Predictors of the use of orthotopic bladderreconstruction after radical cystectomy forbladder cancer: data from a pilot study of1756 cases 2004–2011Luke S. Hounsome, Gary A. Abel*, Julia Verne, David E. Neal† andGeorgios Lyratzopoulos*South West Public Health Observatory, Bristol, *Cambridge Centre for Health Services Research, Department ofPublic Health and Primary Care, University of Cambridge, and †Cancer Research UK Cambridge ResearchInstitute, Department of Uro-oncology, Addenbrooke's Hospital, Cambridge, UK

What’s known on the subject? and What does the study add?• How often orthotopic reconstruction should be used after radical cystectomy is uncertain. Male sex, younger age,

affluence, white ethnicity and treatment in specialist hospitals may be associated with more frequent use. More evidenceabout the level and likely variation in the use of orthotopic surgery is needed to establish whether there are inequalitiesand unmet need.

• In England during the study period orthotopic bladder reconstruction was likely to be used in about one in 15 patientstreated by radical cystectomy. This is lower than previously reported in US series or European studies. Men and youngerpatients were more likely to be treated by orthotopic reconstruction, as were more affluent patients and those with lessadvanced disease. Whether clinical reasons or patient choice can explain some of this variation is unclear. There was noevidence for variation between different English cancer networks.

• A specific procedure code to allow routine analysis of population-based nationwide data would be invaluable forongoing monitoring of potential inequalities and unmet need.

Objective• To examine variation in the use of orthotopic bladder

reconstruction. Variability in the use of orthotopicreconstruction may indicate potential for qualityimprovement.

Patients and Methods• We analysed data from the British Association of

Urological Surgeons Cancer Registry ComplexOperations data set and Hospital Episode Statistics,covering the period 2004–2011.

• Three-level (patient, consultant and cancer network)mixed effect logistic regression models were used toexamine sociodemographic and organizational variationin use of orthotopic reconstruction.

• The primary outcome was the odds ratio for use oforthotopic reconstruction for different patient groups.

Results• The final analysis sample included 1756 patients with

bladder cancer who were treated by cystectomy by 121

consultants in 17 cancer networks. Of these, 120 (6.8%)were treated by orthotopic bladder reconstruction by 49consultants in 14 cancer networks.

• In multivariable analysis, use of orthotopic surgerywas higher in younger patients (odds ratio [OR] = 0.37per increasing 10-year age group from 30–39 to �70,P � 0.001) and men (OR = 2.31, P = 0.005). Therewas also some evidence of less frequent use amongmore deprived patients (OR per decreasing deprivationquintile 1.17, P = 0.037) and those with advanceddisease (OR per increase in stage category 0.8,P = 0.037).

• After accounting for patient- and consultant-levelvariation, there was very limited variation in the use oforthotopic reconstruction between different cancernetworks.

Conclusions• Within the study context, use of orthotopic surgery was

relatively rare and variable between patients withdifferent characteristics but not between different cancernetworks.

© 2013 BJU International | doi:10.1111/j.1464-410X.2012.11644.x 1

• The extent by which this variation reflects variation inquality of care or patient choice is uncertain.

• Examining the dissemination of orthotopic surgery useusing nationwide data is advisable.

Keywordsorthotopic reconstruction, cystectomy, multi-levelregression

IntroductionRadical cystectomy can be a life-saving procedure forpatients with muscle-invasive bladder cancer. In England,approximately 1600 patients every year are treated bycystectomy [1] and in those patients reconstruction of theoutlet of the urinary tract is also required, frequentlyachieved through ileal conduit urostomy. Orthotopicreconstruction (also termed neobladder formation) isanother reconstructive option that offers potentialadvantages, as near-normal urethral voiding can beachieved after intensive rehabilitation and training.Continence rates of >90% during daytime and >75% atnight have been reported in large case series of selectedpatients with adequate follow-up, without compromisingpostoperative safety and longer-term recurrence rates[2–6]. Nevertheless, despite suggestions to the contrary [7],conclusive evidence that orthotopic reconstruction offersquality of life advantages over ileal conduit diversion islacking [8–13].

Orthotopic bladder reconstruction is a specialist operationrequiring access to urological surgeons with specifictraining and experience. The relative paucity of urologicalsurgery expertise combined with evidential uncertaintyabout the quality of life benefits may be responsible forvariable use of the procedure between different patientgroups and hospitals. Analysis of record-linked MedicareSurveillance, Epidemiology and End Results data hassuggested that affluence, younger age, white ethnicity, malesex and treatment at large/specialist hospitals are importantpredictors of continent reconstruction [14]. We know of nostudies examining predictors of the variation in use oforthotopic bladder reconstruction surgery in the UK orother European countries. Given the importance attachedby modern healthcare systems to patient choice andpotential quality of life benefits from the procedure it isdesirable that potential variation in the use of orthotopicreconstruction is further examined.

PatientsIdentifying patients using Hospital Episodes Statistics datawas not possible because there are currently no appropriateprocedure codes for post-cystectomy reconstruction typewithin the Office for Population Censuses and Surveysclassification [15]. We therefore identified patients treatedby cystectomy using data from the British Association ofUrological Surgeons (BAUS) Cancer Registry Complex

Operations data set, comprising data submitted voluntarilyby BAUS members [16]. Information on type ofreconstruction, patient and tumour characteristics andtreating consultant and hospital were obtained by linkinginformation from the BAUS Complex Operations data setto Hospital Episode Statistics or the main BAUS cancerregister using patient identifiers such as patient date ofbirth and postcode. For patient deprivation status, quintilecategories of the nationwide distribution of the incomedomain of the Indices of Deprivation 2007 were used.Characteristics of linked and unlinked patients werecompared to examine potential lack of representativenessof the linked patients and significance of differenceswas assessed by the chi-squared test. Using hospitalinformation, we assigned each patient into a ‘cancernetwork’. In England, cancer networks are organizationsresponsible for improving and assuring the quality ofcancer services. Each one of 29 networks typically serves ageneral population of 1 to 2 million and includes severaldistrict general hospitals and at least one larger, usuallyteaching, hospital [17].

MethodsWe initially described use of orthotopic bladderreconstruction by age, sex, stage and deprivation.Subsequently, multivariable mixed effect logistic regressionmodels were used to identify independent predictors oforthotopic reconstruction, adjusting for patientcharacteristics and organizational variation. This modellingstrategy recognizes the hierarchical nature of the data, withpatient-level observations being nested within consultantsand within cancer networks. Therefore the model estimatespatient-level variation (e.g. between patients of differentdeprivation status) unconfounded by the potentialconcentration of some patient groups in cancer networkswith higher or lower rates of orthotopic reconstruction.Furthermore, it allows us to quantify the degree ofvariation in orthotopic rates between cancer networks, alsoadjusting for variation between consultants withinnetworks. We focused on cancer networks as opposed toindividual hospitals because each NHS hospital hasarrangements to direct suitable patients to specializedcentres for orthotopic reconstruction if expertise forperforming this procedure is not available locally.

The models were run only on those networks whichperformed at least one orthotopic reconstruction, as

Hounsome et al.

2 © 2013 BJU International

recorded in the Complex Operations data set. The modelwas adjusted for sex, age group, income deprivationquintile, tumour stage and number of cystectomyprocedures performed by the cancer network during thestudy period. Because for some cancer networks thenumber of procedures recorded in the data setcystectomies was small, this number was log-transformedusing a base of 2: the respective odds ratio (OR) valueindicates the change in use of orthotopic reconstructionthat could be expected if the number of cystectomies in agiven network were to be doubled. Log-likelihood ratiotests indicated no benefit from including nonlinear(squared or cubed) terms for the number of networkcystectomies. Further, we included in the model randomeffect variables for consultant and cancer network (inaddition to the fixed effect variables described above).Lastly, to test whether variation by patient or tumourcharacteristic was ‘stable’ across cancer networks with eitherhigh or low number of cystectomies, we examinedinteractions between the number of network cystectomiesand sex, age, stage and deprivation groups by including inthe model each interaction term in turn. STATA 10 wasused for regression analysis.

ResultsInformation was initially available for 3832 patients treatedby cystectomy during 2004–2011 in English NHS hospitals.Records linkage in order to assign information onsociodemographic and tumour characteristics was possiblefor a total of 2946 patients (77% of the original sample).Unlinked patients (n = 886) were more likely to be men(P < 0.001); orthotopic reconstruction rates, however, werenot significantly different between linked and unlinkedpatients (P = 0.09, Appendix).

In univariable analysis, orthotopic reconstruction was morefrequently used in men (8% vs 4% in women, P = 0.006),younger patients (e.g. 1% in patients aged � 70 vs 23% inpatients aged 30–39, P < 0.001) and those with lessadvanced disease (7% in patients with stage 1 disease vs 3%in patients with stage 4 disease, P = 0.029) (Table 1).

Subsequent analysis relates to 1756 patients with bladdercancer who were treated by cystectomy by 121 consultantsin 17 cancer networks with at least one recordedorthotopic reconstruction operation. Of these 1756patients, 120 (6.8%) were treated by orthotopic bladder

Table 1 Orthotopic reconstruction use by patient characteristics – univariable analysis(n = 2946).

Patient characteristic All patients Patients treatedby orthotopic reconstruction,

number (%)

P*

SexMen 2175 143 (7) 0.006Women 743 21 (3)Unknown sex 28 2 (7)

Age group30–39 25 4 (16) <0.00140–49 104 21 (20)50–59 415 61 (15)60–69 1027 64 (6)�70 1382 17 (1)No age information 3 0 (0)

DeprivationMost deprived 456 30 (7) 0.0652 552 21 (4)3 522 29 (6)4 529 30 (6)Least deprived 497 37 (7)No deprivation information 400 20 (4)

Stage0 243 15 (6) 0.0291 459 34 (7)2 1172 72 (6)3 298 17 (6)4 217 6 (3)Unknown stage 557 23 (4)

*From univariable logistic regression models (age group, deprivation and stage treated as continuous variables)excluding patients with missing values and cancer networks with no orthotopic cases.

Variation in bladder reconstruction after radical cystectomy

© 2013 BJU International 3

reconstruction by 49 consultants in 17 hospital networks(Table 2).

In multivariable analysis there was evidence thatorthotopic reconstruction use was associated with fourvariables: sex (P = 0.005), age (P < 0.001), stage of disease(P = 0.037) and income deprivation (P = 0.037). Morespecifically, men were substantially more likely to betreated by orthotopic reconstruction (OR 2.31, 95% CI1.28–4.16). Orthotopic reconstruction use decreasedsharply with increasing age (OR 0.37 per increasing10-year age group from 30–39 to �70, 95% CI 0.30–0.45)but increased with decreasing income deprivation (OR1.17 per decreasing quintile of deprivation, 95% CI1.01–1.35). Each increase in stage category decreased thelikelihood of orthotopic reconstruction (OR 0.80, 95% CI

0.66–0.99) (Table 3). There was evidence that networkswith a greater number of cystectomies during the studyhad lower use of orthotopic surgery (P = 0.05) with anOR of 0.8 for a doubling in the network number ofcystectomies (95% CI 0.64–1.00).

There was very large variation between individualconsultants as indicated by a 95% mid-range of ORcompared with the ‘average’ consultant of 0.18–5.46. Thissuggests a nearly 30-fold variation between consultants inthe 95% mid-range around the mean.

After adjusting for the very substantial degree of variationbetween consultants, we observed very limited variation inorthotopic surgery use between different cancer networks,as indicated by the 95% mid-range of OR values of 0.999and 1.0001 compared with the ‘average’ network; this

Table 2 Summary statistics of consultant and cancer network activity (analysis restricted topatients treated in cancer networks with at least one recorded reconstruction operation).

Consultants Cancer networks

All cystectomies* (n = 1756) n = 121 n = 17Mean 14.7 103.3Median 6 58IQR 2–16 35–141Range 1–121 2–406

Orthotopic reconstruction (n = 120) n = 49 n = 17Mean 2.5 7.1Median 1 5IQR 1–3 2–8Range 1–15 1–29

IQR, interquartile range. *Includes patients with any type of reconstruction, including orthotopic.

Table 3 Predictors of orthotopic reconstruction use, from a mixed effects model for patient and organizational variables – multivariable analysis(n = 1756).

OR Lower 95% CI Higher 95% CI P

Patient characteristic variables (fixed effects)Sex

Men 2.31 1.28 4.16 0.005Women Reference

Age group 0.37 0.30 0.45 <0.001Deprivation quintile (1 is most deprived)* 1.17 1.01 1.35 0.037Stage category* 0.80 0.66 0.99 0.037

Organizational variables – fixed effectsTwofold increase in number of network cystectomies 0.80 0.64 1.00 0.05

OR – lower 95%mid-range limit‡

OR – higher 95%mid-range limit‡

Organizational variables – random effectsCancer network (random effect, SD = 0.000045) 0.99 1.0001 <0.001#

Consultant (random effect, SD = 0.87) 0.18 5.46

*The respective OR values indicate the change in OR for moving from one ordinal category to the one immediately higher, e.g. from age group 40–49 to 50–59, or fromdeprivation group 3 to deprivation group 4, or from stage 1 to stage 2 etc. Log-likelihood ratio tests of models treating these variables either as categorical or continuousindicated no evidence that either approach provided additional information (P values of 0.55 and 0.24, respectively). Therefore deprivation and stage were treated ascontinuous variables to optimize power. ‡=exp(�1.96 ¥ SD) where SD is the standard deviation of the random effect on the log OR scale. These limits denote the 95% mid-rangeof ORs around the mean hospital network or consultant (as applicable). #Joint test of random effects for cancer network and consultant.

Hounsome et al.

4 © 2013 BJU International

indicates hardly any degree of variation in orthotopicreconstruction between different networks (P < 0.001,Table 3).

There was no evidence that the observed associationsbetween sex, deprivation and stage and the use oforthotopic reconstruction varied between networks with ahigh and low number of cystectomies (interaction term P =0.111, 0.057, 0.368, respectively). There was evidence ofinteraction between age and network activity, with an ORof 0.91 for an ordinal increase in age group and doublingof network activity (P = 0.001). This indicates that innetworks with a lower level of activity the observed overallage gradients in orthotopic reconstruction use were lesssteep.

DiscussionSummary of Main Findings and Comparison withthe Literature

Within the study context, orthotopic reconstruction wasused in about one in every 15 patients undergoingcystectomy for bladder cancer. Its use was lower amongwomen, older patients, patients with a more advanced stageof disease and also among more deprived patients. Cancernetworks with a high volume of cystectomy procedures hada lower proportion of patients undergoing orthotopicbladder reconstruction. There was very limited variation inorthotopic surgery use among different cancer networks,once variation in orthotopic use between consultants wastaken into account.

There is little population-based evidence on the use oforthotopic bladder reconstruction. Large hospital-basedcase series in specialist centres in Germany and Mexicoindicate rates between 66% and 14% [18,19] whilstmembers of a WHO expert working party held in 2007reported a rate of 47% among a pool of >7000 cystectomypatients [20]. A population-based study from 11 US statesindicated a rate of 20% during the 1990s [14]. It is possiblethat orthotopic reconstruction in England was used morerarely than in other countries during the study period.However, establishing appropriate rates of orthotopicreconstruction a priori is difficult, as there are no uniformlyaccepted standards [21]. Our findings could serve as abenchmark for future prospective monitoring of orthotopicbladder reconstruction use in England.

Younger persons and men were more likely to haveorthotopic reconstruction after cystectomy for bladdercancer. It is difficult to decipher from our analysis thecauses of these differences. Age gradients may reflectstringent but appropriate patient selection of patients withlower levels of comorbidity and of those judged mostlikely to successfully undertake the intensive training and

rehabilitation required after orthotopic bladderreconstruction [11]. One should be aware thatcomorbidity burden per se is unlikely to be the soleexplanatory factor of the observed age differences becauseradical cystectomy is in itself a major operation,associated with substantial risk of postoperative mortality:very unfit patients will not be treated by cystectomy [10].Sex differences may reflect anatomical differences thatmake orthotopic reconstruction more difficult in women.However, although orthotopic neobladder formation istechnically more difficult in women, efficacy outcomescomparable to men have been reported [12]. It istherefore impossible using the available data to concludewhether the observed age and sex differences representvariation in the quality of surgical care. A higher useamong patients living in more affluent neighbourhoodsmay indicate greater preference for orthotopicreconstruction among these patients, or may reflect lowerlevels of comorbidity. Higher use of orthotopicreconstruction among patients of higher socioeconomicstatus has been reported in the USA [14].

Treatment in large specialist hospitals was associated withhigher use of orthotopic surgery in the USA [14], but inour study we observed weak evidence (P = 0.05) for anassociation in the opposite direction (i.e. for lower useamong higher volume cancer networks). Similarly, afteradjusting for patient case-mix variables and variationbetween different consultants, we found overall very limitedvariation in orthotopic surgery use between differentcancer networks. We have observed (and accounted for)variation between individual consultants – which was verylarge: this is an expected finding, given the degree ofsub-specialization required for orthotopic reconstruction.The large amount of variation between consultantsindicates the need to adjust for this variable in futurestudies examining patient and organizational variation inorthotopic reconstruction use.

LimitationsA major limitation of our study population is that it relatesto about 16% of all (about 11 250) patients treated bycystectomy in English hospitals during the 7-year studyperiod 2004–2011 [1]. Further, it is possible that submissionof data to the BAUS Complex Operations data set differedsystematically between hospital networks with either higheror lower than average orthotopic reconstruction. We werenot able within the present study to examine otheroutcomes (such as the efficacy and safety of the procedure,recurrence, survival and quality of life). A crucial barrier toresearching orthotopic reconstruction use in the UKcurrently is the lack of a specific procedure code. Theadvent of such a code will enable relatively effortlessmonitoring of the dissemination and variation in the use of

Variation in bladder reconstruction after radical cystectomy

© 2013 BJU International 5

this procedure and should be considered a priority bypolicy makers. Importantly, we had no data on patientpreference for either orthotopic or other reconstructionmethods. Examination of the variation in use of orthotopicsurgery may be considered unnecessary if it is possibleto measure the quality of decision-making aboutreconstructive option use [22].

ConclusionWithin the study context, we observed limited and variableuse of orthotopic reconstruction. Rates of orthotopicreconstruction varied between different patient groups.Given evidential uncertainty about quality of life benefitsof orthotopic reconstruction and the observed variationin its use, re-examining this question in a nationwidepopulation-based context would be useful. The advent ofappropriate procedure codes encompassing different typesof bladder reconstruction could help underpin such astudy and provide for a surveillance mechanism of thedissemination and use of the procedure in the future.

AcknowledgementsWe acknowledge the BAUS Cancer Registry and allsurgeons who submitted information to it. No externalfunding beyond authors’ salaries was received specificallyfor this study. GL is supported by a postdoctoral researchfellowship award by the National Institute for HealthResearch (NIHR).

Ethical ApprovalThe study was approved by the Scientific Committee of theBAUS Cancer Registry. Data linkage took place within theSouth West Public Health Observatory. Cancer registrationin the UK is supported and regulated by relevantlegislation.

Conflict of InterestNone declared.

References1 HES Online. Main procedures and interventions.

Available at: http://www.hesonline.nhs.uk/Ease/servlet/ContentServer?siteID=1937&categoryID=205.Accessed June 2012

2 Studer UE, Burkhard FC, Schumacher M et al. Twentyyears experience with an ileal orthotopic low pressurebladder substitute – lessons to be learned. J Urol 2006;176: 161–6

3 Meyer JP, Blick C, Arumainayagam N et al. Athree-centre experience of orthotopic neobladder

reconstruction after radical cystectomy: revisiting theinitial experience, and results in 104 patients. BJU Int2009; 103: 680–3

4 Verleyen P, Billiet I, Mattelaer J, Hardeman M,Werbrouck P. Cystectomy and orthotopic ilealneobladder construction. Evaluation of continence andcomplications in a regional hospital. Urol Int 2003; 71:255–61

5 Gore JL, Yu HY, Setodji C, Hanley JM, Litwin MS,Saigal CS. Urologic Diseases in America Project.Urinary diversion and morbidity after radicalcystectomy for bladder cancer. Cancer 2010; 116: 331–9

6 Stein JP, Penson DF, Lee C, Cai J, Miranda G,Skinner DG. Long-term oncological outcomes inwomen undergoing radical cystectomy and orthotopicdiversion for bladder cancer. J Urol 2009; 181: 2052–8;discussion 2058–9

7 Nagele U, Anastasiadis AG, Stenzl A, Kuczyk M.Radical cystectomy with orthotopic neobladder forinvasive bladder cancer: a critical analysis of long-termoncological, functional, and quality of life results.World J Urol 2012; 30: 725–32

8 Porter MP, Penson DF. Health related quality of lifeafter radical cystectomy and urinary diversion forbladder cancer: a systematic review and critical analysisof the literature. J Urol 2005; 173: 1318–22

9 Dutta SC, Chang SC, Coffet CS, Smith JA Jr, Jack G,Cookson MS. Health related quality of life assessmentafter radical cystectomy: comparison of ileal conduitwith continent orthotopic neobladder. J Urol 2002; 168:164–7

10 Autorino R, Quarto G, Di Lorenzo G et al. Healthrelated quality of life after radical cystectomy:comparison of ileal conduit to continent orthotopicneobladder. Eur J Surg Oncol 2009; 35: 858–64

11 Gerharz EW, Månsson A, Hunt S, Skinner EC,Månnsson W. Quality of life after cystectomy andurinary diversion: an evidence based analysis. J Urol2005; 174: 1729–36

12 Evans B, Montie JE, Gilbert SM. Incontinent orcontinent urinary diversion: how to make the rightchoice. Curr Opin Urol 2010; 20: 421–5

13 Somani BK, Nabi G, Wong S et al. How close are weto knowing whether orthotopic bladder replacementsurgery is the new gold standard? Evidence from asystematic review update. Urology 2009; 74: 1331–9

14 Gore JL, Saigal CS, Hanley JM, Schonlau M, LitwinMS. Variations in reconstruction after radicalcystectomy. Cancer 2006; 107: 729–37

15 NHS Connecting for Health. OPCS-4 classification.Available at: http://www.connectingforhealth.nhs.uk/systemsandservices/data/clinicalcoding/codingstandards/opcs4/index_html. Accessed June 2012

Hounsome et al.

6 © 2013 BJU International

16 The British Association of Urological Surgeons.BAUS data and audit publications. Available at:http://www.baus.org.uk/AboutBAUS/publications/data-and-audit. Accessed October 2011

17 Department of Health. National Cancer Action Team.Cancer Networks. Available at: http://ncat.nhs.uk/cancer-networks. Accessed December 2012

18 Hautmann RE, de Petriconi RC, Volkmer BG.Lessons learned from 1000 neobladders: the 90-daycomplication rate. J Urol 2010; 184: 990–4; quiz 1235

19 Martínez-Cornelio A, Hernández-Toriz N,Quintero-Becerra J, Flores-López D, Moreno-PalaciosJ, Vázquez-Martínez E. Management of bladder cancerwith Studer orthotopic neobladder: 13-year experience.Cir Cir 2009; 77: 411–7

20 Hautmann RE, Abol-Enein H, Hafez K et al. World

Health Organization (WHO) Consensus Conferenceon Bladder Cancer. Urinary diversion. Urology 2007;69: 17–49

21 Cooperberg MR, Porter MP, Konety BR. Candidatequality of care indicators for localized bladder cancer.Urol Oncol 2009; 27: 435–42

22 Sox HC, Greenfield S. Quality of care – how good isgood enough? JAMA 2010; 303: 2403–4

Correspondence: Luke Hounsome, South West PublicHealth Observatory, 149 Whiteladies Road, BristolBS8 2RA, UK.

e-mail: luke.hounsome@swpho.nhs.uk

Abbreviations: BAUS, British Association of UrologicalSurgeons; OR, odds ratio.

AppendixCharacteristics of unmatched patients (n = 886) and the group available for initial analysis (n = 2946)

Patient-level characteristic Matched group Unmatched group P

GenderMen 2175 74% 596 67% <0.001*Women 743 25% 263 30%No gender information 28 1% 27 3%

Stage0 243 8% 17 2% 0.70*I 459 16% 73 8%II 1172 40% 243 27%III 298 10% 42 5%IV 217 7% 339 38%Unknown stage 557 19% 172 19%

Reconstruction typeIleal conduit 2285 78% 710 80% 0.09*Orthotopic 167 6% 65 7%Continent cutaneous diversion 12 0% 9 1%Other 29 1% 7 1%Unknown 453 15% 95 11%

*Chi-squared test for men vs women, known vs unknown stage or orthotopic vs non-orthotopic reconstruction comparisons, respectively.

Variation in bladder reconstruction after radical cystectomy

© 2013 BJU International 7