Short-term morbidity and mortality following radical cystectomy: a systematic review

Abstract

Objective

To study short-term (<90 days) morbidity and mortality following radical cystectomy (RC) for bladder cancer and identify modifiable risk factors associated with these.

Design

Systematic review.

Methods

The systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. PubMed and EMBASE were searched for relevant papers on 11 June 2019 and rerun on 27 May 2020. Studies reporting complications, reoperations, length of stay and mortality within 90 days were included. Studies were reviewed according to criteria from the Oxford Centre for Evidence-Based Medicine and the quality of evidence was assessed using the Newcastle–Ottawa Scale.

Results

The search retrieved 1957 articles. Sixty-six articles were included. The quality of evidence was poor to good. Most studies were retrospective, and no randomised clinical trials were identified. Of included studies a median of 6 Martin criteria for reporting complications after surgery were fulfilled. The Clavien-Dindo classification for grading complications was most frequently used. The weighted overall complication rate after RC was 34.9% (range 28.8–68.8) for in-house complications, 39.0% (range 27.3–80.0) for 30-day complications and 58.5% (range 36.1–80.5) for 90-day complications. The most common types of complications reported were gastrointestinal (29.0%) and infectious (26.4%). The weighted mortality rate was 2.4% (range 0.9–4.7) for in-house mortality, 2.1% (0.0–3.7) for 30-day mortality and 4.7% (range 0.0–7.0) for 90-day mortality. Age and comorbidity were identified as the best predictors for complications following RC.

Conclusion

Short-term morbidity and mortality are high following RC. Reporting of complications is heterogeneous and the quality of evidence is generally low. There is a continuous need for randomised studies to address any intervention that can reduce morbidity and mortality following RC.

PROSPERO registration number

104937.

Strengths and limitations of this study.

• This systematic review can provide as a reference paper for future studies and when measuring quality of care.

• This systematic review emphasises the continuous need to identify and moderate risk factors for complications and optimise postoperative management plans to reduce both morbidity and mortality associated with radical cystectomy.

• This review is limited by heterogeneity in outcome measures of morbidity with a lack of clear definitions of surgical complications making a direct comparison between studies difficult.

Introduction

Radical cystectomy (RC) with pelvic lymph node dissection and urinary diversion is the preferred treatment for non-metastatic muscle-invasive bladder cancer (BC), and for some cases of high-risk non-muscle-invasive BC, in patients fit for major surgery.¹ RC is a comprehensive procedure that involves surgery to several organ systems and as a result it is associated with high postoperative morbidity and mortality. Attempts have been made over the years to reduce postoperative complications such as the introduction of Enhanced Recovery After Surgery (ERAS) programmes. However, addressing morbidity and mortality associated with RC across surgical cohorts remains important for preoperative counselling, planning of treatment, identification of modifiable risk factors to reduce morbidity and mortality, future clinical trial design and for assessment of surgical quality. Several measures of morbidity are clinically important such as complication rate, reoperation rate, length of stay (LOS), readmission rate and mortality. In this paper, we conducted a contemporary systematic review of the prevalence of short-term (<90 days) morbidity and mortality following RC for BC.

Methods

Search strategy and study selection

This systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.² A published protocol (PROSPERO) with prespecified outcomes, inclusion criteria and search strategy is accessible online.

A systematic literature search in PubMed and EMBASE was conducted on 11 June 2019 and rerun on 27 May 2020. A search string was created with the help of an information specialist (online supplemental appendix 1).

Articles were screened in a two-stage selection process. In the first stage, two authors (SLM and MAR) reviewed the abstracts. All prospective and retrospective studies on short-term (<90 days) morbidity and mortality after RC were included. Trials with less than 100 participants, indications for cystectomy other than BC, extended procedure (eg, nephroureterectomy), salvage/palliative cystectomy, organ sparing cystectomy (eg, partial cystectomy, prostate sparing cystectomy, vaginal sparing cystectomy, seminal vesicles sparing cystectomy), selected patient group (eg, certain age groups, women only), feasibility studies, surgical technique-only papers, animal series and studies not published in English were excluded. Conference papers, case reports, book chapters, review papers, editorials, comments, letters to the editors and abstracts were also excluded. When in doubt, studies were maintained for further review. In the second stage, the full text of all included articles was obtained and read by the same two authors. An agreement was reached through consensus using Covidence Systematic Review software.³ Any disagreement was resolved by discussion and the final decision was based on a consensus. In case of duplicate data/study the following criteria were applied in the selection: (1) outcome (studies reporting on complications were prioritised over LOS, mortality), (2) size of the cohort (larger studies were prioritised over smaller studies), (3) methodology (prospective studies were prioritised over retrospective studies and extraction of data from medical/hospital records over record linkage (eg, International Classification of Diseases codes in a database)), and (4) study period (studies with the most recent study period were prioritised).

Data extraction and quality assessment

The following data were extracted from all studies where possible: first author, data source (eg, single centre, multicentre, database), institution/country of origin, study period, year of publication, number of cases, study design, length of follow-up, the classification system used for grading complications, use of fast track/ERAS protocol, demographics (age, gender, body mass index, Charlson Comorbidity Index (CCI), American Society of Anesthesiologists (ASA) score, pT-stage, N-stage, neoadjuvant therapy, previous radiation therapy, prior abdominal/pelvic surgery), outcomes (urinary diversion, number of total complications, complication rate, segregated complications, complication reasons, mortality rate, LOS, reoperations, risk factors for outcomes).

The quality of reporting complications was estimated using the Martin criteria providing a score from 0 to 10.⁴ Furthermore, the level of evidence was rated according to criteria from the Oxford Centre for Evidence-Based Medicine.⁵ The methodological quality of the studies was assessed using the Newcastle-Ottawa Scale (NOS) for observational comparative studies.⁶

Outcome measures

The primary outcome was the overall complication rate: the number of patients with one or more complication(s) within 90 days after RC regardless of the classification system used. Secondary outcomes were the following: rate of graded complications according to severity grade used; frequencies of types of complications; LOS, reoperation rate; mortality rate; and risk factors for the development of outcomes of morbidity (eg, complications, death, reoperations).

Statistical analysis

Descriptive statistics were used. A weighted average and range were calculated for all rates. A meta-analysis on risk factors for morbidity was not possible due to the high heterogeneity of reporting in the multivariate analysis across studies.

Patient and public involvement

No patients were involved in conducting this review.

Results

The literature search retrieved 1957 articles after removing duplicates. Of these, 66 studies met the inclusion and exclusion criteria.^7–72 The process is outlined in figure 1.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart. From: Moher D et al.¹⁰⁵

Characteristics of included studies

The characteristics of the included studies are summarised in table 1. Twenty-nine studies (43.9%) were single-centre studies and 37 studies (56.1%) were register or multicentre database studies. Most studies (71.2%) were retrospective, retrospective studies of prospectively maintained databases (12.1%) or combined retrospective and prospective studies (4.5%). Only eight (12.1%) were purely prospective surgical series. Patients were operated in the period 1990–2018. Of included studies, only two reported that an ERAS protocol was used for the entire cohort,^{16 48} and in six studies an ERAS protocol was used in a part of the cohort.33 39 42 67 69 72 In the rest of the included studies, an ERAS protocol was not used, or the authors did not report on perioperative care.

Table 1.

Summary of patient characteristics

		Number of patients with data available (sum of references)	References
Demographics
Percentage of males (weighted average, % (range))	80.8% (71.1–99.1)	194 769	7–14 16–23 25 26 28–36 39 41–52 54–72
Age
Weighted median (range)	69 years (63–73)	69 076	7 9–11 13 14 16 17 19–22 26 27 29 30 43 45 47 48 50–52 55–58 60 61 63 65 68–72
Weighted mean (range)	68.2 years (56.2–72.1)	104 373	8 12 19 23 29 44 47 48 62 64 70 71
BMI
Weighted median (range)	26.1 (22.3–27.8)	10 332	9 10 14 16 21 22 26 27 30 35 45 47 48 61 63 65 69 71 72
Weighted mean (range)	27.6 (20.4–29.7)	13 187	8 12 17–19 28 29 44 47 48 54 59 62 64 67 70 71
ASA score (weighted average, % (range))
I	8.0% (0–35.1)	15 202	10 11 14 19 26 28–31 33 37 42 45 49 53 55 60 63 65 69–71
II	39.0% (1.7–81.9)	15 435	10 11 14 19 26 28–31 33 37 42 45 46 49 53 55 60 63–65 69–71
III	54.0% (7.9–94.0)	13 490	10 14 19 22 26 28–31 33 42 45 46 49 53 55 59 63–65 70 71
IV	4.8% (0–16.3)	12 287	10 14 19 26 28 29 31 33 42 46 49 53 55 59 64 65 70 71
CCI (weighted average, % (range))
0	45.6% (6.3–68.1)	114 334	7 16 20 36 42 50 58 60 70
1	26.7% (4.0–30.6)	85 875	16 20 34 36 50 58 60 70
≥2	20.9% (2.5–69.4)	88 159	16 20 22 34 36 48 52 58 60 64 66 70
Prior abdominal surgery (weighted average, % (range))	41.4% (5.1–55.1)	4214	8 12 16 18 21 29 35 45 61 62 71
Previous pelvic radiation* (weighted average, % (range))	5.5% (1.3–22.1)	3910	8 16 18 21 29 33 51 61 62 71
Neoadjuvant chemotherapy (weighted average, % (range))	13.2% (0–50.8)	23 678	8 10 12 14 16–18 22 25 26 33 35 37 41 45 46 49 53 59 60 62 63 65 68–72
Perioperative details
Surgical approach (weighted average, % (range))		107 822	8 11 13 16–22 25–31 33 35 36 39 41 43 45 47–51 53 54 60 62–64 70–72
Open	86.5% (0–100)
Laparoscopic	1.8% (0–100)
Robot-assisted laparoscopic	10.9% (0–100)
Type of diversion (weighted average, % (range))
Ileal conduit	85.0% (31.4–93.8)	80 675	8 10–14 16–19 21 22 26 29–31 33 35 36 42–51 53–55 60 61 63–72
Neobladder	10.5% (2.6–62.1)	65 307	8 10 12–14 16–19 21 22 26 29–31 33 35 36 42–49 51 53–55 60 61 63–67 69–72
Continent cutaneous diversion	1.25% (0–29.6)	59 853	8 12–14 16 18 26 29 30 36 43 44 46 48 49 51 54 55 60 61 63 65–67 69 71 72
Ureterocutaneostomy	1.35% (0–26.7)	58 210	8 17–19 22 30 31 36 44 47 48 54 60 63 64 66 67 69 72
Nephrostomy	0.01% (0–0.5)	56 718	8 18 19 22 31 36 44 49 51 54 60 63 64 66 67 69 72
Pathological tumour stage (weighted average, % (range))
≤T1	27.1% (6.4–54.9)	31 317	10–12 14 15 17–21 25–27 30 31 35 37 42 43 45 46 48 52 54 55 59 60 62–64 71 72
T2	29.1% (11.9–56.7)	28 916	10–12 14 15 17–21 25–27 30 31 35 37 43 45 46 48 49 52 54 55 60 62–64 71
T3	28.5% (10.8–42.4)	26 537	10 12 14 17–21 25–27 30 31 35 43 46–49 52 54 55 60 62–64 71
T4	13.2% (2.4–25.9)	26 537	10 12 14 17–21 25–27 30 31 35 43 46–49 52 54 55 60 62–64 71
Lymph node-positive disease (weighted average, % (%-range))	19.1% (6.3–44.4)	29 615	10–12 14–16 18–21 25–27 29–31 35 43 45–49 52 54 60 62–64 71 72
LOS
Weighted median (%-range)	11 days (4–39)	77 038	7 10 12–14 16 18 20–22 24 26 28 29 31 33 35 39 40 42 43 45 46 48 50 51 53 55 56 58 59 61–63 65 66 69 71 72
Weighted mean (%-range)	12.5 days (8.2–27.6)	39 562	8 19 25 38 39 44 47 48 50 53 54 58 64 67 68 70

*Not external beam radiation therapy due to bladder cancer.

ASA, American Society of Anesthesiologists; BMI, body mass index; CCI, Charlson Comorbidity Index; LOS, length of stay.;

Complications

Fifty-two studies reported on short-term complications as outlined in table 2. The most frequently reported follow-up period was 90 days. Three studies reporting short-term complications did not state the exact follow-up period and were therefore excluded from the complication rate analysis.^{33 40 67} During the primary hospitalisation, the overall complication rate was 34.9% (28.8–68.8). The complication rate increased with longer follow-up to 39.0% (27.3–80.0) 30 days and 58.5% (36.1–80.5) 90 days postoperatively. Minor complications accounted for 40.0% (19.9–77.4) and 38.2% (19.0–80.8) of the complications reported at 30 and 90 days of follow-up, respectively. Major complications after RC occurred in 15.5% (4.9–24.8) and 16.9% (13.4–32.0) of patients after 30 and 90 days, respectively. Rates of complications according to the Clavien-Dindo classification and reoperations are further outlined in table 2.

Table 2.

Complications and reoperations

Outcome	Complication rate, weighted average (%-range)	Number of patients with data available (sum of references)	References
In-hospital complication rate	34.9%* (28.8–68.8)	76 171	19 32 39 50 58 59 61
30-day complication rate	39.0%† (27.3–80.0)	19 160	9 18 23 28 30 43 44 46 47 51 53 55 60–62 70–72
CD grade I	9.2% (6.0–16.1)	1291	30 35 45 70
CD grade II	29.8% (20.6–52.5)	1291	30 35 45 70
CD grade IIIa+b	6.9% (5.6–14.4)	8749	28 30 35 45 70
CD grade IVa+b	7.8% (0.7–11.0)	8749	28 30 35 45 70
CD grade V	1.7% (0.0–2.1)	8982	28 30 35 45 46 70
Minor complication rate‡ (%)	40.0% (19.9–77.4)	2536	13 18 43 44 51 55 60 62
Major complication rate§	15.5% (4.9–24.8)	4499	13 18 30 43 44 46 51 55 60 62 70 72
90-day complication rate	58.5%¶ (36.1–80.5)	10 625	8 10 12 14 16 17 21 22 26 29–31 42 48 49 54 59–61 63–65 69 71 72
CD grade I	15.0% (4.0–31.6)	4442	29 30 54 59 61 64 69
CD grade II	38.9% (27.0–67.4)	4442	29 30 54 59 61 64 69 72
CD grade IIIa+b	20.5% (8.5–39.2)	5548	29–31 54 59 61 64 69 72
CD grade IVa+b	3.0% (0.2–8.5)	5548	29–31 54 59 61 64 69 72
CD grade V	3.5% (0.1–3.9)	55 440	29–31 36 48 54 59 61 64 69 72
Minor complication rate‡	38.2% (19.0–80.8)	56 955	8 12 16 17 21 26 31 36 42 59–61 63 69
Major complication rate§	16.9% (13.4–32.0)	59 068	8 12 14 16 17 22 26 29–31 36 42 49 59–61 64 69 72
Reoperation rate
30 days	5.8% (3.0–8.7)	11 598	9 21 27 28 30 44–46 53 62 71
90 days	12.3% (9.3–18.9)	1533	10 26 30 54 69

*One study²⁵ did not report on overall complication rate.

†Three studies^{13 35 45} did not report on overall complication rate.

‡Minor complications defined as Clavien-Dindo grades I–II, MSKCC grades 1–2 or minor complications.

§Major complications defined as Clavien-Dindo grades III–V, MSKCC grades 3–5 or major complications.

¶One study³⁶ did not report overall complication rate.

CD, Clavien-Dindo; MSKCC, Memorial Sloan Kettering Cancer Center.

Thirty-four studies8 10 12 14 16 18 19 22 26 28–31 33 35 36 42 45 46 48 49 54 55 59–64 67 69–72 classified complications according to the Clavien-Dindo classification,⁷³ six9 13 17 43 44 51 studies classified complications as minor and major complications, three studies^{21 27 72} used the Memorial Sloan Kettering Cancer Center-modified Clavien-Dindo classification,⁶¹ one study⁶⁵ used Common Terminology Criteria for Adverse Events⁷⁴ and nine studies23 25 32 39 40 47 50 53 58 did not use any system for grading complications.

Type of complications

Twenty-one studies reported on types of 90-day complications (table 3). Gastrointestinal (GI) (29.0%) and infectious (26.4%) complications were the most frequent.

Table 3.

Categories and type of 90-day complications

Category/type	Rate, weighted average (%-range)	Number of patients with data available (sum of references)	References
Gastrointestinal	29.0% (6.7–42.7)	6188	16 21 22 26 27 30 48 49 54 59 61 69
Ileus	16.5% (3.8–33.7)	5073	12 14 21 22 30 31 42 48 49 54 61 64 65 69
Small bowel obstruction	4.6% (1.7–9.0)	3193	14 21 26 31 48 49 54 61 64 65
Constipation	3.3% (0.5–11.4)	2491	12 14 22 48 49 61
Clostridium difficile colitis	2.3% (0.7–3.8)	2574	21 49 61 64 69
Diarrhoea	1.7% (0.6–5.6)	2392	12 22 48 49 54 61
Anastomotic bowel leak	1.1% (0.3–1.9)	3254	12 21 22 26 61 69
Gastrointestinal bleeding	1.0% (0.3–1.3)	2757	12 21 22 61 69
Infectious	26.4% (10.9–46.2)	5270	14 16 21 22 26 27 30 48 49 54 61 69
UTI/pyelonephritis	14.1% (1.1–29.7)	4297	12 21 22 26 42 48 49 54 61 64 65 69
Sepsis	4.2% (1.5–8.5)	3812	21 22 26 42 48 49 54 61 65 69
Fever of unknown origin	3.1% (0.6–4.8)	2966	12 21 22 49 61 69
Pelvic/intra-abdominal abscess	2.4% (0.1–4.3)	2836	21 22 48 61 65 69
Genitourinary	16.0% (6.0–23.5)	5697	17 21 22 27 30 54 59 61 69
Ureter stenosis	3.2% (1.7–7.0)	2539	12 21 22 26 48 61 65
Ureter leakage	3.1% (0.4–5.3)	4282	12 14 21 22 26 48 49 61 64 65 69
Wound	13.1% (5.6–27.0)	6424	12 16 17 21 22 26 27 30 48 59 64 69
Dehiscence	4.0% (1.3–4.9)	2722	26 31 49 61 69
Fascial dehiscence	1.6% (0.4–3.5)	2139	12 21 54 61 65
Infection	10.5% (2.4–19.3)	3827	14 21 26 31 42 49 54 61 69
Cardiac	6.1% (0.6–16.9)	5366	12 16 17 21 30 48 49 59 61 64 69
Myocardial infarction	1.1% (0.2–3.5)	4170	12 14 21 26 48 49 54 61 65 69
Arrhythmia	4.2% (0.2–14.4)	2923	12 21 48 49 61 69
Bleeding	3.5% (0.5–17.8)	2814	16 30 61 64 69
Haematoma	0.9% (0.7–1.2)	1096	12 14 65
Transfusion	23.2% (8.1–45.3)	2606	12 14 31 48 54 61 64
Respiratory	5.0% (1.3–11.5)	6845	12 14 16 17 21 27 30 48 49 54 59 61 69
Pneumonia	2.8% (0.6–5.9)	3639	12 21 26 31 42 48 54 61 69
Thromboembolic	3.6% (0.2–8.1)	4933	12 14 16 21 26 30 31 48 49 54 61 65 69
Neurological	2.8% (0.6–7.7)	4557	12 16 17 21 27 48 49 54 61 69
Renal failure	2.3% (0.5–6.7)	4070	12 14 21 22 48 49 61 65 69
Other
Fistula	1.1% (0.6–1.4)	1560	12 21 26 48 49 64 65
Lymphocele	2.1% (1.3–4.7)	3381	12 14 21 26 48 49 54 61 64 65

UTI, urinary tract infection.

Mortality

Fifty-three studies were included in the mortality analysis (table 4). The weighted average for the in-hospital mortality rate was 2.4% (0.9–4.7), the 30-day mortality rate 2.1% (0.0–3.7) and the 90-day mortality rate 4.7% (0.0–7.0). A total of 17 of 53 studies reporting on mortality stated the causes of death with 183 deaths reported.10 16 21 22 26 29–31 35 45 52 53 57 61 64 69 71 The most frequent cause of death was cardiopulmonary events accounting for 30% followed by progression of BC (15%) and sepsis (11%).

Table 4.

In-hospital, 30-day and 90-day mortality

	Mortality rate, weighted average (%-range)	Number of patients with data available (sum of references)	References
In-hospital mortality	2.4% (0.9–4.7)	87 848	32 41 50 56 58 61 66
30-day mortality	2.1% (0.0–3.7)	61 299	7–9 13–15 20 24 28 30 33 35 37 38 41 43–46 51 53 57 60–62 69–72
90-day mortality	4.7% (0.0–7.0)	108 717	7 8 10–12 14 16 17 20–22 24 26 29–31 33 36–38 41 48 49 52 54 56 59–62 64 65 67–69 71 72

Quality of studies

Only three of the included studies^{22 61 75} met 10 of 10 Martin criteria (online supplemental appendix 2). The median number of fulfilled Martin criteria was 6 (range 2–10). The only criterion fulfilled by all studies was defining the method of accruing data. The level of evidence according to criteria from the Oxford Centre for Evidence-Based Medicine was rated as 3 or 4. The methodological quality across studies was ‘poor’ to ‘good’ assessed using the NOS (online supplemental appendix 2).

Discussion

We systematically reviewed the literature to accurately describe short-term morbidity and mortality following RC and identify modifiable risk factors associated with these. The aim was to identify factors that could form the basis for the design of future randomised trials on postoperative interventions that can reduce the risk of complications.

Main results

RC is an extensive urological procedure and associated with a high risk of short-term minor and major morbidity. Mortality within 90 days of primary surgery is not negligible and occurs in 4.7% according to our review. Our systematic review underlines that complications occur in one in three patients during hospitalisation and that one in five patients have major complications during the first 30 days after RC. This emphasises the continuous need to identify and moderate risk factors for complications and optimise postoperative management plans to reduce both morbidity and mortality associated with RC.

Our analysis identified GI and infectious complications as the most frequently reported complications after RC. Overall, GI complications occurred in 29.0% with a postoperative ileus rate of 15.6%. Urinary tract infections (UTI) were the most frequently occurring infectious complications occurring in 14.1% of patients.

Risk factors for the development of GI complications

Based on the literature reviewed, it was not possible to identify the most important risk factors for GI complications and thus define whether these were potentially modifiable. The risk of ileus, which is often most clinically relevant, seems to be affected by many factors, most importantly increasing age. One study reported an increasing age as a statistically significant risk factor for ileus with an OR of 1.30 (95% CI 1.1 to 1.5) per 10-year increase of age.²³ This finding is supported by a large retrospective study (not included in this review) of 41 498 patients that found an increased OR of developing ileus with per 1-year increment in age (OR 1.012, 95% CI 1.009 to 1.014, p<0.05).⁷⁶ The study also found that the risk of ileus increased with several chronic conditions such as chronic pulmonary and neurological diseases. This underlines that reducing GI complication rates relies primarily on an overall medical assessment and that alternative treatment options should be considered in medically ill patients. Surgeons performing RC must be aware that poor general health status increases the risk of GI complications and entails a poor short-term outcome. Several studies of RC have promoted the implementation of ERAS protocols, which originate from colorectal surgery where ERAS reduces GI complications. In this review, no studies investigated the impact on GI complications with the use of an ERAS protocol versus a non-ERAS protocol in the perioperative care. Generally, there is limited evidence for ERAS in an RC setting.⁷⁷ Only the use of postoperative gum chewing, the use of alvimopan (a peripherally acting μ-opioid receptor antagonist currently not available in Europe) and controlled administration of perioperative fluid management (goal-directed fluid therapy) to avoid both fluid excess and hypovolaemia have been shown to reduce GI complications after RC in randomised clinical trials (RCT).^78–80 Comparative studies indicate that omitting the nasogastric tube and mechanical bowel preparation result in lower GI complications after RC.^{81 82} ERAS offers good practical guidelines, but the various elements such as early mobilisation, omitting pelvic drainage, perioperative body temperature monitoring and early oral diet are not studied individually but introduced in different modified versions with several components used together. Consequently, it is difficult to derive which factor has the largest impact on reducing GI complications. Four studies in this review investigated the impact on overall complications, but the results were conflicting.33 39 67 69 A meta-analysis of ERAS protocols versus traditional protocols has found a faster return of bowel function and lower overall complication rate in the group managed on an ERAS compared with a standard of care protocol, but the overall level of evidence in RC remains low with regard to ERAS implementation.^{83 84} A previous study described that only 20% of surgeons who endorse ERAS guidelines actually practised all interventions recommended by the ERAS society.⁸⁵

Risk factors for the development of infectious complications

The anatomical reconstruction of the urinary tract with the use of bowel as urinary diversion following RC will naturally increase the risk of UTI, which can prolong LOS and is leading to readmittance. Only three of the included studies identified multivariable risk factors that were statistically significant predictors of infectious complications 30 and 90 days after RC.^{23 28 49} Two studies found that continent reservoirs were associated with a higher risk of UTI compared with ileal conduits. Johnson et al found that any continent urinary diversion increased the risk of infectious complications compared with an incontinent urinary diversion (OR 1.68, p<0.001).²⁸ Nazmy et al found that an Indiana pouch increased the risk of UTI compared with ileal conduit (OR 3.55, 95% CI 1.33 to 9.44, p=0.01); however, an orthotopic bladder substitute did not increase the risk of UTI compared with an ileal conduit.⁴⁹ Other studies not included in this review investigating this association have shown conflicting results.^86–89 Hollenbeck et al found preoperative bleeding disorder, poor functional status, preoperative acute renal failure and a >10% weight loss preoperatively to be associated with an increased risk of UTI.²³ In general, the comorbid patient may be at the highest risk for UTI. A large retrospective study of 1133 patients found that a CCI >2 was associated with a higher 90-day postoperative UTI rate (OR 1.8, 95% CI 1.1 to 2.9, p=0.05) compared with a CCI 0–2.⁸⁶ It remains unclear if UTI can be prevented. No studies in this review investigated this question which in general is not well investigated. Pariser et al demonstrated that a change in prophylactic antibiotic protocol from a narrow to a broader coverage did not reduce the UTI rate, although the 30-day risk of overall infectious complications was reduced following RC from 41% to 30% (p=0.043).⁹⁰ A population-based American study reported a lower infectious event rate when using a combination of antibiotic prophylaxis compared with a single-agent antibiotic (OR 0.79, 95% CI 0.70 to 0.89, p<0.001).⁹¹ The authors also investigated if extended antibiotic treatment >24 hours after RC decreased the risk of infectious complications, but no such association was found. Currently, international guidelines recommend that broad-spectrum antibiotics are used in the prophylactic regimen considering the local microbiological environment.^{92 93} However, RCTs addressing antibiotic prophylaxis regarding type, timing and duration for the risk of UTI are lacking and warranted.

Risk factors for mortality

In several studies, patient-related factors, age and comorbidity, were identified as the most important factors for mortality at index hospitalisation, as well as 30 and 90 days following surgery.11 17 24 27 32 38 41 52 66 68

Other risk factors

In addition to patient-related factors, the impact of hospital volume, surgical experience and surgical technique has been addressed. Studies included in this review investigating the impact of hospital volume have shown conflicting results.15 32 34 37 41 56 66 However, a previous meta-analysis of studies not all included in the present review found that the risk of postoperative mortality after RC was decreased by 45% when performed at a high-volume centre compared with a low-volume centre (pooled estimated effect OR 0.55, 95% CI 0.44 to 0.69).⁹⁴ Studies also show that complications are reduced with both increasing hospital and surgeon volume.^{32 36} Unfortunately, the distinction between low volume versus high volume is not well defined and thus not comparable between studies. The European Association of Urology (EAU) Muscle-Invasive and Metastatic Bladder Cancer Guideline Panel recommends RC to be performed at centres with at least 10 RC/year and preferably >20 RC/year.⁹⁵ The surgical technique has been investigated in 12 of the included studies in this review.18 20 29 31 33 35 45 48 62 64 Open RC was compared with robotic-assisted RC in 10 of these non-randomised papers. In seven of these publications, a significantly reduced complication rate was found. However, five RCTs comparing open and robotic surgeries (not included in the review) did not find a difference in complication rates.^96–100 All RCTs have been conducted with extracorporeal urinary diversion performed and it is speculated that robot-assisted RC with intracorporeal urinary diversion may have a lower complication rate compared with open RC. The question in hand is currently being studied in the ongoing iROC trial (robot-assisted radical cystectomy with intracorporeal urinary diversion versus open radical cystectomy) ¹⁰¹.

Limitations

There are limitations of this review that must be addressed. Most of the included studies were retrospective which limits the clinical utility. It is important to notice that we excluded studies with less than 100 patients and studies investigating subgroups of patients such as certain age groups, types of urinary diversion or T-stages of BC. Since most RCTs on RC have less than 100 participants and often exclude patients with certain characteristics they were not included in this review.

Challenges in the comparison of RC studies

The difficulties of comparing RC studies are manifold. First, selection bias between cohorts must be expected. This is reflected by the wide range for the estimates of the weighted averages for ASA score and CCI (table 1). The selection of patients fit for RC is known to be associated with great variation among centres.¹⁰² Second, there was no standardised reporting of complications. Most used different classification systems for severity grade of complications with the Clavien-Dindo classification being the most frequent. Third, even when using a grading system as Clavien-Dindo with certain criteria, the scale can be interpreted differently or modified in some way. For example, some studies do not calculate blood transfusions as a complication even though it could be argued to be a grade II complication. Fourth, measures of morbidity can be defined differently across studies. For example, ileus is reported in up to 20% of patients undergoing RC. However, the reporting of ileus may be questioned as a previous systematic review found that ileus was defined differently across studies, and in the majority of included studies it was not defined at all.¹⁰³ There is an increased focus on more uniform reporting of morbidity following RC and the EAU has proposed authors to use quality criteria originally proposed by Martin et al.^{4 104} In the present review, only three studies fulfilled all the Martin criteria. A previous non-systematic review from 2007 found no study reporting on complications after RC fulfilling all the Martin criteria. Lastly, publication bias must be emphasised as an important limitation.

We refrained from a meta-analysis of predictors of morbidity and mortality in this review as the number of studies investigating the same risk variable and outcome was too small for analysis. Identifying clinical predictors may aid in the prevention of postoperative morbidity and mortality. Currently, there is no risk assessment tool to predict postoperative outcomes after RC, and little correlation is found between the most frequently used risk-scoring systems (eg, CCI and ASA score) and postoperative outcomes.⁷⁰ Nevertheless, in the included studies of this review, comorbidity was in multivariate logistic regression analysis consistently associated with a significantly increased OR of both complications12 17 22 23 32 49 51 58 60 61 72 and mortality.11 17 24 32 41 52 66 Surprisingly, there is a paucity of prospective studies studying the subjects of the most common complications after RC in order to identify clinical predictors of these. Furthermore, prospective randomised studies comparing different interventions/regimens are lacking.

Conclusion

This review shows that RC is associated with a high risk of morbidity and mortality. However, with thorough patient selection, experienced surgeons, treatment at a high-volume hospital and the implementation of an ERAS protocol morbidity and mortality can likely be reduced. Trials addressing medical or surgical interventions to reduce short-term complication are needed.

Data availability statement

Data are stored and available from the corresponding author (SLM) upon reasonable request and can be reused without any further permission.