3.1 Study selection and characteristics of included studies
The literature search identified 4392 studies totally and 112 references from CENTRAL, 2481 references from MEDLINE (OvidSP), and 1799 studies from Embase (Ovidsp). Eleven studies fulfilled our inclusion criteria after screening [13-23]. The details of retrieval was shown in figure. 1. The various characteristics of the included studies were shown in Table 1. Seven trials applied magnesium sulfate as a bolus followed by a continuous infusion [13-19], while the remaining four trials applied a bolus injection [20-23]. In eight studies enrolled participants undergoing Lower abdominal or lower limb surgeries [13,14,16,18-21,23], with participants administered epidural anesthesia. Participants in Mohammad 2015 and Radwan 2017 underwent unilateral thoracic surgery and spine surgery respectively with general plus epidural anesthesia. Two studies enrolled participants undergoing cesarean section, with participants in Sun 2012 administered combines spinal epidural anesthesia, and participants in Elsharkawy 2018 administered epidural anesthesia. Total enrollment ranged from 40 to 100 participants, with the number of participants in each study epidural bupivacaine and magnesium sulfate combination ranging from 20 to 50, control 20 to 50.
3.2 Risk of Bias of the included trials
Figure. 2 and Figure. 3 presents the risk of bias of the included studies.We assessed 3 of these trials as low risk of bias [13,14,17], while the remaining 8 trials as high risk of bias according to our pre-specified criteria [15,16,18-23].
3.3. Primary outcome: the time to the first request for rescue analgesics
Nine studies reported the time to the first request for rescue analgesics, in which the data of six studies were reported as mean ± standard deviation (SD) [13-16,18,21], while the remaining two studies reported the data in figure [17,20,23]. As a result, only six studies involving 400 patients were included for analysis [13-16,18,21]. The meta-analysis showed that the time to the first request for rescue analgesics was prolonged significantly in the magnesium sulfate group compared with the control group (SMD 4.96; 95% CI [2.75, 7.17], P<0.00001, I2 = 98%; Figure. 4). We deemed the quality of the evidence to be very low because: (1) four studies had ‘high’ risk of bias; (2) the result was imprecise (wide confidence interval); and (3) there was significant heterogeneity among studies.
Subgroup analysis was conducted according to the administration of magnesium sulfate. Five studies administrated magnesium sulfate by bolus injection [13-16,18], and only one study by bolus followed a continuous injection [21]. We noted the time to the first request for rescue analgesics was also prolonged significantly in the magnesium sulfate group compared with the control group when the study by bolus followed a continuous injection was excluded (SMD 3.67; 95% CI [1.75, 5.58], P=0.0002, I2 = 97%). We also found significant difference between the subgroups (P<0.0001) and the bolus followed a continuous injection subgroup had a greater effect on the time to the first rescue analgesics.
3.4. Primary outcome: the number of patients required postoperative rescue analgesics
Five studies including 304 patients reported the number of patients required rescue analgesics [13,17,19,20,22]. The merged effect analysis showed that it was significantly less in magnesium group compared with control group (RR 0.38; 95% CI [0.20, 0.74], P = 0.004, I2 = 75%, Figure. 5). We judged, the quality of the evidence to be very low based on the GRADE framework: (1) three studies had a ‘high’ risk of bias; (2) there was significant heterogeneity among studies.
Subgroup analysis was conducted according to the administration of magnesium sulfate. Patients required postoperative rescue analgesics was significantly less in magnesium group when magnesium sulfate was administrated by bolus injection (RR 0.49; 95% CI [0.24, 0.97], P = 0.04, I2 = 70%). No significant difference was shown between the two groups when magnesium sulfate was administrated by bolus followed a continuous injection (RR 0.17; 95% CI [0.01, 1.97], P = 0.16, I2 = 84%)
3.5. Primary outcome: requirement for rescue analgesics
The requirement for rescue analgesia was reported in five studies involving 300 participants [13,15,18,20,23]. The meta-analysis showed that the magnesium sulfate group had significantly lower consumption of rescue analgesics than the control group (SMD -2.65; 95% CI [-4.23, -1.06], P=0.001, I2 = 96%; Figure. 6). We rated the quality of the evidence to be very low: (1) four studies had ‘high’ risk of bias; (2) the result was imprecise; and (3) there was significant heterogeneity among studies.
Subgroup analysis was conducted according to the administration of magnesium sulfate. A significant reduction was shown in the magnesium group when magnesium sulfate was administrated by bolus injection (SMD -2.92; 95% CI [-5.43, -0.42], P=0.02, I2 = 97%). No significant difference was shown between two groups when magnesium sulfate was administrated by bolus followed a continuous injection (SMD -2.31; 95% CI [-5.26, 0.65], P=0.13, I2 = 98%).
3.6. Secondary outcome: duration of motor block
Two studies reported duration of motor block [13,16]. The result showed that the duration of motor block in magnesium sulfate group was significantly longer than contol group (SMD 6.29; 95% CI [0.33, 12.24], P=0.04, I2 = 97%; Figure. 7). We rated the quality of the evidence to be very low: (1) the result was imprecise; (2) there was significant heterogeneity among studies; and (3) limited availability of evidence.
3.7. Secondary outcome: adverse events related to postoperative analgesia protocols
A meta-analysis of adverse events was shown in Figure. 8, such as hypotension (RR 0.80; 95% CI [0.64,1.00], P = 0.05, I2=0%), bradycardia (RR 0.80; 95% CI [0.47, 1.36], P = 0.41, I2=0%), nausea and vomiting (RR 0.57; 95% CI [0.31, 1.06], P = 0.08, I2=0%) and pruritus (RR 0.67; 95% CI [0.11, 3.91], P = 0.65, I2=0%) showed no statistically significant differences, except that the incidence of shivering in magnesium sulfate group was significantly lower than control group (RR 0.31; 95% CI [0.18, 0.53], P <0.0001, I2=24%). We rated the evidence for adverse events to be moderate quality.
3.8. Sensitivity analysis and assessment of publication bias
Based on the prior definition, there were only three studies with a low risk of bias [13,14,17], so we did not conduct the sensitivity analysis based on the risk of bias. Sensitivity analyses of primary outcomes using the fixed effect model yielded stable overall results. Consider that each outcome included fewer than 10 studies, there were insufficient data for any publication bias analysis.