European Journal of Cancer 95 (2018) 52e58

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.ejcancer.com

Original Research

Bladder cancer survival: Women better off in the long run

Bettina Kulle Andreassen a,*, Tom Kristian Grimsrud a,

Erik Skaaheim Haug b,c

a
Department of Research, Cancer Registry of Norway, PB 5313, Majorstuen, 0304, Oslo, Norway
b
Department of Urology, Vestfold Hospital Trust, Tønsberg, Norway
c
Institute of Cancer Genetics and Informatics, Oslo University Hospital, Oslo, Norway

Received 8 February 2018; received in revised form 7 March 2018; accepted 8 March 2018
Available online 7 April 2018

KEYWORDS Abstract Aim: Mortality among patients with bladder cancer is usually reported to be higher
Bladder cancer; for women than men, but how the risk differs and why remain largely unexplained. We also
Gender; described gender-specific differences in survival for patients with bladder cancer and estimated
Gender difference; to what extent they can be explained by differences in T-stage distribution at the first diag-
Prognosis; nosis.
Survival; Methods: The present study comprised all 15,129 new cases of histologically verified invasive
Urothelial cancer and non-invasive urothelial carcinoma of the urinary bladder diagnosed between 1997 and
2011 as registered in the Cancer Registry of Norway. Gender-specific excess mortality risk
rates and risk ratios were calculated based on a flexible parametric relative survival model ad-
justing for T-stage and age, allowing the effect of gender to vary over time. We also present
gender-specific relative survival curves for different T-stage patterns adjusted for age.
Results: Risk rates were significantly higher for women than men up to 2 years after bladder
cancer diagnosis, particularly for muscle-invasive cancers. Thereafter, risk rates appeared to
be higher in men. Adverse T-Stage distribution in women explained half of the unfavourable
survival difference in female patients 2 years after diagnosis.
Conclusion: The common view of worse bladder cancer prognosis in women than in men needs
to be revised. Norwegian women have a less favourable prognosis solely within the first 2 years
after diagnosis, particularly when diagnosed with a muscle-invasive tumour; parts of this
discrepancy can be attributed to more severe initial diagnoses in women.
ª 2018 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

* Corresponding author.
E-mail address: b.k.andreassen@kreftregisteret.no (B.K. Andreassen).

https://doi.org/10.1016/j.ejca.2018.03.001
0959-8049/ª 2018 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58
1. Introduction
Gender differences in cancer survival have been exam-
ined in many studies [1e6], and for most cancer types,
survival is found to be better for women than for men.
Possible explanations for this phenomenon were gender-
specific biological differences [2,3] and differences in
environmental exposures and lifestyle [2], as well as
tumour biology, gender hormones and clinical man-
agement [4]. In bladder cancer, however, most studies
and literature reviews have indicated a worse survival
for female patients [7e16].
Recently, several reviews have summarised and
interpreted results from published studies on the gender
difference in prognosis of bladder cancer [12,17e20]. It
is suspected that multiple factors influence the gender-
specific prognostic outcome [17]. Different anatomy,
delays from the first symptoms to diagnosis in females
[18,21], variations in hormone receptors as well as
tumour biology and less optimal treatment for women
[20] were suggested to play a role in the observed dif-
ferences. However, even after adjustment for some of
these factors, the gender difference in survival remained.
Although there is a widespread perception of a
disadvantageous prognosis of female patients with
bladder cancer, well-performed analyses of evident key
factors are lacking. To our knowledge, studies focussing
on gender differences so far did not properly describe
how gender-specific risk profiles differ over time. There
are also methodological challenges when describing the
risk difference in survival between male and female pa-
tients with bladder cancer. In many publications, the
gender difference is described by the hazard ratio (men
versus women) assuming that the hazard ratio is con-
stant throughout follow-up time (proportional hazards
assumption).
In this article, we will use data from the Cancer
Registry of Norway (CRN) to describe gender-specific
differences in survival for patients with bladder cancer
and evaluate how the risk ratio between male and female
patients with bladder cancer varies over time since
diagnosis. We will also estimate to what extent the
survival difference between men and women can be
explained, by gender differences in T-stage at the first
diagnosis.
2. Material and methods
2.1. Patients
Since 1953, the CRN has, compulsory by law, registered
virtually all new cancer diagnoses in Norway. The reg-
istry receives notifications from three independent
sources (clinicians, pathology laboratories and from the
Cause of Death Registry). Comparisons with data on
diagnoses of hospital discharge and consultations,
53
recorded in The Norwegian Patient Registry, are done
for quality control and to ensure completeness [22].
Patients are identified through the unique national per-
sonal identification number assigned to all newborns
and residents in Norway since 1960.
The present study comprises all new cases of histolog-
ically verified invasive and non-invasive urothelial carci-
noma of the urinary bladder including papillary tumours
and carcinoma in situ, excluding dysplasia, diagnosed
between 1997 and 2011 and registered in the CRN. In
total, 15,129 patients with bladder cancer were included in
this study. Participants were followed up until death,
migration or end of follow up on the 19th of April 2017,
whichever came first. The total follow-up time was 101,518
person-years with a median follow-up time of 12 years.
Morphology, clinical T-, N- and M-categories and
grade were defined, based on the most severe diagnosis
within a 4-month window from the initial bladder cancer
diagnosis. We grouped the patients based on the avail-
able clinical T-, N- and M-categories, morphology and
grade information: low- (TaLG) and high-grade (TaHG)
non-invasive papillary carcinoma (low-grade: TaG1G2;
high-grade: TaG3; World Health Organisation [WHO]
1973 [23]), non-invasive flat carcinoma (Tis), invasive
carcinoma (in the connecting tissue, not muscle) (T1) as
nonemuscle-invasive carcinoma (NMIBC) and muscle-
invasive carcinoma (T2-4) (MIBC). Therefore, strictly
speaking, not all patients with grade II (WHO 1973)
would be low-grade patients after the current definition
(WHO 2004).
2.2. Statistical analysis
The CRN internal coding system is not able to distin-
guish between T1 and MIBC tumours without addi-
tional information, which is not always available.
Therefore, information on clinical T-stage was missing
in 16% (2,369 of 15,129) of the patients in our study.
This incompleteness could not be assumed to be missing
completely at random because the oldest patients
(age > 80) had a significantly higher proportion of T-
stage missing (20%) than younger patients (age 65e79,
16%; age 50e64, 11%; age < 50, 10%; p < 0.001).
Therefore, additional information about morphology,
metastases, lymph node status, survival time, status
(death or alive), gender and age was used in an ordered
logistic regression to predict T-stage for patients with
missing values. This was performed by using multiple
imputation (‘mi’ function in STATA) to generate 10
imputed datasets [24,25]. Finally, the method of Rubin
[26,27] was used to achieve combined point estimates
and confidence intervals from the results of these
imputed datasets in all analyses depending on T-stage.
Relative survival compares the observed survival in
the bladder cancer patient group to the expected sur-
vival of the general Norwegian population. Relative
survival was estimated using the age-standardised
54 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58

Ederer II method applying national population life ta-

bles by gender, age and year of diagnosis [28] and
compared with standard KaplaneMeier estimates [29].
Further on, relative survival models were used, and
therefore, no cause of death information had to be
included in the calculations.
Risk rates (excess mortality risk rates) and risk ratios
(excess mortality risk ratios) were calculated based on
flexible parametric relative survival models [30,31] where
gender, T-stage and age were included as categorical
variables and the gender effect on survival was modelled
as time-dependent covariate with 3 degrees of freedom
(df). The baseline hazard was modelled using 5 df for the
spline variables using the Stata command ‘stpm2’ [32].
Age was modelled with splines (3 df). The interpretation
of excess mortality risk rates and excess mortality risk
ratios in relative survival models is similar to the inter-
pretation of well-known hazard rates and hazard ratios
in cause-specific survival models.
The ‘meansurv’ function in STATA was used to
calculate gender-specific relative survival curves for
different T-stage patterns. The percentage explained by T-
stage was evaluated by dividing the relative survival
improvement for women (T-stage adjusted) by the sur-
vival difference between men and women. These pre-
dictions were estimated for 73-year old patients. However,
gender differences were largely unaffected by age (results
not shown).
All statistical analyses were performed in STATA Fig. 1. KaplaneMeier (KM, dashed lines) and relative survival
[33]. (RS, solid lines) rates for men (black) and women (grey) with
bladder cancer diagnosis (A). Risk rates (excess mortality rates)
3. Results including confidence intervals for men (black) and women (grey)
diagnosed with bladder cancer (B).
Overall, Norwegian men have a better prognosis than
women after a bladder cancer diagnosis (Fig. 1A). This 0.99e1.36) within the follow-up timeframe from 2 to 10
applied both when relying on cause of death information years. Therefore, male patients had a 21% significantly
(KaplaneMeier-curves) and when using a relative survival lower risk of death from their cancer than women within
approach. When calculating the risk ratio (ratio between the first 2 years of follow up and a 16% higher risk of death
the risk rates for men versus women: excess mortality rate when considering the timeframe 2e10 years since diag-
ratio) in a relative survival model by assuming a constant nosis. Moreover, risk rates were significantly higher for
risk ratio over time, we evaluated an adjusted (T-stage, women than men within the first 2 years after diagnosis
age) risk ratio of 0.85 (confidence interval [CI]: 0.78e0.92) (Fig. 1B). After this time point, the risk rates were slightly
(Table 1). Therefore, male patients had a 15% significantly higher in men than in women. The time-dependent risk
lower risk to die from their cancer than women throughout ratio presented in Fig. 2 further illustrates how the risk
the first 10 years of follow-up time. By allowing discrete ratio varied over time. We revealed that the relationship
risk ratios for the follow-up timeframes 0e2 and 2e10 between the gender-specific risk rates of bladder cancer-
years, we found that the risk ratio is 0.79 (CI: 0.71e0.86) erelated death was most unfavourable for women
within the first 2 years after diagnosis and 1.16 (CI: compared with men at the time of diagnosis. Thereafter,

Table 1
Risk ratios (excess mortality rate ratios) and corresponding confidence intervals (CIs) for bladder cancererelated death for male compared with
female patients dependent on different timeframes since bladder cancer diagnosis. Both unadjusted and adjusted (T-stage, age) risk ratios are
presented. Significance against the hypothesis of equal risk rates for both genders is stated by*.
Follow-up period 0e10 years 0e2 years 2e10 years
Risk ratio (M/W) 0.80* (0.73e0.88) 0.71* (0.64e0.79) 1.15 (0.96e1.39)
Risk ratio (M/W) adjusted 0.85* (0.78e0.92) 0.79* (0.71e0.86) 1.16 (0.99e1.36)
M, men; W, women.
 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58 55

life expectancy across genders were taken into account, by

Fig. 2. Risk ratio (excess mortality rate ratio) including confidence
intervals for men versus women with bladder cancer diagnosis. The
blue/red-shaded area indicates the timeframe after diagnosis where
men/women have a lower risk of bladder cancererelated death.
(For interpretation of the references to color/colour in this figure
legend, the reader is referred to the Web version of this article).
an increasing risk ratio over the following 4 years gradu-
ally inverted this relationship leading to a more favourable
prognosis for women, which stays stable throughout the
follow-up period.
To understand the change in gender difference of sur-
vival prognosis over time, we had to look at the gender
differences with respect to other available risk factors,
such as age and T-stage at the first diagnosis. Men and
women differ with respect to severity (T-stage) of the
initial diagnosis and age distribution (Table 2). Women
were significantly more often diagnosed with MIBC (men:
29%; women: 31%; p < 0.001) and were significantly older
(p < 0.001) by the time of diagnosis. Once differences in
using age-specific incidence rates, there was no difference
(in the age-specific incidence rate distribution) between
men and women (p Z 0.91). Men and women diagnosed
with MIBC also differed with respect to severity within
this group of patients. There were less male patients with
MIBC diagnosed with metastatic disease (33% versus
36%; p Z 0.27) and positive lymph node status (23%
versus 33%; p < 0.001) than female patients.
When stratifying the risk ratio analysis for T-stage
(MIBC versus NMIBC), a higher risk for women within
the first 2 years after diagnosis could be seen in both
groups (Supplementary Fig. 1). However, this trend was
more pronounced in MIBC than in patients with
NMIBC. There is no change in this trend across the
diagnostic time points (results not shown).
To quantify the impact of gender-specific T-stage
distributions on the observed gender differences in sur-
vival, we evaluated how much of the gender difference in
relative survival could be explained by the difference in
T-stage at the first diagnosis (Fig. 3). Table 3 shows that
35% of the gender differences in survival, 2 years after
diagnosis, were explained by differences in T-stage at
diagnosis. The corresponding estimates for 5 and 10
years were 52% and 97%, respectively.
4. Discussion
We found that overall survival for Norwegian patients
with bladder cancer is better for men than for women.
We also showed that over the whole follow-up time, the
risk of bladder cancererelated death is independent of
time since diagnosis. These results are in concordance
with many other studies [17,34]. However, we showed

Table 2
Number and percentage of men and women with bladder cancer diagnosis with respect to T-stage and age at diagnosis. Age-specific incidence
rates for male and female patients with bladder cancer.
NMIBC MIBC
TaLG TaHG Tis T1* T2-4*
Men 5081 985 329 2133 2899
44.5% 8.6% 2.9% 18.7% (
0.5%) 25.4% (
0.5%)
Women 1758 217 86 597 1044
47.5% 5.9% 2.3% 16.1% (
0.8%) 28.2% (
0.9%)
Age at diagnosis Association
0e49 50e64 65e79 80 T-stage:
Men 457 2468 5561 2941 p Z 8.3$108
4.0% 21.6% 48.7% 25.7% Age:
Women 139 773 1618 1172 p Z 3.6$1011
3.8% 20.9% 43.7% 31.7% Incidence-rates:
Age-specific incidence rates p Z 0.914
0e49 50e64 65e79 80
Men 2.5 44.4 173.1 317.8
Women 0.8 15.0 47.5 76.3
MIBC Association
Metastases Positive lymph nodes status Metastases:
Men 22.6% 33% p Z 3.7$106
Women 32.9% 36% Lymph node status: p Z 0.266
56 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58
Fig. 3. Relative survival for men, women and women assuming the
same T-stage distribution as men. Black (grey) lines: mean survival
curve for men (women); Dashed grey line: survival curve for
women when assuming men’s covariate pattern.
that the use of time-constant risk ratios implying the
proportional hazard assumption is misleading. Allowing
the gender-specific risk ratios to vary over follow-up
time led to the conclusion that female patients with
bladder cancer have a less favourable prognosis only
within the first 2 years after diagnosis; thereafter,
women have a slightly better prognosis. This pattern
was particularly pronounced in MIBC.
In our dataset, we were able to specify how the
gender-specific differences in T-stage distribution may
influence gender-specific survival differences. We found
that female survival improved immediately when
assuming the male patient’s T-stage distribution for
women. After 5 years since diagnosis, the survival dif-
ference between men and women was halved, and after
10 years, survival of male and female patients with
bladder cancer was approximately the same. This con-
firms that more severe T-stage in female patients with
bladder cancer explain a substantial part of the unfav-
ourable survival when compared with male patients.
It has been argued that the diagnostic delay partly
explains gender differences in survival of patients with
bladder cancer [18,21], and it is obvious that such a
delay would be particularly disadvantageous in more
Table 3
Relative survival rates (in %) and corresponding confidence intervals
for male, female and T-stageeadjusted female patients with bladder
cancer 2, 5 and 10 years after diagnosis. The percentage explained by
gender-specific T-stage differences at diagnosis is presented in bold.
Relative survival in %
2 years 5 years 10 years
Men 84.0 (83.2e84.8) 75.1 (74.0e76.2) 66.3 (64.6e68.0)
Women 79.6 (78.3e80.9) 71.7 (70.1e73.4) 64.5 (62.3e66.9)
Women 81.2 (80.0e82.4) 73.5 (71.9e75.1) 66.2 (64.0e68.5)
(adjusted)
% explained 35.4% 51.5% 96.4%
severe cancers. This could also imply that the proportion
of severe cancers is higher for women than men within
the group of patients with MIBC, which is supported by
the higher proportion of metastases and positive lymph
nodes in female compared to male patients with MIBC.
Our results are in line with a recent Austrian cohort
study [7] showing that survival is almost the same for T1
stage cancers, while the higher the T-stage, the more
unfavourable the prognosis for female patients when
compared to male patients.
Although almost half of the gender-specific differ-
ences in prognosis at 5 years since diagnosis in our study
were explained by a difference in T-stage, half of the
difference remained unexplained. Gender differences in
treatment of bladder cancer, particularly for more severe
bladder cancer diagnoses, could theoretically influence
prognosis. The latter is supported by two Swedish
studies [20,35] based on comprehensive data from the
Swedish National Register of Urinary Bladder Cancer
[36] that reported less optimal treatment in women with
a bladder cancer diagnosis.
Pros and cons have been widely discussed with a
general agreement on the use of relative survival for
population-based studies. The main concern about using
relative survival is the comparability of the general
population with the study population, also with respect
to certain confounders such as smoking. However, the
expected bias is negligible [37].
Data on smoking habits and patients-related risk fac-
tors are not included in this study; therefore, possible
gender differences could not been taken into account.
Smoking is a key risk factor not only for the development
of bladder cancer but also for its impact on prognosis
[38,39]. Rink et al. [40] suggested a larger effect of smoking
on bladder cancererelated survival for women than men.
Other factors, which could possibly differ between genders
and thus explain parts of the remaining survival gap be-
tween male and female patients with bladder cancer,
include occupation, medication, hormone levels and ge-
netic and biologic differences [17,18].
The challenge of incompleteness of the clinical T-
stage variable was solved by imputation, using infor-
mation about morphology, age, survival time, grade and
gender. To capture the variation of our estimates
dependent on the imputation, we generated 10 imputed
datasets and derived the presented estimates by pooling
the results from all 10 imputed datasets.
We showed that former reports assuming a constant
risk ratio have led to the misleading conclusion that fe-
male patients with bladder cancer have a worse prognosis
than men independent of the time since diagnosis. Based
on our Norwegian cohort data, we revealed inferior sur-
vival prognosis for female patients with bladder cancer
solely for the first 2 years after diagnosis, particularly for
patients with MIBC. Thereafter, female patients with
bladder cancer had lower risk of death, resulting in fe-
male’s survival gradually approaching that of men as
 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58
follow-up time increased. We also showed that adverse T-
Stage distribution in women may explain half of the
unfavourable survival in female patients with bladder
cancer 5 years since diagnosis.
Conflict of interest statement
None.
Appendix A. Supplementary data
Supplementary data related to this article can be found
at https://doi.org/10.1016/j.ejca.2018.03.001.
References
[1] Cook MB, McGlynn KA, Devesa SS, Freedman ND,
Anderson WF. Sex disparities in cancer mortality and survival.
Cancer Epidemiol Biomark Prev 2011;20(8):1629e37.
[2] Edgren G, Liang L, Adami HO, Chang ET. Enigmatic sex dis-
parities in cancer incidence. Eur J Epidemiol 2012;27(3):187e96.
[3] Micheli A, Ciampichini R, Oberaigner W, Ciccolallo L, de
Vries E, Izarzugaza I, et al. The advantage of women in cancer
survival: an analysis of EUROCARE-4 data. Eur J Cancer 2009;
45(6):1017e27.
[4] Radkiewicz C, Johansson ALV, Dickman PW, Lambe M,
Edgren G. Sex differences in cancer risk and survival: a Swedish
cohort study. Eur J Cancer 2017;84:130e40.
[5] Innos K, Padrik P, Valvere V, Aareleid T. Sex differences in
cancer survival in Estonia: a population-based study. BMC
Cancer 2015;15:72.
[6] Oberaigner W, Siebert U. Do women with cancer have better
survival as compared to men after adjusting for staging distri-
bution? Eur J Public Health 2011;21(3):387e91.
[7] Waldhoer T, Berger I, Haidinger G, Zielonke N, Madersbacher S.
Sex differences of pT1 bladder cancer survival in Austria: a
descriptive, long-term, nation-wide analysis based on 27,773 pa-
tients. Urol Int 2015;94(4):383e9.
[8] Kamat AM, Hahn NM, Efstathiou JA, Lerner SP,
Malmstrom PU, Choi W, et al. Bladder cancer. Lancet 2016;
388(10061):2796e810.
[9] Ellison LF. Differences in cancer survival in Canada by sex.
Health Rep 2016;27(4):19e27.
[10] Mungan NA, Aben KK, Schoenberg MP, Visser O,
Coebergh JW, Witjes JA, et al. Gender differences in stage-
adjusted bladder cancer survival. Urology 2000;55(6):876e80.
[11] Ristau BT, Davies BJ. Disparity in bladder cancer outcomes:
what’s sex got to do with it? Cancer 2014;120(4):461e3.
[12] Zaitsu M, Toyokawa S, Tonooka A, Nakamura F, Takeuchi T,
Homma Y, et al. Sex differences in bladder cancer pathology and
survival: analysis of a population-based cancer registry. Cancer
Med 2015;4(3):363e70.
[13] Antoni S, Ferlay J, Soerjomataram I, Znaor A, Jemal A, Bray F.
Bladder cancer incidence and mortality: a global overview and
recent trends. Eur Urol 2017;71(1):96e108.
[14] Andreassen BK, Aagnes B, Gislefoss R, Andreassen M,
Wahlqvist R. Incidence and survival of urothelial carcinoma of
the urinary bladder in Norway 1981e2014. BMC Cancer 2016;
16(1):799.
[15] Andreassen BK, Myklebust TA, Haug ES. Crude mortality and
loss of life expectancy of patients diagnosed with urothelial car-
cinoma of the urinary bladder in Norway. Scand J Urol 2017;
51(1):38e43.
57
[16] Marcos-Gragera R, Mallone S, Kiemeney LA, Vilardell L,
Malats N, Allory Y, et al. Urinary tract cancer survival in Europe
1999e2007: results of the population-based study EUROCARE-
5. Eur J Cancer 2015;51(15):2217e30.
[17] Wolff I, Brookman-May S, May M. Sex difference in presentation
and outcomes of bladder cancer: biological reality or statistical
fluke? Curr Opin Urol 2015;25(5):418e26.
[18] Dobruch J, Daneshmand S, Fisch M, Lotan Y, Noon AP,
Resnick MJ, et al. Gender and bladder cancer: a collaborative
review of etiology, biology, and outcomes. Eur Urol 2016;69(2):
300e10.
[19] Mitra AP, Skinner EC, Schuckman AK, Quinn DI, Dorff TB,
Daneshmand S. Effect of gender on outcomes following radical
cystectomy for urothelial carcinoma of the bladder: a critical
analysis of 1994 patients. Urol Oncol 2014;32(1). 52 e1e9.
[20] Thorstenson A, Hagberg O, Ljungberg B, Liedberg F, Jancke G,
Holmang S, et al. Gender-related differences in urothelial carci-
noma of the bladder: a population-based study from the Swedish
National Registry of Urinary Bladder Cancer. Scand J Urol 2016;
50(4):292e7.
[21] Henning A, Wehrberger M, Madersbacher S, Pycha A, Martini T,
Comploj E, et al. Do differences in clinical symptoms and referral
patterns contribute to the gender gap in bladder cancer? BJU Int
2013;112(1):68e73.
[22] Larsen IK, Smastuen M, Johannesen TB, Langmark F,
Parkin DM, Bray F, et al. Data quality at the Cancer Registry of
Norway: an overview of comparability, completeness, validity and
timeliness. Eur J Cancer 2009;45(7):1218e31.
[23] Mostofi F, Sobin L, Tosoni I. Histological typing of urinary
bladder tumours. International histological classification of tu-
mours. No 19. Geneva: World Health Organization; 1973.
[24] Raghunathan TE, Lepkowski JM, Van Hoewyk J, Solenberger P.
A multivariate technique for multiply imputing missing values
using a sequence of regression models. Surv Methodol 2001;27(1):
85e96.
[25] Van Buuren S. Multiple imputation of discrete and continuous
data by fully conditional specification. Stat Methods Med Res
2007;16(3):219e42.
[26] Rubin DB. Multiple imputation for nonresponse in surveys. New
York: Wiley; 1987.
[27] Rubin DB. Multiple imputation after 18þ years. J Am Stat Assoc
1996;91(434):473e89.
[28] Ederer F, Heise H. Instructions to IBM 650 programmers in
processing survival computations: methodological note. 1959.
[29] Kaplan EL, Meier P. Nonparametric estimation from incomplete
observations. J Am Stat Assoc 1958;53(282):457e81.
[30] Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression
models for relative survival. Stat Med 2004;23(1):51e64.
[31] Rutherford MJ, Crowther MJ, Lambert PC. The use of restricted
cubic splines to approximate complex hazard functions in the
analysis of time-to-event data: a simulation study. J Stat Comput
Simul 2015;85(4):777e93.
[32] Lambert PC, Royston P. Further development of flexible para-
metric models for survival analysis. Stata J 2009;9(2):265.
[33] LLC S, editor. STATA statistical software: release 15. Texas,
USA: College Station; 2017.
[34] Patel MI, Bang A, Gillett D, Cheluvappa R, Smith DP. Poor
survival of females with bladder cancer is limited to those aged 70
years or over: a population-wide linkage study, New South Wales,
Australia. Cancer Med 2015;4(8):1145e52.
[35] Jahnson S, Damm O, Hellsten S, Holmang S, Liedberg F,
Ljungberg B, et al. A population-based study of patterns of care
for muscle-invasive bladder cancer in Sweden. Scand J Urol
Nephrol 2009;43(4):271e6.
[36] Haggstrom C, Liedberg F, Hagberg O, Aljabery F, Strock V,
Hosseini A, et al. Cohort profile: the Swedish National Register of
Urinary Bladder Cancer (SNRUBC) and the bladder cancer data
base Sweden (BladderBaSe). BMJ Open 2017;7(9):e016606.
58 B.K. Andreassen et al. / European Journal of Cancer 95 (2018) 52e58

[37] Hinchliffe SR, Rutherford MJ, Crowther MJ, Nelson CP,
Lambert PC. Should relative survival be used with lung cancer
data? Br J Cancer 2012;106(11):1854e9.
[38] Rink M, Furberg H, Zabor EC, Xylinas E, Babjuk M, Pycha A,
et al. Impact of smoking and smoking cessation on oncologic
outcomes in primary non-muscle-invasive bladder cancer. Eur
Urol 2013;63(4):724e32.
[39] Al-Zalabani AH, Stewart KFJ, Wesselius A, Schols AMWJ,
Zeegers MP. Modifiable risk factors for the prevention of bladder
cancer: a systematic review of meta-analyses. Eur J Epidemiol
2016;31(9):811e51.
[40] Rink M, Xylinas E, Trinh QD, Lotan Y, Margulis V, Raman JD,
et al. Gender-specific effect of smoking on upper tract urothelial
carcinoma outcomes. BJU Int 2013;112(5):623e37.