Reducing the dose of neuromuscular blocking agents with adjuncts: a systematic review and meta-analysis

Abstract

Background

Acute global shortages of neuromuscular blocking agents (NMBA) threaten to impact adversely on perioperative and critical care. The use of pharmacological adjuncts may reduce NMBA dose. However, the magnitude of any putative effects remains unclear.

Methods

We conducted a systematic review and meta-analysis of RCTs. We searched Medline, Embase, Web of Science, and Cochrane Database (1970–2020) for RCTs comparing use of pharmacological adjuncts for NMBAs. We excluded RCTs not reporting perioperative NMBA dose. The primary outcome was total NMBA dose used to achieve a clinically acceptable depth of neuromuscular block. We assessed the quality of evidence using the GRADE (Grades of Recommendation, Assessment, Development, and Evaluation) criteria. Data are presented as the standardised mean difference (SMD); I² indicates percentage of variance attributable to heterogeneity.

Results

From 3082 records, the full texts of 159 trials were retrieved. Thirty-one perioperative RCTs met the inclusion criteria for meta-analysis (n=1962). No studies were conducted in critically ill patients. Reduction in NMBA dose was associated with use of magnesium (SMD: –1.10 [–1.44 to –0.76], P<0.001; I²=85%; GRADE=moderate), dexmedetomidine (SMD: –0.89 [–1.55 to –0.22]; P=0.009; I²=87%; GRADE=low), and clonidine (SMD: –0.67 [–1.13 to –0.22]; P=0.004; I²=0%; GRADE=low) but not lidocaine (SMD: –0.46 [–1.01 to –0.09]; P=0.10; I²=68%; GRADE=moderate). Meta-analyses for nicardipine, diltiazem, and dexamethasone were not possible owing to the low numbers of studies. We estimated that 30–50 mg kg⁻¹ magnesium preoperatively (8–15 mg kg h⁻¹ intraoperatively) reduces rocuronium dose by 25.5% (inter-quartile range, 14.7–31).

Conclusions

Magnesium, dexmedetomidine, and clonidine may confer a clinically relevant sparing effect on the required dose of neuromuscular block ing drugs in the perioperative setting.

Systematic review registration

PROSPERO: CRD42020183969.

Editor's key points.

• •Shortages of drugs used in the perioperative period (including neuromuscular blocking agents) are increasingly frequent. Strategies to reduce their usage may alleviate potential supply issues, including the use of pharmacological adjuncts to reduce the dose required for effective neuromuscular block.
• •This systematic review and meta-analysis of 31 RCTs suggests that magnesium, dexmedetomidine, and clonidine reduce the total dose of neuromuscular blocking agents required to maintain satisfactory neuromuscular block without compromising safety and recovery.

Although the number of surgical procedures continues to increase every year,^1,2 global shortages of drugs in critical care³ and perioperative medicine⁴ are becoming increasingly frequent. Inadequate drug supplies burden health systems with additional costs and may increase adverse events in patients.⁵ In combination with the novel coronavirus disease 2019 (COVID-19) pandemic, unprecedented manufacturing pressures have impacted on drug supplies, including neuromuscular blocking agents (NMBAs), worldwide.⁶

Emergent shortages of non-depolarising NMBA may be mitigated by the use of pharmacological adjuncts that reduce NMBA dosing in the perioperative and critical care settings. Several adjuncts may reduce the dose of non-depolarising NMBA. High plasma magnesium concentrations inhibit the release of acetylcholine from the presynaptic nerve terminal and enhance the effect of NMBA.7, 8, 9 Calcium channel blockers¹⁰ and local anaesthetic agents^11,12 may prolong neuromuscular block by impairing presynaptic release of acetylcholine and hence potentiate NMBAs. Alpha2-adrenergic agonists potentiate NMBA action by increasing NMBA plasma concentration.^13,14 In addition, the complex effect of steroids on the physiology of the neuromuscular junction suggests that commonly administered glucocorticoid drugs may also impact on the dosing of nondepolarising NMBAs (and/or reversal agents).15, 16, 17

We therefore performed a systematic review and meta-analysis of the published literature to explore whether pharmacological adjuncts may usefully reduce the total amount of NMBAs required in both perioperative and critical care settings.

Methods

Protocol and registration

We registered the systematic review prospectively with PROSPERO: CRD42020183969. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines¹⁸ and the guidelines from the Cochrane Handbook for Systematic Reviews of interventions¹⁹ for this review. Ethical approval was not required for this study.

Inclusion criteria

This study was open to all adult patients requiring neuromuscular block in the perioperative setting, critical care setting, or both. The intervention was the use of pharmacological adjuncts that may alter depth of neuromuscular block, compared with the administration of a control.

Exclusion criteria

The following studies were excluded: (1) studies with no full English text available or accessible; (2) studies conducted in vitro or in non-human subjects; (3) preclinical studies or studies in which pharmacological adjuncts were tested with another intervention; (4) studies that did not provide information about total perioperative NMBA requirements; and (5) studies in which doses for both intervention and control groups were not reported.

Information sources and search

We conducted a systematic, computerised search on Medline, Embase, Web of Science, and the Cochrane Database of Controlled Trials (CENTRAL) from January 1, 1970 to August 15, 2020. The last search was conducted on August 15, 2020 using different combinations of the following terms: ‘magnesium’, ‘magnesium sulfate’, ‘magnesium sulphate’ ‘nicardipine’, ‘nimodipine’ ‘calcium antagonist’, ‘dexmedetomidine’, ‘clonidine’, ‘lidocaine’, ‘dexamethasone’, ‘corticosteroid’, ‘adjuvant’, ‘rocuronium’, ‘cisatracurium’, ‘atracurium’, ‘vecuronium’, ‘mivacurium’, ‘suxamethonium’, ‘neuromuscular blockade’, ‘neuromuscular’, ‘anaesth∗’ ‘anesth∗’ ‘perioperative’ ‘general anaesth∗’, ‘general anesth∗’, ‘requirement’, ‘induction’, ‘surg∗’ (Supplementary material S1). Results were combined by the Boolean operator ‘AND’ or ‘OR’ with search terms. The formula used for our Medline database search is provided below as an example:

• -((magnesium OR magnesium sulphate OR magnesium sulfate OR nicardipine OR nimodipine OR dexmedetomidine OR clonidine OR lidocaine OR dexamethasone OR corticosteroid OR adjuvant) AND (rocuronium OR cisatracurium OR atracurium OR vecuronium OR mivacurium OR suxamethonium OR neuromuscular blockade) AND (anaesth∗ OR anesth∗ OR perioperative OR general anaesth∗ OR general anesth∗ OR requirement OR induction OR surg∗)) AND (("1970/01/01"[Date - MeSH]: "2020"[Date - MeSH]))

The search was completed by all authors and the results were compared. No search filters or language restrictions were applied. We extracted records to Endnote (Thomson Reuters, Philadelphia, PA, USA) to sort and remove duplicates.

Study selection

Studies were selected for inclusion by two authors (GLA, VW) acting independently. All studies were screened based on title and abstract, followed by full-text review of sub-set articles to determine eligibility. We screened the bibliographies of included studies and consulted experts to identify studies that were missed by the search. Differences in opinion were resolved through discussion and referral to a third investigator (TA). The final list of selected studies was derived by consensus (Table 1). RCTs in which the doses for both the adjuvant drug and the NMBA used during the perioperative period were reported against control were considered for meta-analysis.

Table 1.

Main characteristics of the studies included in the review. ASA, ASA physical status classification system; CABG, coronary artery bypass graft; Clon, clonidine; Dex, dexmedetomidine; Dexa, dexamethasone; Dilt, diltiazem; Ket, ketamine; Lido, lidocaine; Mag, magnesium; Mid, midazolam; N, nicardipine; NMBA, neuromuscular blocking agent; NR, not reported; NS, normal saline; VATS, video-assisted thoracoscopic surgery.

Author, year	ASA	Procedure	Sample size (groups)	Sex (M/F)	Mean (sd)		NMBA	Adjunct(s)	Loading dose (unit)	Maintenance dose (unit)	Comments
					Weight (kg)	Operation time (min)
Memis and colleagues, 200814	1/2	Various elective operations	20 (Dex) 20 (Control: NS)	11/9 14/6	Dex group: 77.8 (13.6) Control: 72.6 (10.5)	NR	ROC	DEX	1 μg kg−1	0.2 μg kg−1 h−1	–
Han and colleagues, 202021	1/2	Robot assisted laparoscopic prostatectomy	34 (Mag) 34 (Control: NS)	NR NR	Mag group: 70.2 (7.0) Control group: 70.0 (7.2)	173 (NR) 160 (NR)	ROC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Bala and colleagues, 201922	1/2	Transsphenoidal Pituitary Surgery	30 (Dex) 30 (Control: NS)	16/14 17/13	Dex group: 67.7 (12.5) Control: 71.2 (14.8)	114.6 (43.6) 130.5 (46.6)	ROC	DEX	1 μg kg−1	0.5 μg kg−1 h−1	–
Ren and colleagues, 201923	1/2/3	Aneurysm embolisation	38 (Dex1) 36 (Dex2) 35 (Control: NS)	19.19 21/15 17/18	Dex1: 67.55 (5.73) Dex2: 66.81 (7.86) Control: 67.29 (6.39)	110.00 (NR) 110.50 (NR) 113.00 (NR)	CIS	DEX	0.5 μg kg−1 0.5 μg kg−1	0.2 μg kg−1 h−1 0.4 μg kg−1 h−1	Three groups were included: saline, Dexmedetomidine 1 and Dexmedetomidine 2. Data from Dex1 group were compared with control.
Walia and colleagues, 201824	1/2	Various elective surgeries (not specified)	40 (Mag) 40 (Dex) 40 (Control: NS)	20/20 12/28 17/23	Mag group: 67.45 (9.05) Dex group: 69.75 (8.48) Control: 71.25 (8.86)	102.5 (24.04) 103.25 (29.47) 103.75 (25.78)	VEC	MAG DEX	30 mg kg−1 0.001 mg kg−1	None	Three groups were included: saline, Mag, and Dexmedetomidine. Only data from the Mag group were compared with control.
Reena and colleagues, 201725	1/2	Elective lumbar spine surgery (not specified)	29 (Mag) 28 (Clon) 30 (Control: NS)	21/9 21/9 19/11	Mag group: 57.6 (6.49) Clon group: 57.9 (6.74) Control: 56.93 (5.29)	134.5 (17.09) 127.33 (16.71) 130.53 (15.36)	VEC	MAG CLON	30 mg kg−1 2 μg kg−1	10 mg kg−1 h−1 1 μg kg−1 h−1	Three groups were included: saline, Mag, and Clonidine. Data from all groups were compared with control.
Sohn and colleagues, 201726	1/2/3	VATS pulmonary lobectomy or segmentectomy	29 (Mag) 33 (Control: NS)	14/15 19/14	Mag group: 61.3 (10.7) Control: 63.1 (9.4)	148.5 (55.2) 151.1 (50.6)	ROC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Aravindan and colleagues, 201627	1/2	Functional endoscopic sinus surgery	15 (Mag) 15 (Dilt) 15 (Control: NS)	8/7 8/7 8/7	Mag group: 58.00 (11.33) Dilt group: 64.33 (11.13) Control: 62.60 (15.60)	NR	VEC	MAG DILT	40 mg kg−1 0.1 mg kg−1`	15 mg kg−1 h−1 0.2 mg kg−1 h−1	Three groups were included: saline, Mag, and Diltiazem. Data from all groups were compared with control.
Mahmoud and colleagues, 201628	1	Right hepatotomy	20 (Mag) 20 (Control: NS)	19/6 18/7	Mag group: 74.88 (5.90) Control: 73.32 (4.02)	451.2 (42) 429.6 (44.4)	ROC	MAG	30 mg kg−1	10 mg kg−1 h−1	–
Ryu and colleagues, 201629	1/2	Laparoscopic gastrectomy	37 (Mag) 37 (Control: NS)	20/17 29/8	Mag group: 65.1 (11.9) Control: 65.6 (9.3)	190 (50) 196 (53)	ROC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Khafagy and colleagues, 201230	1/2	Open cholecystectomy	30 (Mag) 30 (Clon) 30 (Ket) 30 (Control: NS)	5/25 6/24 4/26 8/22	Mag group: 81.5 (8.8) Clon group: 80.0 (5.6) Ket group: 81.5 (6.8) Control: 82.0 (6.9)	129.3 (21.7) 121.9 (16.5) 120.2 (25.3) 122.2 (17.6)	ROC	MAG CLON KET	50 mg kg−1 0.003 mg kg−1 0.4 mg kg−1	8 mg kg−1 h−1 0.002 mg kg−1 h−1 0.2 mg kg−1 h−1	Four groups were included: saline, Mag, Clonidine, and Ketamine. Only data from the Mag and Clonidine groups were compared with control.
Bostankolu and colleagues, 201231	1/2	Elective lower extremity surgery	30 (Dex) 30 (Control: NS)	22/8 18/12	Dex group: 71.9 (11.9) Control: 76.5 (11.6)	112.4 (40.5) 112.2 (36.7)	ROC	DEX	1 μg kg−1	0.7 μg kg−1 h−1	–
Manaa and colleagues, 201232	1/2	Craniotomy/Spinal surgery	30 (Mag) 30 (Control: NS)	19/11 21/9	Mag group: 74.65 (8.02) Control 72.50 (7.97)	NR	ROC	MAG	20 mg kg−1	10 mg kg−1 h−1	–
Saadawy and colleagues, 201033	1/2	Laparoscopic cholecystectomy	40 (Mag) 40 (Lido) 40 (Control: NS)	5/35 6/34 8/32	Mag group: 78.2 (7.2) Lido group: 80.1 (12.4) Control: 77.9 (10.3)	86.2 (14.2) 80.3 (15.8) 79.5 (16.8)	CIS	MAG LIDO	50 mg kg−1 2 mg kg−1	25 mg kg−1 h−1 2 mg kg−1 h−1	Three groups were included: saline, Mag, and Lidocaine. Data from all groups were compared with control.
Lee and colleagues, 200934	1	Elective C-section	25 (Mag45) 23 (Mag30) 24 (Control: NS)	0/25 0/23 0/24	Mag45: 69.7 (7.4) Mag30: 68.7 (10.6) Control: 67.7 (8.6)	67.3 (10.5) 64.4 (11.7) 65.2 (9.7)	ATRA	MAG	45 mg kg−1 30 mg kg−1	15 mg kg−1 h−1 10 mg kg−1 h−1	Three groups were included: saline, Single Mag bolus of 30 mg kg−1 followed by 10 mg kg−1 h−1, and a single Mag bolus of 45 mg kg−1 followed by 15 mg kg−1 h−1. Only data from Mag45 group were compared with control.
Ryu and colleagues, 200835	1/2	Total abdominal hysterectomy	25 (Mag) 25 (Control: NS)	0/25 0/25	Mag group: 57.4 (5.9) Control: 57.5 (8.0)	169.2 (32.3) 162.6 (33.7)	ROC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Seyhan and colleagues, 200636	1/2	Elective hysterectomy with or without salpingooopherectomy	20 (Mag) 20 (Mag10) 20 (Mag20) 20 (Control: NS)	0/20 0/20 0/20 0/20	Mag group: 75.2 (13.1) Mag10 group: 71.5 (12.4) Mag20 group: 73.9 (12.5) Control: 72.7 (11.9)	114.4 (17.9) 117.6 (20.3) 120.1 (17.7) 120.9 (18.3)	ATRA	MAG	40 mg kg 40 mg kg 40 mg kg	0 mg kg−1 h−1 10 mg kg−1 h−1 20 mg kg−1 h−1	Four groups were included: saline, single Mag bolus, and single Mag bolus followed by 10 or 20 mg kg−1 h−1. Only data from Mag20 group were included and compared with control.
Elsharnouby and colleagues, 200637	1/2	Functional endoscopic sinus surgery	30 (Mag) 30 (Control: NS)	16/14 19/11	Mag group: 74.7 (10.6) Control: 72.1 (11.3)	68 (15) 88 (10)	VEC	MAG	40 mg kg−1	15 mg kg−1 h−1	–
Mavrommati and colleagues, 200438	1/2	Abdominal hernioplasty	21 (Mag) 21 (Control: NS)	15/6 13/8	Mag group: 68 (12) Control: 65 (13)	75 (18) 72 (21)	ROC	MAG	30 mg kg−1	6 mg kg−1 h−1	–
Bhatia and colleagues, 200339	1/2	Open cholecystectomy	25 (Mag) 25 (Control: NS)	7/18 5/20	Mag group: 59.40 (10.8) Control: 54.20 (10.8)	77.33 (17.61) 76.24 (15.16)	VEC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Pinard and colleagues, 200340	2/3	Cardiac surgery (CABG/others)	10 (Mag) 10 (Control: NS)	9/1 7/3	Mag group: 71 (10) Control: 77 (14)	247 (42) 271 (51)	CIS	MAG	70 mg kg−1	30 mg kg−1 h−1	–
Choi and colleagues, 200241	1/2	Total abdominal hysterectomy	27 (Mag) 27 (Control: NS)	0/27 0/27	Mag group: 57.84 (7.51) Control: 57.57 (7.03)	142.70 (21.42) 124.19 (22.10)	VEC	MAG	50 mg kg−1	8 mg kg−1 h−1	–
Ko and colleagues, 200142	1/2	Elective total abdominal hysterectomy	29 (Mg) 29 (Control: NS)	0/29 0/29	Mg group: 56.7 (7.2) Control: 58.4 (6.7)	109.3 (23.2) 113.0 (21.1)	VEC	MAG	50 mg kg−1	15 mg kg−1 h−1	–
Schulz-Stubner and colleagues, 200143	NR	Elective pars plana vitrectomies	25 (Mg) 25 Control: NS)	12/13 15/10	NR	NR	MIV	MAG	50 mg kg−1	None	–
Tramer and colleagues, 199644	1/2	Elective abdominal hysterectomy	21 (Mg) 21 (Control: NS)	0/21 0/21	Mag group: 63 (11) Control 66 (11)	88 (21) 108 (34)	VEC	MAG	3 g	0.5 g h−1	–
Kawabata and colleagues, 199445	1/2	Elective otolaryngeal surgery	8 (N1) 15 (N2) 9 (N3) 15 (Control: NS)	3/5 5/10 5/4 6/9	N1: 52.1 (1.7) N2: 57.0 (2.6) N3: 61.7 (3.6) Control: 58.7 (2.1)	NR	VEC	NIC	NR	N1: 1 μg kg−1 min−1 N2: 2 μg kg−1 min−1 N3: 3 μg kg−1 min−1	–
Taittonen and colleagues, 199846	1	Various elective plastic surgeries (Not specified)	10 (Clon) 10 (Midaz) 10 (control: NS)	2/8 6/4 3/7	Clon group: 74 (14) Mid group: 75 (10) Control group: 71 (10)	50 (10) 47 (22) 49 (37)	Mivacurium	CLON MID	2 μg kg−1 70 μg kg−1	None	Three groups were included: saline, Clonidine, and Midazolam. Only data from the Clonidine group were compared with control.
Omar, 201247	1/2/3	Various elective surgeries	23 (Lido) 23 (Control: NS)	11/12 13/10	Lido group: 81.7 (19.1) Control group: 79.5 (16.9)	109 (NR) 114 (NR)	ROC	LIDO	1.5 mg kg−1	1.5 mg kg−1 h−1	–
Lauwick and colleagues, 200848	1/2	Laparoscopic cholecystectomy	25 (Lido) 25 (Control: Fentanyl)	5/20 12/13	Lido group: 66.9 (13.8) Control group: 75.0 (14.8)	60 (NR) 70 (NR)	ROC	LIDO	1.5 mg kg−1	2 mg kg−1 h−1	–
Batistaki and colleagues, 201949	1/2/3	Laparoscopic cholecystectomy	22 (Dexa) 22 (Control: NS)	9/13 7/15	Dexa group: 81.86 (16.65) Control group: 81.13 (17.11)	86.86 (26.78) 72.31 (25.6)	ROC	DEXA	5 mg (1 mg ml−1)	None	–
Rezonja and colleagues, 201650	1/2/3	Elective abdominal or urological surgery	31 (Dexa) 31 (Control: Granisetron)	16/15 16/15	NR	NR	ROC	DEXA	0.15 mg kg−1	None	–

Data collection process and data items

Data were extracted from selected papers by two independent reviewers (GLA, VW) to a pre-formatted Excel worksheet (Microsoft Corporation, Redmond, WA, USA) containing the following characteristics: authors, year, ASA score, type of surgical intervention, sample size, sex, mean weight, mean procedure time, type of NMBA used, type of adjuvant used, study aim (depth, speed of onset of neuromuscular block, or both), initial bolus dose of adjuvant, and maintenance dose of adjuvant (Table 1). Means (standard deviation [sd]) values were extracted for continuous outcomes and numbers of events were extracted for dichotomous outcomes.

Primary outcome

The primary outcome was a reduction in the total dose of NMBA required to achieve clinically acceptable neuromuscular block (as defined by depth of neuromuscular block).

Secondary outcomes

Secondary outcomes were adverse events after administration of adjuncts, induction time, time to extubation, and residual neuromuscular block.

Risk of bias in individual studies

Risk of bias was assessed using the Cochrane Risk of Bias Tool for RCTs.¹⁹ Risk of bias was assessed under the following five domains: selection, performance, attrition, reporting, and other. Two review authors (GLA, VW) independently assessed the risk of bias, and difference of opinions, if encountered, were resolved through discussion and referred to a third reviewer (TA) if a consensus could not be reached.

Statistical analysis

The meta-analysis was conducted using Review Manager software (RevMan version 5.3; The Nordic Cochrane Centre, Copenhagen, Denmark). Data entry was carried out by two investigators (GLA and VW) acting independently. Where possible, the results of studies were pooled if at least two studies reported comparable outcomes.

For primary outcomes, an inverse variance/random effects model, with the standardised mean difference (SMD) as the effect measure, was used for all analyses as methodological and clinical heterogeneity across studies was expected. SMD expresses the size of the intervention effect in each study relative to the variability observed in that study and allows for group comparisons independently of specific units of measures and adjuvants/NMBAs used.¹⁹ Between-study heterogeneity was assessed with the I² statistic test using P<0.1 as the pre-defined threshold for statistical significance. In the presence of multiple intervention groups, we selected one pair of interventions and excluded the others to prevent double counting of participants and to avoid error.¹⁹ Results are presented as SMD with 95% confidence interval (CI), associated P-values, and forest plots.

We also used an inverse variance/fixed effects model, with mean differences as the effects measure, to compare data with similar units of measure. If heterogeneity was found to be >40%, a random effect modelling was used. Results are presented as mean differences with 95% CI, associated P-values, and forest plots.

For secondary outcomes, dichotomous data were analysed using relative risk (RR) with 95% CI. The pre-specified threshold for statistical significance was P<0.05. Publication bias was evaluated by funnel plots generation and assessment of asymmetry.

The Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) working group criteria were used to rate the body of evidence (GRADEpro GDT software; Evidence Prime Inc., Hamilton, Ontario, Canada).²⁰ The evidence was classified as very low, low, moderate, or high. The assessment of evidence was based on study design, risk of bias, inconsistency, indirectness, imprecision, and other considerations using GRADEpro GDT.¹⁵

Results

Study selection

We identified 3082 studies published between 1970 and 2020. After title and abstract screening, we determined that 158 full-text articles may be eligible, 138 of which were excluded. A further 10 articles were identified through hand searching of included articles and review papers. One additional paper was included on the suggestion of a peer reviewer during the review process; this paper was published during the review process.²¹ In total, 31 RCTs^14,21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50 published between 1994 and 2020 met the inclusion criteria (Fig. 1).

Fig 1.

PRISMA flow diagram showing literature search results. Thirty-one randomised controlled trials were included. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis.

Study characteristics

A total of 1085 patients received pharmacological adjuncts, compared with 877 patients who received other drugs or normal saline/control (Table 1). We identified seven different adjuncts (magnesium sulphate, dexmedetomidine, clonidine, nicardipine, diltiazem, lidocaine, and dexamethasone), with the majority of studies comparing these agents with placebo in noncardiac and cardiac surgery. No studies were conducted in critically ill patients. The majority of non-depolarising NMBAs were aminosteroids (rocuronium [n=15]^14,21,22,26,28, 29, 30, 31, 32^,35,38,47, 48, 49, 50; vecuronium [n=9]^{24,25,27,37,39,41,42,44,45} rather than tetrahydroisoquinoline derivatives (cisatracurium [n=3]^23,33,40; atracurium [n=2]^34,36; mivacurium [n=2]).^43,46

Risk of bias within and across studies

The risk of selection bias was low to moderate, as several trials did not report how random sequence generation and allocation concealment were achieved (Fig. 2). The risk of performance and detection bias was moderate to high, with a high percentage of the included studies failing to report if and how personnel and participants were blinded during the intervention and outcome assessment. The risks of attrition, reporting, and other bias were low for most studies. Six studies had design issues consistent with a high risk of bias (Supplementary material S2).41, 42, 43^,45,46,50

Fig 2.

Cumulative risks of bias by category: review authors' judgements about each risk of bias item presented as percentages across all included studies (green: low risk; yellow: insufficient data; red: high risk).

Funnel plot analysis showed symmetrical shapes for all primary and secondary outcomes (Supplementary material S3). However, recent studies have suggested that funnel plots including less than 10 studies may not be sufficient to distinguish real asymmetry from chance and to accurately detect publication bias.^51,52

Pooled results of included studies

Primary outcome: NMBA dose used to achieve satisfactory neuromuscular block

Magnesium

Magnesium was the most prevalent adjunct (21/31 studies).^21,24, 25, 26, 27, 28, 29, 30^,32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44 Magnesium dose, with or without loading dose, varied across the 21 studies (Table 1). Because of the large ranges in dose explored, we based the optimal dose on seven studies that used magnesium primarily to reduce perioperative NMBA use (30–50 mg kg⁻¹ preoperatively and 8–15 mg kg h⁻¹ intraoperatively).^{21,25,29,30,34,35,40} (Supplementary material S4). Twenty out of 21 studies^21,24, 25, 26, 27, 28, 29, 30^,32, 33, 34, 35, 36, 37, 38, 39, 40^,42, 43, 44 reported a NMBA dose reduction after administration of magnesium (Fig. 3), as compared with normal saline (SMD: –1.10 [–1.44 to –0.76], P<0.001; I²=85%; GRADE=moderate). The quality of evidence for this outcome was moderate according to the GRADE criteria. For magnesium, we estimated a median reduction in rocuronium use of 25.5% (inter-quartile range, 14.7–31; Supplementary material S5, S6, and S7).

Fig 3.

Perioperative consumption of NMBAs according to subgroups – magnesium vs control. The difference in mean values is attributable to heterogeneity of units. CI, confidence interval; sd: standard deviation; NMBAs, neuromuscular blocking agents.

Alpha2-adrenergic agonists

Dexmedetomidine

Dexmedetomidine was administered preoperatively (0.5–1 μg kg⁻¹) in five studies^14,22, 23, 24^,31 and intraoperatively (0.2–0.7 μg kg h⁻¹) in four studies^14,22,23,31 (Table 1). Four studies^14,22,23,31 reported that dexmedetomidine reduced NMBA dose used a loading dose of 1 μg kg⁻¹ and a maintenance dose above 0.2 μg kg h⁻¹ (Supplementary material S4). For dexmedetomidine (Fig. 4), there was a dose reduction across each NMBA (SMD: –0.89 [–1.55 to –0.22]; P=0.009; I²=87%). The quality of evidence for this outcome was low by GRADE criteria. Five studies^14,22, 23, 24^,31 comprising 313 patients showed an overall reduction of 6.25 mg ([1.90–10.60]; P=0.005; I²=94%; GRADE Moderate; Supplementary material S8).

Fig 4.

Perioperative consumption of NMBAs according to subgroups – dexmedetomidine vs control. The difference in mean values is attributable to heterogeneity of units. CI, confidence interval; sd: standard deviation; NMBAs, neuromuscular blocking agents.

Clonidine

Clonidine was administered preoperatively (2–3 μg kg⁻¹) in all three studies^25,30,46 and intraoperatively (1–2 μg kg h⁻¹) in two studies (Table 1).^25,30 For clonidine (Fig. 5), there was a dose reduction across each NMBA (SMD: –0.67 [–1.13 to –0.34]; P=0.004; I²=0%). The quality of evidence for the effect of clonidine on NMBA dose was low by GRADE criteria. Unit-specific subgroup analysis was not possible for clonidine because of the heterogeneity in reported units.

Fig 5.

Perioperative consumption of NMBAs according to subgroups. (a) Clonidine vs control. (b) Lidocaine vs control. The difference in mean values is attributable to heterogeneity of units. CI, confidence interval; sd: standard deviation; NMBAs, neuromuscular blocking agents.

Calcium antagonists

Calcium antagonists diltiazem and nicardipine were only used as adjuncts in 2/31 studies (Table 1).^27,45 Diltiazem use had no effect (P=0.14) on mean perioperative vecuronium use (1.29 [sd, 0.15 μg kg min⁻¹]) compared with normal saline (1.30 [0.33] μg kg min⁻¹) (GRADEpro GDT software; Evidence Prime Inc.). Nicardipine was associated with a dose-dependent reduction (P<0.02) in vecuronium infusion requirement (0.55 [0.04] μg kg min⁻¹) compared with normal saline (0.70 [0.03] μg kg⁻¹ min⁻¹).⁴⁵ Meta-analysis for these adjuncts could not be performed because of the small number of trials.

Lidocaine

Lidocaine was administered intravenously both preoperatively (1.5–2 mg kg⁻¹) and intraoperatively (1.5–2 mg kg h⁻¹) and compared with a control in three/30 studies (Table 1).^33,47,48 Lidocaine had no effect on the total perioperative NMBA consumption compared with control (SMD: –0.46 [–1.01 to –0.09]; P=0.10; I²=68%) (Fig. 5). The quality of evidence for the effect of lidocaine on NMBA dose was moderate according to GRADE criteria.

Dexamethasone

The corticosteroid dexamethasone was used in two/31 studies and only administered preoperatively (Table 1).^49,50 One study⁴⁹ showed a significant increase (P=0.01) in the total dose of rocuronium in the dexamethasone group (mean [sd]: 105 [26.5] mg) compared with normal saline (81.14 [8.54] mg), whereas another study showed no difference.⁵⁰

Secondary outcomes – adverse effects

Thirteen studies comprising 697 patients reported adverse events systematically way and were included in the meta-analysis.22, 23, 24, 25, 26, 27^{,29,30,35,36,38,42,44}

Hypotension

The risk of hypotensive episodes was similar in the presence, or absence, of magnesium (n=9)^{24,27,29,30,35,37,38,42,44} (RR=1.00 [0.46, 2.21], P=0.99, I²=0%; GRADE=moderate) and dexmedetomidine (n=3)22, 23, 24 (RR=0.71 [0.30, 1.64], P=0.42, I²=0%; GRADE=moderate), compared with normal saline.

Bradycardia

Nine studies24, 25, 26, 27^{,29,35,36,42,44} reported bradycardia in a way that could be comparable by meta-analysis. There was no difference in the risk of bradycardia with magnesium (RR=2.17 [0.63, 7.50], P=0.22, I²=0%; GRADE: moderate) (Supplementary material S9). For other adverse events and adjuncts, it was not possible to perform meta-analyses owing to lack of uniformity in outcome reporting.

Induction time

One study³⁰ reported data on induction time. The mean induction time was significantly shorter in both the magnesium arm (67.1 [9.3] s) and the clonidine arm (59.8 [6.3] s) compared with normal saline (88.1 [10.2] s; P<0.05).

Time to extubation

Twelve studies^22,25,28, 29, 30^,32,34, 35, 36^,38,47,49 (n=533 patients) reported data on time to extubation. There was no difference in time to extubation for either magnesium (n=9)^25,28, 29, 30^,32,34, 35, 36^,38 (mean difference, 0.27 min [–1.88 to 2.42], P=0.81, I²=99%; GRADE=high) or clonidine (n=2)^25,30 (mean difference, –0.18 min [–0.36 to –0.00], P=0.05, I²= 0%; GRADE= moderate) (Supplementary material S9). Meta-analyses for dexmedetomidine, lidocaine, and dexamethasone could not be performed as this outcome was only reported in three trials.^22,47,49 There was a reduction in extubation time after dexmedetomidine²² (7.7 min [3.1], vs normal saline: 10.8 min (3.8); P=0.01) and lidocaine⁴⁷ (7.2 min [3.6] vs control: 11 min [5.8]; P=0.01). Conversely, dexamethasone had no effect on extubation time, compared with normal saline (2.91 [1.58] min vs 2.32 [1.23] min; P=0.17).⁴⁹

Residual neuromuscular block

Two studies^26,47 reported data on train-of-four ratio on arrival to PACU. For magnesium, the TOF ratio after was greater than 0.9 for both intervention and normal saline,²⁶ whereas lidocaine was associated with a higher TOF ratio.⁴⁷ No other studies reported either clinical or electromyographic data regarding residual neuromuscular block.

Discussion

This systematic review and meta-analysis suggests that magnesium, dexmedetomidine, and clonidine reduce the dose requirement for non-depolarising NMBAs during the perioperative period. The quality of evidence for alpha-2 agonists was, however, low in comparison with magnesium. No association was found between the use of lidocaine and a reduction of total perioperative NMBA dose. We could not conduct meta-analyses to detect any effect of nicardipine, diltiazem, or dexamethasone used as adjuncts to reduce perioperative non-depolarising NMBA dose because of the limited number of studies. No studies addressed this hypothesis in critical care.

The chief issue surrounding the question of pharmacological adjuncts is whether our findings are clinically relevant in terms of their estimated sparing effect. We used well-established statistical approaches to estimate this effect, using the SMD. However, because of the disparity in outcome reporting, we could only estimate the relative reduction afforded by the co-administration of magnesium, dexmedetomidine, or clonidine. This estimate suggests a clinically relevant sparing effect of about 25% reduced NMBA use is possible with magnesium, which would be likely to have a substantial impact on drug supplies in perioperative practice at least. Interestingly, an RCT published during the review process supported this estimate.²¹ This trial reported a 20% lower requirement for rocuronium intraoperatively in patients randomised to receive 50 mg kg⁻¹ of magnesium, followed by a continuous intraoperative infusion at 15 mg kg h⁻¹. There was no difference in recovery time or the incidence of nausea and vomiting.²¹ However, given the lack of data in critical care, plus the likelihood that drugs that are not eliminated by Hofmann degradation may have a prolonged effect in critically ill patients, a similar reduction of ∼25% may not be as clinically relevant in the ICU.

Shortages of NMBAs have been reported across different healthcare systems,⁵³ leading to calls in the USA to develop an essential medicines strategy.⁵⁴ Most recently, the UK Tier 3 Supply Disruption Alert report published in April 2020 highlighted an acute shortage of non-depolarising NMBAs, which prompted the Faculty for Intensive Care Medicine to warn clinicians of the need to preserve non-depolarising NMBAs stocks where possible.⁵⁵ Despite the urgency of that notice, there is little evidence to indicate whether pharmacological adjuncts could play a clinically relevant role in reducing the use of non-depolarising NMBAs. We cannot, however, be sure that adjuncts could also be in short supply, as has been noted from major manufacturers recently.⁵⁶

Changes in drug therapy are well-established triggers for adverse clinical events. Apart from the notable example of norepinephrine shortages in the USA,⁵ head-to-head comparisons of NMBAs in mechanically ventilated, critically ill patients are few. Using a propensity analysis approach, cisatracurium and vecuronium did not appear to be associated with a difference in mortality for patients requiring ventilation for acute respiratory distress syndrome.⁵⁷ However, patients treated with cisatracurium had fewer ventilator and ICU days and were equally likely to be discharged home. Although the mechanism(s) underlying these associations are unclear, prolonged neuromuscular block is independently linked to critical care neuropathy, which in turn delays recovery. Thus, there may be benefit in preserving stocks of benzylisoquinoline NMBAs through the use of adjuncts in the ICU setting.

Our systematic review also found that adverse events were reported infrequently. However, for the adjunct explored most commonly, adverse cardiorespiratory effects related to magnesium sulphate infusion appeared to be no more frequent than normal saline. These data suggest that the magnesium doses used in perioperative studies that decrease NMBA use are well within the margin of safety.²⁹ The emergence of other possible adjuncts with established adverse risk profiles, such as remifentanil, appear to have made little impact on reducing doses of NMBAs.⁵⁸

A strength of this prospectively registered systematic review is that it adhered to international recommendations for this type of article. We used the GRADE criteria to assess the overall quality of evidence (Supplementary material S10). We acknowledge that this systematic review and meta-analysis also has several potential limitations. First, our meta-analysis chiefly comprises data from small, single-centre trials. Consequently, there was an imbalance in the quantity and quality of evidence for each intervention. Second, only four of our studies had a sample size of >100 patients, which may subject our study to small study effect bias. Third, despite the high number of included studies, we noted high heterogeneity amongst analyses. This was largely attributable to disparities in NMBA type, doses of adjuncts, study populations, and design. As a result, the SMD was the most reliable outcome to focus on. Fourth, only 12 of our included studies focused on neuromuscular block as their primary outcome.

In summary, these data suggest that there is, at best, moderate quality evidence for the use of magnesium sulphate, dexmedetomidine, or clonidine in reducing the requirement for NMBA in the perioperative period. The absence of data in critically ill patients suggests that large-scale observational data relating plasma magnesium levels with doses of NMBAs used may provide more rapid insight in this clinical arena.

Authors' contributions

Study design/planning: GLA, VW

Study conduct: all authors

Data analysis: VW, GLA

Writing of manuscript: all authors

Declarations of interest

GLA is an editor for Intensive Care Medicine Experimental, editor for British Journal of Anaesthesia, and has undertaken consultancy work for GlaxoSmithKline. TEFA is an associate editor for British Journal of Anaesthesia. There are no other relationships or activities that could appear to have influenced the submitted work.

Funding

Royal College of Anaesthetists/British Oxygen Company research chair grant in anaesthesia (to GLA), National Institute for Health Research Advanced Fellowship (NIHR300097); Rod Flower Scholarship, WHRI, QMUL (VW).

Handling editor: Jonathan Hardman

Appendix A. Supplementary data

The following are the Supplementary data to this article: