SEARCH

SEARCH BY CITATION

Keywords:

  • cystectomy;
  • bladder cancer;
  • quality of health care;
  • outcomes;
  • bladder neoplasm

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Hospital and surgeon (provider) volume are associated with clinically significant outcomes for many types of surgery. Volume-outcome studies in patients undergoing radical cystectomy for bladder cancer have focused primarily on postoperative mortality. In the current study, the authors assessed the effect of cystectomy provider volume on long-term mortality.

METHODS

Using administrative databases, 2535 patients who underwent cystectomy by 199 surgeons in 90 hospitals in Ontario, Canada, between 1992 and 2004 were identified. The impact of provider volume on overall survival (OS) was assessed using Cox proportional hazards models fully adjusted for patient and tumor characteristics. Separate models were fit to examine the effect of surgeon and hospital volume. To confirm that the impact of volume on OS was independent of the effect of volume on short-term mortality, analyses were repeated excluding those patients experiencing postoperative deaths.

RESULTS

Of 2535 patients, 1796 (70.9%) died during the study period. Both higher hospital volume (hazards ratio [per unit increase in average annual number of procedures], 0.995; 95% confidence interval, 0.990-1.000 [P = .044]) and higher surgeon volume (hazards ratio, 0.984; 95% confidence interval, 0.975-0.994 [P = .002]) were found to be significantly associated with improved OS. Excluding post-operative deaths did not alter the results. Further analyses revealed that the benefit of high volume was attained by receiving care from either high-volume hospitals or high-volume surgeons.

CONCLUSIONS

High-volume providers were associated with improved long-term mortality rates compared with low-volume providers. This finding was independent of the effect of volume on perioperative mortality, suggesting that provider volume effects continue to manifest long after surgery. Cancer 2013;119:3546–3554.. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Patients who undergo radical cystectomy for bladder cancer are at significant risk of postoperative death and have a poor life expectancy. Postoperative mortality rates and 5-year overall survival rates typically range between 1% and 4% and 40% and 60%,[1-6] respectively. One way to improve on these outcomes is to optimize the quality of care for patients undergoing radical cystectomy. Several studies have demonstrated that hospitals and surgeons with higher cystectomy volumes generally have better postoperative outcomes compared with lower-volume providers.[1, 7, 8]

In terms of long-term survival, higher case loads may lead to improved surgical technique and performance of an optimal tumor-clearing surgery (which has been shown to improve long-term survival[9-11]); a more standardized approach to chemotherapy; and improvements in other nonsurgical approaches such as tumor surveillance, medical oncology involvement, and/or screening for complications of urinary diversion. However, to the best of our knowledge, the relationship between hospital and surgeon volume and long-term survival after cystectomy remains unclear. Therefore, we investigated the impact of both hospital and surgeon cystectomy volume on long-term outcome.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Cohort Identification

After local research ethics board approval, patients undergoing radical cystectomy in the province of Ontario, Canada between 1992 and 2004 were identified from the Canadian Institute for Health Information Discharge Abstract Database (CIHI DAD) (Canadian Classification of Diagnostic, Therapeutic and Surgical Procedures code 69.51 and Canadian Classification of Health Interventions codes 1.PM.91 and 1.PM.92). The CIHI DAD contains information regarding all inpatient hospital admissions in Ontario including information on age, sex, comorbidity, urgency of admission, region of residence, and vital status for each cystectomy patient. The Charlson Comorbidity Index was derived from CIHI DAD International Classification of Diseases (ICD) diagnostic codes from each patient's index admission and from any hospital admissions in the year before cystectomy.[12-14]

Because radical cystectomy can be performed for both bladder cancer and nonbladder malignancies (eg, as part of larger exenterative procedures for patients with colorectal, prostate, or gynecological malignancies), we linked the CIHI data to the Ontario Cancer Registry (OCR) to select only those cystectomy patients with a diagnosis of bladder cancer. The OCR contains information concerning all incident cancers detected in the province of Ontario with 97% capture of incident cases of bladder cancer.[15] A total of 3296 patients undergoing cystectomy for bladder cancer were identified. Because of the importance of pathological variables in assessing survival outcomes, we limited analyses assessing long-term outcomes to the 2535 individuals (77%) who had pathology reports available for review in the OCR. The reports were reviewed for variables including pathologic stage, grade, surgical margin and lymph node status, and the presence of lymphovascular invasion or perineural invasion. Pathologic staging was based on the 2002 American Joint Committee on Cancer staging system.[16]

Outcomes

The primary outcome of the current study was overall survival. Patients were followed from the date of cystectomy to March 31, 2007.

Volume Definitions

Hospital volume was defined as the average annual number of cystectomy cases performed at an institution during the study time period. Hospitals were identified using CIHI DAD institution-unique identifiers. Surgeon volume was defined as the average annual number of cystectomy cases performed by a surgeon during his/her active years of clinical activity. Surgeons were identified using Ontario Health Insurance Plan (OHIP) unique identifiers. Each cystectomy identified in CIHI is linkable by a combination of unique identifiers and procedure type and date to an OHIP billing fee code (S484, S485, S453, and S440) and thus a specific surgeon. A minority of physicians in Ontario are salaried and therefore do not submit billing codes. Consequently, 160 cystectomy cases (4.9%) were missing surgeon identifiers and therefore were omitted from analyses of surgeon volume.

For descriptive analyses, we also divided patients into quartiles of provider (hospital and surgeon) volume. Briefly, on calculating the average annual volume, patients were divided into 4 relatively equal groups based on volume category such that patients who were operated on by a specific surgeon (or hospital for hospital-volume quartiles) would be allocated to the same group. Our operationalization of volume thus resulted in slightly different numbers of patients per volume quartile.

Potential Confounding Variables

All multivariable analyses were risk-adjusted for patient factors including age, sex, admission status (urgent/emergent vs elective), Charlson comorbidity score, patient location of residence at the time of surgery (operationalized as 1 of 14 Local Health Integration Networks in Ontario), year of surgery, and socioeconomic status (SES). SES was derived from Canadian census neighborhood-specific quintiles of income. Analyses were also adjusted for tumor characteristics and the use of adjuvant chemotherapy. Adjuvant chemotherapy was defined by the initiation of chemotherapy within the first 6 months postoperatively. We chose a 6-month time period because this allowed ample time for patient discharge, postoperative follow-up, referral to medical oncology, and the initiation of chemotherapy. We did not account for the use of neoadjuvant chemotherapy because this treatment was not widely used during the study time period (< 1% of patients).

Statistical Analyses

All statistical analyses were performed using SAS statistical software (version 9.1.3; SAS Institute Inc, Cary, NC). A 2-sided P value of .05 was defined as being statistically significant. Characteristics of the patients were compared across quartiles of both hospital volume and surgeon volume. Kaplan-Meier curves depicting survival across quartiles of volume were constructed for both hospital and surgeon volume.

Multivariable Cox proportional hazards modeling was performed to assess the effect of volume (surgeon or hospital) on overall mortality. We used marginal Cox proportional hazards models with robust variance estimation to account for the lack of independence due to the clustering of patients by either hospitals or surgeons.[17, 18] In all analyses, volume was modeled as a continuous variable. Multicollinearity, defined as a variance inflation factor (VIF) of > 10, was determined for all variables to ensure collinearity did not affect the regression models.[19] To avoid survivor treatment bias when adjusting for adjuvant chemotherapy, we modeled the use of adjuvant chemotherapy as a time-dependent covariate.[20]

Because the association between provider volume and surgical mortality has been established by others,[1, 7, 8] we performed a secondary analysis in which patients experiencing postoperative mortality were excluded. We used 2 different definitions of postoperative mortality: 1) death occurring within 30 days of cystectomy or before discharge or 2) death occurring within 90 days of cystectomy, because both have been used in publications reporting cystectomy volume–short-term mortality associations. These analyses removed the impact of operative deaths and allowed us to determine the true effect of volume on long-term outcomes after the perioperative period.

Because risk adjustment using administrative data sets may not be fully accurate, we performed sensitivity analyses, reproducing the multivariable Cox proportional hazards models using only the healthiest patients (those with a Charlson score of 0 or 1) to decrease the likelihood of unmeasured confounding.[21] A second sensitivity analysis was performed to assess the impact of not omitting patients without available pathology information on the association between provider volume and overall survival. Adjusting only for those covariates derived from administrative data, all 3296 patients were analyzed. We hypothesized that consistent demonstration of a volume-outcome relationship in the entire population would support the results of the fully adjusted model in which 761 patients with missing pathology reports were omitted.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Patient and Provider Demographics

From 1992 to 2004, radical cystectomy was performed by 199 surgeons in 90 hospitals across Ontario. Baseline information for the entire cohort, divided into quartiles of hospital volume and surgeon volume, are presented in Tables 1 and 2, respectively. The average annual hospital volumes per quartile were 2.13 (range, 0.77-3.22), 4.49 (range, 3.23-5.85), 10.45 (range, 6.00-17.00), and 26.12 (range, 19.43-32.63) cases per year for quartiles 1, 2, 3, and 4, respectively. Likewise, the average annual surgeon volumes per quartile were 0.96 (range, 0.77-1.54), 2.05 (range, 1.67-2.54), 4.44 (range, 2.63-8.08), and 11.56 (range, 8.11-16.71) cases per year for quartiles 1, 2, 3 and 4, respectively. The annual surgeon and hospital volume increased over the duration of the cohort from 2.36 per surgeon and 3.45 per hospital in 1992 to 3.58 per surgeon and 6.76 per hospital in 2004. Patients treated by high-volume providers were more likely to receive adjuvant chemotherapy. Higher rates of lymphadenectomy and lymph node-negative disease were associated with high-volume hospitals and surgeons.

Table 1. Patient-Level and Pathologic Variables by Hospital Volume Quartilea
VariableHospital VolumeP
Quartile 1 (n = 639)Quartile 2 (n = 604)Quartile 3 (n = 598)Quartile 4 (n = 694)
  1. Abbreviations: LHIN, Local Health Integration Network; LVI, lymphovascular invasion; SES, socioeconomic status.

  2. Percentages may not add to 100 due to rounding.

  3. a

    Hospital volume increases with quartiles. Values are shown as counts (percentages) or means (standard deviations). The Kruskal-Wallis test was used for continuous variables and the chi-square or Fisher exact test was used for categorical variables.

  4. b

    Comorbidity scale was based on Charlson scores: none indicates a Charlson score of 0; mild, Charlson score of 1; moderate, Charlson score of 2; severe, Charlson score >2.

  5. c

    Quintile 5 refers to the highest socioeconomic (neighborhood income) status.

  6. d

    Refers to a patient's region of residence modeled as a 14-level categorical variable. Patient allocation by LHIN is not shown for space considerations.

  7. e

    Percentages refer to those patients who have undergone a lymphadenectomy.

Patient level
Age, y68.4 (9.3)67.4 (10.0)68.5 (9.2)66.8 (10.8).035
Sex    .781
Male519 (81.2%)496 (82.1%)488 (81.6%)555 (80.0%) 
Comorbidityb    .196
None212 (33.2%)193 (32.0%)188 (31.4%)190 (27.4%) 
Mild59 (9.2%)58 (9.6%)51 (8.5%)56 (8.1%) 
Moderate147 (23.0%)140 (23.2%)149 (24.9%)158 (22.8%) 
Severe221 (34.6%)213 (35.3%)210 (35.1%)290 (41.8%) 
SESc    <.001
Quintile 1118 (18.5%)83 (13.7%)127 (21.2%)117 (16.9%) 
Quintile 2130 (20.3%)141 (23.3%)140 (23.4%)137 (19.7%) 
Quintile 3134 (21.0%)118 (19.5%)125 (20.9%)113 (16.3%) 
Quintile 4111 (17.4%)116 (19.2%)106 (17.7%)121 (17.4%) 
Quintile 5128 (20.0%)131 (21.7%)86 (14.4%)186 (26.8%) 
Admission status    .184
Urgent/emergent84 (13.2%)81 (13.4%)100 (16.7%)111 (16.0%) 
Adjuvant chemotherapy82 (12.8%)87 (14.4%)46 (7.7%)133 (19.2%)<.001
LHINd    <.001
Pathology
Tumor stage    .054
Tx3 (0.5%)1 (0.2%)4 (0.7%)0 (0%) 
T013 (2.0%)7 (1.2%)13 (2.2%)14 (2.0%) 
Ta13 (2.0%)13 (2.2%)13 (2.2%)12 (1.7%) 
Tis28 (4.4%)38 (6.3%)37 (6.2%)24 (3.5%) 
T165 (10.2%)47 (7.8%)58 (9.7%)68 (9.8%) 
T2163 (25.5%)165 (27.3%)147 (24.6%)171 (24.6%) 
T3237 (37.1%)228 (37.8%)197 (32.9%)234 (33.7%) 
T4117 (18.3%)105 (17.4%)129 (21.6%)171 (24.6%) 
Grade    .641
Not specified42 (6.6%)42 (7.0%)46 (7.7%)51 (7.4%) 
Grade 16 (0.9%)13 (2.2%)12 (2.0%)12 (1.7%) 
Grade 274 (11.6%)79 (13.1%)87 (14.6%)90 (13.0%) 
Grade 3516 (80.9%)470 (77.8%)453 (75.8%)541 (78.0%) 
Positive margin status106 (16.6%)92 (15.2%)101 (16.9%)115 (16.6%).866
LVI258 (40.4%)241 (39.9%)221 (37.0%)299 (43.1%).168
Perineural invasion122 (19.2%)79 (13.1%)83 (13.9%)114 (16.4%).014
Lymphadenectomye324 (50.8%)363 (60.3%)337 (56.4%)556 (80.4%)<.001
Positive lymph node status    <.001
Nx254 (39.8%)210 (34.8%)209 (35.0%)105 (15.1%) 
N0244 (38.2%)263 (43.5%)282 (47.2%)404 (58.2%) 
N+141 (36.6%)131 (33.3%)107 (27.5%)185 (31.4%) 
Table 2. Patient-Level and Pathologic Variables by Surgeon Volume Quartilea
VariableSurgeon VolumeP
Quartile 1 (n = 640)Quartile 2 (n = 560)Quartile 3 (n = 594)Quartile 4 (n = 581)
  1. Abbreviations: LHIN, Local Health Integration Network; LVI, lymphovascular invasion; SES, socioeconomic status.

  2. Percentages may not add to 100 due to rounding.

  3. a

    Surgeon volume increases with quartiles. Values are shown as counts (percentages) or means (standard deviations). The Kruskal-Wallis test was used for continuous variables and the chi-square or Fisher exact test was used for categorical variables.

  4. b

    Comorbidity scale was based on Charlson scores: none indicates Charlson score of 0; mild, Charlson score of 1; moderate, Charlson score of 2; severe, Charlson score >2.

  5. c

    Quintile 5 refers to the highest socioeconomic (neighborhood income) status.

  6. d

    Refers to a patient's region of residence modeled as a 14-level categorical variable. Patient allocation by LHIN is not shown for space considerations.

  7. e

    Percentages refer to those patients who have undergone a lymphadenectomy.

Patient level
Age, y68.1 (9.2)67.8 (9.7)68.3 (10.0)67.2 (10.4).319
Sex    .114
Male528 (82.5%)435 (77.7%)489 (82.3%)476 (81.9%) 
Comorbidityb    .267
None212 (33.1%)179 (32.0%)170 (28.6%)166 (28.6%) 
Mild69 (10.8%)50 (8.9%)51 (8.6%)46 (7.9%) 
Moderate145 (22.7%)121 (21.6%)145 (24.4%)138 (23.8%) 
Severe214 (33.4%)210 (37.5%)228 (38.4%)231 (39.8%) 
SESc    .070
Quintile 1101 (15.8%)107 (19.1%)114 (19.2%)92 (15.8%) 
Quintile 2155 (24.2%)110 (19.6%)134 (22.6%)115 (19.8%) 
Quintile 3140 (21.9%)103 (18.4%)105 (17.7%)111 (19.1%) 
Quintile 498 (15.3%)116 (20.7%)102 (17.2%)109 (18.8%) 
Quintile 5128 (20.0%)108 (19.3%)122 (20.5%)144 (24.8%) 
Admission status    .056
Urgent/emergent84 (13.1%)84 (15.0%)78 (13.1%)105 (18.1%) 
Adjuvant chemotherapy89 (13.9%)55 (9.8%)78 (13.1%)103 (17.7%).002
LHINd    <.001
Pathology
Tumor stage    .197
Tx0 (0%)3 (0.5%)3 (0.5%)0 (0%) 
T09 (1.4%)13 (2.3%)9 (1.5%)15 (2.6%) 
Ta9 (1.4%)12 (2.1%)15 (2.5%)15 (2.6%) 
Tis32 (5.0%)25 (4.5%)26 (4.4%)33 (5.7%) 
T158 (9.1%)54 (9.6%)62 (10.4%)55 (9.5%) 
T2181 (28.3%)138 (24.6%)145 (24.4%)141 (24.3%) 
T3234 (36.6%)214 (38.2%)205 (34.5%)186 (32.0%) 
T4117 (18.3%)101 (18.0%)129 (21.7%)136 (23.4%) 
Grade    .130
Not specified36 (5.6%)41 (7.3%)48 (8.1%)46 (7.9%) 
Grade 110 (1.6%)9 (1.6%)8 (1.4%)14 (2.4%) 
Grade 284 (13.2%)80 (14.3%)62 (10.4%)91 (15.7%) 
Grade 3509 (79.7%)430 (76.8%)476 (80.1%)430 (74.0%) 
Positive margin status106 (16.6%)83 (14.8%)101 (17.0%)94 (16.2%).770
LVI250 (39.1%)225 (40.2%)247 (41.6%)230 (39.6%).826
Perineural invasion93 (14.6%)105 (18.8%)90 (15.2%)90 (15.5%).207
Lymphadenectomye341 (53.5%)286 (51.3%)394 (66.4%)457 (78.7%)<.001
Positive lymph node status    <.001
Nx255 (39.8%)224 (40.0%)166 (28.0%)86 (14.8%) 
N0243 (38.0%)226 (40.4%)284 (47.8%)357 (61.5%) 
N+142 (36.9%)110 (32.7%)144 (33.6%)138 (27.9%) 

Long-Term Mortality

Kaplan-Meier survival curves for the entire cohort divided by quartiles of hospital and surgeon volume are shown in Figure 1. High-volume providers had improved outcomes compared with lower-volume providers. The 5-year overall survival rate was 35%. The total number of deaths from lowest to highest hospital-volume quartile was 468 (73%), 434 (72%), 438 (73%), and 456 (66%), respectively. The total number of deaths from lowest to highest surgeon-volume quartile was 480 (75%), 405 (72%), 414 (70%), and 383 (66%), respectively.

image

Figure 1. Kaplan-Meier survival plots are shown by (A) hospital volume and (B) surgeon volume quartiles between 1992 and 2004. Increasing quartiles indicate increasing hospital and surgeon volume.

Download figure to PowerPoint

Of the 2535 patients for whom pathology information was available, 1796 died during the study time period. The mean and median follow-up for the cohort was 1260 days (standard deviation, 1276 days) and 786 days (range, 0 days-5441 days), respectively. Both hospital and surgeon volume were found to be statistically significantly associated with long-term mortality in both unadjusted and adjusted Cox proportional hazards models (Tables 3 and 4). The adjusted hazards ratio (HR) for hospital volume (HR, 0.995) means that for every additional cystectomy performed at a hospital per year, the risk of long-term mortality decreased by 0.5%. Likewise, for every additional cystectomy performed per year by an individual surgeon, the risk of long-term death diminished by 1.6% (HR, 0.984). The decrease in the instantaneous hazard of death for selected cystectomy volume thresholds, based on results from the Cox multivariate model, is presented in Table 5. Removal of patients who experienced an operative death resulted in HRs that were nearly identical to those in the primary analysis. Specifically, omitting patients with mortality within 30 days of cystectomy or before discharge resulted in HRs of 0.995 (95% confidence interval [95% CI], 0.991-1.000; P = .051) and 0.986 (95% CI, 0.976-0.996; P = .007), respectively, for hospital volume and surgeon volume. Using a varied definition of postoperative mortality (mortality within 90 days of cystectomy) also was not found to alter the results (HR for hospital volume, 0.995 [95% CI, 0.990-1.000; P = .045] and HR for surgeon volume, 0.982 [95% CI, 0.972-0.992; P < .001]).

Table 3. Effect of Hospital Volume on Long-Term Mortality
VariableHR95% CIPa
  1. Abbreviations: 95% CI, 95% confidence interval; HR, hazards ratio; LVI, lymphovascular invasion; PNI, perineural invasion; SES, socioeconomic status.

  2. a

    P values derived from Cox proportional hazards model after accounting for clustered data at the hospital level.

  3. b

    Also adjusted for county of residence and year of surgery.

  4. c

    Comorbidity scale was based on Charlson scores: none indicates a Charlson score of 0; mild, Charlson score of 1; moderate, Charlson score of 2; and severe, Charlson score >2.

  5. d

    Quintile 5 refers to the highest SES (neighborhood income) whereas quintile 1 is the lowest.

Unadjusted hospital volume0.9940.989–0.999.015
Adjustedb hospital volume0.9950.990–1.000.044
Age (per y)1.0241.017–1.030<.001
Sex (female)0.9190.789–1.072.282
Comorbidityc
None (referent)
Mild1.1320.966–1.326.126
Moderate1.1280.973–1.307.111
Severe1.4311.220–1.678<.001
Admission status1.1691.006–1.358.042
SES quintiled
11.2001.037–1.390.015
21.1160.994–1.252.063
30.9570.848–1.080.480
41.0110.871–1.174.883
5 (referent)
Tumor stage
T0, Ta, Tis (referent)
T11.3501.112–1.638.002
T21.4941.208–1.849<.001
T32.3591.898–2.932<.001
T42.7682.243–3.418<.001
Positive margin status1.4951.332–1.679<.001
Lymph node status
N0 (referent)
N+1.3161.122–1.544<.001
Nx1.2361.024–1.493.027
Adjuvant chemotherapy0.8690.750–1.008.063
LVI1.6141.484–1.754<.001
PNI1.0030.881–1.142.963
Tumor grade
1 (referent)
21.1160.714–1.746.630
31.1700.778–1.761.451
X (missing/pT0)1.2540.790–1.989.337
Table 4. Effect of Surgeon Volume on Long-Term Mortality
VariableHR95% CIPa
  1. Abbreviations: 95% CI, 95% confidence interval; HR, hazards ratio; LVI, lymphovascular invasion; PNI, perineural invasion; SES, socioeconomic status.

  2. a

    P values derived from the Cox proportional hazards model after accounting for clustered data at the surgeon level.

  3. b

    Also adjusted for county of residence and year of surgery.

  4. c

    Comorbidity scale was based on Charlson scores: none indicates a Charlson score of 0; mild, Charlson score of 1; moderate, Charlson score of 2; and severe, Charlson score >2.

  5. d

    Quintile 5 refers to the highest SES (neighborhood income) whereas quintile 1 is the lowest.

Unadjusted surgeon volume0.9810.973–0.990<.001
Adjustedb surgeon volume0.9840.975–0.994.002
Age (per y)1.0251.018–1.032<.001
Sex (female)0.9160.801–1.048.200
Comorbidityc
None (ref)
Mild1.1260.945–1.343.185
Moderate1.1681.006–1.356.042
Severe1.4671.266–1.700<.001
Admission status1.1661.014–1.342.032
SES quintiled
11.2121.034–1.420.018
21.1100.973–1.267.119
30.9680.842–1.111.641
41.0270.878–1.201.742
5 (referent)
Tumor stage
T0, Ta, Tis (referent)
T11.3281.049–1.681.018
T21.4511.146–1.836.002
T32.3031.838–2.885<.001
T42.7532.128–3.562<.001
Positive margin status1.5471.347–1.777<.001
Lymph node status
N0 (referent)
N+1.3071.113–1.536.001
Nx1.2231.008–1.484.042
Adjuvant chemotherapy0.8550.743–0.984.029
LVI1.6401.480–1.817<.001
PNI1.0000.867–1.155.995
Tumor grade
1 (referent)
21.1480.734–1.794.545
31.2040.775–1.870.410
X (missing/pT0)1.3060.802–2.126.284
Table 5. Decrease in Hazard of Overall Death by An Incremental Increase in the Number of Cystectomies Performed at the Hospital or Surgeon Level
Incremental Increase in the Annual No. of Cystectomy ProceduresHospital Volume HRSurgeon Volume HR
  1. HR indicates hazards ratio.

10.9950.984
50.9740.924
100.9480.853
200.8990.728

Incorporating both hospital volume and surgeon volume in the same Cox regression model nullified the significance of both variables (HR for hospital volume: 0.998 [95% CI, 0.990-1.006; P = .61]; HR for surgeon volume: 0.988 [95% CI, 0.971-1.005; P = .17]). Since the VIF for both of these variables suggested noncollinearity (VIF < 5), these variables were not measuring the same construct. This suggests that the benefits of high volume can be achieved with either a high-volume surgeon or a high-volume hospital because neither variable dominates its counterpart.

Repeating the analyses in only low-risk patients (ie, those with Charlson scores of 0 and 1) did not appear to alter the conclusions (HR for hospital volume: 0.985 [95% CI, 0.976-0.994; P = .002]; HR for surgeon volume: 0.972 [95% CI, 0.953-0.993; P = .008]). A second sensitivity analysis of all 3296 patients without adjustment for pathology variables yielded results that were similar to those of the pathology-adjusted analyses (HR for hospital volume: 0.993 [95% CI, 0.989-0.998; P = .003]; HR for surgeon volume: 0.979 [95% CI, 0.970-0.989; P < .001]). Overall, patients with missing pathology data were younger (mean, 67.1 years vs 67.9 years; P = .055), healthier (mean Charlson score of 2.2 vs 2.6; P < .001), less likely to be admitted emergently/urgently for cystectomy (10.9% vs 14.8%; P = .006), and less likely to be treated at a high-volume hospital (20.4% vs 27.4% at the highest-volume hospital; P < .001)

Finally, given the varied use of lymphadenectomy in the current study cohort, with 38% of patients not undergoing a formal pelvic lymph node dissection, we assessed the impact of lymphadenectomy on the relationship between provider volume and mortality. Incorporating the performance of lymphadenectomy as a variable into the fully adjusted model did not appear to alter the significant hospital or surgeon volume results (data not shown). In a subgroup analysis of the 1580 patients who underwent lymph node dissection, the HRs for volume remained largely unchanged (HR for hospital volume: 0.996 [95% CI, 0.990-1.002; P = .148]; HR for surgeon volume: 0.982 [95% CI, 0.969-0.996; P = .011]). Although hospital volume was not found to be statistically significant in this analysis, its associated HR suggests a protective effect of increased volume on long-term mortality, even in centers performing lymphadenectomy.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Using a data set with full population coverage from a publicly funded health care system, we demonstrated that higher hospital and surgeon volumes were significantly associated with increased overall survival. The association between volume and long-term outcome persisted even when operative deaths were excluded, suggesting that the impact of provider volume on overall survival is independent of perioperative death.

To the best of our knowledge, a consistent association between provider volume and long-term survival after cystectomy has not previously been reported. For example, Fairey et al demonstrated a paradoxical improvement in outcome with an annual surgeon volume in the range of 5 to 9 per year (HR, 0.66; 95% CI, 0.48-0.92) but not with a volume of ≥ 10 cases annually.[22] With respect to hospital volume, an analysis of Netherlands Cancer Registry data did reveal a significant mortality detriment for an annual cystectomy volume of < 10 (HR, 1.17; 95% CI, 1.01-1.35).[23] However, these analyses did not account for clustering of data or adjuvant chemotherapy in a time-dependent manner, both of which may contribute to false-positive findings.[17, 18, 20] Finally, Birkmeyer et al assessed the association between hospital volume and late survival in 2514 patients who underwent cystectomy using Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data but failed to demonstrate a significant association in adjusted analyses.[24] However, their HR of 0.90 suggested a 10% risk reduction associated with high-volume hospitals, which is similar to the findings of the current study.

To the best of our knowledge, the reasons for the relationship between volume and long-term outcome are currently unclear. We were able to explore one hypothesis (ie, the performance of lymph node dissection) in our data set. Analyzing only those patients who underwent a lymph node dissection did not appear to alter the surgeon-volume results. The HR for hospital volume was also similar to the HR in the primary analysis, although the HR was no longer statistically significant, likely due to a smaller sample size. Although this may suggest that the hospital volume effect may be due to the performance of a lymphadenectomy, an analysis incorporating the performance of lymph node dissection as a variable in our model (and thus using our full data set) yielded statistically significant and unchanged volume parameters, suggesting that the hospital volume finding in our subgroup analysis may in fact be due to a lack of statistical power. Nevertheless, high-volume surgeons may have better outcomes because of superior operative technique (eg, optimal cancer resections or improved lymph node harvesting and staging), vigilant follow-up and surveillance for recurrent disease, and/or timely and appropriate use of nonsurgical interventions such as chemotherapy. High-volume hospitals may have improved pathways to detect and treat patients with recurrent disease, superior access to chemotherapy, and/or better treatment of comorbid diseases. Although these explanations remain speculative, our analyses suggest that the high-volume benefit is obtainable at either the surgeon or hospital level. When incorporated within the same model, neither hospital nor surgeon volume was found to be statistically significant, suggesting that patients treated at either a high-volume center or by a high-volume surgeon can expect improved outcomes.

There are limitations to the current study. First, the analysis was limited to 2535 patients for whom pathology data were available. We restricted our patient population because of the importance of adjusting long-term cancer survival outcomes for pathologic variables such as tumor stage and grade. Patients without pathology data available for analysis tended to be healthier, younger, and less likely to be admitted on an emergent basis. They also were more likely to be treated at lower-volume hospitals. However, an analysis including all patients but not adjusting for pathological variables found similar HRs. Second, although we were able to assess the effect of volume on overall survival, our administrative data precluded the assessment of other outcome measures such as disease-specific or recurrence-free survival because these data elements were not available in the database.[25] Third, although we have identified a difference in mortality based on provider volume, the data from the current study do not explain why high-volume surgeons or hospitals have better outcomes. Increasingly, evidence suggests that volume is a surrogate for underlying structures and/or processes of care that affect quality of care.[26] A search for relevant volume-related structures and processes of care is currently under way.

The results of the current study raise the important health policy question of how to narrow the quality-of-care gap between high-volume and low-volume providers. It should be noted that our finding of a relationship between volume and outcome for patients undergoing cystectomy was identified in a health care system in which physicians and hospitals are publicly funded by a single payer. To our knowledge to date, the vast majority of such studies have arisen from private health care systems. The current study results suggest that the processes and structures of care responsible for any volume-based quality of care discrepancy may be similar in both public and private health care models. Regionalization of health services has been proposed as a possible solution. Private insurers in the United States have already promoted volume-based referral patterns, implementing minimum volume thresholds for several complex operations based on published volume-outcome studies.[27] However, many question the usefulness of regionalization, given the additional burden of excess travel time for patients,[28] the potential marginalization of lower-volume physicians, and the logistical difficulties inherent to implementing system-wide change.[29] Identifying relevant structures and/or processes of care, and improving them, is another way to close the quality-of-care gap. Specifically, initiatives such as the National Surgical Quality Improvement Program have been successful at identifying deficiencies in care at underperforming hospitals for a variety of disease sites.[30] The adoption of such a programatic approach for radical cystectomy in patients with bladder cancer may help to alleviate some of our observed discrepancies in mortality. Other approaches, such as nonpunitive incentives promoting excellence in care or public reporting of results, could also modify physician behavior to achieve lower mortality rates.

Conclusions

In Ontario, Canada both hospital and surgeon volume were found to be significantly associated with long-term mortality in patients undergoing radical cystectomy. This finding was independent of the effect of provider volume on operative mortality. The mechanisms causing the relationship between volume and long-term survival remain unclear. Research into the structures and processes of care underlying the volume-outcome relationship may help to reduce disparities and improve outcomes after cystectomy.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No endorsement by the ICES or the Ontario MOHLTC is intended or should be inferred. Dr. Kulkarni received salary support via a Canadian Institutes of Health Research clinical research fellowship. Dr. Austin was supported in part by a Career Investigator Award from the Heart and Stroke Foundation of Ontario. Dr. Laupacis holds a Canada Research Chair in Health Policy and Citizen Engagement.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Dr. Fleshner has acted as a consultant for Amgen, Janssen, Astellas, and Eli Lilly; has received grants from the Canadian Cancer Society Research Institute and Prostate Cancer Canada; and has performed clinical trials for Ferring, Astellas, Amgen, and Janssen.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  • 1
    Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346:11281137.
  • 2
    Elting LS, Pettaway C, Bekele BN, et al. Correlation between annual volume of cystectomy, professional staffing, and outcomes: a statewide, population-based study. Cancer. 2005;104:975985.
  • 3
    Hollenbeck BK, Dunn RL, Miller DC, Daignault S, Taub DA, Wei JT. Volume-based referral for cancer surgery: informing the debate. J Clin Oncol. 2007;25:9196.
  • 4
    Mahmud SM, Fong B, Fahmy N, Tanguay S, Aprikian AG. Effect of preoperative delay on survival in patients with bladder cancer undergoing cystectomy in Quebec: a population based study. J Urol. 2006;175:7883.
  • 5
    Stein JP, Lieskovsky G, Cote R, et al. Radical cystectomy in the treatment of invasive bladder cancer: long-term results in 1,054 patients. J Clin Oncol. 2001;19:666675.
  • 6
    Ghoneim MA, el-Mekresh MM, el-Baz MA, el-Attar IA, Ashamallah A. Radical cystectomy for carcinoma of the bladder: critical evaluation of the results in 1,026 cases. J Urol. 1997;158:393399.
  • 7
    Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med. 2003;349:21172127.
  • 8
    Konety BR, Dhawan V, Allareddy V, Joslyn SA. Impact of hospital and surgeon volume on in-hospital mortality from radical cystectomy: data from the health care utilization project. J Urol. 2005;173:16951700.
  • 9
    Herr H, Lee C, Chang S, Lerner S; Bladder Cancer Collaborative Group. Standardization of radical cystectomy and pelvic lymph node dissection for bladder cancer: a collaborative group report. J Urol. 2004;171:18231828; discussion 1827–1828.
  • 10
    Herr HW, Faulkner JR, Grossman HB, et al. Surgical factors influence bladder cancer outcomes: a cooperative group report. J Clin Oncol. 2004;22:27812789.
  • 11
    Leissner J, Ghoneim MA, Abol-Enein H, et al. Extended radical lymphadenectomy in patients with urothelial bladder cancer: results of a prospective multicenter study. J Urol. 2004;171:139144.
  • 12
    Quan H, Parsons GA, Ghali WA. Validity of procedure codes in International Classification of Diseases, 9th revision, clinical modification administrative data. Med Care. 2004;42:801809.
  • 13
    Preen DB, Holman CD, Spilsbury K, Semmens JB, Brameld KJ. Length of comorbidity lookback period affected regression model performance of administrative health data. J Clin Epidemiol. 2006;59:940946.
  • 14
    Nuttall M, van der Meulen J, Emberton M. Charlson scores based on ICD-10 administrative data were valid in assessing comorbidity in patients undergoing urological cancer surgery. J Clin Epidemiol. 2006;59:265273.
  • 15
    Cancer Care Ontario Bladder ICD9 Code 188. Toronto, Ontario, Canada, 2003.
  • 16
    Green FL, Page DL, Fleming ID, et al, eds. AJCC Cancer Staging Manual. 6th ed. New York: Springer-Verlag; 2002.
  • 17
    Ying G, Liu C. Statistical analysis of clustered data using SAS System. Proceedings of the North East SAS Users Group (NESUG) Conference; September 17–20, 2006; Philadelphia, PA. Available at: http://www.nesug.org/Proceedings/nesug06/an/da01.pdf.
  • 18
    Chuang SK, Tian L, Wei LJ, Dodson TB. Kaplan-Meier analysis of dental implant survival: a strategy for estimating survival with clustered observations. J Dent Res. 2001;80:20162020.
  • 19
    Christensen LA. Introduction to building a linear regression model. Proceedings of the Twenty-Second Annual SAS Users Group International Conference; March 16–19, 1997; San Diego, CA.
  • 20
    Austin PC, Mamdani MM, van Walraven C, Tu JV. Quantifying the impact of survivor treatment bias in observational studies. J Eval Clin Pract. 2006;12:601612.
  • 21
    Naylor CD, Guyatt GH. Users' guides to the medical literature. X. How to use an article reporting variations in the outcomes of health services. The Evidence-Based Medicine Working Group. JAMA. 1996;275:554558.
  • 22
    Fairey AS, Jacobsen NE, Chetner MP, et al. Associations between comorbidity, and overall survival and bladder cancer specific survival after radical cystectomy: results from the Alberta Urology Institute Radical Cystectomy database. J Urol. 2009;182:8592; discussion 93.
  • 23
    Goossens-Laan CA, Visser O, Hulshof MC, et al. Survival after treatment for carcinoma invading bladder muscle: a Dutch population-based study on the impact of hospital volume. BJU Int. 2012;110:226232.
  • 24
    Birkmeyer JD, Sun Y, Wong SL, Stukel TA. Hospital volume and late survival after cancer surgery. Ann Surg. 2007;245:777783.
  • 25
    Hall S, Schulze K, Groome P, Mackillop W, Holowaty E. Using cancer registry data for survival studies: the example of the Ontario Cancer Registry. J Clin Epidemiol. 2006;59:6776.
  • 26
    Urbach DR, Baxter NN. Does it matter what a hospital is “high volume” for? Specificity of hospital volume-outcome associations for surgical procedures: analysis of administrative data. Qual Saf Health Care. 2004;13:379383.
  • 27
    Shahian DM, Normand SL. The volume-outcome relationship: from Luft to Leapfrog. Ann Thorac Surg. 2003;75:10481058.
  • 28
    Birkmeyer JD, Siewers AE, Marth NJ, Goodman DC. Regionalization of high-risk surgery and implications for patient travel times. JAMA. 2003;290:27032708.
  • 29
    Casebeer A. Regionalizing Canadian healthcare: the good–the bad–the ugly. Healthc Pap. 2004;5:8893; discussion 96–99.
  • 30
    Hall BL, Hamilton BH, Richards K, Bilimoria KY, Cohen Me, Ko CY. Does surgical quality improve in the American College of Surgeons National Surgical Quality Improvement Program: an evaluation of all participating hospitals. Ann Surg. 2009;250:363376.