Population-based trends in urinary diversion among patients undergoing radical cystectomy for bladder cancer

Authors


Correspondence: Simon P. Kim, Mayo Clinic, Department of Urology, 200 First Street Southwest, Rochester, MN 55905, USA.

e-mail: simkim@me.com

Abstract

What's known on the subject? and What does the study add?

  • Variations in the type of urinary diversion exist for patients undergoing radical cystectomy.
  • Although its use has been increasing from 2001 to 2008, patients who are older, female, and primary insured by Medicaid are less likely to receive continent diversions. Furthermore, patients treated surgically at high-volume and teaching hospitals are more likely to receive continent diversions.

Objective

  • To describe the contemporary trends in urinary diversion among patients undergoing radical cystectomy (RC) for bladder cancer; and elucidate whether socioeconomic disparities persist in the type of diversion performed in the USA from a population-based cohort.

Patients and Methods

  • Using the Nationwide Inpatient Sample, we identified patients who underwent RC for bladder cancer between 2001 and 2008.
  • Multivariable regression models were used to identify patient and hospital covariates associated with continent urinary diversion and enumerate predicted probabilities for statistically significant variables over time.

Results

  • Overall, 55 635 (92%) patients undergoing RC for bladder cancer received incontinent urinary diversion, while 4552 (8%) patients received continent diversion from 2001 to 2008.
  • Receipt of continent urinary diversion increased from 6.6% in 2001–2002 to 9.4% in 2007–2008 (P < 0.001 for trend).
  • Patients who were older (odds ratio [OR] 0.93; P < 0.001), female (OR 0.52; P < 0.001) and insured by Medicaid (OR 0.54; P = 0.002) were less likely to receive continent urinary diversion.
  • However, patients treated at teaching (OR 2.14; P < 0.001) and high-volume hospitals (OR 2.39; P = 0.04) had higher odds of continent urinary diversion.
  • Predicted probabilities of continent diversion remained lower for female patients, Medicaid insurance status, and non-teaching and medium/low-volume hospitals over time.

Conclusions

  • In this nationally representative sample of hospitals from 2001 to 2008, the use of continent diversion in RC gradually increased.
  • Although variations in urinary diversion exist by hospital teaching status, case volume, patient gender and primary health insurance, increased attention in expanding the use of continent diversions may help reduce these disparities for patients undergoing RC for bladder cancer.
Abbreviations
GEE

generalised estimating equations

HRQL

health-related quality of life

NIS

Nationwide Inpatient Sample

OR

odds ratio

RC

radical cystectomy

Introduction

In 2011, bladder cancer represented the sixth most lethal malignancy in the USA with an estimated 69 250 incident cases and 14 490 cancer-related deaths [1]. Radical cystectomy (RC) remains the primary treatment method for patients diagnosed with muscle-invasive or high-risk non-muscle-invasive bladder cancer [2, 3]. A critical treatment decision in the surgical management of bladder cancer is the type of urinary diversion to be performed. At present, patients diagnosed with bladder cancer deliberate the risks and benefits of each type of diversion with their urologists, and choose either a continent or incontinent urinary diversion after RC. Pathological factors (tumour location), clinical considerations (renal function and performance status), and patient preference, among others, influence this complex decision concerning which type of urinary diversion would be best for patients treated with surgical extirpation [3-5]. Moreover, different types of urinary diversion carry important implications for health-related quality of life (HRQL) and functional outcomes as well as risk of postoperative morbidity [3, 5-11].

For patients who are considering RC, the choice of urinary diversion should be largely based on the feasibility of performing the operation safely and the ability to provide long-term care according to each patient's overall health and preference. Ideally, patients diagnosed with bladder cancer and similar clinical characteristics should have access to choose a urinary diversion that best fits their preferences, HRQL, and ability to care for an externalised or internalised urinary diversion. However, previous studies suggest that racial, gender and socioeconomic disparities exist in the receipt of different types of urinary diversion [12-14]. Furthermore, with greater regionalisation of complex oncological operations [15], high-volume and teaching hospitals have been shown to be associated with higher rates of continent urinary diversion and lower rates of postoperative morbidity and mortality for RC [12, 16, 17]. Yet, racial disparities involving access to higher volume hospitals and providers are increasingly recognised as a health policy concern [3, 18-20]. Therefore, socioeconomic and hospital factors may be influencing the type of urinary diversion performed in patients with bladder cancer. To the extent that if there have been changes, if any, in the rates for continent or incontinent urinary diversion and persistent racial, gender and socioeconomic disparities in surgical care for bladder cancer remain unknown [20]. In the present study, we sought to describe the contemporary trends of urinary diversion in patients undergoing RC for bladder cancer in the USA; and elucidate whether racial, gender, and socioeconomic disparities persist concerning continent urinary diversion in a nationally representative sample.

Patients and Methods

Data for all patients who underwent RC for bladder cancer were abstracted from the 2001–2008 Nationwide Inpatient Sample (NIS). The NIS from the Healthcare Cost and Utilization Project (HCUP) represents the largest all-payer inpatient care database and contains ≈ 20% of all admissions in the USA [21]. We identified our analytic cohort from a similar method described previously using International Classification of Disease Modification, 9th edition (ICD-9) codes from the hospital claims data in the NIS [11, 17]. We first identified patients who had a bladder neoplasm (188.0–188.9) from the primary diagnostic codes and concomitant procedure codes for RC (57.71 and 57.79). Secondary procedure codes were then used to assign whether patients received incontinent (56.51, 56.52, 56.71) or continent (57.87) urinary diversion. Using this approach, we identified 12 007 patients who underwent RC for bladder cancer from 2001 to 2008. Our final analytic cohort comprised 11 214 patients after excluding 793 paediatric cases (aged < 18 years) and those missing secondary codes to classify urinary diversion. The NIS weights were then used to determine national estimates of RC with urinary diversion.

Patient and Hospital Covariates

Demographic information included as independent covariates were patient age, gender, race, primary health insurance, median zip code income by quartile, and time interval of surgery (2001–2002, 2003–2004, 2005–2006, and 2007–2008). With a quarter of individuals missing a race designation, we created an indicator variable for ‘missing’ in order to include these cases. Secondary diagnostic codes were used to define an Elixhauser Comorbidity Index [22]. We also evaluated hospital teaching status (non-teaching vs teaching), location (urban vs rural), region (Northeast, Midwest, South and West), and hospital volume as independent hospital-level variables. To account for the possible relationship of hospital volume and study outcome, we dichotomized RC hospital volume into the top 90th percentile of annual mean RC for high volume (≥14 RCs/year) and <90th percentile for the low/medium RC volume (<14 RCs/year).

Statistical Analysis

The primary outcome of this study was the receipt of a continent or incontinent urinary diversion after RC. Descriptive statistics were used to summarise patient and hospital characteristics. Bivariate associations of patient and hospital variables with type of urinary diversion were tested by the Pearson's chi-square test. Temporal trends across time intervals for continent and incontinent urinary diversion were evaluated by the Wilcoxon rank-sum test. We then fitted multi-level/mixed effects multivariable logistic regression models to assess whether patient and hospital variables were associated with receipt of continent urinary diversion, adjusting for clustering of patient covariates to the hospital level.

To better characterise temporal changes in continent urinary diversion for patient and hospital covariates identified as statistically significant on the multivariable logistic regression model, generalised estimating equations (GEE) were used to enumerate predicted probabilities for each 2-year time interval from 2001 to 2008, adjusting for patient and hospital covariates and clustering of patients to the hospital level. A two-sided P ≤ 0.05 was considered to indicate statistical significance.

Results

From the NIS, we identified 11 214 patients who underwent RC for bladder cancer at 1264 hospitals from 2001 to 2008, thereby resulting in a national weighted estimate of 55 187 patients. Overall, the mean (SD) age among patients in our analytic cohort was 68.6 (10.5) years. In all, 50 635 (91.8%) patients treated surgically with RC received incontinent urinary diversion, while 4552 (8.2%) patients received continent diversion. In comparison with incontinent urinary diversion after RC, patients reconstructed with a continent urinary diversion were more likely to be younger, male and with fewer comorbidities at the time of surgery (Table 1). Moreover, continent urinary diversion occurred in half as many female patients as compared with male patients (4.4% vs 9.1%; P < 0.001). Patients also insured with Medicare (4.6%) and Medicaid (8.5%) were associated with a lower proportion of continent diversion than those with private health insurance (15.2%; P < 0.001). However, patients who received surgical care at teaching and high-volume (top 90th percentile) hospitals were more likely to have received continent diversion. Over the 8-year study interval, the proportion of patients treated with continent urinary diversion increased from 6.6% in 2001–2002 to 9.4% in 2007–2008 (Fig. 1; P < 0.001 for trend).

Figure 1.

Temporal trends in urinary diversion in patients undergoing RC for bladder cancer.

Table 1. Patients' characteristics by type of diversion after RC for bladder cancer (n = 55 187)
FeatureContinent diversion (n = 50 635), %Incontinent diversion (n = 4 552), %P
Age, years:  <0.001
<5520.279.8 
55–6414.985.1 
65–746.693.4 
≥752.297.8 
Race:  0.99
White8.391.7 
Black7.692.4 
Hispanic8.291.8 
Other8.591.5 
Missing8.391.7 
Gender:  <0.001
Female4.495.6 
Male9.190.9 
Elixhauser Comorbidity Index:  <0.001
0–111.288.8 
2–36.893.2 
≥44.295.8 
Primary insurance:  <0.001
Private15.284.8 
Medicare4.695.4 
Medicaid8.591.5 
Other13.386.7 
Annual household income:  0.07
1 (lowest quartile)7.192.9 
27.992.1 
38.491.6 
4 (highest)9.091.0 
Hospital teaching status:  <0.001
Teaching10.389.7 
Non-teaching3.796.3 
Hospital location:  0.23
Rural6.993.1 
Urban8.391.7 
Hospital region:  <0.001
Northeast9.790.3 
Midwest7.392.7 
South7.093.0 
West10.189.9 
RC volume:  <0.001
Low7.792.3 
High13.087.0 

On multivariable analysis, receipt of continent urinary diversion was associated with specific patient and hospital characteristics (Table 2). Increasing age correlated with a lower receipt of continent urinary diversion. For example, patients aged ≥ 75 years had a 91% lower likelihood of receiving continent urinary diversion than those aged < 55 years (odds ratio [OR] 0.09; P < 0.001). While race was not associated with the type of diversion, female patients (OR 0.52; P < 0.001) and unhealthier patients with more comorbidities at the time of the index surgery were less likely to receive a continent diversion in comparison with male and healthier patients, respectively. Interestingly, patients primarily insured with Medicaid had a 46% lower likelihood of continent diversion than those with private health insurance (OR 0.54; P < 0.002). However, among independent hospital variables, patients treated surgically at teaching hospitals (OR 2.14; P < 0.001) and high-volume centres (OR 2.39; P = 0.04) had a higher OR for receipt of continent diversion than non-teaching and low/medium-volume hospitals, respectively.

Table 2. Multivariable logistic regression for patient and hospital features associated with receipt of continent urinary diversion in patients undergoing RC for bladder cancer
Feature (reference)OR (95% CI)P
Age, years (<55 years):
55–640.67 (0.54–0.83)<0.001
65–740.30 (0.23–0.40)<0.001
≥750.09 (0.07–0.13)<0.001
Race (White):
Black1.04 (0.67–1.61)0.85
Hispanic0.84 (0.50–1.41)0.53
Other0.78 (0.49–1.25)0.31
Missing1.00 (0.76–1.31)0.98
Gender (male):
Female0.52 (0.41–0.67)<0.001
Elixhauser Comorbidity Index (0–1):
2–30.76 (1.60–2.87)0.001
≥40.59 (0.57–0.79)0.001
Primary insurance (private):
Medicare0.79 (0.63–1.02)0.07
Medicaid0.54 (0.36–0.80)0.002
Other0.79 (0.55–1.13)0.20
Annual household income quartile (1 = lowest):
21.12 (0.86–1.46)0.37
31.20 (0.93–1.57)0.16
4 (highest)1.27 (0.97–1.65)0.09
Hospital teaching status (non-teaching):
Teaching2.14 (1.60–2.87)<0.001
Hospital location (rural):
Urban1.02 (0.57–1.83)0.93
Hospital region (Northeast):
Midwest0.95 (0.62–1.45)0.83
South0.93 (0.62–1.39)0.73
West1.66 (1.09–2.53)0.02
RC volume (low):
High2.39 (1.04–5.56)0.04
Year of surgery (2001–2002):
2003–20041.11 (0.85–1.47)0.42
2005–20061.16 (0.88–1.53)0.28
2007–20081.29 (0.99–1.69)0.06

To assess for temporal trends in different types of urinary diversion over time by statistically significant patient and hospital covariates, predicted probabilities were ascertained for each time interval (Table 3). While the predicted probability of continent urinary diversion was lower among females than male patients over the entire 8-year study period (4.3% vs 7.7%; P < 0.001), there were trends of increasing, yet disparate probabilities by gender for each time interval from 2001 to 2008. For instance, the predicted probability of continent diversion increased from 3.7% in 2001–2002 to 4.8% in 2007–2008 for female patients undergoing RC for bladder cancer. The predicted probabilities for continent diversion according to primary health insurance also showed some notable differences. Patients insured with Medicaid had a lower predicted probability of continent urinary diversion (4.6% vs 8.2%; P = 0.001) than those insured with all other types of health insurance. However, there were increasing predicted probabilities of continent diversion across all primary health insurances. Similar trends in increasing use of continent urinary diversion, yet disparate differences were found with non-teaching vs teaching hospitals (4.3% vs 8.1%; P < 0.001) and low/medium vs high-volume hospitals (6.4% vs 11.7%; P = 0.01) in the predicted probabilities of continent diversion for each time interval from 2001 to 2008.

Table 3. Multivariable logistic regression for patient and hospital features associated with receipt of continent urinary diversion in patients undergoing RC for bladder cancer
FeaturePredicted probability % (95% CI)
2001–20022003–20042005–20062007–2008Overall
Gender:
Female3.7 (2.6–4.8)4.2 (3.0–5.5)4.4 (3.2–5.6)4.8 (3.6–6.1)4.3 (3.3–5.4)
Male6.6 (5.3–7.9)7.6 (6.2–8.9)7.7 (6.3–9.2)8.4 (7.1–9.8)7.7 (6.7–8.6)
Primary health insurance:
Private6.9 (5.4–8.3)7.8 (6.2–9.4)8.0 (6.4–9.6)8.8 (7.2–10.3)7.9 (6.8–9.1)
Medicare5.6 (4.3–6.9)6.4 (4.9–7.8)6.5 (5.0–8.1)7.2 (5.7–8.6)6.5 (5.3–7.7)
Medicaid4.1 (2.5–5.7)4.7 (2.9–6.4)4.8 (3.0–6.7)5.3 (3.4–7.2)4.8 (3.1–6.4)
Hospital teaching status:
Non-teaching3.6 (2.7–4.6)4.2 (3.1–5.2)4.4 (3.2–5.5)4.8 (3.6–5.9)4.3 (3.3–5.2)
Teaching7.0 (5.6–8.4)8.0 (6.5–9.5)8.3 (6.7–9.8)9.0 (7.5–10.4)8.1 (7.0–9.2)
Hospital volume:
Low5.6 (4.5–6.6)6.3 (5.2–7.4)6.5 (5.4–7.7)7.1 (6.0–8.2)6.4 (5.7–7.2)
High10.2 (6.1–14.3)11.5 (7.0–16.1)11.9 (7.1–16.5)12.9 (8.1–17.5)11.7 (7.4–15.9)

Discussion

A principal finding in the present study was that approximately 90% of surgically treated patients with bladder cancer received incontinent diversion in the USA from 2001 to 2008. The present results are similar to previously published series. For example, Gore et al. [12] described that 81% of patients received incontinent diversion using SEER-Medicare data from 1992 to 2000. From a 5% random sample using similar population-based data for 1998–2005, a more contemporary study suggested that national rates for continent diversion were lower at 7% [13]. Lowrance et al. [23] reported that 37% of patients received neobladders at a high-volume tertiary referral centre, but the proportion of patients receiving a continent diversion in fact decreased from 47% to 25% from 2000 to 2004. The present study details the contemporary trends regarding the distribution of urinary diversion among patients undergoing RC for bladder cancer in the USA. Although a minority of patients received a continent diversion overall, a key finding in the present study suggests that its use is indeed gradually increasing.

More importantly, the present study suggests that disparities exist in the receipt of different urinary diversion, which have important implications for patient-centred outcomes (i.e. HRQL) and morbidity, by patient age, gender, primary health insurance, hospital volume and teaching status. Few studies have critically examined the relationship of these socioeconomic determinants in relation to the receipt of different urinary diversion in a bladder cancer patient population. A recent study assessed possible sources of variations in urinary diversion for RC using SEER-Medicare data from 1992 to 2000 [12]. In that study, patients who were older, black, female gender, and with more comorbidities were associated with lower receipt of continent urinary diversion; while high-volume, academic and NCI-designated hospitals were associated with greater use. Similarly, other studies have documented that female patients diagnosed with bladder cancer are more likely to undergo incontinent urinary diversion [23, 24]. Taken together, these studies suggest that race, gender, socioeconomic status, and healthcare access may be influencing the type of urinary diversion performed. The present study contributes to answering whether these disparities persist in a contemporary nationally representative sample. Indeed, while a minority patients had similar rates of continent urinary diversion; female patients continue to have lower unadjusted and adjusted rates of continent diversion for RC. We also found that those patients with Medicaid insurance were associated with higher use of incontinent diversion, while variation in the type of urinary diversion continued by hospital teaching status and case volume. An inference from these results may be that treatment choice of continent or incontinent urinary diversion impacted by socioeconomic considerations other than patient functional status, health, and preference. In other words, there may be a relationship of non-clinical factors that limit access to the full choice of different urinary diversion available to patients diagnosed with bladder cancer treated with RC.

While it is informative to assess the extent in which these disparities affect the surgical management of bladder cancer, it is also essential to acknowledge the limitations of the present study. First, treatment decisions in urinary diversion are highly complex surrounded by a confluence of issues ranging from bladder tumour pathology to patient preference, some of which are not routinely captured by hospital claims data. Claims data have known limitations that may mitigate the generalizability of the present findings [25]. In particular, hospital claims lack clinical information about key determinants in the choice of urinary diversion, e.g. functional status, bladder tumour location, multifocality, cancer staging. It is also necessary to recognise that the NIS only captures information about the hospitalisation with limited data on long-term outcomes. Furthermore, it has been well-recognised that many of these risk factors for disparities in healthcare, such as race, gender and Medicaid health insurance, also adversely affect bladder cancer presentation (higher stage or metastatic disease) and survival [26, 27]. It is plausible that the relationship of advanced presentation bladder cancer due to disparities in healthcare access may have partially explained the present findings regarding the observed differences in urinary diversion. In addition, the findings of Medicaid patients having lower odds of receiving continent diversions may be mitigated by access to high-volume medical centres. However, to the best of the authors' knowledge there have not been previous studies evaluating specific contraindications or patient preferences for continent urinary diversion by race, gender, or socioeconomic status. Additionally, locally advanced disease or use of secondary chemotherapy or radiation after RC does not preclude the use of continent diversion [3]. Second, claims data are unable to ascertain patient and surgeon preferences, especially as continent and incontinent diversion have salient differences in body image and acceptance of externalised or internalised urinary reconstruction [8]. Although assessing HRQL for bladder cancer and RC is a nascent research area, patient preference is of particular importance for RC and urinary diversion given the differences in HRQL and functional outcome [7, 8, 10, 27]. We acknowledge that these differences in urinary diversion as articulated in the present study may be attributable to patient and surgeon preferences. Third, we recognise that the present findings may be due to the differential risks of postoperative morbidity and readmission by urinary diversion in RC. Given the greater complexity in continent urinary diversion (as evident in that patients with advanced age are unlikely to receive a neobladder), studies are mixed whether the type of urinary diversion is indeed a risk factor for adverse postoperative outcomes, or more a matter of selection bias [9, 11, 13].

Nonetheless, the present study presents concerning findings about the extent to which the type of urinary diversion is attributable to patient age, gender, and socioeconomic disparities. Furthermore, with greater regionalisation of complex oncological operations, such as RC, to high-volume academic medical centres [15, 28, 29], disparities in the access to tertiary referral centres may further widen the disparities in type of diversion performed. Socioeconomic disparities in access to high-volume hospitals and surgeons for prostate cancer and differences in oncological outcomes for bladder cancer are well known [20, 26]. The present results have policy implications, as reducing disparities in health and healthcare is considered a priority in improving the outcomes quality of medical care [18, 19]. Increased attention in expanding the use of continent diversions may help reduce these disparities for patients undergoing RC for bladder cancer.

Funding

Healthcare Delivery Research Scholars Program, Mayo Clinic, Rochester, MN, USA.

Conflict of Interest

None declared.

Ancillary