Professional Documents
Culture Documents
Bju 14047
Bju 14047
Corresponding Author:
Funding Sources: BJ is supported by a Big Data to Knowledge training grant under the
supervision of Drs. Kosorok and Ferris. AS is supported by an AHRQ K08 grant.
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1111/bju.14047
This article is protected by copyright. All rights reserved.
Abstract
Accepted Article
Objectives: To identify HRQOL changes before to after diagnosis in older adults with bladder
cancer and to compare their changes to non-cancer controls.
Patients and Methods: Data from the Surveillance, Epidemiology, and End Results registries
were linked with Medicare Health Outcomes Survey (MHOS) data. Medicare beneficiaries >= 65
years from 1998-2013 who were diagnosed with bladder cancer between baseline and follow-
up MHOS were matched with non-cancer controls using propensity scores. Linear mixed models
were used to estimate predictors of HRQOL changes.
Results: After matching, 535 bladder cancer patients (458 non-invasive and 77 invasive) and
2770 non-cancer controls were identified. Both non-invasive and invasive cases (respectively)
reported significant declines in HRQOL over time when compared to controls: Physical
Component Summary (-2 and -5.3 vs. -0.4), Bodily Pain (-1.9 and -3.6 vs. -0.7), Role Physical (-
2.7 and -4.7 vs. -0.7), General Health (-2.4 and -6.1 vs. 0), Vitality (-1.2 and -3.5 vs. -0.1) and
Social Functioning (-2.1 and -5.7 vs. -0.8). All scores range from 0 to 100. When stratified by
time since diagnosis, HRQOL improved over 1 year for some domains (Role Physical) but
remained lower across most domains.
Introduction
Bladder cancer is the 5th most common cancer, with 77,000 new cases in 2016, and an
estimated 587,426 living with the disease in the United States.1 Invasive disease requires radical
surgical treatment or chemo-radiation with intensive surveillance. Though less severe, non-
invasive disease requires long-term surveillance with periodic cystoscopy and intravesical
therapy. Intense treatment and surveillance may negatively impact health-related quality of life
(HRQOL).
To date, bladder cancer HRQOL research has been limited, with retrospective studies
investigating treatment differences within cohorts of patients with non-invasive and invasive
disease. Among patients with muscle invasive bladder cancer (MIBC) undergoing cystectomy,
HRQOL has been compared between diversion types2,3 while HRQOL for non-muscle invasive
bladder cancer (NMIBC) patients has been studied between various intravesical treatments.4 A
limitation in the current understanding of HRQOL among patients with bladder cancer is how
patients’ HRQOL changes following diagnosis. This is critical for patients with bladder cancer
given that it is prevalent in older adults that face other threats (i.e. comorbidities) to HRQOL. A
prior study investigating the impact of cancer diagnosis on HRQOL in several cancers reported
To our knowledge, no prospective study has investigated the impact of bladder cancer
diagnosis on HRQOL, using health status assessments before cancer diagnosis and assessing the
impact of time since diagnosis, treatment type, and stage. The objective of this study was to
compare prospective HRQOL changes for a group of patients diagnosed with bladder cancer
with changes among matched non-cancer control individuals drawn from the same population.
We hypothesized that bladder cancer patients, regardless of disease stage or treatment, would
self-report HRQOL declines, most pronounced among those with invasive disease receiving
aggressive treatment.
Study participants
This study included new incident bladder cancer patients 65 years and older who were enrolled
in Medicare Advantage managed care plans. Survey data from the Centers for Medicare and
Medicaid Services’ (CMS’) Medicare Health Outcomes Survey (MHOS) were linked with the
Surveillance, Epidemiology, and End Results (SEER) cancer registries provided by the National
Cancer Institute. The MHOS is an annual survey administered to a random sample of 1000-
1,200 Medicare beneficiaries from each managed care plan.6 Each participant is asked to
complete a baseline survey and a follow-up survey 2 years later if they are enrolled in the same
plan. Cancer-specific information is collected on patients with newly diagnosed cancer within
specified geographic regions through the SEER program. The SEER-MHOS linkage is described
elsewhere.7 We obtained Institutional Review Board permission for secondary data analysis
from the University of North Carolina at Chapel Hill.
Our study included 14 MHOS cohorts with baseline assessments from 1998-2011 and follow-up
assessments from 2000-2013.8 We identified 535 individuals who completed both baseline and
follow-up MHOS and were diagnosed with bladder cancer between assessments. To
prospectively evaluate the impact of a bladder cancer diagnosis on HRQOL, we selected non-
cancer controls who responded to both baseline and follow-up MHOS, resided in the same SEER
region, and participated in the same managed care plans as the cancer patients. Using
propensity scores, we matched 5 controls to each case to balance the proportion of patient and
survey characteristics between the groups. Propensity score matching was performed using
baseline age, race, education, region, marital status, proxy status, assessment mode, smoking
status, high blood pressure, heart disease, stroke, COPD, inflammatory bowel disease, arthritis,
sciatica, diabetes, and depression.9 This resulted in selection of 2,770 controls matched to 535
bladder cancer patients.
Data Collection
The MHOS provides self-reported data on demographics (age, gender, race, smoking status,
marital status), survey characteristics (self-report, proxy status, mail or telephone report),
Statistical Analysis
Analyses were conducted in R version 3.3.0.13 Numeric variables were scaled14 using sample
estimates. Bivariate associations were evaluated using t-test for continuous variables and chi-
square for categorical variables. Using HRQOL before and after diagnosis for each subject as
longitudinal outcomes necessitated a statistical approach accounting for within-subject
correlation and heterogeneity amongst subjects. Hence, the linear mixed model15,16 for
longitudinal data was employed using restricted maximum likelihood estimation.17 Variable
interactions with time allowed estimation of both synchronous and lagged effects18 (e.g. the
effect of comorbidities at baseline on HRQOL at follow-up). Model selection was facilitated by
information criteria19 and several measures of multivariate association.20 Least-squares mean21
estimates facilitated group-wise comparison of HRQOL domains. Confounding effects were
minimized by propensity matching using a five-to-one control to case ratio based on patient
demographics and preexisting conditions. Sensitivity analyses were conducted by separately
fitting a model using the full data (i.e. without propensity matching) and comparing conclusions
drawn. Results from the full data model were in complete agreement with inferences made by
the propensity score model (Supplemental Table 1 and Figures 2-4). Additional logistic
Study Sample
Patients were stratified by non-invasive (n=458) and invasive (n=77) disease and compared to
matched controls (n=2,770). After propensity score matching, no statistically significant
differences between cases and controls were observed based on demographic characteristics,
survey characteristics, or pre-existing comorbidities (Table 1). Mean time (and standard
deviation) between surveys was 2.7 years (1.9) for cases and 2.1 years (0.1) for matched
controls.
Physical HRQOL was evaluated by treatment and time since diagnosis. Patients undergoing
cystectomy reported significant HRQOL decrements across nearly all physical domains,
including PCS, Physical Functioning, Role Physical and General Health (Figure 2). Role Physical
scores improved over time whereas Physical Functioning deteriorated (Figure 3). PCS and
General Health also appeared to decline after diagnosis and did not return to baseline.
Predictors of a significant decrease in PCS included living in the South (OR 2.6; p=0.02) and a
diagnosis of recent depression (OR 4.2; p<0.0001) (Table 3). Patients with hypertension and
atherosclerosis appeared to be less likely to experience significant decreases in PCS.
Predictors of a significant decrease in MCS included recent depression (OR 2.8; p=0.002), being
a former or never smoker (OR 2.9, p=0.009; OR 2.2, p=0.02) (Table 3). Patients with distant
depression (OR 0.2; p=0.001) or chronic sciatica (OR 0.59; p=0.04) appeared to be less likely to
experience significant decreases in MCS.
Discussion
Until recently, our understanding of HRQOL in bladder cancer patients has been limited. A
systematic review evaluating HRQOL in patients with bladder cancer concluded that most
studies were retrospective, cross-sectional, and with methodologic limitations such as lack of
pre-diagnosis data, insufficient follow-up duration, and small sample size.22 While specific
HRQOL differences between urinary diversion were evaluated, differences by time since
diagnosis and management strategies were not considered.23 Since publication of the
systematic review, additional studies have emerged, which shed light on general and cancer-
specific HRQOL among bladder cancer patients.5,24 Notably, HRQOL of bladder cancer patients
was previously evaluated using SEER-MHOS, but in a limited capacity.5 These studies
investigated 9-12 cancer sites (including bladder) and compared cancer patients with non-
cancer controls, with limited detail regarding specific cancer types.5,25 Results were similar in
that physical health status decreased more than mental health across all cancer domains;5 and
physical health status following diagnosis was worse among cancer patients.25 In another study
which evaluated bladder cancer specifically in the SEER-MHOS, differences in PCS and MCS
scores were investigated comparing two cohorts—patients who completed a MHOS before
diagnosis and individuals who completed the survey after diagnosis.24 Although significant
declines in PCS and MCS were observed, a limitation of this cross-sectional study was potential
confounding through comparison of two independent cohorts (pre-diagnosis and post-
diagnosis).24
Other prospective studies have investigated HRQOL differences between bladder cancer cases
and representative community samples of non-cancer controls. Among 823 cases and 2,037
controls, patients with NMIBC and MIBC disease reported similar worse functioning than
controls across all domains using the EORTC QLQ-C30 cancer-specific instrument.27 This is in
contrast to our findings, which revealed significant differences between non-invasive and
invasive disease for physical function, with non-invasive patients having a 2-point decrement
and invasive a 5-point decrement between baseline and follow-up. Differences between our
findings may be due to similar of treatment strategies in the EORTC QLQ-C30 study, with the
majority of patients in that study undergoing cystectomy. In our study, non-invasive patients
were more likely to undergo non-cystectomy treatment, which likely influence HRQOL.
Aside from disease stage and treatments, HRQOL can be impacted by time since diagnosis given
the potential for adaptation and coping. A comparison between 259 patients who underwent
cystectomy or non-cystectomy treatment revealed no differences in long-term (>2 year) HRQOL
using the cancer and disease-specific FACT-Bl questionnaire.29 Given that our study evaluated
short-term HRQOL (within 2 years of diagnosis), the impact of diagnosis and treatment on
HRQOL may differ if follow-up was extended. The majority of domains did not differ based on
time since diagnosis, although Role Physical appeared to improve by 1 year. This domain
involves role limitations due to physical health such as limiting time at work, difficulty
performing work, or accomplishing less than you would like. Recovery may be prolonged after
surgery (or intravesical treatments/intense surveillance), and therefore improvements to
baseline may not occur until one year post-diagnosis in this specific subdomain. To further
investigate longitudinal trends, future research should investigate long-term HRQOL in bladder
cancer patients. Finally, an additional consideration is the presence or absence of prior cancer
diagnoses. A subset of patients with non-invasive disease had an additional cancer diagnosis,
which may impact their HRQOL when compared to new cancer patients. Although this subset of
patients was too small to compare, this consideration should be acknowledged.
While our study represents the first comprehensive evaluation of HRQOL from before bladder
cancer diagnosis to after treatment, some limitations should be noted. First, our cohort was
limited to cancer patients who were enrolled in Medicare Advantage managed care plans.
MHOS participants may not be representative of all older adults residing in SEER areas. While
some studies indicate that managed care enrollees are healthier than fee-for-service
beneficiaries,30,31 another study indicated similarities between the groups.32 Second, the sample
size was relatively small among bladder cancer patients, which may reduce our ability to detect
statistically significant HRQOL differences between treatment groups. Third, different HRQOL
instruments were used during the study period (SF-36 from 1998-2005 and VR-12 from 2006-
2013). Furthermore, these data do not capture cancer patients or controls who dis-enroll from
their plan or die before follow-up HRQOL survey.33 HRQOL instruments were not cancer-specific
and may lack sensitivity with respect to bladder cancer specific treatment effects. Furthermore,
limited health literacy may be a potential barrier to collecting HRQOL from all patients. Fourth,
Our current understanding of HRQOL among bladder cancer patients has several implications
for clinical practice. First, clinicians should inform patients who receive treatment that their
short-term HRQOL in several domains is likely to decrease, even among those undergoing non-
cystectomy procedures. HRQOL decrements are expected for patients with NMIBC and MIBC.
Setting appropriate expectations is critical to improving patient satisfaction, which has been
shown to be linked in a variety of conditions including joint replacement.34-36 Second, shared
decision making should incorporate an honest discussion regarding the impact of HRQOL when
comparing non-cystectomy treatment with cystectomy. A large study investigating patient
participation in shared decision-making showed that the majority of patients wished to engage
directly if quality of life was incorporated into the decision-making process.37 Multidisciplinary
clinics should provide all possible options in an unbiased and straightforward manner. Finally,
understanding that physical HRQOL is most impacted by the diagnosis and treatment of bladder
cancer irrespective of stage informs future interventions wishing to address and optimize
physical health in bladder cancer patients. Physical health decrements may be addressed
through judicious use of physical therapy and other supportive care interventions, even among
NMIBC patients.
In conclusion, our study is the first to report treatment and stage-specific HRQOL changes by
bladder cancer stage and treatment. Our results provide valuable information for researchers
interested in future interventions aimed at improving HRQOL in older adults with bladder
cancer.
Table 1: Propensity score adjusted bladder cancer cases and controls in SEER-MHOS cohort
Table 2: Adjusted HRQOL measures of baseline and follow-up of non-cancer controls and
bladder cancer patients
Table 3: Multivariable model with predictors of meaningful decrease in MCS or PCS among
patients with bladder cancer
Figure 1: Adjusted HRQOL changes from baseline to follow-up of controls and bladder cancer
patients stratified by stage
Figure 2: Adjusted HRQOL changes from baseline to follow-up of controls and bladder cancer
patients stratified by treatment
Supplemental Figure 1: Adjusted HRQOL changes from baseline to follow-up of controls and
bladder cancer patients stratified by time since diagnosis (3 month increments)
Supplemental Figure 2: Adjusted HRQOL changes from baseline to follow-up of controls and
bladder cancer patients stratified by stage – full, non-propensity matched model
Supplemental Figure 3: Adjusted HRQOL changes from baseline to follow-up of controls and
bladder cancer patients stratified by treatment- full, non-propensity matched model
Supplemental Figure 4: Adjusted HRQOL changes from baseline to follow-up of controls and
bladder cancer patients stratified by time since diagnosis- full, non-propensity matched model
REFERENCES
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA: a cancer journal for clinicians
2016;66:7-30.
2. Yang LS, Shan BL, Shan LL, et al. A systematic review and meta-analysis of quality of life
outcomes after radical cystectomy for bladder cancer. Surgical oncology 2016;25:281-97.
3. Porter MP, Penson DF. Health related quality of life after radical cystectomy and urinary
diversion for bladder cancer: a systematic review and critical analysis of the literature. J Urol
2005;173:1318-22.
4. Schmidt S, Frances A, Lorente Garin JA, et al. Quality of life in patients with non-muscle-
invasive bladder cancer: one-year results of a multicentre prospective cohort study. Urol Oncol
2015;33:19 e7-5.
5. Reeve BB, Potosky AL, Smith AW, et al. Impact of cancer on health-related quality of life
of older Americans. J Natl Cancer Inst 2009;101:860-8.
6. Jones N, 3rd, Jones SL, Miller NA. The Medicare Health Outcomes Survey program:
overview, context, and near-term prospects. Health Qual Life Outcomes 2004;2:33.
7. Ambs A, Warren JL, Bellizzi KM, Topor M, Haffer SC, Clauser SB. Overview of the SEER--
Medicare Health Outcomes Survey linked dataset. Health Care Financ Rev 2008;29:5-21.
8. Healthcare Delivery Research Program- SEER- MHOS Data File Response Rates.
(Accessed December 23, 2016, at https://healthcaredelivery.cancer.gov/seer-
mhos/aboutdata/table.response.rates.html.)
9. Reeve BB, Smith AW, Arora NK, Hays RD. Reducing bias in cancer research: application
of propensity score matching. Health Care Financ Rev 2008;29:69-80.
10. Ware JE, Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I.
Conceptual framework and item selection. Medical care 1992;30:473-83.
Table 2: Adjusted HRQOL measures of baseline and follow-up of non-cancer controls and
bladder cancer patients
Southern
--- 2.651 (1.152, 6.267) 0.0235
region
Graduate
--- 0.574 (0.297, 1.098) 0.0955
Education
0.0022 (MCS)
Recent < 0.0001 (PCS)
2.768 (1.461, 5.394) 4.188 (2.136, 8.618)
Depression
Chronic
1.759 (0.963, 3.253) --- 0.0682
Depression
Distant
0.210 (0.073, 0.516) --- 0.0015
Depression
Former
2.900 (1.322, 6.645) --- 0.0094
Smoker
Chronic
0.588 (0.348, 0.983) --- 0.0446
Sciatica