Data and meta-analysis for choosing sugammadex or neostigmine for routine reversal of rocuronium block in adult patients

This meta-analysis was conducted to define clinical efficacy and side effects (bradycardia and post-operative nausea and vomiting [PONV]) in trials comparing sugammadex with neostigmine or placebo for reversal of rocuronium-induced neuromuscular blockade in adult patients. A search of PubMed, Google Scholar, and Cochrane Library electronic databases identified 111 clinical trials for potential inclusion. We performed a meta-analysis of 32 studies that quantitatively compared the efficacy and side effects of sugammadex with either neostigmine or placebo in adult patients requiring general anesthesia. Analyzed outcomes were reversal time, anesthesia time, duration of stay in the post-anesthesia care unit (PACU), and the occurrence of bradycardia or PONV. Odds ratios and 95% confidence intervals (CI) were calculated for binary data. Mean differences and 95% CI were calculated for continuous outcome data. Meta-analyses were performed using random and fixed-effects models. Heterogeneity across studies was assessed using Cochran's Q test and the I2 statistic. Quantification of these outcomes can better inform anesthetists and health systems of the relative costs and benefits of the two reversal agents. This information also forms a basis for a comparative cost analysis in a co-submitted manuscript [1].


a b s t r a c t
This meta-analysis was conducted to define clinical efficacy and side effects (bradycardia and post-operative nausea and vomiting [PONV]) in trials comparing sugammadex with neostigmine or placebo for reversal of rocuroniuminduced neuromuscular blockade in adult patients. A search of PubMed, Google Scholar, and Cochrane Library electronic databases identified 111 clinical trials for potential inclusion. We performed a meta-analysis of 32 studies that quantitatively compared the efficacy and side effects of sugammadex with either neostigmine or placebo in adult patients requiring general anesthesia. Analyzed outcomes were reversal time, anesthesia time, duration of stay in the post-anesthesia care unit (PACU), and the occurrence of bradycardia or PONV. Odds ratios and 95% confidence intervals (CI) were calculated for binary data. Mean differences and 95% CI were calculated for continuous outcome data. Meta-analyses were performed using random and fixed-effects models. Heterogeneity across studies was assessed using Cochran's Q test and the I 2 statistic. Quantification of these outcomes can better inform anesthetists and health systems of the relative costs and benefits of the two reversal agents. This information also forms a basis for a comparative cost analysis in a co-submitted manuscript [1] .
© 2020 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license.
( http://creativecommons.org/licenses/by/4.0/ ) Table   Subject Anesthesiology and Pain Medicine Specific subject area Reversal of rocuronium neuromuscular blockade with sugammadex or neostigmine. Type of data Table  Chart Figure How data were acquired Systematic review and meta-analysis Data format Raw Analyzed Parameters for data collection

Specifications
The primary outcomes recorded were: time to recovery of the train-of-four ratio to > 0.9; total anesthesia time; time from admission to the post-anesthesia recovery unit (PACU) until the patient was ready for discharge from the unit; occurrence of bradycardia; occurrence of post-operative nausea and vomiting (PONV). Description of data collection A search of PubMed, Google Scholar, and Cochrane Library electronic databases identified 111 clinical trials for potential inclusion. We performed screening of citations, data extraction, and quality assessment in duplicate. We performed a meta-analysis of 32 studies that quantitatively compared the efficacy and side effects of sugammadex with either neostigmine or placebo in adult patients requiring general anesthesia. Data

Value of the Data
• This meta-analytic data quantifies and updates our current level of understanding of the comparative efficacy and side effects of a newer, more expensive reversal drug, sugammadex, with its generic counterpart, neostigmine. • Quantification of these outcomes can inform anesthetists and health systems of the relative costs and benefits of the two reversal agents. • This information provides a basis for undertaking comparative cost analyses that can inform clinical and administrative decisions within hospitals and health systems.

Study flow and description of data
The initial literature search identified 117 reports for potential inclusion ( Fig. 1 ). Forty-one of these reports were reviews or qualitative descriptions. The remaining 76 reports were screened;  15 were excluded as not being relevant. Of the 61 studies assessed for eligibility, 17 did not meet inclusion criteria and were excluded. The remaining 32 reports were submitted to quantitative meta-analysis. The characteristics of the included studies are listed in Table 1 . Table 2 lists the 17 excluded studies and the reasons for exclusion. All data are provided in the attached supplemental Excel file.
The assessment of possible study bias in the included studies is outlined in Fig. 2 . Two of the reports studied two distinct subject samples. Sparr et al. separated their dose-finding study into subjects with either a deep ( n = 6) or shallow ( n = 9) levels of neuromuscular blockade prior to reversal [2] . Woo and colleagues reported two separate subject samples: a caucasian group ( n = 59) and a Chinese group ( n = 130) [3] . In both studies, the groups were analyzed separately since merged data were not available in the original articles.       Fig. 3 outlines a meta-analysis of 22 studies that quantified the mean difference in trainof-four (TOF) recovery time to at least 90% of complete reversal. The mean difference between therapies was 11.7 min (95% confidence interval [CI] −15.6 to −7.8 min, P < 0.0 0 01; I 2 = 95%). Sensitivity analyses, omitting each study in turn, produced random-effect mean differences ranging between −10.1 to −12.2 min ( P < 0.0 0 01 for all analyses). Two studies, Carron et al. and DeRobertis et al., visually appeared to be outliers [ 4 , 5 ]. These reports were also the only two retrospective studies included in the analysis. Omitting both these studies in a sensitivity analysis resulted in a random-effects mean difference of −9.3 min (95% CI −11.8 to −6.9 min, P < 0.0 0 01; I 2 = 92%).  Subgroup analyses were conducted on the depth of blockade ( Fig. 4 ). The mean difference for reversing deep blockade (1 to 2 twitches of TOF present after post-tetanic stimulation; n = 2 studies) [ 4 , 6 ] was −24.9 min (95% CI −38.0 to −11.9 min, P = 0.0 0 08), for moderate block (2 of 4 twitches present in TOF; n = 13 studies) −11.8 min (95% CI −16.8 to −6.7 min, P = 0.0 0 01), and for shallow block (4 of 4 twitches present in TOF; n = 7 studies) was −8.0 min (95% CI −14.8 to −1.2 min, P = 0.023).

Train-of-four recovery (TOFR) -sugammadex compared to placebo
Six studies compared reversal with sugammadex to a placebo ( Fig. 5 ). One study (Sparr, 2007) compared reversal of TOF at two different levels of block prior to reversal [2] . These groups were analyzed separately in the meta-analysis. The mean difference in reversal time between sugammadex and placebo was −46.7 min (95% CI −68.4 to −24.9 min, P < 0.0 0 01; I 2 = 89%). Sensitivity analyses, omitting each study in turn, produced random-effects mean differences ranging between −35.2 and −51.9 min ( P ranging from < 0.0 0 01 to 0.0015 for all analyses).

Anesthesia time
Anesthesia time represented the time in minutes between induction of anesthesia and extubation. Nine studies provided adequate data on anesthesia time ( Fig. 6 ). Reversal with sugam- madex compared with neostigmine resulted in a random-effects mean difference of −18.6 min (95% CI −37.3 to + 0.2 min, P = 0.056).

Post-anesthesia recovery unit (PACU) time
PACU time represented the time in minutes between a patient's admission to the PACU and the time that the patient was deemed ready for discharge. Six studies met inclusion criteria for this outcome measure ( Fig. 7 ). Reversal with sugammadex compared with neostigmine resulted in random-effects mean difference of −12.0 min (95% CI −24.7 to + 0.6 min, P = 0.063). Fig. 8 shows the difference in the occurrence of bradycardia, as defined by the investigators, after either sugammadex or neostigmine administration. The random-effects odds ratio was 0.22 (95% CI 0.10 to 0.50, P = 0.0 0 03) for the comparison between sugammadex and neostigmine. Fig. 9 outlines the difference in the occurrence of PONV, as defined by the investigators, after either sugammadex or neostigmine administration. The random-effects odds ratio was 0.64 (95% CI 0.46 to 0.87, P = 0.0065) for the comparison between sugammadex and neostigmine.

Search strategy
PubMed, Google Scholar, and the Cochrane library electronic databases were searched for articles published between January 1, 2005 (the publication year of the first description of sugammadex [8] ) and June 1, 2019. Using the AND function, the search terms "sugammadex" OR "srba" OR "selective relaxant binding agent" were combined with "neostigmine OR placebo" and "rocuronium." The search population then was limited to "human," and "adult." Titles and abstracts for all articles returned by the search strategy were screened. The reference lists of each article, as well as previously-published reviews and meta-analyses, were manually searched for additional references of potential interest. The full texts of each article were then retrieved to assess suitability for inclusion. This manuscript adheres to applicable PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines ( http://www.prisma-statement.org , accessed May 15, 2020).

Inclusion/exclusion criteria
We identified randomized clinical trials and cohort studies with the following inclusion criteria: sugammadex, at a dose between 1 mg/kg and 4 mg/kg, was used for reversal of rocuroniuminduced neuromuscular blockade; the effects of sugammadex were directly compared with ei- ther neostigmine, at a dose of 30 to 80 mcg/kg, or placebo; the depth of neuromucular blockade was objectively quantified and was similar in each group; patients were at least 18 years of age and undergoing a procedure requiring general anesthesia; adequate data were present either in English or an understandable graphic format. Exclusion criteria included: pediatric studies; case series or pre-post time series studies; inability to retrieve a full-text version or English abstract; lack of outcome data of interest.

Quality assessment
We assessed the study design of articles. Both cohort and randomized studies were included. Assessment for potential bias was performed according PRISMA methodology. No studies were excluded for a specific level of potential bias.

Data extraction
Data were extracted from the original texts and summarized in an Excel database. We performed screening of citations, data extraction, and quality assessment in duplicate. Prior to finalization, we verified the database against the original reports and corrected as necessary. The primary outcomes recorded were: time to recovery of the train-of-four ratio to ≥ 0.9; total anesthesia time; time from admission to the post-anesthesia recovery unit (PACU) until the patient was ready for discharge from the unit; occurrence of bradycardia, as defined by the investigators; occurrence of post-operative nausea and vomiting (PONV), as defined by the investigators.
Studies were categorized as to patient population (general surgical versus special population [study sample limited to a specific high-risk procedure or patient population]) and depth of neuromuscular blockade (deep -post-tetanic facilitation only; moderate -return of two of four twitches in a train-of four [TOF] stimulus; or shallow -return of a TOF ratio of 0.1 to 0.9).

Statistical methods
We calculated odds ratios (OR) and 95% confidence intervals (CI) for binary data. Mean difference (MD) and 95% CI were calculated for continuous outcome data. Time-based data in which median and ranges were reported were converted to estimate means and standard deviations according to the techniques outlined by Hozo et al. [9] Meta-analysis was performed using random-and fixed-effects models. The random-effects model appeared more appropriate since it was expected that variation among studies would occur beyond that associated with sampling variation. When calculating odds ratios, 0.5 was added to the frequencies of each cell in studies with a zero number of events within a cell. Heterogeneity across studies was assessed using Cochran's Q test and the I 2 statistic. The I 2 statistic estimated the percentage of total variation among the study effects attributable to heterogeneity among studies rather than sampling error.