[Skip to Content]
[Skip to Content Landing]
Figure 1.  Flow Diagram of Cohort Formation
Flow Diagram of Cohort Formation

The final cohort comprised 1278 patients with low-risk bladder cancer and 2115 patients with high-risk bladder cancer. VA indicates Department of Veterans Affairs.

Figure 2.  Facility-Level Variation in Adjusted Frequency of Cystoscopy Procedures for Low-Risk and High-Risk Patients
Facility-Level Variation in Adjusted Frequency of Cystoscopy Procedures for Low-Risk and High-Risk Patients

Facilities are ranked from lowest frequency to highest frequency of cystoscopy for patients with low-risk (A) and high-risk (B) bladder cancer. The mean frequency across all facilities is indicated on the y-axis. Error bars indicate 95% confidence intervals. Frequency of cystoscopy was adjusted for age, comorbidity, year of diagnosis, and length of follow-up.

Figure 3.  Facility-Level Correlation of Cystoscopy Frequency Between Low-Risk and High-Risk Patients
Facility-Level Correlation of Cystoscopy Frequency Between Low-Risk and High-Risk Patients

Each dot represents 1 facility. The line represents the same cystoscopy frequency for low-risk patients (x-axis) and high-risk patients (y-axis). The shaded area represents facilities where low-risk and high-risk patients undergo cystoscopy at comparable rates (ie, absolute difference of less than 1 cystoscopy over 2 years). Orange dots represent facilities with a statistically significantly higher frequency for high-risk vs low-risk patients.

Table 1.  Demographic Characteristics of Patientsa
Demographic Characteristics of Patientsa
Table 2.  Characteristics of the 85 Facilities Included
Characteristics of the 85 Facilities Included
1.
The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: kidney cancer. Version 2.2018. https://www.nccn.org/professionals/physician_gls/pdf/kidney.pdf. Published November 30, 2017. Accessed December 4, 2017.
2.
The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: non-small cell lung cancer. Version 1.2018. https://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf. Published November 17, 2017. Accessed December 4, 2017.
3.
The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: prostate cancer. Version 2.2018. https://www.nccn.org/professionals/physician_gls/pdf/prostate.pdf. Published March 8, 2018. Accessed April 4, 2018.
4.
Lieberman  DA, Rex  DK, Winawer  SJ, Giardiello  FM, Johnson  DA, Levin  TR.  Guidelines for colonoscopy surveillance after screening and polypectomy: a consensus update by the US Multi-Society Task Force on Colorectal Cancer.  Gastroenterology. 2012;143(3):844-857. doi:10.1053/j.gastro.2012.06.001PubMedGoogle ScholarCrossref
5.
Centers for Medicare & Medicaid Services Quality Payment Program. Merit-based incentive payment system (MIPS) measure #185 & #320. https://qpp.cms.gov/mips/quality-measures. Accessed September 29, 2017.
6.
Kattan  MW, Hess  KR, Amin  MB,  et al; members of the AJCC Precision Medicine Core.  American Joint Committee on Cancer acceptance criteria for inclusion of risk models for individualized prognosis in the practice of precision medicine.  CA Cancer J Clin. 2016;66(5):370-374. doi:10.3322/caac.21339PubMedGoogle ScholarCrossref
7.
Howlader  N, Noone  AM, Krapcho  M,  et al. Seer Cancer Statistics Review, 1975-2014. Bethesda, MD: National Cancer Institute; 2017. http://seer.cancer.gov/csr/1975_2014/. Accessed March 1, 2018.
8.
Chang  SS, Boorjian  SA, Chou  R,  et al. Diagnosis and treatment of non-muscle invasive bladder cancer: AUA/SUO joint guideline. https://www.auanet.org/education/guidelines/non-muscle-invasive-bladder-cancer.cfm. Accessed May 3, 2016.
9.
Snyder  C, Harlan  L, Knopf  K, Potosky  A, Kaplan  R.  Patterns of care for the treatment of bladder cancer.  J Urol. 2003;169(5):1697-1701. doi:10.1097/01.ju.0000056727.30546.b7PubMedGoogle ScholarCrossref
10.
Holmäng  S.  Follow-up of patients with noninvasive and superficially invasive bladder cancer.  Semin Urol Oncol. 2000;18(4):273-279.PubMedGoogle Scholar
11.
Schroeck  FR, Smith  N, Shelton  JB.  Implementing risk-aligned bladder cancer surveillance care.  Urol Oncol. 2018;36(5):257-264. doi:10.1016/j.urolonc.2017.12.016PubMedGoogle ScholarCrossref
12.
The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: bladder cancer. Version 1.2017. https://www.nccn.org/professionals/physician_gls/pdf/bladder.pdf. Published December 21, 2016. Accessed January 29, 2017.
13.
Nieder  AM, Brausi  M, Lamm  D,  et al.  Management of stage T1 tumors of the bladder: International Consensus Panel.  Urology. 2005;66(6)(suppl 1):108-125. doi:10.1016/j.urology.2005.08.066PubMedGoogle ScholarCrossref
14.
Oosterlinck  W, van der Meijden  A, Sylvester  R,  et al; European Association of Urology. Guidelines on TaT1 (non-muscle invasive) bladder cancer. http://uroweb.org/wp-content/uploads/EAU-Guidelines-TaT1-Bladder-Cancer-2006.pdf. Updated March 1, 2006. Accessed June 15, 2017.
15.
Brausi  M, Witjes  JA, Lamm  D,  et al.  A review of current guidelines and best practice recommendations for the management of nonmuscle invasive bladder cancer by the International Bladder Cancer Group.  J Urol. 2011;186(6):2158-2167. doi:10.1016/j.juro.2011.07.076PubMedGoogle ScholarCrossref
16.
Kassouf  W, Traboulsi  SL, Kulkarni  GS,  et al.  CUA guidelines on the management of non-muscle invasive bladder cancer.  Can Urol Assoc J. 2015;9(9-10):E690-E704. doi:10.5489/cuaj.3320PubMedGoogle ScholarCrossref
17.
National Institute for Health and Care Excellence. Bladder cancer: diagnosis and management. NICE guideline NG2. https://www.nice.org.uk/guidance/ng2/evidence. Published February 25, 2015. Accessed July 24, 2017.
18.
Kassouf  W, Traboulsi  SL, Schmitz-Dräger  B,  et al.  Follow-up in non–muscle-invasive bladder cancer—International Bladder Cancer Network recommendations.  Urol Oncol. 2016;34(10):460-468. doi:10.1016/j.urolonc.2016.05.028PubMedGoogle ScholarCrossref
19.
Hollingsworth  JM, Zhang  YS, Miller  DC,  et al.  Identifying better practices for early-stage bladder cancer.  Med Care. 2011;49(12):1112-1117. doi:10.1097/MLR.0b013e3182353bafPubMedGoogle ScholarCrossref
20.
Shulkin  DJ.  Beyond the VA crisis—becoming a high-performance network.  N Engl J Med. 2016;374(11):1003-1005. doi:10.1056/NEJMp1600307PubMedGoogle ScholarCrossref
21.
Schroeck  FR, Sirovich  B, Seigne  JD, Robertson  DJ, Goodney  PP.  Assembling and validating data from multiple sources to study care for veterans with bladder cancer.  BMC Urol. 2017;17(1):78. doi:10.1186/s12894-017-0271-xPubMedGoogle ScholarCrossref
22.
Babjuk  M, Boehle  A, Burger  M,  et al.  EAU guidelines on non–muscle-invasive urothelial carcinoma of the bladder: update 2016.  Eur Urol. 2017;71(3):447-461. doi:10.1016/j.eururo.2016.05.041PubMedGoogle ScholarCrossref
23.
Schroeck  FR, Patterson  OV, Alba  PR,  et al.  Development of a natural language processing engine to generate bladder cancer pathology data for health services research.  Urology. 2017;110:84-91. doi:10.1016/j.urology.2017.07.056PubMedGoogle ScholarCrossref
24.
Hollenbeck  BK, Ye  Z, Dunn  RL, Montie  JE, Birkmeyer  JD.  Provider treatment intensity and outcomes for patients with early-stage bladder cancer.  J Natl Cancer Inst. 2009;101(8):571-580. doi:10.1093/jnci/djp039PubMedGoogle ScholarCrossref
25.
Klabunde  CN, Potosky  AL, Legler  JM, Warren  JL.  Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53(12):1258-1267. doi:10.1016/S0895-4356(00)00256-0PubMedGoogle ScholarCrossref
26.
Quan  H, Sundararajan  V, Halfon  P,  et al.  Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.  Med Care. 2005;43(11):1130-1139. doi:10.1097/01.mlr.0000182534.19832.83PubMedGoogle ScholarCrossref
27.
Schroeck  FR, Pattison  EA, Denhalter  DW,  et al.  Early stage bladder cancer: do pathology reports tell us what we need to know?  Urology. 2016;98:58-63. doi:10.1016/j.urology.2016.07.040PubMedGoogle ScholarCrossref
28.
Schroeck  FR, Montie  JE, Hollenbeck  BK.  Surveillance strategies for non–muscle invasive bladder cancer.  AUA Update Ser. 2012;31:313-323.Google Scholar
29.
MacKenzie  TA, Grunkemeier  GL, Grunwald  GK,  et al.  A primer on using shrinkage to compare in-hospital mortality between centers.  Ann Thorac Surg. 2015;99(3):757-761. doi:10.1016/j.athoracsur.2014.11.039PubMedGoogle ScholarCrossref
30.
Schoen  RE, Pinsky  PF, Weissfeld  JL,  et al.  Utilization of surveillance colonoscopy in community practice.  Gastroenterology. 2010;138(1):73-81. doi:10.1053/j.gastro.2009.09.062PubMedGoogle ScholarCrossref
31.
Makarov  DV, Desai  R, Yu  JB,  et al.  Appropriate and inappropriate imaging rates for prostate cancer go hand in hand by region, as if set by thermostat.  Health Aff (Millwood). 2012;31(4):730-740. doi:10.1377/hlthaff.2011.0336PubMedGoogle ScholarCrossref
32.
Makarov  DV, Soulos  PR, Gold  HT,  et al.  Regional-level correlations in inappropriate imaging rates for prostate and breast cancers: potential implications for the choosing wisely campaign.  JAMA Oncol. 2015;1(2):185-194. doi:10.1001/jamaoncol.2015.37PubMedGoogle ScholarCrossref
33.
Makarov  DV, Loeb  S, Ulmert  D, Drevin  L, Lambe  M, Stattin  P.  Prostate cancer imaging trends after a nationwide effort to discourage inappropriate prostate cancer imaging.  J Natl Cancer Inst. 2013;105(17):1306-1313. doi:10.1093/jnci/djt175PubMedGoogle ScholarCrossref
34.
Kulkarni  GS, Urbach  DR, Austin  PC, Fleshner  NE, Laupacis  A.  Longer wait times increase overall mortality in patients with bladder cancer.  J Urol. 2009;182(4):1318-1324. doi:10.1016/j.juro.2009.06.041PubMedGoogle ScholarCrossref
35.
Koo  K, Zubkoff  L, Sirovich  BE,  et al.  The burden of cystoscopic bladder cancer surveillance: anxiety, discomfort, and patient preferences for decision making.  Urology. 2017;108:122-128. doi:10.1016/j.urology.2017.07.016PubMedGoogle ScholarCrossref
36.
Pruthi  RS, Baldwin  N, Bhalani  V, Wallen  EM.  Conservative management of low risk superficial bladder tumors.  J Urol. 2008;179(1):87-90. doi:10.1016/j.juro.2007.08.171PubMedGoogle ScholarCrossref
37.
Han  DS, Swanton  A, Lynch  KE,  et al.  Overuse of cystoscopic surveillance among patients with low-risk non–muscle-invasive bladder cancer—a national study of patient, provider, and facility factors.  J Urol. 2018;199(4S):e1018. doi:10.1016/j.juro.2018.02.2570Google Scholar
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Oncology
    September 28, 2018

    早期膀胱癌患者接受癌症复发风险相关监测的程度

    Author Affiliations
    • 1Norris Cotton Cancer Center, Dartmouth Hitchcock Medical Center, Lebanon, New Hampshire
    • 2The Dartmouth Institute for Health Policy and Clinical Practice, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire
    • 3Section of Urology, Dartmouth Hitchcock Medical Center, Lebanon, New Hampshire
    • 4White River Junction VA Medical Center, White River Junction, Vermont
    • 5VA Salt Lake City Health Care System, Salt Lake City, Utah
    • 6University of Utah, Salt Lake City
    JAMA Netw Open. 2018;1(5):e183442. doi:10.1001/jamanetworkopen.2018.3442
    关键点 español English

    问题  早期膀胱癌患者是否在接受与癌症复发风险有关的癌症监测?

    结果  这项全国性队列研究纳入了美国85例退伍军人事务部设施机构治疗的早期膀胱癌患者,按可比频率对其中70个设施机构的低风险和高风险患者进行了监测,2年来,各个机构的膀胱镜检查相差不到1次。在各个设施机构中,这些研究结果体现在中等强度且明显相关的高风险和低风险患者膀胱镜检查频率方面。

    意义  早期膀胱癌风险相关监测没有广泛使用,膀胱癌患者和建议执行风险相关监测的其他肿瘤患者的护理医师,都应当意识到这一点。

    Abstract

    Importance  Cancer care guidelines recommend aligning surveillance frequency with underlying cancer risk, ie, more frequent surveillance for patients at high vs low risk of cancer recurrence.

    Objective  To assess the extent to which such risk-aligned surveillance is practiced within US Department of Veterans Affairs facilities by classifying surveillance patterns for low- vs high-risk patients with early-stage bladder cancer.

    Design, Setting, and Participants  US national retrospective cohort study of a population-based sample of patients diagnosed with low-risk or high-risk early-stage bladder between January 1, 2005, and December 31, 2011, with follow-up through December 31, 2014. Analyses were performed March 2017 to April 2018. The study included all Veterans Affairs facilities (n = 85) where both low- and high-risk patients were treated.

    Exposures  Low-risk vs high-risk cancer status, based on definitions from the European Association of Urology risk stratification guidelines and on data extracted from diagnostic pathology reports via validated natural language processing algorithms.

    Main Outcomes and Measures  Adjusted cystoscopy frequency for low-risk and high-risk patients for each facility, estimated using multilevel modeling.

    Results  The study included 1278 low-risk and 2115 high-risk patients (median [interquartile range] age, 77 [71-82] years; 99% [3368 of 3393] male). Across facilities, the adjusted frequency of surveillance cystoscopy ranged from 3.7 to 6.2 (mean, 4.8) procedures over 2 years per patient for low-risk patients and from 4.6 to 6.0 (mean, 5.4) procedures over 2 years per patient for high-risk patients. In 70 of 85 facilities, surveillance was performed at a comparable frequency for low- and high-risk patients, differing by less than 1 cystoscopy over 2 years. Surveillance frequency among high-risk patients statistically significantly exceeded surveillance among low-risk patients at only 4 facilities. Across all facilities, surveillance frequencies for low- vs high-risk patients were moderately strongly correlated (r = 0.52; P < .001).

    Conclusions and Relevance  Patients with early-stage bladder cancer undergo cystoscopic surveillance at comparable frequencies regardless of risk. This finding highlights the need to understand barriers to risk-aligned surveillance with the goal of making it easier for clinicians to deliver it in routine practice.

    Introduction

    Guidelines for cancer surveillance routinely recommend aligning care with patients’ underlying cancer risk. Such risk-aligned surveillance entails more frequent surveillance for patients at high vs low risk of recurrence. For example, recommendations for computed tomography surveillance after treatment for kidney cancer range from every 3 to 6 months for stage II or III disease to every year for stage I disease.1 Similarly, after treatment for lung and prostate cancer, more frequent surveillance is recommended for patients at higher risk for recurrence.2,3 After treatment of colorectal adenoma, risk-aligned surveillance4 has already become a Medicare quality measure.5 Risk-aligned surveillance is likely to become even more relevant in the future, as the advent of personalized medicine increases the possibilities to predict each patient’s cancer risk.6

    Surveillance after treatment of early-stage bladder cancer is a prime example of risk-aligned surveillance. Bladder cancer is the fourth most prevalent noncutaneous cancer in the United States.7 When cancer is identified, patients typically undergo transurethral resection.8 Based on pathologic findings, three-quarters of patients are diagnosed with early-stage disease and then undergo periodic cystoscopic surveillance with close inspection of the bladder mucosa.9,10 Broad consensus holds that the frequency of cystoscopic surveillance should align with each patient’s cancer risk,11 with risk-aligned surveillance recommended by 8 national and international panels.8,12-18 Specifically, low-risk patients are recommended to receive no more than 3 cystoscopy procedures during the first 2 years after diagnosis; high-risk patients should receive 6 to 8.8,11 Previous studies, including those using Surveillance Epidemiology and End Results Medicare data,19 have demonstrated that bladder cancer surveillance is often not aligned with underlying cancer risk. However, studies were limited to assessing care delivered in a fee-for-service environment and by lack of the longitudinal pathology data needed to accurately assign risk.

    We assessed the extent to which risk-aligned surveillance is practiced across and within Department of Veterans Affairs (VA) facilities by classifying surveillance patterns for low- vs high-risk patients with early-stage bladder cancer. Taking advantage of national data—including longitudinal pathology data—from the largest integrated health care system in the United States,20 we sought to study national patterns of care and simultaneously identify local models of best-practice risk-aligned surveillance.

    Methods

    Our overall goal was to compare the frequency of cystoscopic surveillance among patients with low-risk vs high-risk bladder cancer across and within all facilities that perform bladder cancer surveillance in the VA. We proceeded along the following 4 steps: (1) identification of a cohort of patients with newly diagnosed bladder cancer who underwent surveillance in the VA, (2) assessment of low- or high-risk cancer status, (3) measurement of cystoscopic surveillance, and (4) analyses focused on comparing surveillance frequency among low- and high-risk patients across and within facilities. The study was approved by the Veteran’s Institutional Review Board of Northern New England and the University of Utah Institutional Review Board and follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Patient informed consent was waived because use of the data involved no more than minimum risk to the privacy of patients and the research could not be practically conducted without waiving informed consent.

    Cohort Identification

    We used VA and Medicare administrative data as well as full-text pathology data supplemented by data abstracted by VA tumor registrars, as previously described and validated,21 to identify all patients older than 65 years who were newly diagnosed with bladder cancer between January 1, 2005, and December 31, 2011. The diagnosis date was assigned using a previously validated claims algorithm.21 Next, we excluded those who underwent no cystoscopic surveillance in the VA, those not eligible for cystoscopic surveillance (based on VA or Medicare evidence of cystectomy or radiotherapy within 6 months after diagnosis), and those without pathology data around the diagnosis date (Figure 1).

    Because our focus was on facility-level patterns of care, we excluded patients who were treated in facilities with a substantial proportion (>80%) of missing pathology reports (256 patients from 39 facilities). After excluding patients with incomplete covariate data (n = 16), 6708 patients with complete clinical information remained. Of these, 2417 had ineligible pathology (no urothelial carcinoma or muscle invasion indicating non–early-stage disease) and thus were not eligible for cystoscopic surveillance. Eight hundred sixty patients had uncertain pathology (grade 2 or missing grade or stage, making classification into low-risk or high-risk impossible), leaving 3431 patients with early-stage bladder cancer from 91 facilities (Figure 1).

    Assessment of Risk

    Patients with early-stage bladder cancer can be stratified into low- or high-risk categories based on the pathology at the time of diagnosis. Using the European Association of Urology risk-stratification guidelines,22 we defined low-risk cancer status as a primary low-grade noninvasive urothelial carcinoma and high-risk cancer status as a urothelial carcinoma that was either high-grade noninvasive, invasive into the lamina propria (T1), or associated with carcinoma in situ. We operationalized these definitions using data extracted from full-text pathology reports via validated natural language processing algorithms.23 We used pathology reports dated 90 days before to 90 days after the diagnosis date. Because of our focus on comparing cystoscopic surveillance for low- vs high-risk patients across and within facilities, we excluded 38 patients from 6 facilities where either only low- or high-risk patients were treated. Thus, 1278 low-risk and 2115 high-risk patients from 85 facilities remained for analyses (Figure 1).

    Measuring Cystoscopic Surveillance

    We used Common Procedural Terminology and International Classification of Diseases, Ninth Revision, procedure codes to identify cystoscopy procedures during the follow-up period.21,24 The follow-up period started with the bladder cancer diagnosis date and ended with cancer recurrence, death, date of cystectomy or radiotherapy, date of last VA contact, or end of study (December 31, 2014), whichever occurred first. We followed patients only until they had a cancer recurrence, because a recurrence increases risk for further recurrences and thus changes a patient’s cancer risk status.22 We ascertained cancer recurrences using data extracted from full-text pathology reports via the validated natural language processing algorithms.23 Next, we enumerated the number of cystoscopy procedures each patient underwent during the follow-up period. We only counted those cystoscopy procedures that occurred at least 30 days following a previous procedure,25 because procedure codes occurring in close proximity to each other are unlikely to indicate routine surveillance (eg, cystoscopy with biopsy following shortly after a surveillance cystoscopy) and may sometimes even refer to the same procedure. The majority of patients (86.6% [2937 of 3393]) received cystoscopy procedures at only 1 facility; the remaining patients were assigned to the facility where they received the plurality of their cystoscopy procedures.

    Statistical Analysis

    Data management was performed from December 2015 to February 2017. Data analyses were performed from March 2017 to April 2018. Descriptive statistics were used to describe both patient and facility characteristics. Patient characteristics included age at the time of bladder cancer diagnosis, sex, race, year of diagnosis, and comorbidity using the enhanced Elixhauser index.26 Facility characteristics were obtained from the Veterans Health Administration’s Support Service Center and included number of hospital beds, unique patients per year, urology outpatient visits per year, urologist full-time equivalents, rural location, and academic affiliation.27

    Negative binomial multilevel models were used to calculate the adjusted cystoscopy frequency for each facility. The outcome was the patient-level number of cystoscopy procedures during the follow-up period. Models contained a random intercept for facility. They were adjusted for age (≥80 years) and comorbidity (>3 comorbidities) to account for differences across facilities in age and comorbidity, which might independently inform surveillance frequency. To account for time trends, we adjusted for year of diagnosis. We also included an indicator variable to denote whether patients were followed for longer than 2 years, because guideline recommendations specify that frequency of follow-up can be decreased at that point.18,28 The logarithm of length of follow-up was included as an offset.

    Separate models were fit for low- and high-risk patients. From these models, we then calculated the frequency of cystoscopic surveillance for each facility with corresponding 95% confidence intervals for an average low- or high-risk patient diagnosed in 2011 using empirical Bayes estimation.29 This approach accounted for differences in the reliability of estimated facility-level cystoscopy frequencies due to differences in sample size by shrinking the estimates of facilities with a small number of patients closer to the overall mean.29

    To assess the strength of correlation between cystoscopy frequencies for low- and high-risk patients across facilities, we calculated the Pearson correlation coefficient. We determined which facilities, on average, performed at least 1 cystoscopy more over 2 years for high-risk than for low-risk patients. We also assessed whether surveillance frequency was statistically significantly different within each facility for high- vs low-risk patients. For this, a separate negative binomial model for each facility was fit, with the main exposure being high- vs low-risk cancer status while adjusting for the same covariates.

    Sensitivity Analysis

    We performed 2 sensitivity analyses. First, we performed a simpler analysis to address the potential concern that the multivariable multilevel modeling, adjustment, and shrinkage may have affected our findings.29 In this analysis, we calculated the number of cystoscopies performed per 2 years for each patient. We only included patients who had at least 1 year of follow-up and followed them for up to 2 years after diagnosis. This was done to limit the effect of very short or very long follow-up on the calculated number of cystoscopies per 2 years. Next, we limited this sensitivity analysis to facilities that had at least 10 low-risk and 10 high-risk patients in the data set (618 low-risk and 709 high-risk patients across 33 facilities). For each facility, we calculated the unadjusted mean frequency of cystoscopy for low- and high-risk patients. Strength of correlation was assessed between these frequencies using the Pearson correlation coefficient.

    Second, to determine whether reliance on pathology data extracted via natural language processing affected our results, the main analyses were repeated using data abstracted by tumor registrars to characterize cancer risk rather than the natural language processing results. This sensitivity analysis included 1290 low-risk and 3987 high-risk patients. All analyses were performed using SAS statistical software version 9.4 (SAS Institute Inc) and Stata MP statistical software version 15 (StataCorp LLC). Statistical significance was set at 2-sided α < .05 for all analyses.

    Results

    We included 1278 low-risk and 2115 high-risk patients with a median (interquartile range [IQR]) age of 77 (71-82) years; 99% (3368 of 3393) were male. Patients received care across 85 VA facilities from 45 states, the District of Columbia, and Puerto Rico. Each facility contributed a mean (range) of 40 (3-150) patients. 1237 of 3393 patients (36.5%) were aged 80 years or older, and 670 of 3393 (19.8%) had more than 3 comorbidities (Table 1). Facilities had a median (IQR) of 135 (81-218) hospital beds, few were in rural locations, and most had an academic affiliation (Table 2). The median (IQR) proportion of low-risk patients among the facilities was 37% (27%-50%).

    Patients with low-risk cancer underwent a mean (SD) of 5.3 (3.4) cystoscopy procedures during a median (IQR) follow-up of 2.6 (0.9-4.7) years. As expected, patients with high-risk cancer had shorter follow-up periods due to a higher recurrence rate (median [IQR], 1.2 [0.6-3.7] years) during which they underwent a mean (SD) of 4.7 (3.7) cystoscopy procedures.

    Across all facilities, the adjusted frequency of surveillance cystoscopy ranged from 3.7 to 6.2 (mean, 4.8) procedures over 2 years for low-risk patients (Figure 2A) and from 4.6 to 6.0 (mean, 5.4) procedures over 2 years for high-risk patients (Figure 2B). For both low-risk and high-risk patients, overlap of 95% confidence intervals around adjusted cystoscopy frequencies suggested that a common mean frequency could be representative of most facilities (Figure 2).

    Within facilities, differences in cystoscopy frequency for high- vs low-risk patients were small (mean [range], 0.6 more over 2 years [0.15 fewer to 1.3 more]) (Figure 3). At most facilities (70 of 85), cystoscopy was performed at a similar frequency for high- and low-risk patients, with a difference of less than 1 cystoscopy over 2 years. At the remaining 15 facilities, there appeared to be more of a distinction between high- and low-risk patients, with a difference surpassing 1 cystoscopy over 2 years (range, 1.0-1.3 cystoscopies more over 2 years for high- vs low-risk patients). However, we found a statistically significantly higher cystoscopy frequency for high- vs low-risk patients only within 4 of the 85 facilities. Across all of the 85 facilities, these findings were reflected in a moderately strong correlation of cystoscopy frequencies for high-risk and low-risk patients (r = 0.52; P < .001) (Figure 3).

    We performed sensitivity analyses using unadjusted data to characterize cystoscopy frequency for high- and low-risk patients at each facility. Again, cystoscopy frequency was moderately strongly correlated (r = 0.45; P = .008) and cystoscopy was performed at a similar frequency for high- and low-risk patients at most facilities (28 of the 33 included) (eFigure in Supplement 1). In a second sensitivity analysis, we used data abstracted by tumor registrars instead of full-text pathology data to characterize risk. Results from these analyses were not substantially different from those of the main analyses; thus, only the latter are presented.

    Discussion

    We found that across the national VA integrated health care network, patients with high-risk bladder cancer undergo cystoscopic surveillance at comparable frequency to those with low-risk cancer, with few exceptions. At only 4 of 85 facilities was cystoscopic surveillance frequency significantly higher for high-risk patients than for low-risk patients, and even at these facilities high-risk patients underwent on average only approximately 1 cystoscopy more over the course of 2 years. This difference is much smaller than would be expected based on guideline recommendations, which recommend 6 to 8 cystoscopies over 2 years for high-risk patients and no more than 3 cystoscopies over 2 years for low-risk patients.11 Thus, we were not able to identify facilities that clearly exhibit local models of best practice risk-aligned care, although it may be possible to learn from facilities that appear to distinguish between low- and high-risk patients in their surveillance practices.

    This is the first study, to our knowledge, to examine whether risk-aligned cancer surveillance is performed in a national health system. As 1 of many cancers for which ongoing surveillance is routinely recommended, early-stage bladder cancer serves as a useful paradigm for assessing surveillance practices because of its high prevalence7 and because surveillance with cystoscopy is identifiable using administrative data. Our findings demonstrate that facilities with appropriately high surveillance rates for high-risk cancers also have inappropriately high surveillance rates for low-risk cancers and vice versa. They highlight how challenging it can be to routinely incorporate underlying cancer risk into cancer surveillance practice and suggest non–cancer-related factors may be driving surveillance rates, regardless of underlying risk of recurrence and progression.

    Prior work has examined the use of surveillance colonoscopy among patients diagnosed with colorectal adenoma. Similar to our findings, the frequency of surveillance colonoscopy was not aligned with the risk for progression to advanced lesions.30 Another example of patients with cancer not receiving risk-aligned care is imaging among low- and high-risk patients recently diagnosed with prostate or breast cancer.31,32 Diagnostic imaging is appropriate among high-risk patients and inappropriate among low-risk patients. Similar to our findings regarding risk-aligned cancer surveillance, appropriate and inappropriate imaging rates were closely correlated.31,32 Furthermore, efforts to decrease inappropriate imaging for patients with low-risk prostate cancer have produced the unintended consequence of decreasing appropriate imaging for high-risk patients.33 Taken together, these findings highlight the need to develop strategies that make it easier for physicians to deliver cancer care that is aligned with underlying cancer risk.

    Limitations

    Although the present study is informed by national data and our findings reflect care received across a large number of distinct facilities, it is not without limitations. First, our data are from the VA and thus generalizability to other settings may be limited. However, previous studies of surveillance of early-stage bladder cancer using Surveillance Epidemiology and End Results Medicare data have also found that risk of cancer recurrence had little impact on the care patients receive.19 Second, the number of patients within each facility affected the power to detect any statistically significant differences in surveillance frequency between low-risk and high-risk patients. Thus, we calculated the shrunken surveillance frequency for each facility in our main analyses—an approach that adjusted for the reduced reliability of estimates in smaller facilities.29 Third, we acknowledge certain limitations of ascertaining cancer risk. While we used a validated natural language processing engine to extract information from pathology reports,23 certain characteristics that may contribute to an increased risk of cancer recurrence among low-risk patients (such as multifocality and large tumor size, classified as intermediate risk in the European Association of Urology risk-stratification guidelines22) could not be assessed. Thus, among low-risk patients, there are likely subgroups of patients who are at lower and somewhat higher risk of cancer recurrence. However, as all low-risk patients had low-grade disease, they are all still at substantially lower risk than the high-risk patients. Finally, the analysis examining differences in surveillance frequency between low- and high-risk patients is predicated on recommendations for risk-aligned surveillance that are based on lower level evidence.11 Nevertheless, 8 panels reviewing current evidence have recommended risk-aligned surveillance,8,12-18 which allows for timely detection of progression to muscle-invasive cancer among high-risk patients while sparing low-risk patients unnecessary procedures. Timely detection is important because delays in diagnosis of muscle-invasive cancer are associated with increased mortality.34 Avoiding unnecessary procedures is relevant for patients as they lead to more anxiety, discomfort, and costs.35,36

    Strengths

    These limitations notwithstanding, our study has important strengths. Because we had access to full-text pathology reports, this is the first population-based study, to our knowledge, on early-stage bladder cancer surveillance in which cancer recurrence could be ascertained. Because recurrence elevates patients’ future risk of further recurrence and thus influences further surveillance, we censored follow-up at the time of cancer recurrence, which has not been possible using other population-based data sets. Second, we used multilevel modeling to account for differences in reliability of facility-level estimates due to differences in the number of patients per facility. Third, to address potential concerns that the use of complex modeling or data obtained via natural language processing affected the validity of our findings, we performed sensitivity analyses using unadjusted data and data abstracted by tumor registrars, which supported our main findings.

    Our study is the first to assess risk-aligned bladder cancer surveillance in the VA, which has important implications for health policy and future research. Some may argue that financial incentives in a fee-for-service environment are an important factor contributing to more cystoscopy procedures than recommended. However, the VA is a capitated health care system in which such financial incentives do not exist, and we still find substantial overuse of cystoscopy among low-risk patients. In fact, we recently performed patient-level analyses to examine the extent of overuse and found that the number of cystoscopy procedures among low-risk patients is about double what it should be.37 Thus, other factors are likely important contributors to the lack of risk-aligned surveillance and may include limited clinician knowledge of the guidelines, clinicians’ habit to continue what they have always done, and complexity and variability of risk stratification across different guidelines.11 Future work should focus on understanding the barriers to risk-aligned surveillance and on developing strategies to improve surveillance. These strategies should facilitate providers’ ability to deliver risk-aligned surveillance, thus reducing adverse consequences of overuse of surveillance among low-risk patients and of underuse of surveillance among high-risk patients.

    Conclusions

    In conclusion, our study highlights that risk-aligned surveillance for early-stage bladder cancer is not widely practiced. On the contrary, patients with high- and low-risk cancer undergo surveillance at comparable frequency, despite recommendations that high-risk patients warrant surveillance at least twice as often.11 Our findings should alert those who care for patients with bladder cancer and those who care for patients with other neoplasms for which risk-aligned surveillance is recommended. While risk factors, natural history, and tumor-specific characteristics differ across neoplasms, the challenges clinicians face to align surveillance with underlying cancer risk are likely similar.

    Back to top
    Article Information

    Accepted for Publication: August 12, 2018.

    Published: September 28, 2018. doi:10.1001/jamanetworkopen.2018.3442

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Schroeck FR et al. JAMA Network Open.

    Corresponding Author: Florian R. Schroeck, MD, MS, VA Outcomes Group, WRJ VA Medical Center, 215 N Main St, White River Junction, VT 05009 (florian.r.schroeck@dartmouth.edu).

    Author Contributions: Drs Schroeck and Lynch had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Schroeck, Lynch, Seigne, Goodney, Sirovich.

    Acquisition, analysis, or interpretation of data: Schroeck, Lynch, Chang, Mackenzie, Seigne, Robertson, Sirovich.

    Drafting of the manuscript: Schroeck.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Schroeck, Chang, Mackenzie, Goodney.

    Obtained funding: Schroeck.

    Administrative, technical, or material support: Schroeck, Seigne, Goodney.

    Supervision: Schroeck, Sirovich.

    Conflict of Interest Disclosures: Dr Schroeck is site primary investigator (without compensation) for a clinical trial of Vicinium, sponsored by Eleven Biotherapeutics. Mr Seigne owns 100 common stock of Johnson & Johnson. No other disclosures were reported.

    Funding/Support: This study was supported by the Department of Veterans Affairs (Veterans Health Administration VISN1 Career Development Award to Dr Schroeck and IIR 15-085 to Dr Goodney), the Conquer Cancer Foundation (Career Development Award to Dr Schroeck), the Department of Surgery at the Dartmouth-Hitchcock Medical Center (Dow-Crichlow Award to Dr Schroeck), and the US Food and Drug Administration (U01FD005478 to Dr Goodney).

    Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Disclaimer: Opinions expressed in this manuscript are those of the authors and do not constitute official positions of the US federal government or the Department of Veterans Affairs.

    Data Sharing Statement: See Supplement 2.

    Additional Contributions: This study was supported using resources and facilities at the White River Junction Veterans Affairs (VA) Medical Center, the VA Salt Lake City Health Care System, and the VA Informatics and Computing Infrastructure. Support for VA and Centers for Medicare & Medicaid Services data is provided by the Department of Veterans Affairs, the Veterans Health Administration, the Office of Research and Development, the VA’s Health Services Research and Development Service, and the VA Information Resource Center. These organizations were not compensated for their assistance.

    References
    1.
    The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: kidney cancer. Version 2.2018. https://www.nccn.org/professionals/physician_gls/pdf/kidney.pdf. Published November 30, 2017. Accessed December 4, 2017.
    2.
    The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: non-small cell lung cancer. Version 1.2018. https://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf. Published November 17, 2017. Accessed December 4, 2017.
    3.
    The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: prostate cancer. Version 2.2018. https://www.nccn.org/professionals/physician_gls/pdf/prostate.pdf. Published March 8, 2018. Accessed April 4, 2018.
    4.
    Lieberman  DA, Rex  DK, Winawer  SJ, Giardiello  FM, Johnson  DA, Levin  TR.  Guidelines for colonoscopy surveillance after screening and polypectomy: a consensus update by the US Multi-Society Task Force on Colorectal Cancer.  Gastroenterology. 2012;143(3):844-857. doi:10.1053/j.gastro.2012.06.001PubMedGoogle ScholarCrossref
    5.
    Centers for Medicare & Medicaid Services Quality Payment Program. Merit-based incentive payment system (MIPS) measure #185 & #320. https://qpp.cms.gov/mips/quality-measures. Accessed September 29, 2017.
    6.
    Kattan  MW, Hess  KR, Amin  MB,  et al; members of the AJCC Precision Medicine Core.  American Joint Committee on Cancer acceptance criteria for inclusion of risk models for individualized prognosis in the practice of precision medicine.  CA Cancer J Clin. 2016;66(5):370-374. doi:10.3322/caac.21339PubMedGoogle ScholarCrossref
    7.
    Howlader  N, Noone  AM, Krapcho  M,  et al. Seer Cancer Statistics Review, 1975-2014. Bethesda, MD: National Cancer Institute; 2017. http://seer.cancer.gov/csr/1975_2014/. Accessed March 1, 2018.
    8.
    Chang  SS, Boorjian  SA, Chou  R,  et al. Diagnosis and treatment of non-muscle invasive bladder cancer: AUA/SUO joint guideline. https://www.auanet.org/education/guidelines/non-muscle-invasive-bladder-cancer.cfm. Accessed May 3, 2016.
    9.
    Snyder  C, Harlan  L, Knopf  K, Potosky  A, Kaplan  R.  Patterns of care for the treatment of bladder cancer.  J Urol. 2003;169(5):1697-1701. doi:10.1097/01.ju.0000056727.30546.b7PubMedGoogle ScholarCrossref
    10.
    Holmäng  S.  Follow-up of patients with noninvasive and superficially invasive bladder cancer.  Semin Urol Oncol. 2000;18(4):273-279.PubMedGoogle Scholar
    11.
    Schroeck  FR, Smith  N, Shelton  JB.  Implementing risk-aligned bladder cancer surveillance care.  Urol Oncol. 2018;36(5):257-264. doi:10.1016/j.urolonc.2017.12.016PubMedGoogle ScholarCrossref
    12.
    The National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: bladder cancer. Version 1.2017. https://www.nccn.org/professionals/physician_gls/pdf/bladder.pdf. Published December 21, 2016. Accessed January 29, 2017.
    13.
    Nieder  AM, Brausi  M, Lamm  D,  et al.  Management of stage T1 tumors of the bladder: International Consensus Panel.  Urology. 2005;66(6)(suppl 1):108-125. doi:10.1016/j.urology.2005.08.066PubMedGoogle ScholarCrossref
    14.
    Oosterlinck  W, van der Meijden  A, Sylvester  R,  et al; European Association of Urology. Guidelines on TaT1 (non-muscle invasive) bladder cancer. http://uroweb.org/wp-content/uploads/EAU-Guidelines-TaT1-Bladder-Cancer-2006.pdf. Updated March 1, 2006. Accessed June 15, 2017.
    15.
    Brausi  M, Witjes  JA, Lamm  D,  et al.  A review of current guidelines and best practice recommendations for the management of nonmuscle invasive bladder cancer by the International Bladder Cancer Group.  J Urol. 2011;186(6):2158-2167. doi:10.1016/j.juro.2011.07.076PubMedGoogle ScholarCrossref
    16.
    Kassouf  W, Traboulsi  SL, Kulkarni  GS,  et al.  CUA guidelines on the management of non-muscle invasive bladder cancer.  Can Urol Assoc J. 2015;9(9-10):E690-E704. doi:10.5489/cuaj.3320PubMedGoogle ScholarCrossref
    17.
    National Institute for Health and Care Excellence. Bladder cancer: diagnosis and management. NICE guideline NG2. https://www.nice.org.uk/guidance/ng2/evidence. Published February 25, 2015. Accessed July 24, 2017.
    18.
    Kassouf  W, Traboulsi  SL, Schmitz-Dräger  B,  et al.  Follow-up in non–muscle-invasive bladder cancer—International Bladder Cancer Network recommendations.  Urol Oncol. 2016;34(10):460-468. doi:10.1016/j.urolonc.2016.05.028PubMedGoogle ScholarCrossref
    19.
    Hollingsworth  JM, Zhang  YS, Miller  DC,  et al.  Identifying better practices for early-stage bladder cancer.  Med Care. 2011;49(12):1112-1117. doi:10.1097/MLR.0b013e3182353bafPubMedGoogle ScholarCrossref
    20.
    Shulkin  DJ.  Beyond the VA crisis—becoming a high-performance network.  N Engl J Med. 2016;374(11):1003-1005. doi:10.1056/NEJMp1600307PubMedGoogle ScholarCrossref
    21.
    Schroeck  FR, Sirovich  B, Seigne  JD, Robertson  DJ, Goodney  PP.  Assembling and validating data from multiple sources to study care for veterans with bladder cancer.  BMC Urol. 2017;17(1):78. doi:10.1186/s12894-017-0271-xPubMedGoogle ScholarCrossref
    22.
    Babjuk  M, Boehle  A, Burger  M,  et al.  EAU guidelines on non–muscle-invasive urothelial carcinoma of the bladder: update 2016.  Eur Urol. 2017;71(3):447-461. doi:10.1016/j.eururo.2016.05.041PubMedGoogle ScholarCrossref
    23.
    Schroeck  FR, Patterson  OV, Alba  PR,  et al.  Development of a natural language processing engine to generate bladder cancer pathology data for health services research.  Urology. 2017;110:84-91. doi:10.1016/j.urology.2017.07.056PubMedGoogle ScholarCrossref
    24.
    Hollenbeck  BK, Ye  Z, Dunn  RL, Montie  JE, Birkmeyer  JD.  Provider treatment intensity and outcomes for patients with early-stage bladder cancer.  J Natl Cancer Inst. 2009;101(8):571-580. doi:10.1093/jnci/djp039PubMedGoogle ScholarCrossref
    25.
    Klabunde  CN, Potosky  AL, Legler  JM, Warren  JL.  Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53(12):1258-1267. doi:10.1016/S0895-4356(00)00256-0PubMedGoogle ScholarCrossref
    26.
    Quan  H, Sundararajan  V, Halfon  P,  et al.  Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.  Med Care. 2005;43(11):1130-1139. doi:10.1097/01.mlr.0000182534.19832.83PubMedGoogle ScholarCrossref
    27.
    Schroeck  FR, Pattison  EA, Denhalter  DW,  et al.  Early stage bladder cancer: do pathology reports tell us what we need to know?  Urology. 2016;98:58-63. doi:10.1016/j.urology.2016.07.040PubMedGoogle ScholarCrossref
    28.
    Schroeck  FR, Montie  JE, Hollenbeck  BK.  Surveillance strategies for non–muscle invasive bladder cancer.  AUA Update Ser. 2012;31:313-323.Google Scholar
    29.
    MacKenzie  TA, Grunkemeier  GL, Grunwald  GK,  et al.  A primer on using shrinkage to compare in-hospital mortality between centers.  Ann Thorac Surg. 2015;99(3):757-761. doi:10.1016/j.athoracsur.2014.11.039PubMedGoogle ScholarCrossref
    30.
    Schoen  RE, Pinsky  PF, Weissfeld  JL,  et al.  Utilization of surveillance colonoscopy in community practice.  Gastroenterology. 2010;138(1):73-81. doi:10.1053/j.gastro.2009.09.062PubMedGoogle ScholarCrossref
    31.
    Makarov  DV, Desai  R, Yu  JB,  et al.  Appropriate and inappropriate imaging rates for prostate cancer go hand in hand by region, as if set by thermostat.  Health Aff (Millwood). 2012;31(4):730-740. doi:10.1377/hlthaff.2011.0336PubMedGoogle ScholarCrossref
    32.
    Makarov  DV, Soulos  PR, Gold  HT,  et al.  Regional-level correlations in inappropriate imaging rates for prostate and breast cancers: potential implications for the choosing wisely campaign.  JAMA Oncol. 2015;1(2):185-194. doi:10.1001/jamaoncol.2015.37PubMedGoogle ScholarCrossref
    33.
    Makarov  DV, Loeb  S, Ulmert  D, Drevin  L, Lambe  M, Stattin  P.  Prostate cancer imaging trends after a nationwide effort to discourage inappropriate prostate cancer imaging.  J Natl Cancer Inst. 2013;105(17):1306-1313. doi:10.1093/jnci/djt175PubMedGoogle ScholarCrossref
    34.
    Kulkarni  GS, Urbach  DR, Austin  PC, Fleshner  NE, Laupacis  A.  Longer wait times increase overall mortality in patients with bladder cancer.  J Urol. 2009;182(4):1318-1324. doi:10.1016/j.juro.2009.06.041PubMedGoogle ScholarCrossref
    35.
    Koo  K, Zubkoff  L, Sirovich  BE,  et al.  The burden of cystoscopic bladder cancer surveillance: anxiety, discomfort, and patient preferences for decision making.  Urology. 2017;108:122-128. doi:10.1016/j.urology.2017.07.016PubMedGoogle ScholarCrossref
    36.
    Pruthi  RS, Baldwin  N, Bhalani  V, Wallen  EM.  Conservative management of low risk superficial bladder tumors.  J Urol. 2008;179(1):87-90. doi:10.1016/j.juro.2007.08.171PubMedGoogle ScholarCrossref
    37.
    Han  DS, Swanton  A, Lynch  KE,  et al.  Overuse of cystoscopic surveillance among patients with low-risk non–muscle-invasive bladder cancer—a national study of patient, provider, and facility factors.  J Urol. 2018;199(4S):e1018. doi:10.1016/j.juro.2018.02.2570Google Scholar
    ×