Introduction

Iatrogenic injury of the bladder or ureter is a known complication of abdominal, pelvic or vaginal surgery. Potential sequelae include haemorrhage, sepsis, renal loss and death [1,2,3]. The majority of such injuries occur secondary to caesarean section and hysterectomy [2, 4, 5], with a rising proportion now due to ureteroscopy [6, 7]. However, there remains great variation in the estimation of the frequency of urological injury during these major obstetric and gynaecological procedures, which limits the mandate for quality improvement exercises. Therefore, the objective of this review is to determine the incidence of urological injury during caesarean section and each type of hysterectomy.

Materials and methods

Search strategy

Systematic searches were performed of the Cochrane Central Register of Controlled Trials (CENTRAL), Embase and Medline. Searches were performed by Title or Abstract, utilising keywords and Boolean operators as follows: (obstetric, gynaecolog*, gynecolog*, caesarean OR hysterectomy) AND (urolog*, kidney, renal, ureter, bladder OR urethra) AND (iatrogenic, accidental, inadvertent, injur* OR trauma). Grey literature was also searched and eligible, by review of the above search results, bibliographies of retrieved articles and proceedings of the 2010–2019 annual scientific conventions of the Royal Australasian College of Surgeons. Inclusion criteria were agreed upon by all authors. Our method for identifying and evaluating data complied with the Preferred Reporting Items for Systematic Reviews and Meta-analyses criteria [8] (Appendix 1, Fig. 1). This included pre-publication of our intended analysis on PROSPERO (CRD42020161389). Note that although this protocol was intended to restrict inclusion to studies of ≥100 women, this was later reduced to ≥50 women, to reduce instances where identification of insufficient studies precluded meta-analysis. After protocol publication, preventative measures was demoted to a secondary outcome, to allow the study to focus on injury incidence as the sole primary outcome. Identified studies were screened by title and abstract, followed by full-text review. Articles then progressed to data extraction, including review of references. Two independent authors performed study screening and data extraction using a pre-defined form, with a third author involved for instances of disagreement (Appendix 2). Data extraction was performed twice to confirm accuracy. The final list of included articles was determined by compliance with the inclusion criteria and with the consensus of all authors.

Fig. 1
figure 1

Preferred Reporting Items for Systematic reviews and Meta-Analyses flow diagram. **Based on title/abstract screen against study eligibility criteria

Study eligibility

Study eligibility was determined utilising the patient population, intervention, comparator, outcome and study (PICOS) method [9]. Eligible studies reported cohorts of ≥50 women (P) undergoing open or laparoscopic pelvic or abdominal surgery by an obstetrician/gynaecologist (I), were not required to have a comparator cohort (C) and stated raw incidence of iatrogenic injury to the urinary bladder or ureter (O). Eligible publications were original full-length articles, published in English between 01 January 2000 and 20 October 2019 (S). All databases and sources were last searched on 30 November 2019. Studies were collated and analysed separately based on procedure type (caesarean section, hysterectomy), access (open trans-abdominal, open vaginal, laparoscopic trans-abdominal, laparoscopically assisted vaginal, vaginal only, robot assisted) and histology (benign, malignant, any). Studies were ranked by cohort size. When >15 studies were identified within a given sub-group, only the largest 15 studies were included. This limitation, not prespecified in the Prospero protocol, was to ensure that meta-analyses were of manageable size.

Intended analyses

The primary outcomes were the incidence of iatrogenic bladder and ureteric injury during caesarean section and each type of hysterectomy. Secondary outcomes were identified risk factors for injury and suggested preventative measures.

Qualitative summary was intended for all data, tabulating the key features of included studies. Raw proportions of each injury type were combined in a random effects meta-analysis using the R package “meta” [10]. Studies with zero events were not excluded from this pooled analysis. All analyses were two-tailed and significance was assessed at the 5% alpha level. Injury rates were reported as events per 100,000 procedures.

Bias

The authors did not anticipate identifying any randomised controlled trials. Consequently, risk of bias was assessed with the Newcastle–Ottawa Scale, in accordance with the Cochrane Handbook [11, 12]. Each study was independently reviewed by two reviewers (GW, NK) against pre-defined criteria (Appendix 3). Instances of disagreement were resolved by consensus. Risk of bias was not used to exclude studies.

Results

Initial database searches returned 2,159 articles. After removing 972 duplicate results and a further 1,034 irrelevant publications based on title and abstract review, 153 articles were retrieved for full text review (Appendix 4). After excluding ineligible studies and applying the pre-specified limitation of the 15 largest cohorts in each procedure sub-category, 96 eligible articles were included, describing caesarean section or hysterectomy in 130 cohorts, totalling 1,741,894 women (Fig. 1, Table 1) [3, 13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107].

Table 1 Enrolled studies

One study was a randomised controlled trial [80], whereas all the others were non-randomised observational studies. All but 12 studies were retrospective in nature. Mean or median age was reported by 67 of the 130 cohorts and ranged from 27 to 63 years. Similarly, mean or median body mass index was available for 42 cohorts, and varied from 19 to 37 kg/m2. Average American Society of Anaesthesiology score or Charlson Comorbidity Index were available for only 4 [34, 36, 95, 103] or 3 studies [3, 56, 87] respectively. Meta-analyses of bladder and ureteric injury rates are presented in Table 2 and Appendix 5.

Table 2 Meta-analyses by procedure sub-group

Caesarean section

As >15 caesarean section cohorts were identified, the largest 15 were selected, representing 144,816 women [22, 27, 30, 35, 44, 59, 61, 70, 77, 80, 88, 90, 99, 105, 107]. In total, 312 bladder and 7 ureteric injuries were reported. Weighted pooled mean injury rates were 267 and 9 events per 100,000 procedures respectively.

Open abdominal hysterectomy (benign histology)

The 15 largest cohorts were selected, comprising 550,784 women [13, 14, 24, 32, 37, 44, 54,55,56, 64,65,66, 82, 91, 100]. Cumulatively, 5,138 bladder and 1,768 ureteric injuries were detected. Weighted pooled mean injury rates were 641 and 255 events per 100,000 cases respectively.

Open abdominal hysterectomy (malignant histology)

Four cohorts were found, representing 1,711 women [24, 28, 31, 102]. Together, 15 bladder and 15 ureteric injuries were described. Weighted pooled mean injury rates were 614 and 577 events per 100,000 cases respectively.

Open abdominal hysterectomy (any histology)

Nine cohorts were located, totalling 707,492 women [3, 15, 16, 63, 75, 78, 84, 97, 103]. Summatively, 6,252 bladder and 1,712 ureteric injuries were reported. Weighted pooled mean injury rates were 473 and 261 events per 100,000 cases respectively.

Open abdominal hysterectomy (peripartum)

Amongst 6,907 women in four cohorts, 477 bladder and 55 ureteric injuries were reported in total [3, 63, 73, 98]. Weighted pooled mean injury rates were 6,279 and 666 events per 100,000 cases, respectively.

Vaginal hysterectomy (benign histology)

The largest 15 cohorts were selected, comprising 144,856 women [14, 17, 21, 24, 33, 44, 54, 65, 66, 71, 82, 83, 95, 100, 106]. In aggregate, 1,323 bladder and 75 ureteric injuries occurred. Weighted pooled mean injury rates were 878 and 39 events per 100,000 cases respectively.

Vaginal hysterectomy (any histology)

From the six identified cohorts representing 52,492 women, 20 bladder and 100 ureteric injuries were noted [3, 15, 16, 75, 78, 103]. Weighted pooled mean injury rates were 295 and 122 events per 100,000 cases respectively.

Laparoscopic hysterectomy (benign histology)

Within the 15 largest cohorts constituting 50,114 women, 988 bladder and 222 ureteric injuries were observed [20, 34, 42, 43, 53, 57, 62, 64, 65, 72, 76, 82, 100, 101, 104]. Weighted pooled mean injury rates were 997 and 262 events per 100,000 cases respectively.

Laparoscopic hysterectomy (malignant histology)

Eight cohorts were found, comprising 1,541 women [36, 49, 50, 60, 85, 87, 89, 94]. Thirty bladder and 20 ureteric injuries were reported. Weighted pooled mean injury rates were 1,553 and 814 events per 100,000 cases respectively.

Laparoscopic hysterectomy (any histology)

Thirteen cohorts incorporating 27,022 women were identified, with 44 bladder and 221 ureteric injuries recorded [3, 19, 38, 40, 45, 46, 74, 75, 78, 93, 96, 97, 103]. Weighted pooled mean injury rates were 375 and 417 events per 100,000 cases respectively.

Laparoscopically assisted vaginal hysterectomy (benign)

Within 14 cohorts totalling 12,077 women, 65 bladder and 13 ureteric injuries were detailed [18, 29, 37, 41, 42, 47, 52, 55, 58, 67, 68, 79, 81, 92]. Weighted pooled mean injury rates were 445 and 87 events per 100,000 cases respectively.

Laparoscopically assisted vaginal hysterectomy (any histology)

Nine cohorts were identified, representing 40,688 women. A total of 81 bladder and 260 ureteric injuries were recorded [3, 23, 25, 26, 39, 48, 51, 69, 75]. Weighted pooled mean injury rates were 506 and 186 events per 100,000 cases respectively.

Robot-assisted hysterectomy (any histology)

Three cohorts were found, constituting 1,394 women [78, 86, 103]. Together, 3 bladder and 8 ureteric injuries were noted. Weighted pooled mean injury rates were 212 and 398 events per 100,000 cases respectively.

Risk factors for urological injury

Fifty-two studies identified one or more risk factors for bladder or ureteric injury (Appendix 6). In descending order of frequency, the most common elements were surgeon inexperience or low volume (18 studies), prior caesarean section (17), other previous pelvic surgery (14), adhesions (10), large uterus or tumour (10), endometriosis (8), cancer (5), radiotherapy (5), above average haemorrhage (5), low or high body mass index (5), placental adhesion disorder (4), concomitant surgery (4) and emergency procedure (2). Note that “surgeon inexperience or low volume” was self-defined by each study, and included undefined [28, 34, 49, 50, 59, 62, 77, 78, 93, 96], a surgical trainee [97], a consultant with <8 years’ experience [41] or having been primary operator for a given hysterectomy approach for fewer than 20 [23], 30 [40, 82], 50 [43, 87] or 100 cases [57].

Preventing urological injury

Strategies to reduce the risk of bladder or ureteric injury during caesarean section or hysterectomy were promoted by 35 studies (Appendix 7). From most to least common, recommended measures included improved anatomical knowledge (12 studies), strong uterine traction (12), careful dissection generally (9), prophylactic identification of ureters (8), bladder distension with fluid to clarify planes (7), avoiding diathermy near ureters (6), actively dissecting (as opposed to blunt traction) the bladder away from uterus (3), urethral catheterisation at the start of the case (3), prophylactic ureteric catheters/stents (3) and shielding the bladder with retractors (2).

Assessment of bias

The Newcastle–Ottawa Quality Assessment Scale suggested that the risk of bias was intermediate (46 studies) or high (35 studies) for most of the 96 works identified (Appendix 8). Key methodological and governance information was frequently absent, including typical post-operative follow-up (missing in 70% of studies), financial disclosure (missing in 73%), conflict of interest (missing in 53%) and ethics approval (missing or explicitly not present in 52%). Eighty four of the 96 studies were retrospective and therefore at an increased risk of selection bias. Publication bias was not assessed, given the studies’ heterogenous methodology and array of procedure sub-types.

Discussion

To our knowledge, this represents the largest systematic review to date of urological injury during major obstetric and gynaecological surgery. For clinicians performing caesarean section or hysterectomy, these findings may aid their daily practice in three ways. First, the pooled mean injury incidence may aid in counselling patients on the risk of bladder or ureteric injury for caesarean section or their specific surgical approach to hysterectomy. Our findings suggest that rates of bladder and ureteric injury are low for caesarean section and most approaches to hysterectomy. It is clear that both bladder and ureteric injury risk are highest during peripartum and laparoscopic radical hysterectomy. We believe that for peripartum hysterectomy this relates to poor visibility from haemorrhage from the gravid uterus and to placental invasion disorders, which may distort anatomical planes and invade the bladder, and for malignant hysterectomy to the desire to dissect widely to achieve negative margins and challenges from tumour infiltration. Analyses between tumour stage and injury incidence were not performed. Second, the collated risk factors can aid in the pre-operative assessment of a specific patient’s risk of urological injury. This information may be used to involve a senior surgeon, prophylactically identify and safeguard the ureters, refer to a tertiary centre or employ other cautionary steps. Third, the advocated preventative strategies can be both incorporated into routine practice and utilised more intensively in settings of known increased risk, as identified above.

This review’s relevance is underlined by the ongoing high volume and changing surgical approaches of major obstetric and gynaecological surgery. Globally, there is a trend towards more caesarean sections [108], with >30 million performed annually, comprising >20% of all births. Well over 1 million hysterectomies are performed annually [109], although the incidence is slowly declining in most [110, 111] but not all nations [112]. Simultaneously, as with other specialties, the approach to hysterectomy continues to shift towards minimally invasive means. Twenty years ago, abdominal (open) hysterectomy was the most common technique in both developed and developing nations [82, 110, 111, 113, 114]. From this baseline, minimally invasive approaches are now the most common in developed nations. Robot-assisted hysterectomy is the most common approach in the United States of America [115], laparoscopic hysterectomy predominates in Australia [114], Denmark [111] and Taiwan [113], whereas the vaginal approach is customary in Austria [116]. In developing nations, abdominal hysterectomy remains the norm [117].

Some authors recommended preventative strategies of routine cystoscopy (often with intravenous indigo carmine [3, 15, 23, 32, 40, 41, 43, 45, 51, 55, 93]) or prophylactic placement of ureteric stents. However, the evidence suggests that neither of these might be sufficiently sensitive or cost effective. Intra-operative cystoscopy seems a logical precaution, allowing prompt inspection for haematuria, intact urothelium, ureteric jets and blood from the ureteric orifices. However, where practised, cystoscopy is diagnostic and not preventative, being performed at the end of the gynaecological procedure to detect an injury that has already occurred. Furthermore, cystoscopy has low sensitivity for both bladder [91] and ureteric [19, 78, 91] injuries. Regarding prophylactic stents, randomised controlled trials of their use in gynaecological procedures have given mixed results regarding reduced rates of ureteric injury [118, 119]. However, ureteric stent use may reduce diagnostic delay and post-operative morbidity [120].

Many clinicians may not appreciate the significant risk of death in women with ureteric injuries. Although most bladder injuries are diagnosed intra-operatively, most ureteric injuries are detected post-operatively, with a typical diagnostic delay of 10–14 days [3, 40, 47, 65]. A study of >200,000 women undergoing hysterectomy found that compared with patients with no ureteric injury, patients with a delayed diagnosis of ureteric injury have significantly lower 1-year overall survival (99.7% vs 91.7%) [3]. The reasons for the reduced survival were not assessed. This 1 in 12 risk of death at 1 year, akin to stage IIIC colorectal cancer [121], is a terrifying prospect for these patients, who are predominantly women aged 30–50 years undergoing hysterectomy for a benign indication [110]. Separate to death is the inconvenience and morbidity of further interventions. Although some ureteric injuries may be managed by minimally invasive means such as ureteric stent insertion, most will require formal repair via ureteric reimplantation (neoureterocystostomy) [40, 46, 47, 122]. Some selected cases will require additional measures such as a psoas hitch, Boari flap, uretero-ureterostomy or bowel interposition [65]. Some require up to six further procedures [91].

The leading risk factor for urological injury was surgeon inexperience or low volume, identified by 18 studies [23, 28, 34, 40, 41, 43, 49, 50, 57, 59, 62, 77, 78, 82, 87, 93, 96, 97]. A strong demonstration of this is Mäkinen et al.’s Finnish study of >10,000 hysterectomies. This found that, compared with surgeons who had performed ≤30 laparoscopic hysterectomies, those with experience of >30 procedures had a significantly lower rate of injury to the bladder (2.2% vs 0.8%) or ureter (2.0% vs 0.5%) [123]. Gynaecological trainees and consultants early in their learning curve may benefit their patients by increasing their supervision during this time, as well as incorporating the most commonly advocated preventative strategies of improving their anatomical knowledge, and intra-operative techniques of strong uterine traction and careful dissection generally.

This review’s strengths are its comprehensive curation and critique of the literature. It is limited by the lack of randomised trials and the heterogeneous methodology of the studies included. Non-randomised studies are more prone to bias; thus, these results should be interpreted with caution. However, as pointed out by some of the identified studies [82], national registry-based observational studies may better reflect clinical reality in the hands of the “average” gynaecological surgeon than randomised controlled trials. Exclusion of non-English publications is another limitation. Additionally, this review’s inclusion criteria sought to balance broad inclusion with manageable data collection. The decision to include the 15 largest studies for each surgical approach, regardless of whether or not they observed any bladder or ureteric complications, may have reduced the number of studies for which sub-group meta-analysis was possible. Furthermore, the small number of eligible works identified for open abdominal hysterectomy for malignant histology (4 studies), vaginal hysterectomy for any histology (6 studies) and robot-assisted hysterectomy for any histology (3 studies) limits confidence in the findings for these sub-groups.

Many studies may have had inadequate follow-up to detect ureteric injuries. Sixty-eight of the 97 studies included did not detail their post-operative follow-up (Table 1). Of those that did, 19 out of 29 studies stated only their planned consultations, without measuring whether these occurred or not [3, 25, 36, 49, 54, 58, 59, 61, 62, 64, 71, 75, 77, 79, 80, 86, 90, 93, 95, 103]. As highlighted by Wang et al., “both short- and long-term follow up are required because complications may occur greater than four weeks after the initial surgery” [62]. Hence, this opaque provision of aftercare limits certainty that all events have been captured, and pooled complication rates may be higher than our findings suggest.

This study’s methodology confined its scope to bladder and ureteric injury. Caesarean section and hysterectomy may cause other urological complications, such as transient urinary retention, nerve injury causing atonic bladder [89], vesico-vaginal fistula or uretero-vaginal fistula [49]. These were not assessed by this review.

Conclusion

Caesarean section and most types of hysterectomy carry low rates of bladder and ureteric injury. Surgeon inexperience represents the leading risk factor for iatrogenic injury. Improved anatomical knowledge is the most commonly suggested preventative strategy. Obstetricians and gynaecologists should counsel the patient for her individual risk of injury, prospectively establish risk factors and implement preventative strategies to minimise risk.