Radical Cystectomy in England from 2013 to 2019 on 12,644 patients: An analysis of national trends and comparison of surgical approaches using Hospital Episode Statistics data

Abstract Introduction We evaluate the data of 12,644 Radical Cystectomies in England (Open, Robotic and Laparoscopic) with trends in the adaption of techniques and post‐operative complications. Methods This analysis utilised national Hospital Episode Statistics (HES) from NHS England. Results There was a statistically significant increase (P < .001) in the number of Robotic assisted radical cystectomies from 10.8% in 2013‐2014 and 39.5% in 2018‐2019.The average LOS reduced from 12.3 to 10.8 days for RARC from 2013 to 2019 similarly the LOS reduced from 16.2 to 14.3 for ORC. The rate of sepsis (0‐90 days) did rise from 5% to 14.5% between 2013‐2014 and 2017‐2018 for the entire cohort (P < .001). Acute renal failure (ARF) increased over the years from 9.5% to 17% (P < .001). The rate for fever, UTI, critical care activity and ARF were higher for ORC than RARC (P < .001).The comparison of all episodes within 90 days for conduit versus non‐conduit diversions showed significantly higher rates of sepsis, infections, UTI and fever in non‐conduit group .Overall complications were significantly higher in non‐conduit group throughout the duration except was year 2016‐17(P < .001).The robotic approach has increased in last 5 years with nearly 40% of the cystectomies now being robotically in 2018‐19 from the initial percentage of 10.8% in 2013‐14. Conclusion This evaluation of the HES data from NHS England for 12,644 RC confirms an increase in the adoption of Robotic Cystectomy. Our data confirms the need to develop strategies with enhanced recovery protocols and post‐operative close monitoring following Radical Cystectomy in order to reduce post‐operative complications.


| ME THODS
This analysis utilized national Hospital Episode Statistics (HES) data from NHS England, containing information on inpatient admissions and outpatient appointments for all English NHS Clinical Commissioning Groups (CCGs) 21 as reported at the time of patients' interaction with the healthcare system. HES data were accessed using Harvey Walsh Health Informatics (Cheshire, UK) as a licensed intermediary. The work was supported by a research grant from Intuitive Surgical (California, USA). The data were pseudonymized at a source, precluding the need for ethical approval. Data are recorded on a real-time basis, avoiding any potential recall bias. Each "episode" is defined as an inpatient admission during which patients are assigned a diagnosis coded for in the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10). 22 The HES-recorded procedure-specific codes (Classification of Intervention and Procedure Codes or  were used to identify patients from 2013 to 2019 and to classify each patient in to an operative group. 23 The exact OPCS-4 codes used are outlined in Annexure 1. Using this information patients were separated into three groups, namely ORC, Laparoscopic cystectomy, and RARC. Patients were also stratified according to urinary diversion technique, namely conduit or non-conduit diversion.
Patients who were either readmitted or attended an accident and emergency department in the first 90 days following their index procedure were also identified, along with the ICD-10 diagnosis code corresponding to the reason for readmission. These data were available from years 2013-2019 only.
For the cost analysis, data for the cost per procedure as well as the cost of re-admission per procedure were evaluated along with the global burden of expenditure in each modality for these past years.    (Table 2).

| Length of stay
Mean LOS for the entire cohort was 12.5 days. Across all surgical approaches, there was a trend toward a decreased average length of spital stay (LOS) between 2013 and 2019 although this failed to reach statistical significance ( Figure 2

| 341
Following propensity-score matching, comparisons between RARC and ORC are shown in Table 2. Median LOS following RARC was 8 days for RARC, compared to 11 days for ORC (P < .001).

| Post-procedure events
Among all 12,644 patients, the proportion for whom there were recorded ICD codes for "post-operative events" within 90 days were consistently in the region of 90%. as well as for The rate of recorded sepsis rose from 5% to 14.5% (P < .001) ( Table 2). The recorded acute renal failure/acute kidney injury (AKI) increased over time from 9.5% to 17% (P < .001). The proportion of patients listed for critical care activity and cardiac episodes also rose (P < .001). However, 90-day in-hospital mortality was stable across time periods (P = .270). A trend toward increasing rates of recorded sepsis, fever, and acute renal failure was seen across all surgical approaches ( Table 3). Comparisons of 90-day events between RARC and ORC, following the propensity score matching, are outlined in Table 3.
The proportion of patients undergoing ORC that required critical care activity increased over time (P < .001) and had an increased rate of cardiac events (P < .007) ( Table 3). Between surgical approaches, there was a consistently higher rate of infection and colostomy among patients undergoing open procedures compared to robotic or laparoscopic (P < .001).
Following propensity score matching, the rates of critical care activity, were significantly lower for RARC patients, compared to ORC (P < .001), and were lower still for patients undergoing laparoscopic RC (P < .001) . Colostomy rates were 11.9% for ORC, compared to 7.8% for RARC (P < .001).
When comparisons were made between those with conduit vs non-conduit diversions, significantly higher rates of sepsis, infections, UTI, and fever were seen in the non-conduit group. Overall complications were significantly higher in the non-conduit group throughout the time period (P < .001).

| Readmissions
The overall rate of 90-day non-elective readmissions (NER) across  (Table 4). Following propensity score matching, the rate of 90-day NER was significantly lower for LRC than either RARC or ORC (P = .015).

| Cost implications
The  (Table 4).  (10,27) This analysis shows RARC is becoming more widely utilized in England. 10 We demonstrate that length following propensity score matching, across the whole data period of stay (LOS) shows a declining trend across all surgical approaches, including following RARC, for which an average LOS of 10.8days was recorded in 2018-19 which is comparable to the established literature. 28 LOS following RARC was shorter than ORC. Other factors can influence and impact the length of stay in NHS practice such as social factors which may affect the ability to compare this data internationally. LOS reductions across all surgical approaches may well be related to enhanced recovery protocols post-surgery, upon which there has been significant emphasis.. The apparently higher rates of colostomy, infection, LOS, and readmission in ORC may be related to more complex cases or more advanced disease. Without data on tumor characteristics, we cannot fully adjust for this here. However, we noted that comorbidity rates were similar in each surgery type.

| D ISCUSS I ON
Non-elective readmissions are a major factor in the cost effectiveness of cystectomy approaches. We observed that the NER rate following RARC declined, and the rate following ORC increased over our study period. However, given their respective starting points, they converged toward a similar rate of around 25%. This may be due to the fact that RARC was in its early days of adoption in the United Kingdom in the initial years of data collection, whereas ORC was well-established. Alternatively, it may relate to changes in patient selection over time. Similar comparisons in other datasets have shown no major difference between the surgical approaches in terms of 30day readmission rates, 13,29 although RC it is the authors' opinion that readmission rates should be quoted until at least 90-days, as per our analysis. Of note, any in-person interactions between a patient and the health care system in the post-operative period will be reflected as NER in HES coding which may, for example, include attendances for catheter complications or stoma nurse clinic visits. It is reassuring that despite this potential over-estimation, overall 90-day NER rates in the United Kingdom are comparable to global data.
It is interesting to note the overall rate of recorded sepsis within 90-days post cystectomy increased across our study period. This appears unlikely to be due to clinical reasons, but rather, may be an effect of increased awareness and reporting of the condition. The highest rate recorded was in 2015-2016, corresponding with an NHS campaign in 2015 to detect and treat sepsis early. 30 The American College of Surgeons National Surgical Quality Improvement Project database emphasizes that post RC, 25% of patients develop infections within 30 days with rates of sepsis being 12.7%. 31 That study reported that operative time >480 minutes was associated with surgical site infections (SSI), sepsis, and UTI and that perioperative blood transfusion positively correlated with higher rates of SSI and sepsis. 31 Indeed, it has been previously established that higher blood loss and the requirement of transfusions are more common in ORC than RARC, which was not further assessed in our work. 9,17,28 Other work has shown that operative times may reduce over a surgeon's learning curve, or may be lower in high-volume centers which may help minimize the infection-risk post RC. 31,32 Antibiotic strategies and levels of resistance may also play a role in this apparent increase in the infections reported by HES. Similarly, increased rates of AKI throughout the cohort may be related to changing awareness, and to evolving electronic blood reporting systems which may translate to increased coding of such conditions. | subjected to incorrect coding and overlapping terms such as UTI, infection, and sepsis resulting in skewed data. Accuracy of recording over time may not be consistent, and cannot be controlled for. This study was unable to assess functional and oncological outcomes, as no data are collected on this via the HES system, which could have helped contextualize metrics we have assessed such as LOS and readmission rate. Also, we were unable to report, or adjust for many patient or tumor characteristics. Propensity score-matching also has its limitations, particularly with regards to comparisons to the relatively small laparoscopic cohort. Nonetheless, the large sample size helps counteract some of these limitations and allows broad comparisons to be made.

| CON CLUS ION
This evaluation of NHS England HES data for 12,644 RC confirms a continued rise in the proportion of cystectomy being performed robotically. This paper emphasizes the need to further-develop enhanced recovery protocols and close post-operative monitoring of patients following radical cystectomy. RARC appears to have potential real-world benefits of reduced LOS and reduced rates of many 90-day post-procedural events including infection, cardiac events, renal failure, and critical care activity.

ACK N OWLED G M ENTS
Harvey Walsh Statistical support and Miss Smruti Mokal, Senior Statistician, Tata Memorial Hospital, Mumbai, India.

CO N FLI C T O F I NTE R E S T
Nothing to disclose.