Marital status: a gender-independent risk factor for poorer survival after radical cystectomy


  • J.D.S. and M.M. contributed equally to the work.

Jesse D. Sammon, Vattikuti Urology Institute, K-9 Henry Ford Health System, 2799 W. Grand Boulevard, Detroit, MI 48202, USA. e-mail:


Study Type – Prognosis (cohort)

Level of Evidence 2a

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

Married individuals have lower morbidity and mortality rates for all major causes of death. Cancer-specific survival is better in married patients with testis cancer, prostate cancer, breast cancer, cervical cancer, as well as head and neck cancers.

We have found the effect of marital status on outcomes after radical cystectomy to be variable, depending on gender and the outcome addressed. Being married is predictive of lower all-cause mortality for both men and women relative to their separated, divorced or widowed (SDW) or never-married counterparts. It is also predictive of lower bladder-cancer-specific mortality relative to SDW individuals. Marriage also exerts a protective effect on men regarding non-organ-confined disease, with those never having married having significantly higher rates.


  • • To examine the effect of marital status (MS) on the rate of non-organ-confined disease (NOCD) at radical cystectomy (RC)
  • • To assess the effect of MS on the rate of bladder-cancer-specific mortality (BCSM) and all-cause mortality (ACM) after RC for urothelial carcinoma of the urinary bladder (UCUB).


  • • A total of 14 859 patients, who underwent RC for UCUB, were captured within the Surveillance, Epidemiology, and End Results database, between 1988 and 2006.
  • • Logistic regression analysis was used to assess the rate of NOCD (T3-4/NI-3/M0) at RC and Cox regression analyses were used to assess BCSM and ACM.
  • • Analyses were stratified according to gender; covariates included socio-economic status, tumour stage, age, race, tumour grade and year of surgery.


  • • Never-married males had a higher rate of NOCD at RC (odds ratio = 1.22, P= 0.004), an effect not found in never-married females.
  • • Separated, divorced or widowed (SDW) males (hazard ratio [HR]= 1.18, P= 0.005) and females (HR = 1.16, P= 0.002) had higher rates of BCSM than their married counterparts.
  • • SDW and never-married males had higher rates of ACM than their married counterparts (HR = 1.22, P < 0.001 and HR = 1.26, P < 0.001, respectively).
  • • SDW and never-married females also had higher rates of ACM than married females (HR = 1.24, P < 0.001 and HR = 1.22, P= 0.01, respectively).


  • • For both men and women, being SDW conveyed an increased risk of BCSM after RC.
  • • SDW and never marrying had a deleterious effect on ACM.
  • • Unfavourable stage at RC was also seen more commonly in never-married males.

marital status


non-organ-confined disease


radical cystectomy


bladder-cancer-specific mortality


all-cause mortality


urothelial carcinoma of the urinary bladder


separated, divorced or widowed


hazard ratio


Surveillance, Epidemiology and End Results


socio-economic status.


Marital status (MS) has been shown to affect the natural history of many diseases, including several cancers [1–4]. Being married has been associated with improved survival in patients with prostate cancer [5], breast cancer [6], cervical cancer [7], and head and neck cancers [8], and there is also an associated decrease in overall and cancer-specific mortality in testis cancer [9]. To date, there has been no consensus regarding the effect of MS on patients with urothelial carcinoma of the urinary bladder (UCUB), although three studies have previously addressed this topic: Nelles et al. [10], in their study of 127 015 patients, failed to show a survival advantage in married patients with UCUB, but found an increased risk of death in widowed men; Datta et al. [11], in their study of 19 982 patients, recorded a survival benefit in married patients with superficial UCUB; and Gore et al.'s study of 5854 patients [12] showed improved survival after radical cystectomy (RC) for UCUB in married patients. However, these authors did not look at women and men independently.

In the present study, we analysed a large population-based cohort of patients, assessing the impact of MS on bladder cancer stage at RC, as well as on cancer-control outcomes after RC. Although there is a relative paucity of analysis, we hypothesized that unmarried patients with UCUB may have a more advanced stage at RC and may fare worse after RC.



The study cohort originated from the Surveillance, Epidemiology and End Results (SEER) database reported by the National Cancer Institute statistics programme [13], which routinely collects patient demographics and publishes cancer incidence and survival data from population-based cancer registries covering ≈26 percent of the US population. These registries include the San Francisco-Oakland, Detroit, Seattle Puget Sound, Atlanta, San-Jose Monterey, Los Angeles, and Georgia metropolitan areas, as well as the states of Connecticut, Hawaii, Iowa, New Mexico, Utah, Alaska, California (excluding San Francisco, Los Angeles, and San Jose), Kentucky, Louisiana and New Jersey.

We identified individuals treated with RC for histologically confirmed UCUB [International Classification of Disease for Oncology (C67.0-C67.9)] from the SEER database between 1988 and 2006. To eliminate most childhood bladder tumours, as well as to account for sociodemographic characteristics, we excluded individuals <18 years of age. We also excluded death certificate and autopsy diagnoses, as well as those with missing or incomplete staging and grading information.


Within the SEER database, the MS of the patient is recorded at the time of surgery, as divorced, widowed or separated (DWS), married, never married or unknown. Patients with unknown MS at the time of RC were excluded from the study (n= 409). Covariates consisted of age groups (18–49, 50–59, 60–69, 70–79 and ≥80 years), gender (male or female), race (white, black or other), tumour stage (organ-confined disease [pT1-2pN0M0] or non-organ-confined disease (NOCD; i.e. locally advanced [pT3-4pN1-3M0]), tumour grade (low [I-II] or high [III-IV]), and year of surgery categories (1988–1997, 1998–2001, 2002–2004 or 2005–2006). A composite variable was created to account for various socio-economic status (SES) components available within the SEER database: median family income, percentage of persons living below the poverty line, and percentage of persons without a high school diploma. All values were standardized. We then generated the sum of these three standardized values to obtain a SES score. SES scores below and above the median were respectively indicative of low and high SES.


Chi-squared and chi-squared trend tests were used to compare proportions and trends between married, SDW and never-married patients. We performed three sets of analyses. First, we used univariable and multivariable logistic regression models to examine the rate of NOCD (pT3-4/pN1-3/M0) at RC. All covariates described above were considered. All analyses were stratified by gender to assess for potential differences between males and females. Second, we focused on the rates of bladder-cancer-specific mortality (BCSM). We relied on the life table method to obtain survival rates specific to each MS category. Subsequently, we performed univariable and multivariable Cox regression analyses for prediction of BCSM. Covariates included: MS, SES, tumour stage, age categories, race, tumour grade and year of surgery categories. Again, all analyses were stratified by gender. In the third set of analyses, we repeated the same steps but assessed all-cause mortality (ACM).

All statistical tests were performed using the R (version 2.11.1) statistical package. All tests were two-sided, and a P value ≤0.05 was considered to indicate statistical significance.


Between 1988 and 2006, 14 859 patients underwent RC for UCUB and met criteria for assessment. Patient characteristics are shown in Table 1. The majority were male (74.7%), white (90.1%) and of high SES (51.2%). The mean age of the population was 68 (median 69) years. Almost half (46.8%) had NOCD at RC and nearly all (88.2%) had high grade tumours.

Table 1. Descriptive characteristics of 14 859 patients treated with RC for UCUB between 1988 and 2006 within the SEER database, stratified according to MS
VariablesOverall, N= 14 589Married, N= 9975SWD, N= 3480Never-married, N= 1404 P
Age, n (%)    <0.001
 18–49 years836 (5.6)486 (4.9)123 (3.5)227 (16.2)
 50–59 years2 305 (15.5)1600 (16.0)365 (10.5)340 (24.2)
 60–69 years4 472 (30.1)3172 (31.8)879 (25.3)421 (30.0)
 70–79 years5 208 (35.0)3588 (36.0)1334 (38.3)286 (20.4)
 ≥80 years2 038 (13.7)1129 (11.3)779 (22.4)130 (9.3)
Gender, n (%)    <0.001
 Female3 765 (25.3)1737 (17.4)1658 (47.6)370 (26.4)
 Male11 094 (74.7)8238 (82.6)1822 (52.4)1034 (73.6)
Race, n (%)    <0.001
 White13 386 (90.1)9110 (91.3)3094 (88.9)1182 (84.2)
 Black772 (5.2)347 (3.5)256 (7.4)169 (12.0)
 Other701 (4.7)518 (5.2)130 (3.7)53 (3.8)
Tumour stage, n (%)    <0.001
 Localized (T1-2/N0/M0)7 905 (53.2)5428 (54.4)1781 (51.2)696 (49.6)
 Locally advanced(T3-4/N1-3)6 954 (46.8)4547 (45.6)1699 (48.8)708 (50.4)
Tumour grade, n (%)    0.2
 Low (I–II)1 746 (11.8)1205 (12.1)391 (11.2)150 (10.7)
 High (III–IV)13 113 (88.2)8770 (87.9)3089 (88.8)1254 (89.3)
SES, n (%)    0.001
 High7 601 (51.2)5196 (52.1)1743 (50.1)662 (47.2)
 Low7 258 (48.8)4779 (47.9)1737 (49.9)742 (52.8)
Year of surgery, n (%)    0.01
 1988–19975 881 (39.6)4011 (40.2)1353 (38.9)517 (36.8)
 1998–20012 998 (20.2)1998 (20.0)695 (20.0)305 (21.7)
 2002–20043 632 (24.4)2385 (23.9)881 (25.3)366 (26.1)
 2005–20062 348 (15.8)1581 (15.8)551 (15.8)216 (15.4)

When MS was stratified according to gender, 74.3, 16.4 and 9.3% of males and 46.1, 44.0 and 9.8% of females were married, SDW or never-married, respectively. Temporal trends in proportions of males and females according to MS showed a lack of significant variation over time (Fig. 1A–C).

Figure 1.

The proportion of A, married, B, SWD and C, never-married males and females over time.

Table 2 shows the comparative ratio of NOCD within covariate groupings, stratified according to gender. Never-married males had a 22% higher rate of NOCD than their married counterparts (P= 0.004). The effect of MS on the rate of NOCD failed to reach significance in females.

Table 2. Multivariable logistic regression analyses for prediction of NOCD (T3-4/N1-3) at RC in 14 859 patients for UCUB of all stages between 1988 and 2006, stratified according to gender
Multivariable OR (95% CI) P Multivariable OR (95% CI) P
  1. OR, odds ratio.

 Married1.0 (Ref.) 1.0 (Ref.) 
 SDW1.07 (0.97–1.19)0.21.06 (0.92–1.23)0.4
 Never-married1.22 (1.06–1.39)0.0041.12 (0.89–1.42)0.3
 High1.0 (Ref.) 1.0 (Ref.) 
 Low1.03 (0.95–1.11)0.40.95 (0.83–1.09)0.5
 18–49 years1.0 (Ref.) 1.0 (Ref.) 
 50–59 years1 (0.83–1.21)1.00.9 (0.63–1.29)0.6
 60–69 years1.08 (0.91–1.29)0.40.93 (0.67–1.3)0.7
 70–79 years1.19 (1–1.41)0.060.88 (0.63–1.22)0.4
 ≥80 years1.25 (1.03–1.52)0.030.78 (0.55–1.1)0.2
 White1.0 (Ref.) 1.0 (Ref.) 
 Black1.03 (0.85–1.25)0.71.32 (1.03–1.7)0.03
 Other1.04 (0.87–1.24)0.70.94 (0.69–1.28)0.7
Tumour grade    
 Low (I–II)1.0 (Ref.) 1.0 (Ref.) 
 High (III–IV)3.84 (3.33–4.43)<0.0013.87 (3–4.99)<0.001
Year of surgery    
 1988–19971.0 (Ref.) 1.0 (Ref.) 
 1998–20011.38 (1.24–1.53)<0.0011.51 (1.25–1.81)<0.001
 2002–20041.49 (1.35–1.65)<0.0011.74 (1.47–2.07)<0.001
 2005–20061.4 (1.25–1.57)<0.0011.57 (1.29–1.91)<0.001

Gender-stratified Kaplan–Meier plots showed that BCSM rates were higher in SDW males than in married males (log-rank P < 0.001 [Fig. 2A]). The same relationship was found in females (log-rank P < 0.001 [Fig. 2B]). ACM rates were higher in SDW than in married and never-married males (all log-rank P < 0.05 [Fig. 2C]). ACM rates were higher in SDW and never-married females than in their married counterparts (log-rank P < 0.001 [Fig. 2D]).

Figure 2.

Kaplan–Meier plots depicting bladder-cancer-specific-mortality-free survival in A, males and B, females, stratified according to MS. Kaplan–Meier plots depicting ACM-free survival in C, males and D, females, stratified according to MS. Legend, red: married, purple: unknown, blue: never-married, green: SWD.

Multivariable Cox regression models assessing independent predictors of BCSM are shown in Table 3. SDW men had a 16% higher rate of BCSM relative to their married counterparts (P= 0.002). Similarly, an 18% higher rate of BCSM was found in SDW females relative to their married counterparts (P= 0.005).

Table 3. Multivariable Cox regression analyses in predicting BCSM in 14 859 patients treated with RC for UCUB of all stages between 1988 and 2006, stratified according to gender
Multivariable HR (95% CI) P Multivariable HR (95% CI) P
 Married1.0 (Ref.) 1.0 (Ref.) 
 SDW1.16 (1.06–1.27)0.0021.18 (1.05–1.33)0.005
 Never-married1.12 (0.99–1.27)0.071.04 (0.86–1.26)0.7
 High1.0 (Ref.) 1.0 (Ref.) 
 Low1.11 (1.03–1.19)0.0041.05 (0.94–1.17)0.4
Tumour stage    
 Localized (T1-2/N0/M0)1.0 (Ref.) 1.0 (Ref.) 
 Locally advanced (T3-4/N1-3)3.2 (2.97–3.45)<0.0012.97 (2.64–3.35)<0.001
 18–49 years1.0 (Ref.) 1.0 (Ref.) 
 50–59 years1.15 (0.96–1.37)0.11.2 (0.88–1.62)0.2
 60–69 years1.32 (1.11–1.56)0.0011.25 (0.94–1.66)0.1
 70–79 years1.49 (1.26–1.76)<0.0011.42 (1.07–1.88)0.02
 ≥80 years2.17 (1.81–2.61)<0.0011.87 (1.39–2.51)<0.001
 White1.0 (Ref.) 1.0 (Ref.) 
 Black1.15 (0.97–1.37)0.11.19 (0.99–1.44)0.07
 Other0.78 (0.66–0.93)0.0070.8 (0.61–1.05)0.1
Tumour grade    
 Low (I–II)1.0 (Ref.) 1.0 (Ref.) 
 High (III–IV)1.61 (1.41–1.84)<0.0011.3 (1.06–1.59)0.01
Year of surgery    
 1988–19971.0 (Ref.) 1.0 (Ref.) 
 1998–20010.91 (0.84–1)0.050.92 (0.8–1.06)0.2
 2002–20040.99 (0.9–1.08)0.80.94 (0.82–1.08)0.4
 2005–20060.84 (0.72–0.99)0.040.89 (0.71–1.1)0.3

When ACM was examined (Table 4), SDW and never-married males had a 22% and 26% higher rate of ACM relative to their married counterparts, respectively (both P < 0.001). The same rates were 24% and 22% higher in SDW (P < 0.001) and never-married (P= 0.01) females relative to their married counterparts.

Table 4. Multivariable Cox regression analyses in predicting ACM in 14 859 patients treated with RC for UCUB of all stages between 1988 and 2006, stratified according to gender
Multivariable HR (95% CI) P Multivariable HR (95% CI) P
 Married1.0 (Ref.) 1.0 (Ref.) 
 SDW1.22 (1.14–1.3)<0.0011.24 (1.13–1.37)<0.001
 Never-married1.26 (1.15–1.38)<0.0011.22 (1.05–1.42)0.01
 High1.0 (Ref.) 1.0 (Ref.) 
 Low1.08 (1.03–1.14)0.0021.04 (0.96–1.14)0.3
Tumour stage    
 Localized (T1-2/N0/M0)1.0 (Ref.) 1.0 (Ref.) 
 Locally advanced (T3-4/N1-3)2.29 (2.17–2.41)<0.0012.29 (2.09–2.5)<0.001
 18–49 years1.0 (Ref.) 1.0 (Ref.) 
 50–59 years1.29 (1.11–1.5)0.0011.24 (0.95–1.61)0.1
 60–69 years1.76 (1.53–2.02)<0.0011.53 (1.2–1.96)0.001
 70–79 years2.42 (2.11–2.78)<0.0011.89 (1.48–2.41)<0.001
 ≥80 years3.83 (3.31–4.44)<0.0012.73 (2.12–3.51)<0.001
 White1.0 (Ref.) 1.0 (Ref.) 
 Black1.21 (1.06–1.37)0.0041.21 (1.03–1.41)0.02
 Other0.83 (0.73–0.94)0.0040.77 (0.62–0.96)0.02
Tumour grade    
 Low (I–II)2.0 (Ref.) 2.0 (Ref.) 
 High (III–IV)1.18 (1.09–1.28)<0.0011.11 (0.96–1.28)0.2
Year of surgery    
 1988–19971.0 (Ref.) 1.0 (Ref.) 
 1998–20010.96 (0.9–1.03)0.20.93 (0.83–1.05)0.2
 2002–20041.02 (0.95–1.1)0.50.93 (0.83–1.05)0.2
 2005–20060.9 (0.8–1.02)0.10.93 (0.77–1.12)0.4


Marital status represents an important determinant of population health in general and of cancer control outcomes in many solid tumours. Hu et al. [1] reported that married individuals have lower morbidity and mortality rates for all major causes of death, as did Johnson et al. [2] in the National Longitudinal Mortality Study. Goodwin et al. [3] showed that married patients with various cancer primaries had better survival, as did Lai et al. [4]. The effect of MS has also been documented for several types of cancers, such as testis cancer [9], prostate cancer [5], breast cancer [6], cervical cancer [7], as well as head and neck cancers [8]. All of these studies found better cancer-specific survival in married individuals.

To better assess the impact of MS in patients with UCUB, we relied on a large population-based cohort, focusing on three specific steps. First, we tested the effect of MS on the rate of NOCD at RC; followed by BCSM and finally, ACM. MS probably exerts different effects on these three outcomes, necessitating three separate analyses. Moreover, we systematically examined outcomes stratified by gender as MS has previously been found to exert a gender-specific effect. To our knowledge, the present study represents the first attempt to simultaneously examine the effect of gender and MS on these three outcomes.

Several of the present study findings are noteworthy. First, analyses of NOCD showed higher rates in never-married males, whereas, in females, MS failed to exert a predictive effect. These data suggest that MS has no predictive effect on cancer stage at RC in females. However, never-married males present with a more advanced stage than those who are married, implying that being married confers a protective effect in males. The spousal effect may lead to more timely presentation, diagnosis and RC. Interestingly, SDW males were not at a higher risk of NOCD than married males. It is possible that previous marriage is sufficient to sensitize males to closer follow-up and more health-oriented behaviour [14]. Regardless of the underlying cause, never-married males represent individuals where particular care should be taken to avoid diagnosis delays. When examined from the perspective of general medical practice, these factors should be taken into account when UCUB is suspected, such as in the presence of microscopic or gross haematuria or in high-risk populations (smokers or populations with occupational exposure).

Second, our results show that SDW patients (both males and females) are at higher risk of BCSM than their married counterparts. This could be explained by the fact that a change in MS causes a great deal of emotional stress. It is known that emotional stress stimulates sympathetic responses that result in compromised DNA repair, increased angiogenesis and decreased natural killer cell function, all of which affect progression of cancer cells [15]. From a clinical perspective, there is also substantial evidence that providing psychosocial support reduces depression, anxiety and pain, and may increase survival time with cancer [16]. Marriage conveys social support through this psychoimmunological pathway. Therefore, lack of marriage, in particular loss of marriage, may predispose a patient to quicker progression of cancer and higher BCSM rates. From a practical perspective, it is important to avoid diagnosis and treatment delays in SDW individuals, with the incentive of reducing the BCSM disadvantage in this MS subgroup. Conversely, it is possible that married individuals are choosing surgeons and institutions where better outcomes, including BCSM, are recorded: Iwashyna et al. [14] reported that spouses might influence medical care and decision-making.

The present results, with regard to BCSM, corroborate the findings of a previous study by Gore et al. [12], also relying on the SEER cancer survival database (n= 5854). However, the authors of that study did not perform a subset analysis where male and female subjects were examined separately. Previous studies have suggested that the health benefits from marriage are more important for men than for women [17]. In the present study cohort, the effect of MS on BCSM was present regardless of gender. Similar gender-stratified findings were previously reported in colon cancer [18].

Our findings contrast with those reported by Nelles et al. [10], also based on data from SEER (n= 127 015), wherein MS did not appear to have a significant survival effect for women. Two important differences in study design can account for this discrepancy. The authors of that study examined all patients with UCUB, while the present study focused exclusively on patients undergoing RC. Perhaps the most important difference is that their analyses focused on death from any cancer, while the models in the present study were fitted to predict BCSM exclusively.

Third, regarding ACM, it is noteworthy that the effect of MS is equally distributed within both genders. Specifically, SDW and never-married males and females had higher ACM rates. This observation implies that the overall health status of these individuals is adversely affected by not being married. Pruthi et al. [19] showed that married patients undergoing RC for UCUB were in better general health and had shorter lengths of hospitalization. Alternatively, in addition to better baseline health status, married patients might enjoy a healthier lifestyle after surgery than their non-married counterparts. For example, although the effect of smoking cessation on BCSM is controversial [20], its effect on the risk of myocardial infarction and stroke is not [21]. After a diagnosis of UCUB, one could speculate that married patients are encouraged by their spouses to stop smoking. It is also possible that married individuals with cancer might exercise more often than their SDW counterparts. In the specific context of postoperative recovery from oncological surgery, the effect of physical activity on ACM is well documented [22].

Several limitations apply to the present study. While the SEER database includes a large number of patients, it does not allow the determination of direct causal relationships of MS on mortality and stage at RC. Also, the SEER database does not contain measures of SES at the individual level, but only contains ecological SES characteristics, which can be described as county attributes. Most importantly, the SEER database does not include details on the MS of each individual. For example, we were unable to assess the quality of the marriage. In addition, many individuals may be in a meaningful relationship and living with a partner at the time of RC, even though they were classified as never-married. Additionally, the SEER database does not contain information on comorbidities. The aforementioned pitfalls of the SEER database could have affected our results and caused us to overestimate or underestimate the effect of MS on stage at RC or BCSM and ACM. Finally, the present study was conducted in a North American cohort, specific to the USA. As such, these findings may not be generalizable to other sociodemographic settings.

In conclusion, we examined the effect of MS on the rate of NOCD at RC, as well as on the rates of BCSM and ACM. Several of our study findings are unique. First, we showed that the effect of MS differs according to gender. Second, we showed that the effect of MS differs according to the endpoint that is addressed. Finally, we quantified the strength of the effect of MS on UCUB outcomes after RC and elucidated the gender variability. Different risk groups exist within each of the examined endpoints and those at highest risk may benefit from a higher index of clinical suspicion.


Pierre I. Karakiewicz is partially supported by the University of Montreal Health Centre Urology Specialists, Fonds de la Recherche en Santé du Quebec, the University of Montreal Department of Surgery and the University of Montreal Health Centre (CHUM) Foundation.


None declared.