3.1 Results of search
We screened 9,822 possible studies from database searches, then we carried out full‐text review of 40 articles after screening based on titles and abstracts. No eligible studies were included by manual retrieval. As a result, 18 studies that met our inclusion criteria were included in our analysis, and the details of study retrieval is shown in Figure 1.
3.2 Characteristics of the included studies
The characteristics of the studies are summarized in table 1. Six of the included studies involved cesarean section [14-19], whereas the remaining thirteen involved surgeries in infraumbilical [20,21]; gynaecological [22, 23]; urologic [24]; and orthopaedic [25-31]. Sample sizes ranged from 50 to 150 patients. The mean age of all the participants wass below 50 years old and American Society of Anesthesiologists Physical Status (ASA ) I to II. The doses of magnesium sulfate administrated ranged from 25 to 100 mg. All of the eighteen included studies were RCTs conducted in India [20, 21, 23, 26, 27, 29, 31]; Turkey [19]; Egypt [14, 22, 24, 25]; and Iran [15-18, 28, 30]. Four studies were identified as being of ‘high’ quality according to our predesign criteria [14, 16, 23, 24], leaving 14 low-quality studies. Figure 2 and Figure 3 show the risks of bias in the included studies.
3.3. Primary outcome
Fourteen studies reported duration of analgesia ( the time to the first complaint of pain or request for analgesics) [14-16, 19-23, 25-27, 28, 30, 31], one study where the date was reported as histogram was excluded from the quantitative analysis [26]. Finally, thirteen studies including 513 participants in the magnesium sulfate group and 411 participants in the control group were included in the quantitative analysis, and the result demonstrated that the addition of intrathecal magnesium sulfate to bupivacaine provided a longer duration of analgesia compared with the control group (SMD 0.99; 95% CI [0.45, 1.52], P = 0.0003, I2 = 93%; Figure 4).
Substantial heterogeneity was identified in study effects (I2=93%), we conducted subgroup analysis according to the type of surgery (cesarean section versus non-casarean section) to identify the source of heterogeneity, and still found significant heterogeneity in the cesarean section subgroup (I2=94%) or the non-casarean section subgroup (I2=93%). We also performed the subgroup analysis according to the doses of magnesium sulfate administrated, heterogeneity still existed in the low dose (magnesium sulfate 50 mg) subgroup (I2=93%) and the high dose ( magnesium sulfate 100 mg) subgroup (I2=84%).
Considering that the number of ‘high’ quality studies was small and only three [14, 26, 23], we did not conduct a sensitivity analysis based on the quality of studies. we used a fixed effect model which did not alter interpretation of the result.
we performed publication bias analysis using a funnel plot, and found that the funnel plot was symmetrical, indicating less risk of publication bias (Figure 5).
We used the GRADE approach to judge the quality of the evidence for duration of analgesia to be very low. We downgraded quality due to unclear risk of bias for several key domains among the included studies; the sample sizes of included studies were small; in addition, there was unexplained heterogeneity among studies and imprecision of results (wide confidence intervals).
3.4 Secondary outcomes
3.4.1. The onset and duration of sensory block
Twelve studies reported the onset of sensory block [14, 16, 17, 19, 22, 23, 24, 26, 28-31], one study where the date was reported as histogram was excluded from the quantitative analysis [26]. The combined statistical results showed that the time of onset of sensory block was delayed in the magnesium sulfate group compared with the control group (SMD 1.20; 95% CI [0.65, 1.75], P =<0.0001, I2 = 91%; Figure 6).
Twelve studies reported the duration of sensory block [14, 16, 17-19, 23, 24, 26-30]. We excluded one study from the quantitative analysis because the date was reported as histogram [26]. The meta-analysis showed that the duration of sensory block were significantly prolonged in the magnesium sulfate group compared with the control group (SMD 0.59; 95% CI [0.14, 1.04], P =0.01, I2 = 89%; Figure 7).
3.4.2. The onset and duration of motor block
Seven studies analysed the onset of motor block [16, 19, 23, 24, 36, 28, 29], In one study date was reported as histogram [26], we excluded it and conducted a meta-analysis showing that there was a significant delay in onset of motor block in the magnesium sulfate group compared with the control group (SMD 1.46; 95% CI [0.23, 2.69], P =0.02, I2 = 96%; Figure 8).
Twelve studies reported the duration of motor block [16, 18-21, 23, 24, 26-30], one study where the date was reported as histogram was excluded from the quantitative analysis [26]. The pooled SMD found no significant difference in the duration of motor block between groups (SMD 0.42; 95% CI [-0.05, 0.89], P =0.08, I2 = 90%; Figure 9).
3.4.3. Rescue analgesics consumption
Rescue analgesics consumption was reported by six studies [22, 23, 25, 26, 28, 30]. We excuded one study from the quantitative analysis because the date was reported as histogram [26]. The combined statistical results showed that the rescue analgesics consumption was reduced in magnesium sulfate group compared with the control group (SMD -0.81; 95% CI [-1.06, -0.56], P < 0.00001, I2 = 11%, Figure 10).
3.4.4. Adverse events
There was no significant difference with regard to hypotension, assessed in seven studies [18, 19, 23, 24, 27, 28, 31]; bradycardia, reported in five trials [ 19, 23, 24, 28, 31]; nausea and vomiting, reported in eleven trials [15, 16, 18, 19, 22-25, 27, 30, 31]; pruritus, reported in two trials [19,30]; shivering, reported in seven trials [17-19, 23-25, 27]. There are six studies reported the neurological deficit [16, 18, 19, 23, 27, 29], all of them observed no neurological deficits in both magnesium and control group.