- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Published population-based data on the recourse to major surgical procedures in urological oncology are mostly limited to the USA. Issues related to overall recourse to cystectomy, adoption of partial instead of radical cystectomy, determinants of perioperative mortality, complications and length of stay, pelvic lymphadenectomy, and type of urinary diversion have been largely investigated in the USA [1–10], and partly in Canada [11,12], England [13–15] and Sweden [16,17], but evidence from western and southern Europe is lacking.
The European Association of Urology guidelines [18,19] recommend radical cystectomy as treatment of muscle-invasive bladder cancer; radical cystectomy can also be considered an immediate treatment option of non-muscle-invasive bladder cancer at high risk of progression. Lymphadenectomy should be an integral part of cystectomy. Before cystectomy, the patient should be counselled adequately regarding all possible alternatives for urinary diversion, and the final decision should be based on a consensus between patient and surgeon. An orthotopic neobladder should be offered to male and female patients lacking any contraindications and who have no tumour in the urethra and at the level of urethral dissection . Partial cystectomy is not recommended, even though it could be adopted in rare selected cases .
Treatment of bladder cancer in southern Europe is a major issue since bladder cancer incidence is far higher in Italy and Spain than in any other country where cancer registration is in place ; age-standardized (world population) incidence rates per 100 000 in males and females were as high as 38.5 and 6.7 in Piedmont and 31.9 and 6.5 in Veneto, two regions located in northwestern and northeastern Italy, respectively . Major trends in urological oncology practice can be surveyed through hospital discharge records (HDRs) ; the aim of the present study is to provide data on the recourse to cystectomy for bladder cancer and to analyse factors associated with the type of urinary diversion by means of analyses of HDRs of the Piedmont and Veneto regions.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
On 1 January 2009 the total populations of Piedmont and Veneto were about 4 400 000 and 4 900 000 inhabitants, respectively, and there were in total almost 2 000 000 discharges from regional hospitals each year. Discharge diagnoses and procedures are recorded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD9-CM). Regional archives of HDRs include all hospitalizations in regional hospitals as well as discharges of residents hospitalized outside the two regions. All discharges from 1 January 2000 to 31 December 2008 of patients aged ≥18 years with a diagnosis of bladder cancer (ICD9-CM diagnostic code 188.x) with intervention codes 57.6 (partial cystectomy), 57.71 (radical cystectomy), 57.79 (other total cystectomy) and 68.8 (exenteration) were extracted from the electronic archives of HDRs for the Piedmont and Veneto regions. The demographic characteristics (age, sex, residency in or outside the region), year of intervention, presence of comorbidities (an adaptation of the Charlson Comorbidity Index computed on the index admission through a program from the National Cancer Institute) (http://healthservices.cancer.gov/seermedicare/program/comorbidity.html) and hospital procedural volume were retrieved from HDRs. Hospital volume was defined as the average number of cystectomies per year performed in each hospital, separately for the three calendar periods 2000–2002, 2003–2005 and 2006–2008. Type of urinary diversion was classified as continent diversion (intervention code 57.87), ureterocutaneostomy (56.61), ileal conduit (56.51, 56.71) and unspecified (other codes or missing data). Lastly, codes for regional or radical lymphadenectomy (40.3 and 40.5×) were extracted from HDRs.
Trends in crude cystectomy rates (including data from residents seeking care outside regional hospitals) and in the recourse to partial (code 57.6) vs radical (other intervention codes) cystectomy were computed for the two regions. All other analyses were restricted to radical cystectomies performed in Piedmont and Veneto hospitals, aiming to investigate trends and determinants of type of urinary diversion, length of stay and in-hospital mortality. Bivariate analyses were conducted using the chi-squared test, and differences in length of stay across study groups were assessed by the Kruskal–Wallis rank test. Thereafter, multilevel models were used to assess the influence of patient-level and hospital-level variables while taking into account clustering of patients within providers (random intercept for the hospital level); in-hospital mortality (rare outcome) was investigated through multilevel logistic regression, and the recourse to continent diversion (common procedure) vs ileal conduit or ureterocutaneostomy through multilevel Poisson regression. Statistical analyses were carried out with the package Stata 9 (College Station, TX, USA). HDRs were kept anonymous and therefore the study was exempt from institutional review board approval.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Overall 8064 cystectomies were performed in Piedmont and Veneto hospitals through 2000–2008, 6347 (79%) on subjects living in the two regions and 1690 (21%) on patients living outside the study area and seeking care in regional facilities. The latter patients were younger (with respect to residents in the study area, 61% vs 43% were aged <70 years), with a lower proportion of females (12% vs 18%), and were more represented (31%) among hospitalizations in Veneto facilities. Data on 292 Veneto and Piedmont residents discharged after a cystectomy performed outside the two regions were available for the years 2001–2007, allowing the computation of population-based crude cystectomy rates equal to 9.8 and 7.2 per 100 000 inhabitants in Piedmont and Veneto, respectively.
Partial cystectomies accounted for 5.5% of all interventions in 2000–2002 and dropped to 3.0% in 2006–2008. Overall 18% of cystectomies were performed in hospitals with low procedural volume (one to nine interventions a year), 35% in medium-volume hospitals (10–19 per year) and 47% in high-volume hospitals (≥20 per year); such proportions remained stable across study periods.
All subsequent analyses were restricted to 7743 radical cystectomies performed in regional hospitals. Median age at radical cystectomy was 70 years (interquantile range 63–76); the male to female ratio was about 5 : 1. Overall, only 8.4% of HDRs missed information on type of reconstruction, especially for smaller hospitals in earlier years (Table 1). Continent urinary diversion (35.0% of all cystectomies) was the preferred technique in patients aged <70 years; ileal conduit (44.7% of all surgeries) accounted for the majority of interventions among older ages; one-third of patients aged ≥80 years with bladder cancer underwent ureterocutaneostomy (11.9% among all ages). The percentage of continent diversion was halved in females with respect to males, was reduced in patients with comorbidities, and was strongly related to hospital volume. All these associations were confirmed by multilevel Poisson regression (Table 2); the higher percentage of continent diversion among non-residents seeking care in regional hospitals was partly explained by their referral to high-volume hospitals and by demographic characteristics. The recourse to continent diversion was similar in the two regions at the beginning of the study period, but afterword remained stable in Veneto (36.5% in 2000–2002 and 39.1% in 2006–2008) and decreased in Piedmont (from 34.5% to 25.7%).
Table 1. Technique of urinary diversion registered in HDRs with radical cystectomy by patients’ characteristics and hospital procedural volume, Veneto and Piedmont regions, 2000–2008
| || n ||Continent, %||Ileal conduit, %||Ureterocutaneostomy, %||Missing, %|
|Calendar period*|| || || || || |
Table 2. Urinary diversion in Piedmont and Veneto hospitals, 2000–2008: risk ratios (RR) with 95% CI for recourse to continent diversion vs ileal conduit or ureterocutaneostomy, obtained by multilevel Poisson regression
| ||RR||CI|| P |
|Age|| || || |
| <55||1.00|| || |
|Gender|| || || |
| Females||1.00|| || |
|Charlson index|| || || |
| Each 1 pt increase (continuous)||0.80||0.73–0.88||<0.001|
|Residency|| || || |
| Non-residents||1.00|| || |
|Hospital volume|| || || |
| Low||1.00|| || |
|Calendar period|| || || |
| 2000–2002||1.00|| || |
|Region|| || || |
| Piedmont||1.00|| || |
Pelvic lymphadenectomy was reported in 58%, 66% and 71% of radical cystectomies performed in 2000–2002, 2003–2005 and 2006–2008, respectively.
Median length of stay decreased over time from 20 to 18 days (P < 0.001), and was related to the type of reconstruction, with continent diversion requiring longer hospitalizations (Fig. 1). Moreover, length of stay increased with comorbidity, in patients treated in low-volume hospitals, and in non-residents (data not shown).
Figure 1. Time trends in median length of stay (LOS) by type of urinary diversion (continent vs ileal conduit or ureterocutaneostomy): Piedmont and Veneto hospitals, 2000–2008.
Download figure to PowerPoint
Overall in-hospital mortality was 2.8%, decreasing from 3.2% in 2000–2002 to 2.2% in 2006–2008 (P= 0.039; see Table 3); a reduction in mortality could be observed in both regions. If non-residents (representing younger selected patients with lower risk of death) were excluded, in-hospital mortality was 3.3% (from 4.0% in 2000–2002 to 2.4% in 2006–2008, P= 0.018). About 60% of in-hospital deaths occurred within 30 days and 30% between 30 and 60 days from admission; among residents in-hospital 30-day mortality was 2.0% (declining from 2.2% to 1.6%, P= 0.306). Age was the main determinant of in-hospital mortality in multilevel regression (Table 3); results did not change if the model was restricted to residents (data not shown).
Table 3. In-hospital mortality for radical cystectomy, probability of error in the chi-squared test, and odds ratio (OR) with 95% CI obtained by multilevel logistic regression
| ||In-hospital mortality, %||Univariate analysis||Multivariate analysis|
| P, chi-squared||OR (CI)|| P |
|Age|| ||<0.001|| || |
| <55||1.0|| ||1.00|| |
| 55–59||1.5|| ||1.38 (0.50–3.82)||0.540|
| 60–64||0.8|| ||0.71 (0.24–2.05)||0.523|
| 65–69||2.3|| ||1.98 (0.82–4.78)||0.128|
| 70–74||3.3|| ||2.79 (1.19–6.54)||0.018|
| 75–79||3.8|| ||3.13 (1.34–7.34)||0.009|
| ≥80||5.4|| ||4.61 (1.95–10.9)||0.001|
|Gender|| ||0.844|| || |
| M||2.9|| ||1.00|| |
| F||2.8|| ||0.89 (0.61–1.28)||0.524|
|Charlson index|| ||0.009|| || |
| = 0||2.6|| ||1.00|| |
| ≥1||4.0|| ||1.34 (1.09–1.65)*||0.005|
|Residency|| ||<0.001|| || |
| Non-residents||1.1|| ||1.00|| |
| Residents||3.3|| ||2.74 (1.64–4.56)||<0.001|
|Hospital volume|| ||0.567|| || |
| <10 cystectomies/year||3.3|| ||1.00|| |
| 10–20 cystectomies/year||2.8|| ||0.79 (0.53–1.17)||0.240|
| ≥20 cystectomies/year||2.7|| ||1.04 (0.70–1.54)||0.843|
|Calendar period|| ||0.039|| || |
| 2000–2002||3.2|| ||1.00|| |
| 2003–2005||3.1|| ||0.97 (0.68–1.37)||0.852|
| 2006–2008||2.2|| ||0.62 (0.42–0.90)||0.012|
|Region|| ||0.047|| || |
| Piedmont||3.2|| ||1.00|| |
| Veneto||2.5|| ||0.96 (0.71–1.29)||0.768|
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Crude cystectomy rates were very high in northern Italy, approaching 10 per 100 000 population in Piedmont where bladder cancer incidence is among the highest registered in the world. Incidence estimates from cancer registries are available only for part of the study populations (about half of the Veneto and one-quarter of the Piedmont population) and are updated to 2004–2006; overall the ratio of cystectomy rates to bladder cancer incidence reported by local cancer registries was between 1:5 and 1:6. In the USA, cystectomy rates decreased from 3.6 to 3.0 per 100 000 in 1988–2000 , with a ratio to bladder cancer incidence of about 1:7. High incidence rates, an ageing population and a frequent recourse to radical surgery make radical cystectomy a relatively common procedure in northern Italy. This could explain the high procedure-specific hospital volumes, with only 19% of patients discharged from hospitals performing <10 cystectomies a year: the corresponding figure in England in 2000–2007 was 35% (declining from 40% to 30%) ; in the USA 54% of patients in 1997–2000 underwent cystectomy in hospitals performing fewer than eight procedures a year .
Partial cystectomy has specific indications (e.g. solitary tumours located in the dome); in spite of decreasing partial cystectomy rates in the USA, in 2000 the procedure still accounted for about 13% of surgeries, suggesting a widespread inappropriate use . Even though recent data from Surveillance, Epidemiology and End Results (SEER) suggest that partial cystectomy could not undermine overall and cancer-specific survival in appropriate selected cases , in northern Italy the adoption of this procedure was already limited at the beginning of the study period and further decreased over time.
The only population-based studies on diversion after cystectomy have been published in the USA and Sweden; usually analyses are restricted to hospital records reporting codes for continent diversion or ileal conduit, available for about 84% of radical cystectomies registered in US discharge data . Analyses of the SEER–Medicare database found that 19.9% of patients underwent continent urinary diversion ; a study on the Nationwide Inpatient Sample reported that the recourse to reconstruction increased from 1998 to 2001, thereafter reaching a plateau below 20%. A single paper on Texas hospitals’ discharges examined all intervention codes: ileal conduit was performed on 86.6% of patients, reconstruction on 4.4%, cutaneous diversion on 0.7%, and 8.3% had unknown urinary diversion . In Sweden the percentage of patients undergoing cystectomy with continent reconstruction decreased from 41% in 1997 to 28% in 2003 . Analyses of HDRs from northern Italy show a high recourse to continent diversion; the ICD9-CM classification (version 2007) of procedures does not allow the distinction between orthotopic bladder substitution and continent cutaneous diversion, but adoption of the latter procedure is very limited in the study area. By contrast, recourse to a cutaneous diversion without bowel interposition, mainly adopted for frail patients undergoing palliative or salvage cystectomy, seems to be well above figures reported by the single US paper examining all types of diversion. Confirming previous findings, the recourse to continent diversion decreased with age and was lower among females [8,9,17], in subjects with comorbidities [8,9] and in low-volume hospitals [6,8,9]. The impact on length of hospital stay (Fig. 1), concerns about an increased risk of postoperative complications, and a rising proportion of elderly patients with multiple comorbidities could be among constraints to the adoption of continent reconstructive techniques, which decreased over time in one of the two regions analysed (Piedmont), similarly to the report from Sweden .
In HDRs lymphadenectomy could only be traced by intervention codes; it is therefore not possible to disentangle increasing trends in completeness of HDRs and actual recourse to the procedure. Nonetheless, the upward trend observed in both regions suggests an increasing awareness of urologists on the role of lymphadenectomy. Similarly, the type of lymph node dissection was missing in 29% of discharge records from Texas hospitals . Also studies on the US SEER database report a proportion of missing data ranging from 10% to 29%[26–28]; among patients with available information, patients with no lymph node sampled at radical cystectomy decreased from 37% in 1988 to 16% in 2004 .
Despite a declining time trend, mean (median) length of stay was still about 20 (18) days at the end of the study period, close to figures from England where it decreased from 20.7 to 18.7 through 1995–2002 . By contrast, in the USA the median hospital stay dropped from 13 days in 1998–1990 to nine in 1997–2000, with a corresponding increase in the recourse to home health services .
In-hospital mortality is a major issue in studies on cystectomy : in the USA in-hospital mortality was 2.9% in the Nationwide Inpatient Sample 1988–1999 , with a decreasing trend from 1988–1990 to 1997–2000 ; more recent analyses on 1998–2003 data report in-hospital mortality equal to 2.6%. In England overall in-hospital mortality dropped from 5.3% to 3.6% through 1995–2002 , and in 2000–2007 30-day in-hospital mortality was about 2.6%. In-hospital mortality is a measure limited by the absence of post-discharge surveillance, being therefore influenced also by trends in length of stay; furthermore a prolonged follow-up (90 days) allows a better evaluation of both post-cystectomy complications and mortality [29–31]. Data from northern Italy confirm that in-hospital mortality is decreasing, but also that a prolonged follow-up is required since the burden of post-cystectomy deaths span well beyond the first 30 days.
HDRs allow a timely and complete coverage of surgical practice on a population basis, with the evaluation of structural (procedural volume), process (type of urinary diversion) and outcome indicators (in-hospital mortality) for the quality of care in patients with bladder cancer . The main limit of the study is that HDRs contain no cancer-specific data such histology, stage or grade. Histology and cancer stage are shown to affect post-surgical mortality , but grade and stage of disease had only a marginal influence on variations in diversion after cystectomy . Furthermore, comorbidities are incompletely reported, especially in the Veneto region, where directives from regional health authorities limited the registration of secondary diagnoses to those with a major impact on hospitalization costs; as a consequence, adjustment for comorbid conditions is only partial in our analyses. By contrast, completeness of procedures listing on type of urinary diversion as well as lymph node dissection is similar to that reported in the USA . Lastly, the study is based on anonymized HDRs without the possibility of record linkage to cancer registry data and outpatient treatments’ archives; therefore information on the time interval between diagnosis and cystectomy, or on prior treatments such as local therapy or radiotherapy, is not available.
In conclusion, this first population-based report on cystectomies from continental Europe highlights a marked recourse to the procedure in northern Italy, with a marginal and declining role for partial cystectomy. The decreasing trend of in-hospital mortality and length of stay observed in the USA and in the UK is confirmed. A high proportion of continent diversion but also of ureterocutaneostomy is observed with respect to US data; it is therefore crucial to continue to survey major trends in urological practice.