Abstract
Background
We conducted this meta-analysis to elucidate whether additional neuraxial magnesium sulfate (MgSO4) can improve postoperative analgesia in women undergoing Cesarean section (CS) delivery.
Method
We searched Pubmed, Embase and ClinicalTrial.gov. We included only randomized controlled trials that have compared the quality of postoperative analgesia with and without additional neuraxial MgSO4 in women undergoing CS delivery with neuraxial anesthesia regimens. The primary outcomes included analgesic duration of neuraxial anesthesia, postoperative pain scores and postoperative consumption of analgesics. The secondary outcomes included patients' satisfaction and adverse effects related to postoperative analgesia.
Results
Nine relevant studies comprising a total of 827 women undergoing CS delivery were included. Analyses revealed that CS women receiving additional neuraxial MgSO4(the MgSO4 group) had longer duration of neuraxial anesthesia (effect size [ES] = 1.920, 95% confidence interval [CI] = 0.999 to 2.842, P < 0.001), longer duration of sensory block (ES = 1.020, 95% CI = 0.463 to 1.577, P < 0.001), lower postoperative pain scores at rest (ES = −1.206, 95% CI = −2.084 to −0.329, P = 0.007), pain scores with motion (ES = −1.435, 95% CI = −2.631 to −0.240, P = 0.019) and consumption of analgesics (ES = −1.620, 95% CI = −2.434 to −0.806, P < 0.001) than CS women without receiving additional neuraxial MgSO4 (the control group). Of note, the MgSO4 group tended to have higher rate on rating satisfaction as “excellent&r
Keywords
Magnesium sulfate; Cesarean section; Postoperative analgesia;
1. Introduction
Providing appropriate postoperative analgesia to women undergoing Cesarean section (CS) delivery is strongly recommended, because it ensures early recovery and mobility to enable them to take care of their newborns after surgery.1 Current analgesic strategies include neuraxial analgesia, systemic opioid and/or non-opioid medications, peripheral nerve blocks, and nonpharmacological therapies (e.g., music therapy or massage).2,3 A growing body of evidence indicates that multimodal analgesia regimen is the most suitable post-CS analgesia strategy.3
Magnesium sulfate (MgSO4) is a widely used medication for seizure prevention in women with preeclampsia.4 MgSO4 is also known to possess certain analgesic property. Among the proposed mechanisms, its effects on antagonizing the N-methyl-d-aspartate (NMDA) receptor5,6 and the calcium channel7,8 have been proven to be crucial in mediating the analgesic property of MgSO4. One meta-analysis has proven that neuraxial MgSO4 could prolong the analgesic effect of neuraxial anesthesia, reduce postoperative pain severity, and reduce postoperative analgesics consumption in below-umbilical surgeries.9
In women during late pregnancy, tolerance to pain significantly increased.10 This phenomenon can be attributed primarily to sex hormones. Progesterone, particularly, is known to possess anti-nociceptive effects. The mechanism may involve its effects on altering expression of the NMDA receptors at spinal cord level11,12 and modulating pro-inflammatory cytokine expression around damaged nerve tissue.13 Because both MgSO4 and progesterone work on NMDA receptor, it is unclear whether the analgesic effect of MgSO4 still exists in high progesterone level conditions, such as full-term pregnancy. In line with this notion, we thus hypothesized that additional neuraxial MgSO4 does not exert significant effects on improving postoperative analgesia in women undergoing CS delivery with neuraxial anesthesia. To test our hypothesis, we conducted this meta-analysis study.
2. Materials and methods
2.1. Sources
This meta-analysis was conducted according to the detailed guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) statement.14
We identified studies from databases of Pubmed, Embase and ClinicalTrial.gov from inception to November 2016. The combination of search terms was ‘cesarean section’ or ‘cesarean delivery’ (in both American and British English spelling) and ‘magnesium’. In the Pubmed and Embase databases, a built-in filter ‘randomized controlled trial’ was applied. We also manually searched for related articles suggested by the databases and the references of the included studies. No language restriction was used.
2.2. Study selection
The search results were screened by one author (SCW). According to the title and abstract, only randomized controlled trials (RCTs) investigating the effects of neuraxial (i.e. epidural or intrathecal) MgSO4 on postoperative analgesia in women undergoing CS delivery were included. We recruited studies employing additional neuraxial MgSO4 in the regimen of neuraxial anesthesia as the treatment arm. Those studies compare neuraxial MgSO4 to the other neuraxial agent, such as benzodiazepine, α2adrenergic receptor agonist or steroid, were excluded. The primary outcomes included duration of sensory block, duration of neuraxial anesthesia, postoperative pain score and postoperative consumption of analgesics. The secondary outcomes included patient's satisfaction and adverse effects related to post-operative analgesia protocols. The included articles had to report at least one of the collected outcomes.
2.3. Validating scoring
The quality of the method of each RCT was rated using the Cochrane Handbook for Systematic Review of Intervention 5.1.0.15 Two authors (SCW and PTP) independently reviewed and rated each trial by using this method. Disagreements in data or scoring were resolved through discussion with a third author (HYC or CJH).
2.4. Data extraction
We developed a data extraction sheet and extracted the data from each included study. Data extraction included the name of the first author and year of publication, number of participants in each group, intervention details (i.e., MgSO4 dosage, administration route and timing, regimen of neuraxial anesthesia and postoperative analgesics) and outcome measurements (i.e., duration of sensory block, duration of neuraxial anesthesia, postoperative pain score, consumption of analgesics, patient's satisfaction and adverse effects). In those studies included more than 1 experimental group, such as different dosing of MgSO4, we split the control groups to create comparison pairs.16 With regard to the adverse effects, we recorded any adverse effect which was reported. Postoperative nausea and vomiting (PONV) was reported either as 1 item or as two separate items in the included RCTs. In the 2-item cases, we included only vomiting in the analysis.
2.5. Data analysis
All analyses were performed using comprehensive meta-analysis version 2.0 (Biostat, Englewood, NJ, USA). We analyzed the results reported in at least two studies. The inverse variance random-effect model was used for data analysis. For continuous variables (i.e., pain score and consumption of analgesics), the effect size (ES) and 95% confidence intervals (CI) were estimated using Hedges's g. If the data were reported as median and interquartile range, we converted the result into mean and standard deviation by using a previously published formula.17 For patients' satisfaction and adverse effects, the odds ratio (OR) and 95% CI were analyzed. If the incidence of an adverse event was 0 in both arms which results in calculation error, 0.5 was added for correction according to the Cochrane handbook.16 Cochran's Q test was used to estimate inter-study heterogeneity, with statistical significance set at P < 0.1. The I2 statistic was used to determine the magnitude of heterogeneity.18 An I2 value of 50% or lower represented “no observed heterogeneity”, whereas an I2value of more than 50% indicated “heterogeneity” among the included studies. We examined the existence of publication bias by Egger's regression test for continuous outcomes, and Harbord's modified test for binary outcomes.19 If publication bias existed, Duval and Tweedie's trim and fill method was applied to correct the publication bias.20
3. Results
3.1. Search results and study descriptions
We initially obtained 103 studies. Among them, 17 duplicates were excluded. The remaining 86 studies were further screened. Only 11 studies were relevant. In the 11 relevant studies, we were not able to find the full text of one study21 and the other one had MgSO4 administered intravenously.22 These two studies were thus excluded. As a result, a total of 9 studies23–31 were included in the analysis (Fig. 1). Of note, the study by Jabalameli et al included 3 experimental groups with various doses.25 Thus we split this study into 3 study pairs, as mentioned previously. Sun et al conducted a study with 4 arms.26 We split this study into 2 study pairs according to the presence of opioid in the neuraxial anesthesia regimen. Among the 9 studies, a total of 827 women were included. Additional neuraxial MgSO4 was administered via epidural route in 2 studies at a dose of 500 mg26,28 and via intrathecal route in the other 7 studies with a dose of either 100 mg,25,29 75 mg25 or 50 mg.23–25,27,30,31 All studies used bupivacaine for neuraxial anesthesia except for one study that used lidocaine instead.31 Four studies and 1 study pair from Sun et al included opioids in neuraxial anesthesia regimens.23,26–29 Details of the included studies were listed in Table 1.
Download full-size image
Eight studies reported the duration of spinal anesthesia.23,24,26–31 Six studies reported the duration of sensory block.23–25,27,29,30 Three studies reported postoperative pain scores.26,27,29 Four studies reported postoperative consumption of analgesics.23,25–27 Four studies have reported patients' satisfaction.23,26,27,29 In terms of adverse effects, PONV and hypotension were reported in 9 studies, pruritus in 6, shivering in 5, sedation in 4, respiratory depression in 3, bradycardia and post-dural puncture headache (PDPH) in 2, hypoxemia, dyspnea and dizziness in 1, respectively. Details of the definition and recording of outcomes were listed in Table 2.
3.2. Methodological quality assessment
Table 3 presents the methodological quality assessment of the included studies. Two studies were rated unclear risk of bias in sequence generation.23,29 Four studies were rated unclear risk of bias in allocation concealment.23,24,30,31 Four studies were rated unclear risk of bias in blinding.23,24,29,30 These unclear risks of bias were all resulted from inadequate description in the methodology section. Six studies were rated high risk in selective outcome reporting, as they had mentioned some outcome recording in the methodology section but did not report in the result section eventually.23–25,28,30
3.3. Duration of neuraxial anesthesia
The duration of neuraxial anesthesia, which defined in most studies as the time period from neuraxial injection to first rescue analgesic postoperatively, was reported in 8 studies.23,24,26–31 However, Sun et al reported the result with Kaplan–Meier analysis rather than the actual average of time.26 As a result, only 7 studies were included in our analysis.20,21,24–28 The result showed that the duration of neuraxial anesthesia in women receiving additional neuraxial MgSO4 (the MgSO4 group) was significantly longer than that in those who did not (the control group) (ES = 1.920, 95% CI = 0.999 to 2.842, P < 0.001; Fig. 2A). Heterogeneity was observed (Q = 109.537, P < 0.001, I2 = 94.522) between the two groups. Egger's regression revealed an intercept of 22.348 with a P of 0.013, indicating publication bias existed. Trim and fill analysis was performed and the adjusted result revealed the same trend (ES = 1.547, 95% CI = 0.548 to 2.545). Sensitivity analysis was not performed as there was no apparent outlier seen on the forest plot (Fig. 2A).
Download full-size image
To identify the source of heterogeneity, we performed moderator analysis by grouping the studies according to the presence of opioid in neuraxial anaesthesia regimen as well as the route of MgSO4 administration. We also performed moderator analysis according to MgSO4 dosage in those studies administered MgSO4 via intrathecal route. The result of moderator analyses showed that heterogeneity still existed after grouping according to the presence of opioid, MgSO4 administration route or dosage (Table 4).
3.4. Duration of sensory block
Six studies were included in the duration of sensory block.23–25,27,29,30 The result showed that the duration of sensory block of the MgSO4 group was significantly longer than that of the control group (ES = 1.020, 95% CI = 0.463 to 1.577, P < 0.001; Fig. 2B). Heterogeneity was observed (Q = 51.907, P < 0.001, I2 = 86.514) between the two groups. Egger's regression revealed publication bias existed (intercept: 9.570, P = 0.027). Trim and fill analysis was performed and the adjust result revealed the same trend (ES = 0.793, 95% CI = 0.165 to 1.420). Sensitivity analysis was not performed as there was no apparent outlier seen on the forest plot (Fig. 2B).
We performed moderator analysis in the same manner, as above-mentioned, to identify the source of heterogeneity. As all the studies reported duration of sensory block administered MgSO4 via intrathecal route, thus we did not perform moderator analysis according to the MgSO4 administration route. The result of moderator analyses showed that heterogeneity still existed after grouping according to the presence of opioid and intrathecal MgSO4 dosage (Table 4).
3.5. Pain scores
There were 6 studies mentioned about pain scores recording in methodology.23,24,26–29 However, only 3 studies reported pain scores in results.26,27,29 As a result, only 3 studies were included for analysis.26,27,29 The analysis result showed that the post-CS pain score at rest in the MgSO4 group was significantly lower than that in the control group (ES = −1.206, 95% CI = −2.084 to −0.329, P = 0.007; Fig. 3A). Heterogeneity was observed (Q = 39.521, P < 0.001, I2 = 92.409) between the two groups. Egger's regression revealed no publication bias (intercept: −16.455, P = 0.317). As there was one apparent outlier on the forest plot (Fig. 3A), we performed sensitivity analysis by removing the outlier.26 Results of sensitivity analysis revealed the same trend (ES = −0.775, 95% CI = −1.234 to −0.315, P = 0.001; Q = 5.586, P = 0.061, I2 = 64.198).
Download full-size image
We also performed moderator analysis to identify the source of heterogeneity, as previously described. After grouping the studies according to MgSO4 administration route, the studies administered MgSO4 via intrathecal route demonstrated no heterogeneity. Heterogeneity still existed when MgSO4 was administered via epidural route (Table 4). Heterogeneity also existed after grouping the studies according to the presence of opioid in the neuraxial regimen (Table 4). Moderator analysis according to MgSO4 dosage in those studies administered MgSO4 via intrathecal route was not performed, because there were only 2 studies in this category and these 2 studies administered 50 mg27 and 100 mg29 MgSO4, respectively.
For post-CS pains score with motion, the MgSO4 group also had significantly lower pain score than the control group (ES = −1.435, 95% CI = −2.631 to −0.240, P = 0.019; Fig. 3B). Heterogeneity was also observed (Q = 37.210, P < 0.001, I2 = 94.265). Egger's regression also revealed no publication bias (intercept: −20.842, P = 0.403). Sensitivity analysis was performed by removing one apparent outlier,26and the adjusted result revealed the same trend (ES = −0.812, 95% CI = −1.291 to −0.333, P = 0.001; Q = 2.194, P = 0.139, I2 = 54.420).
Moderator analysis showed heterogeneity existed after grouping the studies according to the presence of opioid in the neuraxial regimen and MgSO4 administration route (Table 4).
3.6. Consumption of analgesics
Six studies reported postoperative consumption of analgesics.23,25–29 However, 2 of them reported the result as the percentage of patient who requested analgesics rather than the amount of consumption.28,29 As a result, only 4 studies were included in the analysis.23,25–27 The MgSO4 group had significantly lower postoperative consumption of analgesics than the control group (ES = −1.620, 95% CI = −2.434 to −0.806, P < 0.001; Fig. 4). Heterogeneity was observed (overall Q = 79.857, P < 0.001, I2 = 92.487). Egger's regression revealed no publication bias (intercept: −8.336, P = 0.052). Sensitivity analysis was performed by removing the apparent outlier,23and the adjusted result revealed the same trend (ES = −1.159, 95% CI = −1.720 to −0.599, P < 0.001; Q = 29.701, P < 0.001, I2 = 83.166).
Download full-size image
We also performed moderator analysis, as above-mentioned. Moderator analysis result showed that heterogeneity still existed after grouping the studies according to the presence of opioid in the neuraxial anesthesia, the MgSO4 administration route and dosage (Table 4). Furthermore, we also performed moderator analysis according to postoperative pain control regimen. The result also showed that heterogeneity existed with either opioid or diclofenac (Table 4)
3.7. Patients' satisfaction
Four studies reported patients' satisfaction.23,26,27,29 All of them reported in a scale of excellent, good, average and poor, or similar scoring system. There was no patient reported average or poor in 2 studies.23,27 We extract the number of patient who rated “excellent” as a positive event in satisfaction. Analysis results revealed that the MgSO4 group tended to report “excellent” than the control group (OR = 3.748, 95% CI = 2.218 to 6.332, P < 0.001; Fig. 5). No heterogeneity was observed (Q = 1.035, P = 0.904, I2 = 0.000). Harbord's modified test revealed no publication bias (coefficient = 0.62, 95% CI = −1.47 to 2.70, P = 0.417). We did not perform sensitivity analysis and moderator analysis as there was no apparent outlier or heterogeneity observed.
Download full-size image
3.8. Adverse effects
PONV was reported in all 9 studies. Two studies reported nausea and vomiting as independent results,30,31 thus the incidence of vomiting in these studies were extracted as aforementioned. The result showed that the risk of having PONV of the MgSO4 group and that of the control group was not significantly different (OR = 1.246, 95% CI = 0.786 to 1.976, P = 0.350). No heterogeneity was observed (Q = 4.550, P = 0.951, I2 = 0.000). Harbord's modified test revealed no publication bias (coefficient = −1.92, 95% CI = −4.75 to 0.92, P = 0.160).
Hypotension was reported in all 9 studies. However, Banihashem at el. reported the result in only the first 21 min after anesthesia, and the result was presented in figure.24 Thus its data was excluded. The result showed that the risk of having hypotension of the MgSO4 group was comparable to that of the control group (OR = 0.771, 95% CI = 0.518 to 1.148, P = 0.200). No heterogeneity was observed (Q = 4.861, P = 0.900, I2 = 0.000). Harbord's modified test revealed no publication bias (coefficient = 1.28, 95% CI = −0.34 to 2.90, P = 0.107).
Pruritus was reported in 6 studies.23,26–30 Malleeswaran et al reported the incidence of patient reported “mild” or “severe”,27 while the other studies simply reported the presence of event. We decided to count both “mild” and “severe” event as presence of event. The result showed that the risk of having pruritus of the MgSO4 group was comparable to that of the control group (OR = 0.848, 95% CI = 0.485 to 1.482, P = 0.563). No heterogeneity was observed (Q = 5.007, P = 0.543, I2 = 0.000) and Harbord's modified test revealed no publication bias (coefficient = 1.02, 95% CI = −1.10 to 3.14, P = 0.271).
Shivering was reported in 5 studies.23,25,26,28,30 The result showed that the risk of having shivering of the MgSO4 group was also comparable to that of the control group (OR = 0.688, 95% CI = 0.309 to 1.530, P = 0.359). No heterogeneity was observed (Q = 11.041, P = 0.137, I2 = 36.598). Harbord's modified test revealed no publication bias (coefficient = −3.54, 95% CI = −7.53 to 0.44, P = 0.073).
For sedation, 4 studies mentioned about recording in the method description.23,25,27,30 However, only 2 of them reported the result. Furthermore, in these 2 studies, Xiao et al23 reported “severe sedation”, which turned out to be none; Malleswaran et al27 reported that all patients were either “awake and respond readily” or “awake but respond lethargic”. Due to the inconsistency of reporting, we did not perform meta-analysis on this item.
For bradycardia, 4 studies mentioned about recording in the method description.24–26,30 However, only 2 studies reported the result.26,30 The result showed that the risk of having bradycardia of the MgSO4 group was also comparable to that of the control group (OR = 1.965, 95% CI = 0.334 to 11.541, P = 0.455). No heterogeneity was observed (Q = 0.372, P = 0.830, I2 = 0.000). Harbord's modified test revealed publication bias existed (coefficient = 5.10, 95% CI = 3.83 to 6.37, P = 0.012). However, we are not able to perform trim-and-fill analysis because there were only 2 studies.
Respiratory depression was reported in 3 studies.23,29,30 The incidences of respiratory depression in both the MgSO4 and the control groups were zero. Headache was reported in 2 studies23,29; Ghrab et al reported 17 cases with headache, no difference between groups as they stated, but the distribution of the cases was not report.29 As a result, we were not able to perform analysis. Hypoxemia,25 dyspnea31 and dizziness28 were reported in 1 study, respectively. There were only sporadic events of these adverse effects. According to the original articles, none of the between-group differences of these side effects reached statistical significance.
4. Discussion
Data from the present study revealed that additional neuraxial MgSO4 exerted significant effects on prolonging the duration of neuraxial anesthesia, reducing postoperative pain scores as well as decreasing postoperative consumption of analgesics in CS women. This study also revealed that additional neuraxial MgSO4significantly increased patients' satisfaction in CS women. Of note, additional neuraxial MgSO4 exerted no significant effects on the incidences of adverse effects related to postoperative analgesia protocols. These data demonstrated the beneficial effects of additional neuraxial MgSO4 in post-CS analgesia. Collectively, these data confirmed that additional neuraxial MgSO4 improved postoperative analgesia quality in women undergoing CS delivery.
As aforementioned, the analgesic effects of progesterone and MgSO4 are both derived from antagonizing the NMDA receptors.5,6,9 Considering that women undergoing CS delivery have high progesterone levels,10 we hypothesized that adding MgSO4 as an adjuvant may not exert significant effects on improving postoperative analgesia in CS women. To our surprise, data from this study denied our hypothesis. The definitive mechanisms remain to be elucidated. However, several possible mechanisms might explain our observation. As aforementioned, MgSO4antagonizes both the NMDA receptors5,6 and calcium channel.7,8 Calcium is a crucial element of nociceptive pathway.7,8 MgSO4 along with several other calcium channel blockers had been proven to have analgesic property.7,8 Therefore, it is possible that, in addition to antagonizing the NMDA receptors, MgSO4 might also block calcium channels and exert significant anti-nociceptive effect in women undergoing CS delivery. In addition, the progesterone level in cerebral spinal fluid (CSF) of full-term pregnancy women is 8 times higher than non-pregnant women.32 However, the CSF progesterone level rapidly dropped in postpartum period.32 Therefore, it is likely that certain portion of the NMDA receptors will be un-antagonized due to the decreased CSF level of progesterone in postpartum period. Adding neuraxial MgSO4 can block the un-antagonized NMDA receptors and, thus, exert significant analgesia effects in post-CS women.
Because heterogeneity was observed in most of our analyses, this study thus performed moderator analysis to identify the source of heterogeneity. We tried to group the studies according to the presence of opioid in neuraxial anesthesia regimen and the route of MgSO4 administration. In studies administered MgSO4 via intrathecal route, we further grouped the studies according to MgSO4 dosage. In the analysis of analgesic consumption, we also grouped the studies according to different postoperative analgesics. However, our result revealed that, except for pain score at rest with MgSO4 administered via intrathecal route and consumption of analgesics with high dose MgSO4 administered via intrathecal route, most of the heterogeneities remained existed with the moderator analyses. Judging from these data, we thus believe that the moderators we studied, to some extent, might contribute to the heterogeneity of the results. However, the major sources of heterogeneity remain un-identified. More studies are needed before conclusions can be drawn.
The present study has certain limitations. Firstly, our findings showed high between-study heterogeneity and thus caution should be used if further interpretation of our findings is intended. Secondly, the reported adverse effects were primarily opioid-related. The adverse effect of MgSO4 (e.g., neuromuscular weakness4) was not reported in any of the included studies. Although severe neuromuscular weakness is rarely seen with normal clinical protocols,4 whether neuraxial MgSO4 can cause significant neuromuscular weakness in women undergoing CS delivery remains un-studied. Thirdly, one of the included studies20 was conducted in women who had severe pre-eclampsia. These subjects thus received both intravenous and neuraxial MgSO4, according to their protocols. It is thus likely that concurrent use intravenous and neuraxial MgSO4 may influence the study results. More studies are needed before we can conclude on the extent of influence. Fourthly, our data demonstrated that additional neuraxial MgSO4 can exert significant analgesia in women undergoing CS delivery, i.e., the assumed high progesterone level condition. However, as data regarding direct progesterone measurements are lacking in this meta-analysis study, the relationship between MgSO4 and progesterone remains un-elucidated. Future studies are required before further conclusion can be drawn.
In conclusion, neuraxial MgSO4 improves postoperative analgesia in women undergoing CS delivery.
Disclosure
This work was supported by a grant from Taipei Tzu Chi Hospital (TCRD-TPE-104-25), awarded to SCW and CJH. The authors declare that they have no conflict of interest. Part of the study data have been presented in Euroanaesthesia 2015, Berlin, Germany, between 30 May and 02 June 2015.