostigmine group, and from 12.6 to 16.2 pmol/L in the sugammadex group; P < 0.05) that returned to baseline after 24 hours in both groups. T3 decreased in both groups postoperatively (from 5.2 to 3.5 in the neostigmine group, and from 4.9 to 3.3 pmol/L in the sugammadex group), with no intergroup differences noted (P > 0.05). Mean thyroid‐stimulating hormone (TSH) after 24 hours was not different between groups (1.32 in the neostigmine group vs 1.27 pmol/L in the sugammadex group; P = 0.49). In conclusion, sugammadex treatment did not change the levels of thyroid hormones and may be used safely in patients undergoing total thyroidectomy.

Cognitive function

Riga 2014 (n = 114) investigated cognitive function in patients receiving sugammadex or neostigmine and found no significant differences between groups when measuring cognitive function with the mini‐mental state evaluation test (P = 0.25), as described in Tombaugh 1992, and the Clock Drawing test (P = 0.06), as described in Agrell 1998.

Postoperative vomiting and nausea (PONV)

Carron 2013 reported higher PONV scores in the neostigmine group than in the sugammadex group (3.2 ± 1.5 vs 1.9 ± 1.3; P = 0.015; n = 40) with no significant difference in antiemetic supplement (7 (35%) vs 3 (15%); P = 0.10).

Tas 2015 compared sugammadex 2 mg/kg (n = 24) versus neostigmine 0.05 mg/kg plus atropine 0.02 mg/kg (n = 26) and reported no differences regarding nausea/vomiting between groups (P = 0.512).

Raziel 2013 (n = 40) observed no differences between sugammadex 2 mg/kg and neostigmine 0.05 mg/kg in nausea/vomiting among morbidly obese participants undergoing bariatric surgery.

Pain

Martini 2014 compared moderate NMB (T1 to T2) induced by atracurium/mivacurium reversed by neostigmine 1 to 2 mg plus atropine 0.5 to 1 mg (n = 12) versus deep NMB (PTC 1 to 2) induced by high‐dose rocuronium and reversed by sugammadex 4 mg/kg (n = 12) and found no significant differences in pain score as measured by a 10‐point scale (2.6 ± 1.6 vs 2.1 ± 2.2, respectively).

Tas 2015 compared sugammadex 2 mg/kg (n = 24) versus neostigmine 0.05 mg/kg plus atropine 0.02 mg/kg (n = 26) and reported no differences regarding postoperative pain between groups (P = 0.280).

Overall signs of postoperative residual paralysis

We chose the following parameters as overall signs of postoperative residual paralysis: inability to perform 5 second head‐lift test and general muscle weakness after extubation and at PACU discharge, amblyopia, asthenia, desaturation < 90%, transitory oxygen supplementation, respiratory distress, respiratory depression, postoperative respiratory complications (evaluated by PRSES – postoperative system evaluation score), moderate dyspnoea, pneumonia, acute lung failure, or symptoms of residual NMB or recurrence of NMB if specifically reported by study authors. The following 15 studies reported any of these adverse events: Balaka 2011; Blobner 2010; Brueckmann 2015; Carron 2013; Flockton 2008; Geldner 2012; Jones 2008; Khuenl‐Brady 2010; Koyuncu 2015; Kvolik 2012b; Lemmens 2010; Mekawy 2012; Schaller 2010; Woo 2013; Wu 2014).

Meta‐analysis of trial results showed significantly reduced risk of overall signs of postoperative residual paralysis (RR 0.40, 95% CI 0.28 to 0.57; I² = 0%; n = 1474; random‐effects model; NNTB 13; Analysis 3.39; GRADE quality of evidence: moderate; summary of findings Table 3) in the sugammadex group when compared with the neostigmine group. We downgraded GRADE quality of evidence one level owing to high risk of bias.

Investigators reported the following data on overall events of postoperative residual paralysis, which were ineligible for meta‐analysis.
Carron 2013 (n = 40) found higher peripheral oxygen saturation levels (SpO₂) levels at recovery admission in the sugammadex group (97 ± 2.3% vs 94.4 ± 4%; P = 0.018), along with faster ability to swallow after extubation (7.1 ± 1.8 minutes vs 12.2 ± 6 minutes; P = 0.0027), and faster ability to get into bed independently (24 ± 9 minutes vs 33.4 ± 12 minutes; P = 0.022) when compared with the neostigmine group.

Foletto 2014 (n = 34) reported that respiratory function was restored more quickly in morbidly obese (MOB) participants who received sugammadex when measured by postoperative forced vital capacity (1.6 ± 0.7 vs 2.41 ± 0.8 L; P < 0.05), forced expiratory volume in one second (1.37 ± 0.7 vs 2.05 ± 0.6 L/s; P < 0.05), and peak expiratory flow 30 minutes postoperatively (2.55 ± 1.7 vs 3.75 ± 1.4 L/s; P < 0.05), but observed no significant differences in spirometry performed 15 minutes postoperatively.

Raziel 2013 (n = 40) observed no differences between sugammadex 2 mg/kg and neostigmine 0.05 mg/kg in respiratory function among morbidly obese participants undergoing bariatric surgery.

Martini 2014 compared moderate NMB (T1 to T2) induced by atracurium/mivacurium reversed by neostigmine 1 to 2 mg plus atropine 0.5 to 1 mg (n = 12) with deep NMB (PTC 1 to 2) induced by high‐dose rocuronium and reversed by sugammadex 4 mg/kg (n = 12), and found no significant difference in saturation in PACU (98.6 ± 1.8 vs 98.2 ± 1.4, respectively) or breathing rate in PACU (14.5 ± 2.2 vs 14.5 ± 2.2, respectively).

Sherman 2014 found lower saturation levels (95.8 ± 0.014 vs 96.72 ± 0.01; P < 0.02), lower minimal saturation (93% vs 94%), and no difference in respiratory complications when comparing neostigmine 2.5 mg (n = 25) versus sugammadex 2 mg/kg (n = 32).

Tas 2015 compared sugammadex 2 mg/kg (n = 24) versus neostigmine 0.05 mg/kg plus atropine 0.02 mg/kg (n = 26) and reported no differences between groups regarding saturation levels after extubation (97.6 ± 0.2 vs 98.0 ± 0.2, respectively; P = 0.280).

Furthermore, several trials conducted postoperative neuromuscular monitoring to quantify the risk of residual neuromuscular blockade, defined as TOFR < 0.9: Brueckmann 2015 found that zero out of 74 (0%) sugammadex participants and 33 out of 76 (43.4%) neostigmine participants had TOFR > 0.9 at PACU admission (odds ratio (OR) 0.0, 95% CI 0.0 to 0.6; P < 0.0001). Of the 33 neostigmine participants, 2 also had clinical evidence of residual NMB.

Sabo 2011 described that 2 out of 50 participants (4%) in the sugammadex group had residual NMB (TOFR < 0.9) at the time of extubation compared with 26 out of 43 participants (60.5) in the neostigmine group, although data provided no clinical evidence (i.e. respiratory problems) of residual NMB in either group.

Gaszynski 2011 described that TOF at PACU was 109.8% versus 85.5% (P < 0.05; n = 70) in the sugammadex and neostigmine groups, respectively, and reached > 90% in every case in the sugammadex group but not in the neostigmine group.

No participants experienced recurrence of neuromuscular blockade based on neuromuscular monitoring in Geldner 2012 (n = 133).

Sugammadex (any dose) versus neostigmine (any dose), drug‐related serious adverse events (SAEs)

Fourteen trials reported serious adverse events (SAEs) possibly, probably, or definitely related to study drug (Adamus 2011; Blobner 2010; Brueckmann 2015; Flockton 2008; Geldner 2012; Hakimoglu 2016; Jones 2008; Kaufhold 2016; Khuenl‐Brady 2010; Koyuncu 2015; Lemmens 2010; Schaller 2010; Woo 2013; Wu 2014). Meta‐analysis of trial results showed no significant differences between sugammadex and neostigmine regarding participants with one or more serious adverse events or for composite adverse events (RR 0.54, 95% CI 0.13 to 2.25; I² = 0%; ten studies; n = 959; random‐effects model; Analysis 3.40; GRADE quality of evidence: low; summary of findings Table 3). We downgraded GRADE quality of evidence two levels owing to high risk of bias and imprecision.

Clearly reported drug‐related serious adverse events included one case of acute myocardial infarction, pneumonia, and inadequate NMB reversal in the neostigmine group (Brueckmann 2015), one case of acute lung failure in the neostigmine group (Schaller 2010), one case of postoperative upper abdominal pain in the neostigmine group (Geldner 2012), one case of postprocedural haemorrhage in the sugammadex group (Brueckmann 2015), and finally one case of respiratory depression in the sugammadex group (Koyuncu 2015).

Trial sequential analysis (TSA)

We applied TSA to several outcome data as described in summary of findings Table for the main comparison, summary of findings Table 2, and summary of findings Table 3.

TSA of all trials comparing neostigmine 0.05 mg/kg versus sugammadex 2.0 mg/kg with regard to recovery time from T2 to TOFR > 0.9 minutes indicates that with a required information size of 106, firm evidence sugammadex in a random‐effects model, with an alfa‐boundary adjusted MD of ‐10.22 (95% CI ‐12.11 to ‐8.33; diversity (D²) = 87%; I² = 84%; random‐effects model; Figure 1). The cumulative Z‐curve crossed the monitoring boundary constructed for the required information size with 80% power and alpha of 0.05. However, none of the included trials had low risk of bias, and given that TSA is ideally designed for trials with low risk of bias and cannot be adjusted for risk of bias, the precision of our findings has to be downgraded. Furthermore, we found a high degree of diversity and heterogeneity, which once again raises questions about the reliability of the calculated required information size.

TSA of dichotomous data on drug‐related risk of adverse events when neostigmine (any dose) was compared with sugammadex (any dose) with continuity adjustment for zero event trials (0.001 in each arm) resulted in an alfa‐boundary adjusted RR of 0.62 (95% CI 0.51 to 0.74; diversity (D²) = 34%; I² = 14%; random‐effects model; Figure 3), with a control event proportion of 27.97%. With the required information size of 502, analyses provided firm evidence in favour of sugammadex, with 2298 participants included, corresponding to a relative risk reduction (RRR) of 38% with 80% power and alpha of 0.05. Despite the fact that the cumulative Z‐curve does not cross the monitoring boundary directly, it is hard to imagine future trials radically changing the overall picture of this analysis. Once again, none of the included trials had low risk of bias and this does downgrade the reliability of our finding.

TSA of dichotomous data on risk of serious adverse events when neostigmine (any dose) was compared with sugammadex (any dose) with continuity adjustment for zero event trials (0.001 in each arm) resulted in an alfa‐boundary adjusted RR of 0.35 (95% CI 0.00 to 3190; diversity (D²) = 0%; I² = 0%; random‐effects model), with a control event proportion of 1.04%. The cumulative Z‐curve does not cross the monitoring boundary constructed for a required information size of 8189 participants, with 11.71% of the required information size included across included trials so far with 80% power and alpha of 0.05. Once again, none of the included trials had low risk of bias and this affects the reliability and precision of our estimates.

TSA of dichotomous data on risk of signs of residual neuromuscular blockade when neostigmine (any dose) was compared with sugammadex (any dose) with continuity adjustment for zero event trials (0.001 in each arm) resulted in an alfa‐boundary adjusted RR of 0.4 (95% CI 0.27 to 0.59; diversity (D²) = 0%; I² = 0%; random‐effects model), with 80% power and alpha of 0.05 (Figure 4), with a control event proportion of 13.08%. The cumulative Z‐curve crosses the monitoring boundary constructed for a required information size of 424 participants, indicating firm evidence in favour of sugammadex. However, as previously described, none of the included trials had low risk of bias and this equally diminishes the reliability and precision of our estimates.

Finally, owing to overall high risks of bias, imprecision, and indirectness involved in assessment of GRADE for the above analysis, one could easily argue that the required power should be 90% ‐ not 80% ‐ by which the required information size would be increased; nevertheless we cannot rule out the direction of results in favour of sugammadex, despite the absence of large trials with low risk of bias.

Discussion

Summary of main results

In this systematic review of 41 randomized controlled trials (RCTs; 4206 participants) comparing the efficacy and safety of sugammadex versus neostigmine in reversing rocuronium‐induced neuromuscular blockade (NMB), we found a large and significant difference in reversal time favouring sugammadex. For meta‐analyses of primary outcomes, 12 studies (n = 949) were eligible.

Meta‐analysis of trial results showed that sugammadex 2 mg/kg reversed NMB from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 in 1.96 minutes, and neostigmine 0.05 mg/kg reversed NMB from T2 to TOFR > 0.9 in 12.87 minutes. Sugammadex 2 mg/kg was therefore on average 10.22 minutes (6.6 times) faster than neostigmine 0.05 mg/kg in reversing NMB at T2 reappearance (mean difference (MD) 10.22 minutes, 95% confidence interval (CI) 8.48 to 11.96; I² = 84%; ten studies; n = 835; random‐effects model; GRADE quality of evidence: moderate; Analysis 1.1). Reversal time from post‐tetanic count (PTC) 1 to 5 to TOFR > 0.9 was not investigated; this was considered clinically irrelevant owing to the doses of sugammadex and neostigmine used for this comparison.

Sugammadex 4 mg/kg reversed NMB from PTC 1 to 5 to TOFR > 0.9 in 2.9 minutes, and neostigmine 0.07 mg/kg reversed NMB from PTC 1 to 5 to TOFR > 0.9 in 48.8 minutes. Sugammadex 4 mg/kg was therefore on average 45.78 minutes (16.8 times) faster than neostigmine 0.07 mg/kg in reversing NMB at PTC 1 to 5 reappearance (MD 45.78 minutes, 95% CI 39.41 to 52.15; I² = 0%; n = 114; random‐effects model; GRADE quality of evidence: low; Analysis 2.1). Reversal time from T2 to TOFR > 0.9 was not investigated since it was deemed clinically irrelevant owing to the doses of sugammadex and neostigmine used for this comparison.

We found 28 trials (n = 2298) eligible for meta‐analysis of the secondary outcomes (risks of adverse events and serious adverse events). We found significantly fewer composite adverse events in the sugammadex group than in the neostigmine group (risk ratio (RR) 0.60, 95% CI 0.49 to 0.74; I² = 40%; 28 studies; n = 2298; random‐effects model; GRADE quality of data: moderate; Analysis 3.1). Specifically, the risk of composite adverse events was 283/1000 in the neostigmine group and 159/1000 in the sugammadex group. Analysis of number needed to treat for an additional beneficial outcome (NNTB) revealed that eight patients should be treated with sugammadex rather then neostigmine to avoid one patient experiencing a single random adverse event. Furthermore, significantly fewer participants had one or more adverse events (RR 0.62, 95% CI 0.48 to 0.81; I² = 0%; n = 1766; random‐effects model; GRADE quality of data: moderate; Analysis 3.5) in the sugammadex group than in the neostigmine group. Review of specific adverse events in the sugammadex group compared with the neostigmine group revealed significantly less risk of the following adverse events: bradycardia (Analysis 3.6), postoperative nausea and vomiting (PONV) (Analysis 3.7), desaturation (Analysis 3.8), and need for transitory oxygen supplementation (Analysis 3.10). Also, a significantly lower number of participants in the sugammadex group were not able to perform 5 second sustained head‐lift at extubation (Analysis 3.11). Data showed no significant differences between sugammadex and neostigmine regarding participants with one or more serious adverse events, nor in composite adverse events (RR 0.54, 95% CI 0.13 to 2.25; I² = 0%; ten studies; n = 959; random‐effects model; GRADE quality of evidence: low; Analysis 3.40). Reversal time from T2 and PTC 1 to 5 to TOFR > 0.9 was not investigated, as it is clinically irrelevant owing to the doses of sugammadex and neostigmine used for this comparison.

Overall completeness and applicability of evidence

For our primary outcome, we performed a comparison of the effects of sugammadex and neostigmine at two depths of NMB: moderate block as indicated by reappearance of T2, and deep block as indicated by reappearance of PTC 1 to 5 on neuromuscular monitoring. However, administration of neostigmine is not recommended for reversal of deep block and absence of any signs of neuromuscular recovery due to the ceiling effect (Caldwell 2009; Plaud 2010), which is seen when maximal acetylcholine concentration is not sufficient to adequately compete with the muscle relaxant. According to the current prescribing information, this is an off‐label indication (www.fda.gov). Nevertheless, our search identified two trials (Carron 2013, Jones 2008) in which sugammadex and neostigmine were used to reverse rocuronium‐induced deep NMB, and one trial (Lemmens 2010) in which sugammadex and neostigmine were used to reverse vecuronium‐induced deep block. As this was not an exclusion criterion in the original protocol and the data were available, we chose to include these three studies in our review. However, for reasons explained above, the clinical importance of these comparatory findings aside from the obvious faster reversal due to sugammadex remains questionable.

In this context, one could argue that a comparison between sugammadex and neostigmine for reversing a shallow NMB would be more relevant. However, this was not a predefined outcome in the original protocol. Furthermore, our search identified five trials in which some degree of shallow block was indicated (Kaufhold 2016; Koyuncu 2015; Pongracz 2013; Schaller 2010; Yagan 2015), but none of these trials obtained comparable data on recovery time to TOFR > 0.9.

The overall quantity of data on which our conclusions can be based is large, and data were drawn from 41 randomized controlled trials with 4206 participants. According to GRADE, the quality of evidence for most of our meta‐analyses is moderate. Most trial participants were adults classified as American Society of Anesthesiologists (ASA) I to III who were undergoing elective surgery, and reported outcomes were relevant in a clinical setting. Primary and secondary outcomes, recovery time to TOFR > 0.9, and adverse effects, were generally well reported. Therefore, on basis of the large number of identified studies and participants, available evidence seems to be applicable to adult patients of ASA classification I to III who are undergoing elective surgery.

According to our meta‐analyses, sugammadex 2 mg/kg given at T2 reverses the NMB within 1.96 minutes and 6.6 times (10.22 minutes) faster than neostigmine 0.05 mg/kg (12.87 minutes). Furthermore, sugammadex 4 mg/kg, given to deep NMB at PTC 1 to 5 reappearance, reverses the block in 2.9 minutes and 16.8 times (45.78 minutes) faster than neostigmine 0.07 mg/kg (48.8 minutes).

The time difference offers several potential advantages in that a patient who is paralysed with a neuromuscular blocking agent has to be out of the NMB with TOFR > 0.9 before undergoing tracheal extubation, to avoid adverse effects due to residual paralysis (Eikermann 2006; Murphy 2008; Murphy 2013).

Sugammadex rapidly reverses NMB. This appears favourable because it reduces required anaesthesia time for the patient. Additionally, unlike neostigmine, sugammadex can be administered at any stage during a surgical procedure and independent of the depth of blockade. A reduced duration of anaesthesia not only may improve recovery time for the patient but could potentially reduce costs by saving the time needed for a prolonged awakening and potentially enabling smoother flow of patients through the operating theatre.

The cost‐effectiveness of sugammadex was not a predefined outcome of this review. To demonstrate cost‐effectiveness of sugammadex, two issues must be established: reduced patient recovery time perioperatively, and translation of any such reduction into resource utilization in terms of freeing up staff to work on productive alternative activities such as caring for other patients. This outcome is very difficult to assess owing to various confounders, such as the organizational structure of each hospital (Dexter 1995), procedural flow, variability of NMB, monitoring and extubation practices, turnover times between procedures, frequency of emergency procedures, operating room overtime resource use, staff payments, productive alternative use of freed resources (Fuchs‐Buder 2012; Paton 2010), and finally the cost of available drugs in each country. Furthermore, it is difficult to calculate whether any reduction in adverse events associated with sugammadex, besides improved quality of care, can readily be translated into cost‐effectiveness.

One systematic review (Paton 2010) compared the cost‐effectiveness of sugammadex versus neostigmine/glycopyrrolate for routine reversal of moderate or profound muscle relaxation produced by rocuronium and vecuronium. Results from included trials (Flockton 2008, Blobner 2010, Lemmens 2010, Jones 2008) indicate that sugammadex 2 mg/kg (4 mg/kg) produces more rapid recovery from moderate NMB than is achieved with neostigmine/glycopyrrolate. Economic assessment indicated that if the reductions in recovery time associated with sugammadex in these trials were replicated in routine clinical practice, sugammadex would be cost‐effective if those reductions were achieved in the operating theatre, but not if they were achieved in the recovery room. Review authors went on to conclude that further research is required to evaluate the effects of sugammadex on patient safety, predictability of recovery from NMB, patient outcomes, and efficient use of resources. A recent Canadian study (Insinga 2016) used a discrete model‐event simulation to investigate the potential impact of substituting sugammadex for neostigmine on operating room efficiency and incidence of residual NMB. Study authors concluded that the principal impact for patients managed by moderate NMB is likely to be seen as a reduction in the risk of residual NMB and associated complications. For patients maintained at a deep level of block until the procedure is completed, sugammadex was likely to both enhance operating room efficiency and reduce residual block complications. Last but not least, the cost per anaesthetic case might increase in case of unrestricted use of sugammadex, as shown in a retrospective observational audit (Ledowski 2012).

In conclusion, considerable uncertainties remain regarding the cost‐effectiveness of sugammadex, and further investigation is needed. Currently, the cost of sugammadex is relatively high as the result of proprietary rights. The price for the smallest vial (100 mg/mL, 2 mL) in Denmark is around 117 euros. In addition, drug patents are set to expire on 27 January 2021 (Drugs.com). How this will affect the price and clinical usage of sugammadex remains to be established.

Another important clinical consideration in the choice of reversing agent is the risk of adverse effects.

The decision to use a drug is based on an overall assessment of its benefits and harms. Monitoring and reporting of adverse events during a clinical trial constitutes a cumbersome and complex task involving many assumptions and choices, such as adequate blinding of study participants and investigators, distinction between adverse and serious adverse events, causality of adverse events to study drugs, reporting by patients, and finally consistent and transparent monitoring, coding, and reporting by investigators.

Trials included in this review defined, monitored, and reported adverse events in many different ways. Some trials (Blobner 2010; Jones 2008; Lemmens 2010) coded all adverse events and serious adverse events described by the investigator in a systematic way using the Medical Dictionary for Regulatory Activities (MeDRA). Other trials reported symptoms related to study drug administration without necessarily defining them as adverse events (Adamus 2011; Mekawy 2012) ‐ an issue most often seen in meeting abstracts (Balaka 2011; Georgiou 2013; Kvolik 2012b) that is probably due to word count restriction. Furthermore, some included trials specifically addressed causality between adverse events and study drugs by presenting not only adverse events observed regardless of relation to study drug but also adverse events possibly, probably, or definitely related to study drug (Blobner 2010; Jones 2008; Lemmens 2010; Woo 2013), although others did not specifically address this issue (Adamus 2011; Castro 2014; Yagan 2015). Smaller trials with few observed adverse events usually presented all observed adverse events (Balaka 2011; Koc 2015; Koyuncu 2015; Yagan 2015), while bigger trials presented the most frequently occurring adverse events (Brueckmann 2015; Jones 2008; Lemmens 2010; Woo 2013). Additionally, some trials used blinded safety outcome assessors (Blobner 2010; Brueckmann 2015; Carron 2013; Flockton 2008; Woo 2013) in contrast to others (Grintescu 2009; Kizilay 2016). Last but not least, very few of the included trials were designed and powered to address safety as a primary outcome (Brueckmann 2015; Rahe‐Meyer 2014).

As explained earlier in the Methods and Results sections, overall clinical signs of postoperative residual paralysis such as inability to perform 5 second head‐lift and general muscle weakness observed in some trials (Blobner 2010; Flockton 2008; Jones 2008; Khuenl‐Brady 2010; Lemmens 2010) were regarded as adverse events in this review. Furthermore, we decided to include reported symptoms related to drug administration as adverse events, even though they were not specifically defined as adverse events, to avoid potentially dismissing good quality data because of lack of correct phrasing. We have addressed and explained under the notes section in Characteristics of included studies any discrepancy in adverse events presented in the original article and in this review due to definitions of adverse events or additional data about adverse events supplied through email correspondence with trial authors. Readers of medical journals and of this review need to be aware of these issues as they appraise this review and the literature critically.

Included trials provided sparse data regarding which body weight dose calculations were based upon (i.e. ideal, correlated, or lean body weight), and we were unable to retrieve additional data that would shed light on this. As a consequence, we have regarded the weight data provided as total body weight.

Our results show an overall significantly lower risk of adverse events in the sugammadex group than in the neostigmine group (Analysis 3.1; Analysis 3.5), along with an NNTB of eight for avoidance of an adverse event.

Data show significantly less risk of the clinically important adverse effect PONV (Analysis 3.7) and less risk of overall signs of postoperative residual paralysis in the sugammadex group (Analysis 3.39), making this treatment preferable because residual blockade increases the risk of serious adverse effects such as acute respiratory failure (Murphy 2008; Sauer 2011). Data also show reduced risk of bradycardia (Analysis 3.6) in the sugammadex group. However, the two groups reported many adverse reactions similarly, as presented in the Results section. Results show that no cases of anaphylaxis were reported.

Our results may not be directly applicable to all groups of patients because sugammadex may have different outcomes for patients with higher ASA classes and for patients with special comorbidities or systemic dysfunction.

These patients are not represented well in the trials included in our meta‐analyses, but lower risk of adverse effects as well as sufficient reversal from neuromuscular blockade may be even more beneficial for this group of patients, and inclusion of these more fragile patients in future trials could potentially reduce the NNTB for avoidance of adverse events. However, this might not be applicable to all patient groups (e.g.. severe renal impairment has been discussed as a possible contraindication to treatment. Sugammadex is excreted unchanged in the urine by the kidneys. Renal clearance of sugammadex is rapid, with most of the dose (70%) excreted within six hours (Golembiewski 2016). None of the included trials enrolled participants with renal dysfunction. However, Isik 2016 (n = 50) investigated the effects of neostigmine and sugammadex on adults of ASA I to II with normal renal function and found that both drugs may impair renal function, but sugammadex was more tolerable than neostigmine. A pharmacokinetic study (Staals 2010) investigated the pharmacokinetics of sugammadex 2 mg/kg and of rocuronium 0,6 mg/kg in 15 participants with renal failure and in 15 healthy controls. Investigators found that urinary excretion of sugammadex was reduced among participants with renal failure. The median quantity of sugammadex excreted in the urine within 72 hours among participants with renal failure was 29%, and 73% in controls. Nevertheless, one has to conclude on the basis of existing evidence that studies on the use of sugammadex for patients with renal impairment are needed to examine safety, preferably with longer follow‐up than 72 hours, because late renal impairment has to be addressed equally.

Sugammadex has been suspected of increasing the risk of specific adverse effects such as QTc prolongation and bleeding events (Bridion 2014). However, we found limited data from few trials in our systematic review on these variables, and presented data were ineligible for meta‐analysis.

The summary of product characteristics provided by Bridion states that the "administration of 4 and 16 mg/kg of Sugammadex in healthy volunteers resulted in maximum and mean prolongations of the aPTT by 17% and 22%, respectively, and PT by 11% and 22%, respectively. These mean aPTT and PT prolongations were limited and of short duration < 30 min" (Bridion 2014). Rahe‐Meyer 2014 (n = 1198) included participants undergoing hip or knee surgery and compared sugammadex 4 mg/kg versus usual care (neostigmine or spontaneous recovery). Study findings indicate that sugammadex induced limited (< 8% at 10 minutes) and transient (< 1 hour) increases in activated partial thromboplastin time (aPTT) and prothrombin time (PT) but was not associated with increased risk or severity of bleeding. Tas 2015 (n = 50) investigated the effects of sugammadex and neostigmine on postoperative coagulation parameters and bleeding after seroplasty and demonstrated that sugammadex increased postoperative bleeding without significantly affecting PT and aPTT values. An RCT of healthy adults reported that after administration of sugammadex at doses of 4 mg/kg and 16 mg/kg, a dose‐dependent, limited, temporary, and clinically irrelevant prolongation in PT and aPTT was observed (De Kam 2014). A one‐year retrospective study (n = 193) performed in participants with high risk of postoperative bleeding (laparotomy for cancer surgery requiring suction drains) did not find sugammadex at doses of 2 and 4 mg·kg^–1 to be associated with increased bleeding as measured by the amount of blood found in suction drains and dressings (Raft 2011).

However, upon review of the published literature, we are unable to refute or reject any safety concern with regard to sugammadex for patients at high risk of bleeding due to existing severe coagulopathy or due to the nature of procedures associated with high risk of transfusion because evidence is inadequate to support or withhold any use of sugammadex.

We found limited evidence with regard to haemodynamic implications of sugammadex use, but Kizilay 2016 compared the haemodynamic effects of sugammadex and neostigmine in participants with cardiac disease undergoing non‐cardiac surgery. Haemodynamic parameters were more prominently increased among participants receiving neostigmine, and cardiac function was noted to be more stable among those given sugammadex. Data show no significant differences between and within groups in terms of QTc values.

Morbidly obese patients make up a high‐risk group (Gaszynski 2011), and because of their often compromised respiratory function, they are considered especially vulnerable to residual curarization in the postoperative period influencing respiratory function (Gaszynski 2011). Three trials (n = 161) investigated the optimal sugammadex dose per kilogram body weight; total body weight (TBW) (Foletto 2014), corrected body weight (CBW) (Gaszynski 2011; Georgiou 2013), and ideal body weight (IBW) (Georgiou 2013). All three studies found sugammadex 2 mg/kg to be significantly faster than neostigmine 0.04 to 0.05 mg/kg in reversing neuromuscular blockade at T2 reappearance, and in reducing the risk of postoperative residual curarization (Foletto 2014; Gaszynski 2011).

Researchers have speculated about the influence of volatile anaesthetics and recovery times when neuromuscular blocking agents (NMBAs) are used (Reid 2001). However, we found no significant differences in recovery time to TOFR > 0.9 when anaesthesia maintained with volatile anaesthetic (eight trials; n = 490) was compared with total intravenous anaesthetic (TIVA) (four trials; n = 381) (Analysis 1.2).

Sugammadex was specifically designed to reverse rocuronium as a non‐depolarizing NMBA, as is demonstrated by most of the trials included in this review. However, two of the included trials (Lemmens 2010; Rahe‐Meyer 2014) used sugammadex to revert vecuronium. Furthermore, Flockton 2008 compared sugammadex following rocuronium versus neostigmine following cisatracurium, and Martini 2014 compared atracurium for induction and mivacurium for maintenance versus rocuronium for both induction and maintenance. Two studies (Castro 2014; Sherman 2014) provided no information on the NMBA used.

Quality of the evidence

This systematic review provides a robust assessment of the efficacy of sugammadex because it includes a large number of trials with large numbers of participants showing a consistent direction of results across all trials and additional confirmation through various exploratory analyses favouring the intervention for our primary outcome.

However, this review also has several potential limitations, as our findings and interpretations are limited by the quality and quantity of available evidence from included RCTs. The RCT is considered the most rigorous method of determining whether a cause‐effect relationship exists between an intervention and an outcome. The strength of the RCT lies in the process of randomization, but several potential risks of bias in trial methods can affect results.

The review authors have judged the risk of bias for each included study by using the recommended risk of bias assessment in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). All of our studies had at least one "high or unclear risk of bias", and we considered the risk that trials may overestimate or underestimate the true intervention effect a serious limitation for all trials. In particular, judgements of performance risk of bias and funding risk of bias were overall high. We judged none of the included studies as having low risk of bias.

Application of the GRADE approach enables us to incorporate risk of bias, directness of evidence, heterogeneity, precision of effect estimate, and risk of publication bias.

The GRADE quality of our findings ranks as moderate for our primary outcome, and from low to moderate across different outcomes. The main limiting factors that accounted for decreased quality of evidence included high risk of bias and imprecision (summary of findings Table for the main comparison; summary of findings Table 2; summary of findings Table 3).

We mainly assessed the risk of bias of included trials using published data, which ultimately may not reflect the truth. We contacted all trial authors; 12 (33.3%) responded and provided further information. Lack of reporting of some of the data may have affected our judgement on risk of bias in either direction.

We applied several statistical methods to explore and reduce the extent of bias, such as complete case analysis, trial sequential analysis (TSA), overall methodological bias assessment, and analyses of various relevant clinical and physiological outcomes.

Application of TSA to our primary outcome indicates that at this stage, sugammadex appears superior to neostigmine. TSA provided firm evidence in favour of sugammadex for outcomes such as recovery time from T2 to TOFR > 0.9 minutes, adverse events, and overall signs of postoperative residual paralysis. However, none of the included trials were at low risk of bias, and as TSA cannot be adjusted for risk of bias, we did not calculate the low risk of bias adjusted information size, which ultimately affects the reliability and precision of our findings.

Evaluated outcomes consistently favoured sugammadex. However, we graded the quality of evidence as moderate because of the high proportion of trials at high risk of bias, large clinical and statistical heterogeneity, and small sample sizes, but we upgraded the level of evidence in favour of sugammadex as indicated by TSA analyses.

On the basis of the criteria mentioned above, we deemed the overall GRADE quality of evidence in this review to be moderate.

Sugammadex was specifically designed to reverse rocuronium as a non‐depolarizing NMBA, as most of the trials included in this review demonstrated this. However, two of the included trials (Lemmens 2010; Rahe‐Meyer 2014) used sugammadex to revert vecuronium.

Potential biases in the review process

We have followed the recommendations provided in the Cochrane Handbook for Systematic Reviews of Interventions as this official guide describes in detail the process of preparing and maintaining Cochrane systematic reviews on the effects of healthcare interventions (Higgins 2011). We have adhered to this handbook in handling the included RCTs.

Meta‐analyses are limited by the quality and quantity of available evidence. Even though our meta‐analyses are based on a large quantity of data, results and methods for some of the included studies were not thoroughly described. Furthermore, some of the included trials were not specifically designed to address the primary or secondary outcomes of this review, leading to possibly biased data. We have addressed this problem by labelling studies with high risk of "other bias", as is shown in Characteristics of included studies, Figure 2, and Figure 6, and by downgrading the GRADE quality of evidence (summary of findings Table for the main comparison, summary of findings Table 2, and summary of findings Table 3). Additionally, we are aware that as we have performed many analyses of specific adverse events, the probability of achieving significant results by chance is high.

We used the same search strategy as was used in the original version of this review (Abrishami 2009), and we found 41 eligible studies for inclusion. We cannot exclude the possibility that we may have missed some of the published literature beyond the electronic databases searched for this review. However, the 41 trials with 4206 participants included in this review appear to provide sufficient data for meta‐analyses, and our TSAs indicate a better safety profile and clinical superiority of sugammadex compared with neostigmine for the population included in this trial.

We found 20 relevant ongoing trials registered at https://clinicaltrials.gov and three trials awaiting classification, but none of these studies have published data within our main search update from 2008 to 2 May 2016. When published, these trials may change the results and conclusions of this review. However, the main strength of this update consists of the quantity of data comparing sugammadex versus neostigmine in reversing NMB. The new search added eight years of research and 38 new trials to the review that was originally published (Abrishami 2009), which comprised three trials comparing sugammadex and neostigmine. Additionally, we have substantially updated and revised the methods of this review compared with methods of the previous one.

As a consequence, this review diverges from intended adherence to the Cochrane Handbook for Systematic Reviews of Interventions by not following the original protocol (Abrishami 2009) prepared for the first version of this review (Abrishami 2009). After several discussions with the editorial team, we made the decision to split the original review in two based on the extensive number of publications using various comparators, interventions, and outcome measures. Therefore, it seemed more appropriate to take the original review in a different direction and place more emphasis on safety issues and efficacy. Although this may be perceived as introduction of post hoc analyses, review authors selected outcomes and subgroup and sensitivity analyses for this review before identifying included trials (search) and extracting data to minimize the risk of bias.

Agreements and disagreements with other studies or reviews

The original published review (Abrishami 2009) found no difference with regard to adverse effects between sugammadex and neostigmine. This review found that sugammadex reduced the risk of adverse events when compared with neostigmine. We updated this review as of 2 May 2016 with regard to the search, adding eight years of research and 38 new trials; the original review (Abrishami 2009) comprised three trials. We re‐ran the search on 10 May 2017. Currently three trials are awaiting classification and 20 studies are ongoing.

Our results on the primary outcome, recovery time, are in accordance with the findings of all RCTs included in the meta‐analyses, as they reflect superiority for sugammadex as a reversing agent over neostigmine. With regard to our secondary outcomes ‐ risks of adverse and serious adverse events ‐ we found more diverging results among the included trials, although overall risk of adverse events was reduced in the sugammadex group (Analysis 3.1; Analysis 3.5). No previous publication has addressed this issue with the same rigour.

A recent systematic review of sugammadex versus neostigmine for reversal of NMB (Abad‐Gurumeta 2015) included 1553 participants across 17 RCTs (all are included in this review).

Abad‐Gurumeta 2015 focused mainly on postoperative residual paralysis and drug‐related adverse events (Abad‐Gurumeta 2015). Review authors found that sugammadex reduced all signs of residual postoperative paralysis (RR 0.46, 95% CI 0.29 to 0.71; P = 0.0004) and risk of minor respiratory events (RR 0.51, 95% CI 0.32 to 0.80; P = 0.0034). However, they reported no differences in critical respiratory events (RR 0.13, 95% CI 0.02 to 1.06; P = 0.06). Sugammadex reduced drug‐related adverse effects (RR 0.72, 95% CI 0.54 to 0.95; P = 0.02) but data show no differences in the rate of postoperative nausea or the rate of postoperative vomiting, Findings of this review were generally in line with the results of our updated review with regard to adverse and serious adverse events.

Another systematic review (Paton 2010), which included four trials (n = 606), compared sugammadex versus neostigmine/glycopyrrolate for routine reversal of NMB with economics evaluation. Researchers found that sugammadex was beneficial in terms of enhanced patient safety and increased predictability of recovery from rocuronium‐induced NMB, with more efficient use of theatre time and staff. Conclusions of review authors on recovery time, adverse events, and cost‐benefit considerations are in line with those of our updated review.

Figure 1

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Figure 4

Figure 5

Study flow diagram.

Figure 6

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Analysis 1.1

Comparison 1 Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg, Outcome 1 Recovery time from T2 to TOFR > 0.9.

Analysis 1.2

Comparison 1 Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg, Outcome 2 Subgroup analysis: TIVA vs volatile anaesthetics.

Analysis 1.3

Comparison 1 Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg, Outcome 3 Sensitivity analysis: meeting abstracts excluded.

Analysis 2.1

Comparison 2 Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg, Outcome 1 Recovery time from PTC 1 to 5 to TOFR > 0.9.

Analysis 3.1

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 1 Risk of composite adverse events.

Analysis 3.2

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 2 Composite adverse events: subgroup analysis for dosage.

Analysis 3.3

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 3 Composite adverse events: subgroup analysis ‐ TIVA vs volatile anaesthetics.

Analysis 3.4

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 4 Composite adverse events: sensitivity analysis ‐ excluding meeting abstracts.

Analysis 3.5

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 5 Participants with ≥ adverse event.

Analysis 3.6

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 6 Bradycardia: subgroup analysis ‐ atropine vs glycopyrrolate.

Analysis 3.7

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 7 PONV: subgroup analysis ‐ TIVA vs volatile anaesthetics.

Analysis 3.8

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 8 Desaturation.

Analysis 3.9

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 9 Procedural complications.

Analysis 3.10

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 10 Transitory oxygen supplementation.

Analysis 3.11

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 11 Not able to perform 5 second head‐lift after extubation.

Analysis 3.12

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 12 General muscle weakness after extubation.

Analysis 3.13

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 13 Nausea.

Analysis 3.14

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 14 Vomiting.

Analysis 3.15

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 15 Postprocedural nausea.

Analysis 3.16

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 16 Headache.

Analysis 3.17

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 17 Hypertension.

Analysis 3.18

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 18 Hypotension.

Analysis 3.19

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 19 Cough.

Analysis 3.20

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 20 Dry mouth.

Analysis 3.21

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 21 Dizziness.

Analysis 3.22

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 22 Tachycardia.

Analysis 3.23

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 23 Pruritus.

Analysis 3.24

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 24 Pyrexia.

Analysis 3.25

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 25 Shivering.

Analysis 3.26

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 26 Chills.

Analysis 3.27

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 27 Rash.

Analysis 3.28

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 28 Supraventricular extrasystoles.

Analysis 3.29

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 29 Laryngospasm.

Analysis 3.30

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 30 Increased upper airway secretion.

Analysis 3.31

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 31 Procedural hypertension.

Analysis 3.32

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 32 Procedural hypotension.

Analysis 3.33

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 33 Abdominal pain.

Analysis 3.34

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 34 Clinical signs of residual NMB.

Analysis 3.35

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 35 Clinical signs of inadequate reversal of NMB.

Analysis 3.36

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 36 Clinical signs of recurrence of residual NMB.

Analysis 3.37

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 37 General muscle weakness at PACU discharge.

Analysis 3.38

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 38 Not able to perform 5 second head‐lift at PACU discharge.

Analysis 3.39

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 39 Overall signs of postoperative residual paralysis.

Analysis 3.40

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 40 Risk of composite serious adverse events.

Analysis 3.41

Comparison 3 Sugammadex (any dose) vs neostigmine (any dose), Outcome 41 Participants with ≥ 1 serious adverse event.

Summary of findings for the main comparison. Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg

Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg
Patient or population: adult patients, ASA I to IV, who received non‐depolarizing NMBAs Setting: elective in‐patient or day‐case surgical procedures performed at centres across Europe and Asia Intervention: sugammadex 2.0 mg/kg Comparison: neostigmine 0.05 mg/kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
	Neostigmine 0.05 mg/kg	Sugammadex 2.0 mg/kg
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Mean recovery time from T2 to TOFR > 0.9 was 12.87 minutes	Mean recovery time from T2 to TOFR > 0.9 was 1.96 minutes Mean recovery time from T2 to TOFR > 0.9 in the sugammadex group was10.22 minutes faster (8.48 to 11.96 minutes faster) than neostigmine	‐	835 (10 studies)	⊕⊕⊕⊝^c Moderate	TSA alfa‐boundary adjusted MD is ‐10.22 (95% CI ‐12.11 to ‐8.33; diversity (D²) = 87%, I² = 84%, random‐effects model, 80% power, alpha 0.05). Cumulative Z‐curve crosses the monitoring boundary (Figure 1)
Recovery time^a from post‐tetanic count (PTC) 1 to 5 to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Outcome not clinically relevant for this comparison
Risks of adverse events and serious adverse events^b, bradycardia, PONV, and signs of residual neuromuscular blockade	Outcome not analysed for this comparison
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High quality:We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in post‐anaesthetic care unit, or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies)

Summary of findings for the main comparison. Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg

Summary of findings 2. Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg

Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg
Patient or population: adult patients, ASA I to IV, who received non‐depolarizing NMBAs Setting: elective in‐patient or day‐case surgical procedures performed in Italy and USA Intervention: sugammadex 4.0 mg/kg Comparison: neostigmine 0.07 mg/kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
	Neostigmine 0.07 mg/kg	Sugammadex 4.0 mg/kg
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Outcome not clinically relevant for this comparison.
Recovery time^a from post‐tetanic count (PTC 1 to 5) to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Mean recovery time from PTC 1 to 5 to TOFR > 0.9 was 48.8 minutes	Mean recovery time from PTC 1 to 5 to TOFR > 0.9 was 2.9 minutes Mean recovery time from PTC 1 to 5 to TOFR > 0.9 in the sugammadex group was 45.78 minutes faster (52.15 to 39.41 minutes faster) than in the neostigmine group	‐	114 (2 studies)	⊕⊕⊝⊝^c Low
Risk of adverse events and serious adverse events^b, bradycardia, PONV, and signs of residual neuromuscular blockade	Outcome not analysed for this comparison
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in the post‐anaesthetic care unit, or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of participants, n = 114)

Summary of findings 2. Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg

Summary of findings 3. Sugammadex (any dose) vs neostigmine (any dose)

Sugammadex (any dose) compared to Neostigmine (any dose)
Patient or population: Adult patients, ASA I‐IV, who received non‐depolarizing NMBAs Setting: Elective in‐patient or day‐case surgical procedures performed in centres across Europe, USA and Asia Intervention: Sugammadex (any dose) Comparison: Neostigmine (any dose)
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
Outcomes	Risk with neostigmine (any dose)	Risk with sugammadex (any dose)	Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
Recovery time^a from second twitch (T2) to train‐of‐four ratio (TOFR) > 0.9 (moderate block)	Outcome not clinically relevant for this comparison
Recovery time^a from post‐tetanic count (PTC) 1 to 5 to train‐of‐four ratio (TOFR) > 0.9 (deep block)	Outcome not clinically relevant for this comparison
Risk of composite adverse events^b	283 per 1000	159 per 1000 (137 to 204)	RR 0.60 (0.49 to 0.74)	2298 (28 studies)	⊕⊕⊕⊝^c Moderate	TSA with continuity adjustment for zero event trials (0.001 in each arm); alfa‐boundary adjusted RR 0.62 (95% CI 0.51 to 0.74; diversity (D²) = 34%, I² = 14%, random‐effects model; 80% power, 0.05 alpha; Figure 3)
Bradycardia	84 per 1000	13 per 1000 (6 to 28)	RR 0.16 (0.07 to 0.34)	1218 (11 studies)	⊕⊕⊕⊝^d Moderate
PONV	131 per 1000	68 per 1000 (33 to 115)	RR 0.52 (0.28 to 0.97)	389 (6 studies)	⊕⊕⊝⊝^e Low
Overall signs of postoperative residual paralysis	131 per 1000	52 per 1000 (37 to 75)	RR 0.40 (0.28 to 0.57)	1474 (15 studies)	⊕⊕⊕⊝^f Moderate	TSA with continuity adjustment for zero event trials (0.001 in each arm): alfa‐boundary adjusted RR 0.4 (95% CI 0.27 to 0.59; diversity (D²) = 0%, I² = 0%, random‐effects model, 80% power, 0.05 alpha, Figure 4). Cumulative Z‐curve crosses the monitoring boundary constructed for a required information size of 424 participants indicating firm evidence in favour of sugammadex
Risk of serious adverse events^b	10 per 1000	6 per 1000 (1 to 23)	RR 0.54 (0.13 to 2.25)	959 (10 studies)	⊕⊕⊝⊝^g Low	TSA with continuity adjustment for zero event trials (0.001 in each arm): alfa‐boundary adjusted RR 0.35 (95% CI 0.00 to 3190; diversity (D²) = 0%, I² = 0%, random‐effects model, 80% power, alpha 0.05), Cumulative Z‐curve does not cross the monitoring boundary constructed for a required information size of 8189 participants with 11.71% of the required information size included
*The risk in the intervention group (and its 95% confidence interval) is based on assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval; OR: odds ratio; RR: risk ratio;
GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aRecovery time was measured in minutes from administration of study drug to TOFR > 0.9 by TOF‐watch assessor using acceleromyography at the same monitoring site in all studies (ulnar nerve and adductor pollicis muscle) ^bAdverse events and serious adverse events were defined by study authors and were observed and assessed by safety outcome assessors in the operating theatre, in the post‐anaesthetic care unit or up to seven days after surgery, depending on each study. Furthermore, overall clinical signs of postoperative residual paralysis reported by trials were regarded as adverse events in this review. Risk of adverse events was measured as number of adverse events per all participants and/or number of participants experiencing one or more adverse events per all participants, depending on the study. Only adverse events that were possibly, probably, or definitely related to study drug were included in risk assessments ^cDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^dDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^eDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of participants‐ n = 389 ‐ and wide confidence interval (CI) ‐ 0.28 to 0.97) ^fDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) ^gDowngraded one level owing to high risk of bias (evidence limited by inclusion of data from open‐label studies and studies with potential funding bias ‐ for details, see Figure 2 and Characteristics of included studies) and by one level owing to imprecision (small number of events ‐ 10/1000 in the neostigmine group vs 6/1000 in the sugammadex group ‐ and wide confidence interval (CI) ‐ 0.13 to 2.25)

Summary of findings 3. Sugammadex (any dose) vs neostigmine (any dose)

Table 1. Table of studies ineligible for meta‐analysis

Study ID	Reasons for ineligibility	Comparisons	Conclusions
Isik 2016	Primary endpoint: acute effects of sugammadex and neostigmine on renal function	Sugammadex 4 mg /kg at reappearance of PTC 1 to 2 or T2 vs neostigmine 40 µg/kg + atropine 10 µg/kg at reappearance of T2	We believe that the use of more specific and sensitive new‐generation markers such as Cystatin C to evaluate kidney function will provide better understanding and interpretation of our results. Sugammadex has more tolerable effects on kidney function than does neostigmine. However, when compared with preoperative values, negative alteration of postoperative values can be seen. Neostigmine and sugammadex do not cause renal failure but may affect kidney function
Kvolik 2012a	TOFR recovery data available only as mean, no data on standard deviation, study author has not replied	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg	Recovery of cough reflexes was faster and respiration more efficient in patients receiving sugammadex. Safe extubation was determined by age, TOFR recovery, and effects of other anaesthetics
Kvolik 2013	TOFR recovery data available only as mean, no data on standard deviation, study author has not replied	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg + atropine 25 µg/kg	An increase in BIS Index registered after reversal of rocuronium effects was faster during the recovery period in patients who were given sugammadex as compared with neostigmine. Although rapid increase in BIS Indices was registered in sugammadex group, more sensitive measurements are needed to confirm clinical value of this observation
Martini 2014	Primary endpoint: influence of depth of the NMB on SRS (surgical rating score)	Neostigmine 1 to 2 mg + atropine 0.5 to 1 mg (for reversal of moderate NMB) vs sugammadex 4 mg/kg (for reversal of deep NMB)	Application of 5‐point SRS showed that deep NMB results in improved quality of surgical conditions compared with moderate block in retroperitoneal laparoscopy, without compromise to patients’ perioperative and postoperative cardiorespiratory conditions
Rahe‐Meyer 2014	Comparison: sugammadex 4 mg/kg vs usual care (neostigmine with glycopyrrolate or atropine, or placebo/spontaneous recovery). Study author has not replied with separate data on neostigmine with glycopyrrolate or atropine or placebo/spontaneous recovery.	Sugammadex 4 mg/kg vs usual care (neostigmine with glycopyrrolate or atropine, or placebo/spontaneous recovery)	Sugammadex produced limited, transient (< 1 hour) increases in activated partial thromboplastin time and prothrombin time but was not associated with increased risk of bleeding vs usual care
Raziel 2013	No useable data available for quantitative meta‐analysis on recovery time or risk of adverse events	Sugammadex 2 mg/kg vs neostigmine 50 µg/kg + atropine 10 µg/kg	Sugammadex facilitates reversal of neuromuscular blockade after bariatric surgery, depending on the depth of neuromuscular blockade induced
Riga 2014	Primary outcome: cognitive function assessed by change in Mini‐Mental State Evaluation test (MMSE), Clock Drawing Test, and Isaacs Set Test, performed preoperatively, 1 hour postoperatively, and at discharge (1 to 15 days postoperatively)	Sugammadex vs neostigmine/atropine	No significant difference was observed regarding cognitive function after neostigmine/atropine combination or sugammadex was received for reversal of rocuronium‐induced neuromuscular blockade for elective surgery
Sherman 2014	Primary outcome: postoperative complications, data not available in useful format	Sugammadex 2 mg/kg vs neostigmine 2.5 mg/kg	Use of sugammadex (compared with neostigmine) as reversal agent following laparoscopic sleeve gastrectomy; surgery was associated with higher postoperative oxygen saturation despite lower TOF count before administration of reversal agent. Lack of differences in other measured variables may stem from the small size of patient groups studied
Sustic 2012	Outcome: gastric emptying evaluated by paracetamol absorption test	Sugammadex 2 mg/kg vs neostigmine 40 µg/kg + atropine group 15 µg/kg	Although study results show a tendency toward faster gastric emptying in sugammadex group, this difference is not significant in most, possibly owing to small sample size in this study
Tas 2015	Aim: to evaluate effects of sugammadex on postoperative nausea‐vomiting, pain, coagulation parameters, and quantity of postoperative bleeding. Data not available in useful format	Neostigmine 0.05 mg/kg + atropine 0.02 mg/kg vs sugammadex 2 mg/kg	Sugammadex was associated with greater postoperative bleeding than neostigmine in septoplasty patients. For surgical procedures with high risk of bleeding, the safety of sugammadex needs to be verified
Acronyms: BIS ‐ Bispectral Index MMSE ‐ Mini‐Mental State Examination NMB ‐ neuromuscular blockade T2 ‐ second twitch in train‐of‐four stimulation TOFR ‐ train‐of‐four ratio PTC ‐ post‐tetanic count SRS ‐ surgical rating score

Table 1. Table of studies ineligible for meta‐analysis

Table 2. Quality variables of neuromuscular monitoring methods among included trials

Study ID	Method of recording	Monitor site	Arm fixation	Supramaximal stimulation	Temperature maintained and recorded	Initial signal stabilization	Twich height calibration	Preload used
Adamus 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Yes	Not mentioned
Blobner 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Yes	Not mentioned
Brueckmann 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Yes	Not mentioned
Carron 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Not mentioned	Not mentioned	Yes	Yes	No
Castro 2014	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Cheong 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Not mentioned	Not mentioned	Not mentioned
Flockton 2008	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Gaszynski 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Not mentioned	Not mentioned	Not mentioned
Geldner 2012	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Yes	Yes	Not mentioned
Hakimoglu 2016	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Illman 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes		Yes	No
Isik 2016	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Not mentioned	Yes	Not mentioned	Not mentioned	Not mentioned
Jones 2008	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Kaufhold 2016	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Not mentioned
Khuenl‐Brady 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Yes	Yes	Not mentioned
Kizilay 2016	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Koc 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Yes	Not mentioned	Yes	Not mentioned
Koyuncu 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	No	Yes	No	No	Yes	Not mentioned
Lemmens 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Yes	Yes	Yes	Not mentioned
Martini 2014	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Not mentioned	Yes	Yes	Yes
Mekawy 2012	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Yes	Not mentioned
Pongracz 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Yes
Sabo 2011	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Schaller 2010	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes	Yes	Yes	Yes	Yes	Not mentioned
Tas 2015	Acceleromyography	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Woo 2013	Acceleromyography	N. ulnaris, M. adductor pollicis	Yes *	No *	Yes *	Yes	Yes	No *
Wu 2014	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Yes	Not mentioned	Yes	Yes	Not mentioned
Yagan 2015	Acceleromyography	N. ulnaris, M. adductor pollicis	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned	Not mentioned
Studies with only abstracts were not included in this table because they did not document information regarding neuromuscular monitoring List of abbreviations: N. ulnaris ‐ ulnar nerve M. adductor pollicis ‐ adductor pollicis muscle

Table 2. Quality variables of neuromuscular monitoring methods among included trials

Table 3. Table of adverse events

	Sugammadex			Neostigmine
Specific adverse events	Number of AEs	Number of participants	Risk of AEs, %	Number of AEs	Number of participants	Risk of AEs, %	RR (95% CI)	Number of studies	Total number of participants
Cough	20	100	20,0	14	100	14,0	1.42 (0.42‐4.81)	3	200
Shivering	13	91	14,3	19	99	19,2	0.75 (0.40‐1.43)	3	190
Desaturation	2	63	3,2	11	71	15,5	0.23 (0.06‐0.83)	2	134
General muscle weakness after extubation	13	142	9,2	22	146	15,1	0.61 (0.31‐1.18)	4	288
Breath‐hold	3	30	10,0	4	30	13,3	‐	1	60
PONV	13	206	6,3	24	183	13,1	0.52 (0.28‐0.97)	6	389
Laryngospasm	2	50	4,0	6	50	12,0	0.34 (0.07‐1.65)	2	100
Not able to perform 5 second head‐lift after extubation	7	193	3,6	23	202	11,4	0.34 (0.15‐0.78)	6	395
Bradycardia	4	621	0,6	50	597	8,4	0.16 (0.07‐0.34)	11	1218
Procedural complications	0	85	0,0	7	83	8,4	0.12 (0.02‐0.97)	2	168
Postprocedural nausea	8	128	6,3	5	122	4,1	1.34 (0.47‐3.81)	3	250
Dry mouth	3	146	2,1	9	143	6,3	0.44 (0.10‐1.87)	3	289
Headache	12	195	6,2	11	193	5,7	1.02 (0.48‐2.18)	4	388
Increased beta‐N‐acetyl‐D‐glucosaminidase	2	34	5,9	0	39	0,0	‐	1	73
Strange taste in mouth	2	35	5,7	0	35	0,0	‐	1	70
Nausea	17	364	4,7	20	355	5,6	0.83 (0.44‐1.56)	9	719
Leukocytosis	1	46	2,2	2	36	5,6	‐	1	82
Albumin present in the urine	0	48	0,0	2	48	4,2	‐	1	96
Vomiting	6	149	4,0	2	148	1,4	2.05 (0.50‐8.48)	4	297
Bronchospasm	2	50	4,0	1	50	2,0	‐	1	100
Chills	3	82	3,7	0	84	0,0	4.04 (0.46‐35.85)	2	166
General muscle weakness at PACU discharge	3	208	1,4	6	202	3,0	0.49 (0.12‐1.90)	5	410
Procedural hypertension	4	133	3,0	2	134	1,5	1.65 (0.33‐8.21)	3	267
Tremor	1	34	2,9	0	39	0,0	‐	1	73
Altered facial sensation	1	34	2,9	0	39	0,0	‐	1	73
Postprocedural hypertension	1	37	2,7	0	38	0,0	‐	1	75
Paraesthesia	1	37	2,7	0	38	0,0	‐	1	75
Increased blood PK	1	37	2,7	0	38	0,0	‐	1	75
Increased upper airway secretions	2	223	0,9	6	219	2,7	0.37 (0.09‐1.59)	2	442
Hyperhidrosis	0	37	0	1	38	2,6	‐	1	75
Decreased blood protein	0	37	0	1	38	2,6	‐	1	75
Restlessness	0	37	0	1	38	2,6	‐	1	75
Chest discomfort	0	37	0	1	38	2,6	‐	1	75
Incision site complication	0	37	0	1	38	2,6	‐	1	75
Procedural hypotension	1	200	0,5	5	191	2,6	0.49 (0.02‐14.1)	2	391
Postprocedural complication	0	37	0,0	1	38	2,6	‐	1	75
Tachycardia	1	165	0,6	4	173	2,3	0.44 (0.09‐2.22)	3	338
Pruritus	2	87	2,3	1	88	1,1	1.62 (0.20‐12.88)	2	175
Intraoperative movement	1	43	2,3	1	51	2,0	‐	1	94
Anxiety	0	46	0	1	46	2,2	‐	1	92
Depression	0	46	0	1	46	2,2	‐	1	92
Fatigue	0	46	0	1	46	2,2	‐	1	92
Hypotension	5	227	2,2	5	238	2,1	1.23 (0.38‐3.96)	4	465
Supraventricular extrasystoles	0	96	0,0	2	93	2,2	0.32 (0.03‐3.05)	2	189
Clinical signs of inadequate reversal of NMB	0	188	0,0	4	180	2,2	0.11 (0.01‐2.02)	4	368
Leukocytosis	1	46	2,2	0	36	0,0	‐	1	82
Ventricular extrasystoles	0	48	0,0	1	45	2,2	‐	1	93
Sleep disorder	0	48	0,0	1	45	2,2	‐	1	93
Increased gamma‐glutamyl‐transferase	0	48	0,0	1	45	2,2	‐	1	93
Retching	1	48	2,1	0	45	0,0	‐	1	93
Airway complication to anaesthesia	1	48	2,1	0	45	0,0	‐	1	93
Hot flush	1	48	2,1	0	45	0,0	‐	1	93
Abdominal pain	1	48	2,1	0	48	0,0	‐	1	96
Severe abdominal pain	1	48	2,1	0	48	0,0	‐	1	96
Pharyngolaryngeal pain	1	48	2,1	0	48	0,0	‐	1	96
Diarrhoea	1	48	2,1	0	48	0,0	‐	1	96
Tinnitus	1	48	2,1	0	48	0,0	‐	1	96
Involontary muscle contractions	0	48	0,0	1	48	2,1	‐	1	96
Visual accomodation disorder	0	48	0,0	1	48	2,1	‐	1	96
Increased B₂‐microglobulin	0	48	0,0	1	48	2,1	‐	1	96
Severe bradycardia	0	48	0,0	1	48	2,1	‐	1	96
Productive cough	0	48	0,0	1	48	2,1	‐	1	96
Pyrexia	2	133	1,5	1	131	0,8	1.43 (0.23‐8.91)	3	264
Hypertension	2	143	1,4	1	144	0,7	1.45 (0.23‐9.05)	3	287
Decreased hematocrit	1	74	1,4	0	77	0,0	‐	1	151
Procedural haemorrhage	1	74	1,4	0	77	0,0	‐	1	151
Delayed recovery from anaesthesia	0	74	0,0	1	77	1,3	‐	1	151
Respiratory distress	0	74	0,0	1	77	1,3	‐	1	151
Dizziness	1	85	1,2	1	83	1,2	0.98 (0.10‐9.23)	2	168
Abdominal pain	1	99	1,0	1	97	1,0	0.98 (0.10‐9.27)	2	196
Rash	2	355	0,6	3	346	0,9	0.83 (0.17‐3.96)	5	701
Severe muscle weakness	0	149	0,0	1	142	0,7	‐	1	291
Mild hypoventilation	1	149	0,7	0	142	0,0	‐	1	291
Clinical signs of recurrence of residual NMB	1	674	0,1	2	615	0,3	0.74 (0.05‐10.7)	13	1289
Clinical signs of residual NMB	0	341	0,0	0	305	0,0	‐	7	646
Not able to perform 5 second head‐lift at PACU discharge	0	205	0,0	0	194	0,0	‐	5	399
Redness at injection site	0	50	0,0	0	50	0,0	‐	1	100
Hypersensitivity	0	60	0,0	0	30	0,0	‐	1	90
Table of reported adverse events possibly, probably, or definitely related to sugammadex or neostigmine, listed in descending order according to risk of adverse events. Furthermore, the number of studies observing for each adverse event is presented List of abbreviations: NMB ‐ neuromuscular blockade PACU ‐ post‐anaesthesia care unit

Table 3. Table of adverse events

Comparison 1. Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Recovery time from T2 to TOFR > 0.9 Show forest plot	10	835	Mean Difference (IV, Random, 95% CI)	‐10.22 [‐11.96, ‐8.48]

2 Subgroup analysis: TIVA vs volatile anaesthetics Show forest plot	11	871	Mean Difference (IV, Random, 95% CI)	‐9.83 [‐11.45, ‐8.20]

2.1 TIVA	3	381	Mean Difference (IV, Random, 95% CI)	‐8.50 [‐10.15, ‐6.86]
2.2 Volitile anaesthetics	8	490	Mean Difference (IV, Random, 95% CI)	‐10.57 [‐12.96, ‐8.18]
3 Sensitivity analysis: meeting abstracts excluded Show forest plot	9	767	Mean Difference (IV, Random, 95% CI)	‐9.27 [‐11.14, ‐7.40]

Comparison 1. Sugammadex 2.0 mg/kg vs neostigmine 0.05 mg/kg

Comparison 2. Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Recovery time from PTC 1 to 5 to TOFR > 0.9 Show forest plot	2	114	Mean Difference (IV, Random, 95% CI)	‐45.78 [‐52.15, ‐39.41]

Comparison 2. Sugammadex 4.0 mg/kg vs neostigmine 0.07 mg/kg

Comparison 3. Sugammadex (any dose) vs neostigmine (any dose)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Risk of composite adverse events Show forest plot	28	2298	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.49, 0.74]

2 Composite adverse events: subgroup analysis for dosage Show forest plot	28	2298	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.49, 0.73]

2.1 Sugammadex 2 mg/kg vs neostigmine 0.04 mg/kg	1	21	Risk Ratio (M‐H, Random, 95% CI)	0.45 [0.05, 4.28]
2.2 Sugammadex 2 mg/kg vs neostigmine 0.05 mg/kg	12	1076	Risk Ratio (M‐H, Random, 95% CI)	0.52 [0.34, 0.80]
2.3 Sugammadex 2 mg/kg vs neostigmine 0.07 mg/kg	2	131	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.57, 1.44]
2.4 Sugammadex 2 mg/kg vs neostigmine 2.5 mg	1	40	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2.5 Sugammadex 3 mg/kg vs neostigmine 0.03 mg/kg	1	90	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2.6 Sugammadex 4 mg/kg vs neostigmine 0.05 mg/kg	4	333	Risk Ratio (M‐H, Random, 95% CI)	0.66 [0.49, 0.88]
2.7 Sugammadex 4 mg/kg vs neostigmine 0.07 mg/kg	3	197	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.25, 0.93]
2.8 Sugammadex, several doses vs neostigmine, several doses	4	410	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.40, 0.90]
3 Composite adverse events: subgroup analysis ‐ TIVA vs volatile anaesthetics Show forest plot	28	2298	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.49, 0.73]

3.1 TIVA	7	748	Risk Ratio (M‐H, Random, 95% CI)	0.51 [0.20, 1.31]
3.2 Volatile anaesthetic	20	1510	Risk Ratio (M‐H, Random, 95% CI)	0.64 [0.55, 0.73]
3.3 No information	1	40	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
4 Composite adverse events: sensitivity analysis ‐ excluding meeting abstracts Show forest plot	24	2091	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.49, 0.73]

5 Participants with ≥ adverse event Show forest plot	19	1766	Risk Ratio (M‐H, Random, 95% CI)	0.62 [0.48, 0.81]

6 Bradycardia: subgroup analysis ‐ atropine vs glycopyrrolate Show forest plot	11	1218	Risk Ratio (M‐H, Random, 95% CI)	0.16 [0.07, 0.34]

6.1 Atropine	6	667	Risk Ratio (M‐H, Random, 95% CI)	0.14 [0.05, 0.36]
6.2 Glycopyrrolate	5	551	Risk Ratio (M‐H, Random, 95% CI)	0.20 [0.06, 0.69]
7 PONV: subgroup analysis ‐ TIVA vs volatile anaesthetics Show forest plot	6	389	Risk Ratio (M‐H, Random, 95% CI)	0.52 [0.28, 0.97]

7.1 TIVA	1	94	Risk Ratio (M‐H, Random, 95% CI)	3.55 [0.15, 84.86]
7.2 Volatile anaesthetics	5	295	Risk Ratio (M‐H, Random, 95% CI)	0.48 [0.25, 0.91]
8 Desaturation Show forest plot	2	134	Risk Ratio (M‐H, Random, 95% CI)	0.23 [0.06, 0.83]

9 Procedural complications Show forest plot	2	168	Risk Ratio (M‐H, Random, 95% CI)	0.12 [0.02, 0.97]

10 Transitory oxygen supplementation Show forest plot	2	76	Risk Ratio (M‐H, Random, 95% CI)	0.24 [0.09, 0.66]

11 Not able to perform 5 second head‐lift after extubation Show forest plot	6	395	Risk Ratio (M‐H, Random, 95% CI)	0.34 [0.15, 0.78]

12 General muscle weakness after extubation Show forest plot	4	288	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.31, 1.18]

13 Nausea Show forest plot	9	719	Risk Ratio (M‐H, Random, 95% CI)	0.83 [0.44, 1.56]

14 Vomiting Show forest plot	4	297	Risk Ratio (M‐H, Random, 95% CI)	2.05 [0.50, 8.48]

15 Postprocedural nausea Show forest plot	2	168	Risk Ratio (M‐H, Random, 95% CI)	1.39 [0.27, 7.12]

16 Headache Show forest plot	4	388	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.48, 2.18]

17 Hypertension Show forest plot	3	287	Risk Ratio (M‐H, Random, 95% CI)	1.45 [0.23, 9.05]

18 Hypotension Show forest plot	4	465	Risk Ratio (M‐H, Random, 95% CI)	1.23 [0.38, 3.96]

19 Cough Show forest plot	3	200	Risk Ratio (M‐H, Random, 95% CI)	1.42 [0.42, 4.81]

20 Dry mouth Show forest plot	3	289	Risk Ratio (M‐H, Random, 95% CI)	0.44 [0.10, 1.87]

21 Dizziness Show forest plot	2	168	Risk Ratio (M‐H, Random, 95% CI)	0.98 [0.10, 9.23]

22 Tachycardia Show forest plot	3	338	Risk Ratio (M‐H, Random, 95% CI)	0.44 [0.09, 2.22]

23 Pruritus Show forest plot	2	175	Risk Ratio (M‐H, Random, 95% CI)	1.62 [0.20, 12.88]

24 Pyrexia Show forest plot	3	264	Risk Ratio (M‐H, Random, 95% CI)	1.43 [0.23, 8.91]

25 Shivering Show forest plot	3	190	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.40, 1.43]

26 Chills Show forest plot	2	166	Risk Ratio (M‐H, Random, 95% CI)	4.04 [0.46, 35.85]

27 Rash Show forest plot	5	701	Risk Ratio (M‐H, Random, 95% CI)	0.83 [0.17, 3.96]

28 Supraventricular extrasystoles Show forest plot	2	189	Risk Ratio (M‐H, Random, 95% CI)	0.32 [0.03, 3.05]

29 Laryngospasm Show forest plot	2	100	Risk Ratio (M‐H, Random, 95% CI)	0.34 [0.07, 1.65]

30 Increased upper airway secretion Show forest plot	2	442	Risk Ratio (M‐H, Random, 95% CI)	0.37 [0.09, 1.59]

31 Procedural hypertension Show forest plot	3	267	Risk Ratio (M‐H, Random, 95% CI)	1.65 [0.33, 8.21]

32 Procedural hypotension Show forest plot	2	391	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.02, 14.15]

33 Abdominal pain Show forest plot	2	196	Risk Ratio (M‐H, Random, 95% CI)	0.98 [0.10, 9.27]

34 Clinical signs of residual NMB Show forest plot	7	646	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

35 Clinical signs of inadequate reversal of NMB Show forest plot	4	368	Risk Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 2.02]

36 Clinical signs of recurrence of residual NMB Show forest plot	13	1289	Risk Ratio (M‐H, Random, 95% CI)	0.74 [0.05, 10.74]

37 General muscle weakness at PACU discharge Show forest plot	5	410	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.12, 1.90]

38 Not able to perform 5 second head‐lift at PACU discharge Show forest plot	5	399	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

39 Overall signs of postoperative residual paralysis Show forest plot	15	1474	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.28, 0.57]

40 Risk of composite serious adverse events Show forest plot	10	959	Risk Ratio (M‐H, Random, 95% CI)	0.54 [0.13, 2.25]

41 Participants with ≥ 1 serious adverse event Show forest plot	10	959	Risk Ratio (M‐H, Random, 95% CI)	0.54 [0.13, 2.25]

Comparison 3. Sugammadex (any dose) vs neostigmine (any dose)