- To evaluate early postoperative morbidity in patients undergoing either robot-assisted (RARC) or open radical cystectomy (ORC) for bladder cancer.
robot-assisted radical cystectomy
open radical cystectomy
systemic inflammatory response syndrome
Radical cystectomy with simultaneous pelvic lymphadenectomy can be considered the standard of care for carcinoma invading the bladder muscle and for selected high-risk non-muscle-invasive bladder cancers [1, 2]. Nevertheless, the well-established open surgical approach, open radical cystectomy (ORC), is still associated with significant peri-operative morbidity (49–68%) and notable peri-operative mortality rates (1.5–6.9%) [3-7]. Since the advent of the robotic surgical platform, a small but increasing number of institutions have introduced robot-assisted radical cystectomy (RARC) as a minimally invasive alternative to ORC, not least to reduce the frequency of early peri-operative complications; however, the body of evidence supporting such a reduction in the complication rate is still limited to the few existing comparative studies between RARC and ORC [8-14], with only three of the referenced six studies having shown a statistically significant difference [9, 11, 12]. In view of this, we evaluated the early postoperative morbidity of RARC and ORC in a single-centre prospective observational study. To our knowledge, this comparative study is one of the largest RARC series published to date.
We chose a study period starting in October 2007 because at this time a prospective observational study on early peri-operative complications in patients undergoing radical cystectomy in the framework of peri-operative ‘fast-track’ management had been initiated in cooperation with our Department of Anaesthesiology, Critical Care Medicine and Pain Therapy. Before August 2009, our routine surgical treatment for carcinoma invading the bladder muscle and recurrent or unresectable high grade non-muscle-invasive bladder cancer and selected cases of primary very-high-risk non-muscle-invasive bladder cancer (e.g. primary T1G3 with concomitant Tis) was ORC. Thereafter, after the installation of a four-arm daVinci surgical system (Intuitive Surgical Inc., Mountain View, CA, USA) in our institution in April 2009, and after achieving proficiency in robot-assisted radical prostatectomy, all patients were offered RARC. The only reasons for rejection of RARC and recommendation of ORC were contraindications for a steep Trendelenburg position, such as previous retinal or cerebral haemorrhage or glaucoma, and contraindications for pneumoperitoneum such as severe chronic obstructive lung disease or severe congestive heart failure. None of the patients in the ORC group and two patients in the RARC group had received neoadjuvant chemotherapy (cisplatin/gemcitabine).
Between August 2009 and August 2012, 100 consecutive patients underwent RARC performed by four surgeons (Table 1). All procedures were performed using a transperitoneal six-port approach (four-arm robotic setting) . Urinary diversion was always performed extracorporeally via a small lower median laparotomy, which was simultaneously used for extraction of the bladder and lymph node specimens. Intra-operatively, these specimens were always secured in retrieval bags to avoid tumour cell spillage. A cohort of 42 consecutive patients, who underwent ORC by four surgeons between October 2007 and July 2009, served as a control group (Table 1). In both the RARC and ORC groups a meticulous pelvic lymphadenectomy was routinely performed. In patients with an ileal conduit or an ileal neobladder, the ileoileal anstomosis was always completed with a single-layer end-to-end technique using interrupted seromuscular 3/0 polyglactin sutures. The uretero-ileal anastomosis in patients receiving an ileal conduit was performed using one of the two techniques described by Wallace in 1966 and 1970 [16, 17], respectively, where the head-to-head or side-to-side conjoined ureters are ‘capped’ to the oral end of the ileal segment. To construct an ileal neobladder, the suturing technique of Hautmann et al. (n = 18)  or Studer et al. (n = 11)  was used with the uretero-ileal anastomoses performed end-to-end or end-to-side. For any kind of uretero-ileal anastomosis, a running 4/0 monofilament (poliglecaprone or polydioxanon) suture with a semicircular technique was always applied.
|Surgeon||Experienced in ORC*||Number of procedures|
|ORC||RARC||RALP preceding first RARC|
In all patients, mid-thoracic epidural anaesthesia and postoperative analgesia, a low intra-operative fluid infusion protocol, removal of nasogastric tube at extubation, early postoperative mobilization, oral fluid intake beginning on the day of surgery, and further rapid return to regular diet (‘fast-track’ management) were applied . The fast-track management followed a well-defined and standardized protocol from the very beginning and remained unchanged throughout the study. We regularly inserted one to two abdominal drains, which were scheduled to be removed between the 3rd and 5th days after surgery when serum creatinine levels in the drainage fluid had been checked. Peri-operative antibiotics (second-generation cephalosporin and metronidazole i.v.) were started at induction of anaesthesia and routinely continued until the 5th postoperative day. Further postoperative treatments, e.g. prophylaxis of venous thromboembolism (in accordance with the guidelines of the German Association of the Scientific Medical Societies) , daily clinical examinations, removal of ureteric stents, laboratory monitoring etc., were carried out as per the department's clinical pathway and did not change during the study period.
Data on patient characteristics, clinical and pathological tumour characteristics and surgical and peri-operative outcomes were collected prospectively in a Microsoft Access database. The general health condition of the patient was appraised using the Charlson comorbidity index without age adjustment . We did not use the age equivalence index because Charlson et al., in their original work, noted that ‘In studies involving less than 5 years of follow-up, age may not be a significant predictor of mortality. Therefore, in studies with a short length of follow-up, the age equivalence index will probably not be useful’ . Follow-up after discharge from hospital was documented using a standardized questionnaire. All postoperative complications were assessed using the Clavien–Dindo classification system (Table 2) . When the study was initiated in October 2007, a follow-up of 60 days after surgery for documentation of postoperative complications seemed to be sufficient, but as an increasing number of publications were reporting 90-day postoperative complications we used the change from ORC to RARC as an opportunity to re-assess our method and likewise extended the routine follow-up to 90 days after surgery. Thus, data on the postoperative course were collected prospectively up to the 90th postoperative day for the RARC group and up to the 60th postoperative day for the ORC group.
The RARC and the ORC groups were compared with regard to patient characteristics, surgical data, tumour characteristics and postoperative complications. In addition to the overall complication rate (grades 1–5), we also analysed the rate of major complications (grades 3a–5) and the rate of only life-threatening complications (grades 4a–5). As different types of urinary diversion might have different effects on the development of complications, we also analysed a subgroup consisting only of patients with an ileal conduit, but a subgroup analysis of only patients with a neobladder did not seem reasonable because of small case numbers. In view of the fact that our RARC group included the initial learning curve of the robotic procedure in our institution whereas the ORC control group represented patients of a mature open-surgical series, we additionally compared the most recent 50 patients in the RARC group with the patients in the ORC group. Continuous variables are presented as mean (sd) values and median values with interquartile ranges (IQRs), and categorical variables as frequencies and percentages. We used the two-sided, two independent sample Wilcoxon rank-sum test for analysis of continuous variables and the two-sided Fisher's exact test or chi-squared test for analysis of categorical variables with two or more characteristics, respectively. A P value of 0.05 was considered to indicate statistical significance. Statistical analysis was conducted with sas software (SAS Institute Inc., Cary, NC, USA).
The RARC and ORC groups were similar with respect to age, body mass index and gender distribution (Table 3). In both groups, 38% of patients were ≥75 years old; however, according to the Charlson comorbidity index, a larger proportion of patients in the RARC group had serious comorbidities than in the ORC group. Table 4 gives a detailed overview of the surgical and peri-operative outcomes. There were no differences between the RARC and ORC groups regarding pathological tumour characteristics or types of urinary diversion, while longer surgery times, lower intra-operative blood loss and higher lymph node yields were seen in the RARC group. Because of the lower blood loss, there were less transfusion requirements in the RARC group in the peri-operative course. Intra-operative complications occurred in three patients undergoing RARC (two rectal lesions, one lesion of the small intestine) and in two patients undergoing ORC (one rectal lesion, one splenic lesion). Time to first bowel movement and postoperative hospital stay were similar in each group (Table 4).
|Gender, n (%)||0.216|
|Female||24 (24)||15 (36)|
|Male||76 (76)||27 (64)|
|Mean (sd)||71.4 (9.4)||69.0 (11.5)|
|Median (IQR) age||72.5 (66.4–76.9)||71.1 (61.8–77.8)|
|Body mass index, kg/m2||0.960|
|Mean (sd)||27.0 (4.6)||27.0 (4.5)|
|Median (IQR) age||27.0 (24.2–29.3)||26.5 (24.0–30.1)|
|Charlson comorbidity index score||0.058|
|Mean (sd)||1.4 (1.5)||0.8 (0.9)|
|Median (IQR)||1 (0–2)||1 (0–1)|
|0–1, n (%)||59 (59)||33 (79)||0.034|
|≥2, n (%)||41 (41)||9 (21)|
|Clinical stage, n (%)||0.478|
|cTis||9 (9)||3 (7)|
|cT1 + Tis||29 (29) + 14 (14)||12 (29) + 6 (14)|
|≥cT2||57 (57)||27 (64)|
|Clinical grade, n (%)||0.526|
|G2||26 (26)||9 (21)|
|G3||70 (70)||33 (79)|
|Pathological stage, n (%)||0.157|
|pTis||11 (11)||2 (5)|
|pT1 + Tis||17 (17) + 10 (10)||7 (17) + 4 (10)|
|pT2||28 (28)||15 (36)|
|pT3||28 (28)||17 (40)|
|pT4||11 (11)||1 (2)|
|Pathological grading*, n (%)||0.512|
|G2||20 (20)||9 (21)|
|G3||70 (70)||33 (79)|
|Pathological lymph node†, n (%) status||0.821|
|pN0||80 (80)||32 (76)|
|pN+||20 (20)||9 (21)|
|Surgical margin status, n (%)||1.00|
|R0||98 (98)||41 (98)|
|R1||2 (2)||1 (2)|
|Surgical duration [min]||<0.001|
|Mean (sd)||410 (68)||351 (92)|
|Median (IQR)||404 (354–455)||333 (296–389)|
|Type of urinary diversion, n (%)||0.111|
|Ileal conduit||76 (76)||31 (74)|
|Orthotopic ileal neobladder||22 (22)||7 (17)|
|Other urinary diversion||2 (2)||4 (10)|
|Modified ureterosigmoidostomy (Mainz Pouch II)||1||1|
|Continent cutaneous diversion with ileocecal reservoir||1||2|
|Lymph node count||<0.001|
|Mean (sd)||27.5 (11.0)||19.6 (8.8)|
|Median (IQR)||26.5 (20–33)||19 (16–25)|
|Estimated intra-operative blood loss, mL||<0.001|
|Mean (sd)||351 (170)||810 (621)|
|Median (IQR)||300 (200–400)||600 (400–1000)|
|Preoperative haemoglobin, g/dL||0.925|
|Mean (sd)||13.7 (1.7)||13.7 (1.6)|
|Median (IQR)||13.7 (12.3–14.7)||13.7 (12.3–15.3)|
|Transfusion requirements, n (%)||<0.001|
|No||73 (73)||17 (40)|
|Yes||27 (27)||25 (60)|
|Time to first bowel movement, days||0.413|
|Mean (sd)||2.3 (1.5)||2.3 (1.1)|
|Median (IQR)||2 (1–3)||2 (2–3)|
|Postoperative hospital stay, days||0.293|
|Mean (sd)||17.1 (7.6)||19.9 (12.0)|
|Median (IQR)||14.0 (12.5–18.5)||15.5 (13.0–21.5)|
Table 5 provides a summary of all postoperative complications observed in both groups in terms of their type and their severity according to the Clavien–Dindo classification system. In total, 141 adverse events occurred in 100 patients in the RARC group and 134 occurred in 42 patients in the ORC group. The mean (sd) number of complications per patient was, therefore, significantly lower in the RARC group: 1.4 (1.8), median 1 (IQR 0–2) vs 3.2 (2.0), median 3 (IQR 2–4); P < 0.001. Also, in those patients who experienced a complicated postoperative course, a mean (sd) of only 2.4 (1.7) adverse events per patient was observed after RARC (median 2 [IQR 1–3.5]) vs a mean (sd) of 3.4 (1.8) after ORC (median 3 [IQR 2–4]; P = 0.003).
|Complication, n (%)||RARC, N = 100||ORC, N = 42||P|
|Clavien–Dindo grade||Total||Clavien–Dindo grade||Total|
|Procedural||9 (9)||24 (24)||33 (33)||9 (21)||22 (52)||31 (74)|
|Superficial wound healing deficit||3 (3)||4 (4)||7 (7)||2 (5)||8 (19)||10 (24)||0.009|
|Deep wound healing deficit (e.g. vaginal or urethral stump)||1 (1)||1 (1)||2 (2)||–||1 (2)||1 (2)||1.000|
|Protracted abdominal pain||–||–||–||5 (12)||–||5 (12)||0.002|
|Incisional hernia||1 (1)||1 (1)||2 (2)||–||–||–||1.000|
|Burst abdomen||–||4 (4)||4 (4)||–||3 (7)||3 (7)||0.422|
|Lymphocele||1 (1)||3 (3)||4 (4)||–||3 (7)||3 (7)||0.422|
|Postoperative bleeding||–||–||–||–||1 (2)||1 (2)||0.296|
|Urinoma/pelvic abscess||–||2 (2)||2 (2)||–||1 (2)||1 (2)||1.000|
|Hydronephrosis due to reflux||1 (1)||–||1 (1)||1 (2)||–||1 (2)||0.506|
|Hydronephrosis due to anastomotic stenosis||–||4 (4)||4 (4)||–||–||–||0.319|
|Suture dehiscence of urinary diversion/tract||1 (1)||1 (1)||2 (2)||–||1 (2)||1 (2)||1.000|
|Urethroileal anastomotic stricture (neobladder)||–||1 (1)||1 (1)||–||–||–||1.000|
|Stoma stenosis||–||1 (1)||1 (1)||–||–||–||1.000|
|Parastomal hernia||–||–||–||–||1 (2)||1 (2)||0.296|
|Rectal lesion||–||1 (1)||1 (1)||–||–||–||1.000|
|Small bowel lesion||–||1 (1)||1 (1)||1 (2)||1 (2)||2 (5)||0.209|
|Rectovaginal fistula formation||–||–||–||–||1 (2)||1 (2)||0.296|
|Inguinal hernia||1 (1)||–||1 (1)||–||–||–||1.000|
|Drainage sutured to the fascia||–||–||–||–||1 (2)||1 (2)||0.296|
|Infectious||30 (30)||5 (5)||35 (35)||14 (33)||7 (17)||21 (50)|
|Febrile UTI||29 (29)||–||29 (29)||12 (29)||–||12 (29)||1.000|
|SIRS/Sepsis||–||3 (3)||3 (3)||1 (2)||5 (12)||6 (14)||0.02|
|Peritonitis||–||2 (2)||2 (2)||–||1 (2)||1 (2)||1.000|
|Pneumonia||1 (1)||–||1 (1)||1 (2)||1 (2)||2 (5)||0.209|
|Miscellaneous||28 (28)||1 (1)||29 (29)||26 (62)||0 (0)||26 (62)|
|Anaemia requiring transfusion||27 (27)||–||27 (27)||25 (60)||–||25 (60)||<0.001|
|Renal failure||–||1 (1)||1 (1)||–||–||–||1.000|
|Other||1 (1)||–||1 (1)||1 (2)||–||1 (2)||0.506|
|Gastrointestinal||16 (16)||4 (4)||20 (20)||27 (64)||4 (9)||31 (74)|
|Constipation||3 (3)||–||3 (3)||5 (12)||–||5 (12)||0.05|
|Diarrhoea||4 (4)||–||4 (4)||4 (10)||–||4 (10)||0.236|
|Paralytic ileus||6 (6)||–||6 (6)||14 (33)||–||14 (33)||<0.001|
|Mechanic ileus||–||2 (2)||2 (2)||–||2 (5)||2 (5)||0.582|
|Enterocolitis||3 (3)||–||3 (3)||2 (5)||–||2 (5)||0.632|
|Gastritis||–||2 (2)||2 (2)||–||1 (2)||1 (2)||1.000|
|Other||–||–||–||2 (5)||1 (2)||3 (7)||0.025|
|Vascular||7 (7)||0 (0)||7 (7)||8 (19)||3 (7)||11 (26)|
|Thromboembolic events||3 (3)||–||3 (3)||5 (12)||3 (7)||8 (19)||0.003|
|Hypertension||–||–||–||1 (2)||–||1 (2)||0.296|
|Hypotension||2 (2)||–||2 (2)||1 (2)||–||1 (2)||1.000|
|Lymphoedema of the leg||2 (2)||–||2 (2)||1 (2)||–||1 (2)||1.000|
|Neurological||5 (5)||2 (2)||7 (7)||4 (10)||0 (0)||4 (10)|
|Apoplectic insult||–||1 (1)||1 (1)||–||–||–||1.000|
|Transitory psychotic syndrome||2 (2)||1 (1)||3 (3)||1 (2)||–||1 (2)||1.000|
|Depression||1 (1)||–||1 (1)||1 (2)||–||1 (2)||0.506|
|Other||2 (2)||–||2 (2)||2 (5)||–||2 (5)||0.582|
|Respiratory||2 (2)||1 (1)||3 (3)||2 (5)||2 (5)||4 (10)|
|Pleural effusion||–||1 (1)||1 (1)||2 (5)||1 (2)||3 (7)||0.078|
|Pneumothorax||1 (1)||–||1 (1)||–||–||–||1.000|
|Respiratory failure||1 (1)||–||1 (1)||–||1 (2)||1 (2)||0.506|
|Cardiac||0 (0)||5 (5)||5 (5)||2 (5)||2 (5)||4 (10)|
|Cardiac arrest||–||1 (1)||1 (1)||–||–||–||1.000|
|Acute coronary syndrome||–||2 (2)||2 (2)||–||1 (2)||1 (2)||1.000|
|Arrhythmia||–||1 (1)||1 (1)||1 (2)||1 (2)||2 (5)||0.209|
|Heart failure||–||1 (1)||1 (1)||1 (2)||–||1 (2)||0.506|
|Death||–||2 (2)||2 (2)||–||2 (5)||2 (5)|
Overall, 39 patients (93%) in the ORC group and 59 patients (59%) in the RARC group experienced one or more postoperative complications of any Clavien–Dindo grade <60 and <90 days after surgery, respectively (Table 6). Major complications (grades 3a–5) were more frequent after ORC and, in addition, life-threatening and fatal complications (grades 4a–5) were more likely after ORC; however, the difference in the rate of grade 4a–5 complications only just failed to reach statistical significance. The relative risk reductions in patients who underwent RARC for any complications, major complications and life-threatening/fatal complications were 0.36, 0.44 and 0.63, respectively. Two patients died in each group. In the subgroup of only those patients with an ileal conduit as urinary diversion, we observed lower rates of complications of any Clavien–Dindo grade, of complications graded 3a–5 and of complications graded 4a–5 after ORC (Table 6). The relative risk reductions in patients in the RARC group for any complications, major complications and life-threatening/fatal complications were 0.37, 0.54 and 0.59, respectively. The comparison of the most recent 50 patients in the RARC group with the patients in the ORC group showed a significant reduction in favour of the RARC group, not only for any complications and major complications, but also for life-threatening/fatal complications (Table 6).
|Complications, n (%)||RARC*||ORC||P|
|Entire study population (100 RARC vs 42 ORC)|
|Grade 1–5||59 (59)||39 (93)||<0.001|
|Grade 3a–5||24 (24)||18 (43)||0.029|
|Grade 4a–5||7 (7)||8 (19)||0.068|
|Subgroup of patients with ileal conduit as urinary diversion(76 RARC vs 31 ORC)|
|Grade 1–5||43 (57)||28 (90)||0.001|
|Grade 3a–5||17 (22)||15 (48)||0.011|
|Grade 4a–5||6 (8)||6 (19)||0.102|
|Last 50 RARC patients vs 42 ORC patients (50 RARC vs 42 ORC)|
|Grade 1–5||26 (52)||39 (93)||<0.001|
|Grade 3a–5||11 (22)||18 (43)||0.043|
|Grade 4a–5||1 (2)||8 (19)||0.010|
In both the RARC and ORC groups, infection-related, procedural, gastrointestinal, miscellaneous and vascularadverse events represented the top five categories of postoperative complications (Fig. 1); however, while the rates of infection-related complications were similar after RARC and ORC (P = 0.132), the rates of gastrointestinal, procedural, miscellaneous and vascular complications were significantly higher after ORC (all P < 0.01). The difference in the rate of gastrointestinal complications between the RARC and ORC groups was mainly driven by higher rates of paralytic ileus and constipation in the ORC group (Table 5). Regarding miscellaneous complications, the difference was almost entirely caused by anaemia requiring transfusion. The rates of procedural complications differed not least because of higher rates of superficial wound healing deficits and protracted abdominal pain in patients in the ORC group. Vascular complications were more frequent after ORC mainly because of a higher rate of thromboembolic events. Systemic inflammatory response syndrome (SIRS) or sepsis and other gastrointestinal adverse events were also more common in the ORC group.
A marked reduction in peri-operative morbidity in patients with bladder cancer undergoing radical cystectomy by a minimally invasive robot-assisted technique could represent a strong argument for performing RARC rather than ORC, given that the oncological and functional outcome is not impaired; however, while it has been proven consistently that estimated blood loss and transfusion requirements can be reduced through the robotic approach [8-11, 13, 14] the existing literature still provides limited evidence for such a reduction in the rate of postoperative complications. Only three comparative studies have been able to demonstrate a statistically significant difference in favour of RARC [9, 11, 12]. In view of this, and bearing in mind further inconsistencies between existing studies, such as differing assessment periods (not specified  vs inpatient  vs 30-day [10, 13] vs 90-day complications [9, 11, 14]) and differing classification of complications (not specified [8, 12] vs Clavien–Dindo classification [9-11, 13, 14]), it becomes obvious that more research in this field is needed. The present study is a prospective observational study comparing early postoperative morbidity in patients undergoing RARC and those undergoing ORC with uniform application of the Clavien–Dindo classification system.
Our RARC and ORC cohorts were well matched with respect to patient characteristics, type of urinary diversion and tumour characteristics (Tables 3, 4). In addition, a higher Charlson comorbidity index score in patients undergoing RARC than in those undergoing ORC corroborates our policy of not selecting healthier patients for the robotic approach.
Like most other comparative studies we observed a longer operation duration [8, 10, 11, 13, 14] and a lower estimated blood loss [8-11, 13, 14] with less transfusion requirements [8, 9, 11, 13, 14] in the RARC group (Table 4). The longer operation duration in robot-assisted cases of ∼60 min in our hands can largely be ascribed to the extra operative steps associated with the robotic set-up, such as trocar placement, docking and undocking of the robot, and conversion to open surgical urinary diversion. The markedly lower intra-operative blood loss in the RARC group can probably be explained by the tamponade effect of the pneumoperitoneum used during laparoscopy [24-26], by more precise suturing of bleeding vessels achieved through optimized visualization  and by the effect of the steep Trendelenburg position on blood volume in the lower abdomen . As preoperative haemoglobin was measured in the patients in both groups, the lower intra-operative blood loss in RARC patients expectedly emered as the most important factor for the lower transfusion requirements in the RARC group in our series.
Regarding all events classified as complications on the basis of the Clavien–Dindo classification system, we observed significantly fewer complications after RARC than after ORC (Table 6). While in both groups most patients with a complicated postoperative course experienced only minor complications, the rate of patients with major complications was still considerable; however, RARC again was associated with significantly fewer major complications. The aforementioned results were verified in the subgroup of only patients with an ileal conduit as urinary diversion. In the subgroup of the 50 most recent patients in the RARC group we even observed a significant reduction in life-threatening/fatal complications. Such a significant reduction in the rate of postoperative complications after RARC has only been reported by Khan et al.  within an observation period >30 days after surgery. Ng et al.  found a significantly lower rate of major complications up to 90 days after RARC compared with ORC, but there was only a nonsignificant difference in the total rate of complications. In the study by Yu et al.  a significant difference in favour of RARC was only observed for inpatient complications.
The overall and major complication rates observed in our RARC cohort correspond well to the rates reported in recent publications (overall complications 33–80%; major complications 8–35% [9-14, 28-31]). The complication rates in our ORC group, however, are markedly higher than recently published rates (overall complications 49–74%; major complications 13–31% [3-6, 9-14]). A potential reason for this discrepancy seems to be the meticulous documentation of all complications according to the Clavien–Dindo classification system in our analysis. Although each of the referenced studies also used the Clavien–Dindo or modified Clavien–Dindo classification, some adverse events originally defined as complications by Dindo et al. are not uniformly recorded. First and foremost, we defined any blood transfusion as a grade 2 complication (Table 2). Such transfusions were necessary in 60% of our ORC group. By contrast, Novara et al.  reported a complication rate of 49% in their ORC series, taking no account of the transfusion rate of 67%. In addition, UTIs and ileuses, which represent two of the most frequent complications in ours and most other studies, were recorded in no case in their analysis. Shabsigh et al. , who reported a complication rate of 64%, likewise took no account of the transfusion rate of 66% in their ORC series. Khan et al.  also did not record blood transfusions that were administered in 58% of patients who underwent ORC as complications because they were defined as part of the standard care pathway. Furthermore, no case of UTI was reported. Ng et al.  reported a complication rate of 62% in their ORC cohort, including a rate of only 2.8% significant transfusions, although a mean of 3.65 transfusion units per patient was administered. In this context, Yuh et al.  likewise explained the higher complication rate in their RARC series compared with other studies by the fact that any blood transfusion was defined as a complication.
Apart from miscellaneous complications the most frequent complications by category in ours and in other studies were infectious, gastrointestinal and procedural (in some publications referred to as wound- and genitourinary-related events) [3, 5, 9, 11, 14, 29, 31]; however, while infectious complications occurred with similar frequency after RARC and ORC, gastrointestinal and procedural complications were more frequent after ORC (Fig. 1). In this context, apart from anaemia requiring transfusion, paralytic ileus had the most considerable impact on the greater complication rate in patients in the ORC group (Table 5). Interestingly, despite the clearly different rates of ileus in the RARC and ORC groups, no difference in time to first bowel movement was observed. Moreover, postoperative ileus even occurred in patients who had previously had a bowel movement. One reason for the high rate of paralytic ileus after ORC could be the greater blood loss that was identified as an independent risk factor in an ORC study . The underlying pathomechanism, however, remains unclear. Conversely, the minimal invasiveness of the robotic approach could have prevented bowel paralysis in patients in the RARC group. This assumption is corroborated by a four-arm randomized trial that proved laparoscopic surgery to be an independent predictive factor of faster recovery of gastrointestinal motility and improved clinical recovery after colorectal surgery . Although it remains speculative as to what extent this observation can be applied directly to patients undergoing cystectomy, the proposed main mechanism of postoperative ileus, namely intestinal inflammation triggered by surgical handling , occurs in all major abdominal surgeries. In this respect, the unequivocal advantage of the minimally invasive robotic approach over the open surgical approach is the fact that the view of the surgical field is facilitated through patient positioning and the pneumoperitoneum, without extensive bowel manipulation.
Further adverse events that significantly contributed to the difference in complication rates between RARC and ORC in our analysis were constipation, superficial wound healing deficit, protracted abdominal pain, thromboembolic event, SIRS/sepsis and other gastrointestinal-related events (Table 5). The lower constipation rate after RARC can be explained on the basis of the aforementioned. The lower rates of superficial wound healing deficits and protracted abdominal pain are most probably the result of the smaller abdominal incision together with the distinctly shorter and less extensive use of wound retractors in RARC cases. Our assumption that the reduction in thromboembolic events and SIRS/sepsis was attributable to the lower overall complication rate in the RARC group (because both events frequently represent sequelae of previous complications) is hypothetical. The only complication that was more frequent in the RARC group, although not statistically significant, was hydronephrosis resulting from ureteroileal anastomotic stenosis. This was also observed by Khan et al.  in patients who underwent laparoscopic cystectomy. The authors ascribed this to inadequate left ureteric mobilization during the initial learning curve. In our series this kind of complication likewise occurred in three patients on the left side and in one patient on both sides.
In the present study the lower complication rate achieved with the robotic approach did not translate into a significantly shorter hospital stay. Conflicting results are reported in the literature, where shorter hospitalization after RARC has been observed [9, 11] or not [10, 12-14]. Nevertheless, comparison of the results is difficult as patient expectations and the availability of outpatient care may vary considerably regionally. Furthermore, different healthcare systems may have different financial incentives for either early or late discharge from hospital. The median hospital stay of 14 days for the RARC cohort in our series might appear quite long to urologists from the USA, but is very similar to the mean hospital stay of 18.8 days reported in a recent large RARC series from a high-volume university hospital in Germany . Consequently, these authors concluded that the long hospital stay in their series cannot be used as a benchmark of the patients' general postoperative health status.
For patients undergoing RARC a favourable outcome in terms of peri-operative morbidity is only of value if at the same time the oncological aims are not compromised. A positive surgical margin rate of 2%, a median lymph node count of 27 and lymph node metastasis rate of 20% in our RARC cohort probably reflect oncologically adequate surgery.
The limitations of the present study include a lack of randomization, a small ORC sample size and differing follow-up periods for the RARC and ORC groups. Furthermore, despite the application of the Clavien–Dindo classification system, comparability with other studies using the same methodology can still be restricted since, as previously discussed, interobserver reproducibility of the classification system might be biased.
We observed that a significant reduction in early postoperative morbidity was associated with the robotic approach. Despite more serious comorbidities and a 30-day longer follow-up in the RARC group, these patients experienced fewer postoperative complications than those in the ORC group. Major complications, in particular, were less frequent after RARC. Clearly, as our analysis reflects a non-randomized single-centre experience, further randomized prospective multicentre studies are needed to confirm our results.
We would like to thank Ms Hilary Coleman (medical translator, MB) for linguistic revision of the English manuscript and Ms Usha Kreaden (Senior Biostatistician, Intuitive Surgical Inc.) for statistical revision, and Ms Michaela Vanberg and Ms Anne Pailliart for data processing.
YD and DK are official proctors for Intuitive Surgical Inc. MM, YD and DK received funding outside the submitted work.