The analgesic efficacy compared ultrasound-guided continuous transverse abdominis plane block with epidural analgesia following abdominal surgery: a systematic review and meta-analysis of randomized controlled trials

We used the recommendations of PRISMA for this systematic review and meta-analysis. We searched the online databases PubMed, Embase and Cochrane Central Register for all relevant studies. Search terms included: epidural anaesthesia OR epidural analgesia OR epidural injection OR epidural administration. The results of this search subsequently combined the following terms: continuous TAP block OR continuous transversus abdominis plane block OR transversus abdominis plane catheters OR TAP block catheters OR abdominal wall block OR transversus abdominal wall block OR nerve block. The search strategy was limited to randomized controlled trials (RCTs) and those performed on humans. No language restriction was applied. The most recent electronic search was completed in June of 2019. We also manually checked the bibliographies of relevant articles for other potentially eligible trials.

This systematic review and meta-analysis is only aimed at female and male adults (18 years or older) who have undergone different types of abdominal surgery.

Ultrasound-guided continuous TAP blocks adopting various approaches (subcostal, oblique subcostal, lateral, or posterior [5, 16]) compared with EA following abdominal surgeries were included in this study.

The pre-specified primary outcome was dynamic pain scores (upon movement) 24 h after abdominal surgery. Secondary outcomes were pain scores at rest after 24 h and pain scores, at rest and dynamic, after 12 h, 48 h and 72 h. Postoperative opioid consumption was measured at 24 h, 48 h, and 72 h following surgery. Meanwhile, we addressed function-related outcomes including time of removal of the urinary catheter, time to first flatus, time to first ambulation, and length of hospital stay. Outcomes of side effects were also evaluated, including hypotension and block complications within the first 24 h postoperatively.

We extracted independent data using established standard data collection forms by two authors (QCS and LYM). Disagreements were resolved by discussion with another author (LJC). If needed, we contacted the corresponding authors of selected articles to obtain the mean and standard deviation of the data. If there was no response, we used the median and quartile ranges to approximate the estimation [17, 18]. Different pain scores assessed with verbal, visual, or numeric rating scales were all converted to a standardized number (on a 10-point scale) for analysis. All opioid analgesic drug usages were converted to equianalgesic doses of intravenous morphine for quantitative evaluations (10 mg of IV morphine = 1.5 mg of IV hydromorphone = 0.1 mg of IV fentanyl = 75 mg of IV pethidine = 100 mg of IV tramadol = 30 mg of oral morphine = 7.5 mg of oral hydromorphone = 20 mg of oral oxycodone) [19].

The quality of the reviewed trials was assessed independently by two authors (QCS and LYM) following the Cochrane Collaboration Risk of Bias Tool for randomized controlled trials [20]. Disagreements were resolved by discussion with another author (LJC). The Cochrane Risk of Bias Tool measured the following: adequacy of sequence generation, allocation concealment, blinding of participants, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other potential sources of bias.

All statistical analyses were performed with the assistance of Review Manager software (RevMan version 5.3.5; The Cochrane Collaboration 2014). For continuous data, when measuring methods were different, the standardized mean difference (SMD) with 95% confidence interval (CI) was calculated; otherwise, the weight mean difference (WMD) with 95% CI was calculated. A risk ratio (RR) with a 95% CI was calculated for dichotomous data. The I² statistic, used for evaluating heterogeneity, was predefined using the following three scales: low (I² < 50%), moderate (I² = 50–74%), and high (I² > 75%) [21]. We pooled outcome data using a fixed effects model in the case of low heterogeneity; otherwise, we chose a random effects model. Predetermined subgroup analysis was conducted according to the type of surgical operation (open surgery or laparoscopy), the method of local anesthetic administration (continuous or intermittent), and whether other anti-inflammatory drugs were regularly used postoperatively. A P-value of < 0.05 was considered statistically significant.

In total, 977 potentially eligible studies were identified through the literature search. We excluded 217 records that were duplicates and a further 738 records for other reasons. After review of the remaining 22 articles in full, 8 RCTs [13, 15, 22‐27] ultimately met the inclusive criteria and 7 RCTs [13, 15, 22‐26] were included in the meta-analysis. A flowchart of this process, including the reasons for excluding studies, is shown in Fig. 1.

Ultimately included trials in this review were published between 2011 and 2017, totaling 453 patients (224 in the TAP block group and 229 in the EA group). Six trials [13, 22‐25, 27] were published in English and the other 2 trials [15, 26] were published in Chinese. The detailed characteristics of the included trials (8 RCTs) are presented in Table 1.

According to our assessment of the Cochrane Collaboration Risk of Bias tool (Fig. 2), most trials have a high risk of bias, which is mainly related to the blindness of participants and evaluators. However, in these trials, it was extremely difficult to blind patients and clinicians.

Seven studies [13, 15, 22‐26], including the meta-analysis with a total of 384 patients (TAP block group:191 patients; EA group: 193 patients), reported dynamic pain scores 24 h postoperatively (Fig. 3). Dynamic pain scores after 24 h were overall equivalent between TAP block and EA groups (7 trials [13, 15, 22‐26]; WMD:0.44; 95% CI: 0.1 to 0.99; I² = 91%; p = 0.11). No significant difference was found in the subgroup analysis of open surgery (4 trials [23‐26]; WMD:0.58; 95% CI: − 0.52 to 1.69; I² = 95%; p = 0.3) and laparoscopic surgery (3 trials [13, 15, 26]; WMD:0.29; 95% CI: − 0.18 to 0.77; I² = 79%; p = 0.22) (Fig. 3), and no difference was found between the continuous local anesthetics subgroup (5 trials [13, 15, 23, 25, 26]; WMD:0.23; 95% CI: − 0.1 to 0.57; I² = 68%; p = 0.17) and intermittent local anesthetics subgroup (2 trials [22, 24]; WMD:0.85; 95% CI: − 1.41 to 3.31; I² = 98%; p = 0.46) (Fig. 4). However, there was a significant difference between the subgroups of regularly administering NSAIDs (4 trials [13, 22, 23, 25]; WMD:-0.11; 95%CI: − 0.32 to 0.09; I² = 0%; p = 0.28) or not (3 trials [15, 24, 26]; WMD:1.02; 95%CI: 0.09 to 1.96; I² = 94%; p = 0.03) for adjuvant analgesics postoperatively (Fig. 5). The pain score in the subgroup not administering NSAIDs was significantly higher. We also performed sensitivity analysis by omitting one study each time, which did not alter the overall combined WMD, and the pooled result was still robust (P > 0.05).

The pain scores at rest showed no significant difference between the two groups 12 h (3 trials [15, 24, 26]; WMD:0.67; 95%CI: − 0.14 to 1.48; I² = 95%; p = 0.1), 24 h (6 trials [15, 22‐26]; WMD:-0.07; 95% CI: − 0.21 to 0.06; I² = 90%; p = 0.3), or 48 h (5 trials [15, 22‐24, 26]; WMD:0.82; 95% CI: − 0.04 to 1.21; I² = 91%; p = 0.06) postoperatively.

With movement, there were no significant differences in pain scores 12 h (3 trials [15, 24, 26], WMD:0.99; 95%CI: − 0.10 to 2.09; I² = 92%; p = 0.07), 48 h (5 trials [15, 22‐24, 26];WMD:0.52; 95% CI: − 0.16 to 1.21; I² = 91%; p = 0.14), or 72 h (2 trials [22, 23]; WMD:-0.21; 95% CI:-0.69 to 0.28; I² = 72%; p = 0.4) postoperatively. There was also no significant difference in morphine consumption postoperatively after 24 h (3 trials [23‐25]; WMD:1.99; 95%CI: − 4.86 to 8.84; I² = 90%; p = 0.57), 48 h (5 trials [13, 23‐26]; WMD:4.12; 95% CI: − 3.13 to 11.16; I² = 93%; p = 0.27) (Fig. 6), and 72 h (3 trials [22, 23, 25]; WMD:7.67; 95% CI: − 3.40 to 18.73; I² = 90%; p = 0.17) compared with the EA group.

Regarding functional recovery, time to first flatus was no different between the two groups (5 trials [13, 15, 24‐26]; WMD:1.57; 95% CI: − 4.7 to 7.84; I² = 52%; P = 0.62). However, time of removal of the urinary catheter measured in the TAP group was significantly shorter (3 trials [13, 15, 26], WMD:-18.95, 95%CI:-25.22 to − 12.71; I² = 0%; p < 0.01), as was time to first ambulation postoperatively (4 trials [13, 15, 24, 26], WMD:-6.61, 95%CI: − 13.03 to − 0.19; I² = 67%; p < 0.05). Meanwhile, length of hospital stay revealed no difference between the TAP and EA groups (4 trials [13, 15, 23, 26], WMD:-0.02; 95%CI: − 0.28 to 0.23; I² = 0%; p = 0.85).

The incidence of hypotension was significantly higher postoperatively in the EA group than in the TAP group (5 trials [15, 23‐26]; RR: 0.16; 95%CI: 0.06 to 0.42; I² = 0%; P = 0.0002). One trial reported that a patient in the TAP group developed a unilateral abdominal wall hematoma immediately after surgery. However, the author was unclear whether this was due to a trauma caused by the insertion of the TAP catheter or surgical puncture.

There are several limitations that must be taken into consideration when interpreting the results of this review. Firstly, because of different surgical procedures, the location of TAP blocks and local anesthetic infusion strategies may be individualized. There are many factors that affect the procedure of continuous TAP block, including puncture location, catheter size, depth of catheter insertion, and various local anesthetic dosage, which will increase the heterogeneity between trails. Secondly, the protocols of anti-inflammatory drugs in the included trials were significantly different (including the type of drug, dosage and delivery speed), which may lead to increased heterogeneity. Such anti-inflammatory drugs may interfere with the overall evaluation of the pain score (somatic and visceral pain). Furthermore, the success of the TAP catheter also depends on the surgeon’s level of experience. Moreover, it was extremely difficult to blind patients and clinicians, when we were conducting a TAP block performance, but we judge that this lack of blindness is unlikely to affect our primary outcomes. In addition, specific criteria for removing the urinary catheter and specific ambulation protocols were not defined in the same way in the included trials. Therefore, more structured and standardised continuous TAP blocking protocols should be developed to compare with EA.