全文

Abstract

Caesarean section (CS) is the most common major surgical procedure performed worldwide. Traditionally, creation of a bladder flap (BF) has been a routine surgical step at CS although recent randomised controlled trials (RCTs) have begun to question its value. We performed a meta-analysis of RCTs examining the benefits of BF formation at CS.

Pubmed, Medline, Embase, CINAHL Plus^®, Web of Science Reference and Cochrane Databases online were searched in March 2012 using combinations of the terms “c(a)esarean”, “bladder”, “flap” and “technique”. Citations identified in the primary search were screened for eligibility. Online clinical registries (www.clinicaltrials.gov, www.controlled-trials.com and www.ukcrc.org.) were also searched. The primary outcome was bladder injury. Secondary outcomes were skin incision-delivery interval, total operating time, blood loss and duration of hospitalisation. Pooled outcome measures (odds ratio [OR] and weighted mean difference [WMD]) were calculated using a random effects model.

Three published RCTs and one unpublished trial identified from an online trial registry were included (n = 581 women). All four trials excluded very preterm and emergency CS. Omission of the BF step at CS reduced the skin incision-delivery interval (WMD 1.27 min; p = 0.0001). No differences were found for bladder injury (pooled OR 0.96), total operating time (WMD 3.5 min), blood loss (WMD 42 ml) or duration of hospitalisation (WMD 0.07 days).

Omission of the BF at elective CS does not appear to increase the rate of peri-operative complications and improves the skin incision-delivery interval. The role of BF formation in very preterm procedures and emergency intrapartum CS needs further study.

Keywords: Bladder flap, Caesarean section, Meta-analysis.

 

1. Introduction
Surgical delivery of the baby by caesarean section dates back almost 4000 years [1]. Today, caesarean section (CS) is the most common major surgical procedure performed worldwide [2]. Indeed, in some parts of the world the rate of caesarean delivery is more than 50% of total births [3]. Recent data from the United States show that 1 in 3 babies are currently delivered abdominally [4]. Given the invasive character of the procedure, there is a need for on-going assessment of optimal caesarean surgical technique, to ensure patient safety and minimise the risk of morbidity and mortality.

Several different surgical techniques for CS have been reported, which have evolved over time [1] and [5]. For many years, particular surgical steps during a routine CS were performed simply because the technique was that originally learned by the surgeon. In recent times, with the advent of evidence-based practice, caesarean technique has been subjected to more rigorous analysis [6] and [7]. Contemporary meta-analyses have examined the options for skin closure at CS [8], the need for routine bladder catheterisation [9], the benefits of uterine exteriorization [10] and the effect of peritoneal closure on subsequent intra-abdominal adhesions [11]. However, the routine creation of a bladder flap (BF) during CS has only recently begun to be examined and there remains genuine clinical equipoise over the benefit of this technique. To the best of our knowledge, formal meta-analysis of the role of BF formation at CS has not previously been published.

The BF (referred to outside the United States as “reflection of the utero-vesical peritoneum”) has traditionally been an integral step of the standard CS. It involves transverse incision of the vesico-uterine peritoneal fold to facilitate retraction of the bladder inferiorly, away from the lower uterine segment [2] and [12]. Recently, however, some authors have begun to question the value of routine BF formation at CS [2] and [12]. Two small randomised clinical trials indicate no benefit of BF formation [13] and [14]. We present a systematic review and meta-analysis of randomised controlled trials (RCTs) to assess the need for BF formation at caesarean section.

 2. Materials and methods
The systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines [15].

2.1. Search strategy
The MEDLINE database was searched in March 2012 using the search terms ‘BF’, ‘(caesarean OR caesarean) bladder’, ‘(caesarean OR caesarean) section bladder’ and ‘(caesarean OR caesarean) flap’. Embase, CINAHL Plus® with Full Text, Web of Science Reference and Cochrane Databases were also searched. No limits were set on the electronic searches and no language restrictions were used. The following clinical trial registries were searched for related entries: www.clinicaltrials.gov, www.controlled-trials.com and www.ukcrc.org. Reference lists from eligible studies were scrutinised to identify further citations.

Potentially eligible reports were reviewed independently by two authors (H.O’N. and C.A.W.). Data were extracted from eligible trials for each of the chosen outcomes and recorded in a Microsoft® Excel (Microsoft, Redmond, Washington, USA) spread-sheet, which was cross-checked by two authors (H.O’N. and S.R.W.).

2.2. Study selection and outcomes
Studies were deemed to be eligible for inclusion in the meta-analysis provided that they met the following criteria: prospective randomised controlled trial, conducted in women aged ≥18 years, directly comparing the inclusion and omission of the BF formation at CS, with at least one clinical endpoint reported by the authors. Studies were excluded from the analysis if any of these criteria were not met. The primary outcome for the present meta-analysis was iatrogenic bladder injury during CS. Secondary outcomes were skin incision-to-delivery time, total operating time, estimated blood loss and duration of hospitalisation.

2.3. Statistical analysis
Pooled outcome measures, with 95% confidence intervals (95% CI) were calculated using the random effects model of DerSimonian and Laird [16]. The weighted mean difference (WMD) was determined for continuous variables and pooled odds ratios (ORs) were calculated for categorical variables. Heterogeneity, the variation in outcomes between studies, was assessed using Cochran's Q and the inconsistency index, I2. Significant statistical heterogeneity between the pooled studies was indicated by Cochran's Q when P < 0.05 and when I2 ≥ 50%. Cochran's Q is the standard test to measure the degree of heterogeneity between primary studies, but it is seen as unreliable when the number of trials is small. In this situation I2 should be used as it does not rely on the number of studies analysed, the treatment effect, or the type of outcome data [17]. The statistical analyses were performed using Statsdirect 2.5.7 (Statsdirect, Altrincham, UK). The 5% level was considered significant and all p-values are two-tailed.

2.4. Assessment of study quality and bias
Trial quality was evaluated using the Jadad score [18], outlined in Table 1. A score between 0 and 5 is assigned, based on the methodology employed by the individual study; the higher the score, the greater the trial quality and the less likelihood of bias. Jadad points are assigned for appropriate randomisation, double-blinding and description of dropouts and withdrawals within the trial. Assessment was compared with the criteria outlined by the Cochrane Collaboration® (2011).

Bias across studies was assessed qualitatively by visualisation of a funnel plot or measured quantitatively using the Egger test [19]. In the absence of bias, the funnel plot should resemble a symmetrical inverted funnel; if there is no publication bias, all studies should theoretically be within the diagonal lines, which represent the 95% confidence interval. Asymmetry of a funnel plot in itself, however, is not proof of bias [17]. The Egger test is a test of asymmetry of the funnel plot; significance was set at 5%. Publication bias could not be entirely avoided in this review.

3. Results
In total, 5085 citations were retrieved from the primary searches. After initial screening, we identified 92 relevant citations, of which 37 were review articles and were excluded (Fig. 1). Of the remaining 55 citations, 17 potentially eligible publications and one registered clinical trial were identified. After comprehensive assessment of these manuscripts, four randomised controlled trials meeting our inclusion criteria were identified [13], [14], [20], and [21], which form the basis of the present analysis (n = 581 women; Table 1). Assessment of Jadad scores indicated lower study quality for two of the included RCTs [13] and [20] and higher study quality for the remaining studies [14] and [21].

Three of the studies included in the present analysis are randomised controlled trials which have been published in full in peer-reviewed journals [13], [14], and [20]. The remaining RCT, by Helms et al., was identified from the clinicaltrials.gov trial registry (NCT00470288). Although this trial (“Randomised Controlled Trial of BF versus None”) has not been published in full, the authors were contacted and provided us with their primary data, to allow inclusion in the present analysis [21]. One of the included studies did not report standard deviation data for several of the outcome measures [20]. The authors of this study were contacted but additional data could not be obtained. Given the similarities in study size between the trials by Hohlagschwandtner et al. [13] and Malvasi et al. [20], we elected to use standard deviation data from the former during analysis of the latter, to avoid exclusion of the Malvasi trial on this basis alone.

3.1. Bladder injury
Three studies reported bladder injury [14], [20], and [21]. Overall, no difference in the rate of bladder injury was found with (0.67%; 2/299) or without (0.7%; 2/282) BF (pooled OR, 0.96; 95% CI 0.19–4.84, p = 0.96; Fig. 2). There was no evidence of heterogeneity (Cochran's Q = 0.004, p = 0.998 and I2 = 0%, 95% CI = 0–72%), and there were insufficient trials to construct a funnel plot or to calculate bias with the Egger test.

3.2. Skin incision-to-delivery time
Three studies reported incision-to-delivery time [13], [14], and [21]. Overall BF formation was found to be associated with a longer incision-to-delivery time (WMD 1.27 min; 95% CI 0.63– 1.92 min, p = 0.0001; Fig. 3). There was no evidence of heterogeneity (Cochran's Q = 3.12, p = 0.21 and I2 = 35.9%, 95% CI = 0–81%), and there were insufficient trials to construct a funnel plot or to calculate bias with the Egger test.

3.3. Total operating time
All four studies reported total operating time. Overall no difference was found (WMD 3.5 min; 95% CI = −0.19 to 7.16 min; p = 0.06; Fig. 4). There was evidence of heterogeneity (Cochran's Q = 17.53, p = 0.0005 and I2 = 82.9%, 95% CI 38.1–91.6%) and bias (Egger = 3.25, p = 0.07). The funnel plot was asymmetrical and two of the studies were plotted outside the diagonal lines.

3.4. Estimated blood loss
Three studies reported estimated blood loss [14], [20], and [21]. Overall no difference was found (WMD 42.4 ml; 95% CI = −32.3 to 117 ml, p = 0.27; Fig. 5). There was evidence of heterogeneity (Cochran's Q = 64.52, p < 0.0001 and I2 = 96%, 95 CI = 94.5–98%), but there were insufficient trials to construct a funnel plot or to calculate bias with the Egger test.

3.5. Duration of hospitalisation
Two studies reported days hospitalised [14] and [21]. Overall, no difference was found (WMD 0.07 days, 95% CI = −0.50 to 0.64 days, p = 0.81). There was no evidence of heterogeneity (Cochran's Q = 0.19, p = 0.66), and there were insufficient trials to construct a funnel plot or to calculate bias with the Egger test.

4. Discussion
Originally, the BF step was described in the pre-antibiotic era when it was believed that closure of the bladder peritoneum would limit spread of any intrauterine infection into the peritoneal cavity [2] and [13]. Since then, many practitioners have abandoned routine closure of the visceral peritoneum, as it appears to increase postpartum febrile morbidity [7] and [22]. Recent evidence-based guidelines from the National Institute for Health and Clinical Excellence (NICE) in the UK recommended that neither the visceral nor parietal peritoneal layers should be routinely sutured during CS [23]. In addition, there are reports where bladder injury has been postulated as being associated with creation of the BF [24], [25], and [26]. Assertions that non-formation of a BF at CS reduces iatrogenic bladder injury have not been formally tested due to the large numbers that this would require given the relative rarity of bladder injury at CS. With this in mind, level I evidence has begun to examine whether or not routine fashioning of the BF at CS is advantageous [14], [20], and [21].

The present analysis finds that omission of the BF step at CS results in a quicker delivery of the infant (by, on average, 76 s) without any difference in iatrogenic bladder injury, estimated blood loss or duration of hospitalisation. Although in most cases the clinical importance of this time saving is questionable, it may confer a benefit in cases of category I (“crash”) CS for prolonged fetal bradycardia. It must be noted, however, that these categories of CS were excluded in at least three of the studies in this meta-analysis [14], [20], and [21]. While it is reassuring to find that routine omission of the BF does not increase peri-operative complications, it is likely that the present meta-analysis is underpowered to detect differences in rare complications. For example, consent guidelines from the Royal College of Obstetricians and Gynaecologists (RCOG) in the UK cite a 0.1% incidence of bladder injury at CS [27]. As such, the present analysis of bladder injury by BF formation in 479 women (Table 2), is hugely underpowered to detect a significant difference in this outcome measure. An RCT to detect a reduction in bladder injury from 0.3 to 0.15% by omission of the BF, with 80% power at the 5% significance level, would require a minimum of 16,970 women in each arm. It is unlikely that such a trial would ever be considered, therefore necessitating extrapolation from current available data.

The first randomised trial testing the need for BF formation at CS was published more than 10 years ago by Hohlagschwandtner et al. [13]. They reported reduced operating time, incision-delivery interval, blood loss and analgesic requirements in 50 women undergoing primary CS without BF compared to a similar number in whom a BF was routinely incorporated. Until very recently, however, no other randomised studies had been published to shed further light on this issue. In 2011, Malvasi et al. reported the relationship between BF formation in 115 primary CS and evidence of intra-abdominal adhesions in a subsequent CS [20]. Adhesions at the time of repeat CS were graded by a surgeon blinded to the primary technique, according to the Adhesion Scoring Method of the American Fertility Society (none, mild or severe) [28]. Use of a BF at primary CS was associated with a significantly higher rate of, and denser, adhesions between the bladder and the uterus in subsequent CS [20]. Most recently, a good quality RCT from Tuuli et al. randomised 258 women to (non-) formation of the BF. The only significant finding was a shortened incision-delivery time in the non-BF arm and the authors concluded that omission of this step does not increase complications [14]. Given the dearth of level I evidence on this clinical issue, we also elected to include data from an unpublished RCT, which was identified from an online trial registry (NCT00470288), in the present review.

Any assessment of the role of BF formation during CS must distinguish between primary and repeat procedures. A review of risk factors for bladder injury at CS reported that two-thirds of cases of bladder injury were in women with a history of previous caesarean delivery. Overall, previous CS increased the risk of intra-operative bladder injury almost fourfold [25]. In two of the trials included here, women with previous CS were excluded [13] and [20], but the largest trial published to date – by Tuuli et al. – stratified outcomes into primary and repeat elective CS. The authors reported no significant interactions between BF technique and primary versus repeat CS except the finding of a shorter incision-delivery interval, which was seen only in the primary procedure sub-group [14].

Although our results are encouraging for clinicians who routinely omit the BF step, there are some populations of patients in whom the issue remains unclear. None of the trials included here recruited women requiring emergency intrapartum CS. Women undergoing intrapartum CS have a higher incidence of bladder injury than pre-labour CS [25], likely due to thinning of the labouring uterus and the need to disimpact the fetal head. Any conclusions regarding the safety of omitting the BF must be interpreted with caution in women undergoing emergency intrapartum CS, particularly in the second stage. The second group of women in whom the role of BF formation remains unclear are the preterm population. None of the trials included here recruited women at <32 weeks’ gestation. Although a clear association between gestational age at CS and bladder injury has not been demonstrated [25], care should be exercised in women undergoing very preterm CS, in whom the lower uterine segment may be poorly formed.

We acknowledge several potential limitations with the present analysis. Firstly, although an assessment of study quality is mandatory in meta-analyses, three of the RCTs included here had a Jadad score of ≤2 (Table 1), indicating poorer quality. Additionally, there were insufficient trials to construct funnel plots or to calculate bias with the Egger test for all but one secondary outcome, ‘total operating time’. For this variable, the funnel plot was asymmetrical and two of the studies were plotted outside the diagonal lines (95% confidence interval). The Egger test was not significantly different and there was an absence of studies in the lower left part of the plot (where negative or null studies would be located). Together these analyses suggest the possibility of publication bias and may point towards the existence of more unpublished studies.

Secondly, it must also be noted that not all eligible RCTs reported the same outcomes: all studies reported ‘total operating time’, three studies reported on three other variables and only two studies reported on ‘days hospitalised’ (Table 2). Furthermore, as discussed earlier, one of the studies included in the meta-analysis did not report on the range or standard deviation associated with their published results [21]. Our efforts to obtain these data directly from the authors were unsuccessful and a decision was taken to use standard deviation data from a similarly sized trial [22]. Thirdly, additional surgical techniques during CS can potentially influence the outcome measures incorporated here. Three techniques routinely omit the BF step – the Pelosi-type caesarean, the modified Misgav–Ladach technique and the modified Joel–Cohen–Stark technique [29], [30], and [31]. These various surgical techniques differ, however, in their approach to tissue dissection, uterine extension and peritoneal closure, all of which may influence outcomes such as operative time, blood loss and post-operative pain [32].

Finally, and perhaps most importantly, caution must be exercised in over-interpreting the results presented in the present analysis to all clinical scenarios. These data do not include those patients who require the most expeditious caesarean delivery, that is emergency CS during labour, particularly at full cervical dilatation.

Nonetheless, to our knowledge this is the first meta-analysis of RCTs assessing the benefits of routine BF formation at caesarean section. Our results find that omission of the BF at non-emergency CS does not appear to increase iatrogenic bladder injury, estimated blood loss or duration of hospitalisation. Furthermore, not creating a BF appears to marginally, but significantly, improve the skin incision-delivery interval. The effect of BF formation on rare complications, in women undergoing preterm CS or in women undergoing emergency CS during labour, remains unclear and warrants further study. Furthermore, long-term effects such as operative morbidity, healing and integrity of the uterine scar, effect on adhesions and incidence of placenta accreta need to be assessed.

5. Conclusions
In women undergoing elective caesarean delivery, the routine creation of a bladder flap does not appear to improve peri-operative outcomes and may be associated with a longer incision-to-delivery interval. Omission of the BF results statistically in a quicker delivery of the neonate. The role of BF formation in women undergoing very preterm and emergency intrapartum CS remains unclear.