Marriage and mortality in bladder carcinoma


  • John L. Gore M.D.,

    Corresponding author
    1. Department of Urology, David Geffen School of Medicine, University of California–Los Angeles, Los Angeles, California
    • Department of Urology, David Geffen School of Medicine, University of California–Los Angeles, 66-124 CHS, 10833 LeConte Avenue, Los Angeles, CA 90095
    Search for more papers by this author
    • Fax: (310) 206-5343

  • Lorna Kwan M.P.H.,

    1. Jonsson Comprehensive Cancer Center, University of California–Los Angeles, Los Angeles, California
    Search for more papers by this author
  • Christopher S. Saigal M.D., M.P.H.,

    1. Department of Urology, David Geffen School of Medicine, University of California–Los Angeles, Los Angeles, California
    Search for more papers by this author
  • Mark S. Litwin M.D., M.P.H.

    1. Department of Urology, David Geffen School of Medicine, University of California–Los Angeles, Los Angeles, California
    2. Department of Health Services, School of Public Health, University of California–Los Angeles, Los Angeles, California
    Search for more papers by this author



Being married confers significant benefits in survival for patients with a variety of chronic conditions including breast and prostate carcinoma. The authors attempted to determine whether marital status is associated with survival in patients undergoing radical cystectomy for bladder carcinoma.


The authors identified 7262 subjects from the Surveillance, Epidemiology, and End Results public-use database who underwent radical cystectomy for transitional cell carcinoma of the bladder. They performed survival analyses using Kaplan–Meier estimates and Cox proportional hazards models. The authors created multivariate models to evaluate the independent association between marital status and survival, controlling for pathologic stage, lymph node status, age, race/ethnicity, and gender.


Married subjects were older and more often male, white, and had earlier disease stage at diagnosis. Married subjects had significantly better survival than did single or widowed subjects (P < 0.001), and married subjects revealed a trend toward better survival than separated/divorced subjects (P = 0.20). Multivariate modeling revealed that compared with single subjects, those who were married had better survival, independent of age at the time of diagnosis, gender, race/ethnicity, disease stage, and lymph node status (P < 0.001).


Marriage was associated with improved survival in patients with bladder carcinoma, independent of other factors known to influence mortality in this population. Although the mechanisms underlying this survival advantage are unknown, possibilities include differences in cancer screening, risk behaviors, and access to medical care. The interaction between psychosocial factors and the body's immune function may further explain the differential survival in this cohort. Cancer 2005. © 2005 American Cancer Society.

Survival after a cancer diagnosis is largely a function of tumor biology. However, socioeconomic and demographic variables may also significantly impact cancer-specific survival. Individuals of lower socioeconomic status (SES) have higher overall mortality rates and decreased survival for a majority of cancers.1 These inflated mortality rates appear to be independent of risk factors known to be more prevalent among persons of lower socioeconomic position, including smoking, alcohol use, obesity, and a sedentary life style.2 Other nonclinical variables associated with mortality include age, gender, race, and insurance status.3, 4

The impact of marital status and social support on mortality has been defined for a variety of cancer diagnoses.5–8 Most of these investigations assessed the association between social support and survival for cancers with high 5-year survival rates. In these indolent malignancies, social support would likely have a more profound influence than in disease processes that follow a more aggressive pattern. In melanoma, the beneficial effect of positive emotional adjustment was negated when patients presented with disease metastatic to regional lymph nodes, suggesting that the salutary effect of positive adjustment could not overcome aggressive tumor biology.

We postulated that patients who require radical cystectomy for bladder carcinoma would benefit from spousal support, despite diminished survival rates compared with other malignancies in which marital status is believed to influence mortality. We analyzed data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) public-use database to determine whether marital status affects survival in patients who undergo radical cystectomy for bladder carcinoma. To determine the independent effect of marital status on survival, we controlled variables known to impact survival after radical cystectomy, including age, race/ethnicity, disease stage, and lymph node status.


Patient Population

We accessed the SEER cancer incidence public-use database (1973–2000) to identify 7272 subjects with a histologic diagnosis of transitional cell carcinoma of the bladder who underwent a radical cystectomy as primary surgical therapy between 1983–2000.9 The SEER cancer registry identifies patients from 11 geographic regions representative of the general U.S. population, including the metropolitan areas of Atlanta, Detroit, Los Angeles, San Francisco/Oakland, San Jose/Monterey, and Seattle/Puget Sound, and the states of Connecticut, Hawaii, Iowa, New Mexico, and Utah. For patients with biopsy-proven cancer, data regarding patient demographics, clinicopathologic staging, and treatment type are abstracted from medical records and logged in deidentified datasets.

The SEER registry catalogs marital status from hospital records and categorizes patients as single, married, separated, divorced, or widowed. We consolidated separated and divorced patients into one category but otherwise maintained the SEER classification. Vital status is obtained through local death certificates, voter registration records, and hospital medical records. Cause of death is determined, when possible, from death certificates and hospital records. Other information abstracted for the purpose of the current study included patient age, gender, race/ethnicity, disease stage, and lymph node status. Age in the SEER registry is a categoric variable, which we collapsed into fewer categories (age < 50 years, ages 50–59 years, ages 60–69 years, ages 70–79 years, and age ≥ 80 years).

We categorized stage at the time of radical cystectomy according to the American Joint Committee on Cancer/International Union Against Cancer TNM classification and stage groupings as follows: Stage I (T1N0M0: tumor extension into the lamina propria but not the muscularis propria of the bladder wall [T1], no regional lymph node metastasis [N0], no distant metastasis [M0]), Stage II (T2N0M0: tumor extension into but not beyond the muscularis propria of the bladder wall [T2], no regional lymph node metastasis [N0], no distant metastasis [M0]), Stage III (T3–T4aN0M0: tumor extension into perivesical fat [T3] or direct extension into the prostate in men and uterus in women [T4a], no regional lymph node metastasis [N0], no distant metastasis [M0]), and Stage IV (T4bN0M0, T1–4N1–3M0–1, T1–4N0–1M1: tumor extension into the pelvic or abdominal wall [T4b], or any tumor with regional lymph node involvement [N1] or distant metastasis [M1]).10 A previous diagnosis of cancer did not preclude inclusion in our analysis.

Statistical Analysis

Descriptive statistics are presented for demographic and clinical data. We compared these data among subjects of different marital status using chi-square analysis. We then performed survival analyses using Kaplan–Meier estimates and Cox proportional hazards models. Multivariate models were created utilizing Cox proportional hazards analysis that incorporated marital status with covariates known to be important predictors of survival in bladder carcinoma. In addition to marital status, we included stage, lymph node status, patient age, race/ethnicity, and gender in our multivariate model. To evaluate whether the effect of marital status on survival was modified by other patient characteristics, we performed a separate analysis in which we incorporated interaction terms into univariate and multivariate models. We evaluated the interaction between marital status and both age and race/ethnicity for survival after radical cystectomy. We further analyzed data for subjects missing marital status and stage information. We evaluated demographic data for this subset and created multivariate models that incorporated missing status as a dummy variable for both stage and marital status to determine whether exclusion of this subset biased our outcomes. All statistical analyses were performed with SAS software, version 8.02 (SAS Institute Inc., Cary, NC).


Of 7272 patients in the SEER database who underwent radical cystectomy for transitional cell carcinoma of the bladder, 5854 had complete information on disease stage and marital status. Table 1 presents demographic and clinical data for the study sample. Married subjects tended to be older, male, white, and to have earlier disease stage at cystectomy. Widowed subjects were older and, along with single subjects, tended to have later stage disease. In addition, a predominance of single subjects was female, the only category in which female gender comprised the majority.

Table 1. Demographic and Clinical Characteristics of the Study Sample (n = 5854)
CharacteristicsNo. of married patients (%)No. of separated/divorced patients (%)No. of widowed patients (%)No. of single patients (%)P value
  1. AJCC/UICC: American Joint Committee on Cancer/International Union Against Cancer.

Overall no. of patients3976531786561 
Age in yrs     
 < 50186 (4.7)57 (10.7)7 (0.9)99 (17.7)< 0.001
 50–59633 (15.9)119 (22.4)26 (3.3)135 (24.1) 
 60–691418 (35.6)199 (37.5)181 (23.0)173 (30.8) 
 70–791458 (36.7)141 (26.6)387 (49.3)113 (20.1) 
 ≥ 80281 (7.1)15 (2.8)185 (23.5)41 (7.3) 
 Male3241 (81.5)355 (66.9)477 (60.7)145 (25.9)< 0.001
 Female735 (18.5)176 (33.1)309 (39.3)416 (74.1) 
 White3436 (86.4)436 (82.1)668 (85.0)446 (79.5)< 0.001
 African American133 (3.4)41 (7.7)50 (6.4)62 (11.1) 
 Hispanic167 (4.2)36 (6.8)30 (3.8)28 (5.0) 
 Other240 (6.0)18 (3.4)38 (4.8)25 (4.4) 
 I566 (14.2)68 (12.8)86 (10.9)80 (14.3)0.015
 II1332 (33.5)160 (30.1)250 (31.8)166 (29.6) 
 III844 (21.2)124 (23.4)184 (23.4)106 (18.9) 
 IV1234 (31.1)179 (33.7)266 (33.9)209 (37.2) 

With a follow-up of 39.2 ± 37.0 months, 2908 of 5854 subjects were alive at the time of last follow-up (49.7%). Of those who died, 1502 died of bladder carcinoma, 683 died of known causes not related to bladder carcinoma, and 761 died of unknown causes. Of the subjects who died of causes unrelated to their bladder carcinoma diagnosis, the most common cause was cardiovascular disease (206 patients [30.2%]). Nonbladder carcinoma primary malignancies were also common causes of death (161 patients [23.6%]), especially lung carcinoma (61 patients [8.9%]). Other common causes of death included cerebrovascular disease (59 patients [8.6%]) and infection (54 patients [7.9%]).

Kaplan–Meier survival analysis revealed that married subjects had better overall survival than subjects of all other marital status categories (Fig. 1). Table 2 presents a univariate comparison of our subjects' risk of death after radical cystectomy based on marital status. Widowed subjects fared the worst (hazards ratio [HR] =1.49; 95% confidence interval [95% CI], 1.36–1.62). Single subjects also exhibited worse survival compared with married subjects (HR = 1.24; 95% CI, 1.11–1.39). Separated/divorced subjects demonstrated a trend toward worse survival, but the difference was not significant.

Figure 1.

Survival after radical cystectomy stratified by marital status. Filled squares: married patients; X: separated/divorced patients; open circles: single patients; filled diamonds: widowed patients.

Table 2. Univariate Cox Regression Analysis of the Association between Marital Status and Mortality after Radical Cystectomy
CharacteristicsHazards ratio (95% CI)P value
  1. 95% CI: 95% confidence interval.

Marital status  
 Separated/divorced1.09 (0.96–1.24)0.195
 Widowed1.44 (1.31–1.60)0.005
 Single1.20 (1.06–1.36)< 0.001

Multivariate modeling revealed that being married, independent of age at diagnosis, gender, race/ethnicity, disease stage, and lymph node status, is associated with better survival after radical cystectomy (Table 3). In our multivariate model, widowed subjects and subjects who were separated/divorced had worse survival than married subjects, but the difference was not significant. Single subjects, however, were found to have significantly worse survival than married patients (HR = 1.26; 95% CI, 1.11–1.43) when controlling all relevant variables. We found no interaction between marital status and race/ethnicity for survival after radical cystectomy. Conversely, age modified the association between marital status and survival, exclusively among widowed subjects. The largest effect was observed among widowed subjects age < 50 years (HR = 4.54, 95% CI, 1.96–10.52; P < 0.001). However, this subset included only 7 subjects and the 95% CIs cross those of other marital status categories in subjects age < 50 years (HR = 1.31, 95%CI, 0.80–2.16; P = 0.281 for separated/divorced subjects, and HR = 1.43, 95% CI, 0.94–2.17 for single subjects).

Table 3. Multivariate Model of Factors Associated with Mortality after Radical Cystectomy
CharacteristicsHazards ratio (95% CI)P value
  1. 95% CI: 95% confidence interval; AJCC/UICC: American Joint Committee on Cancer/International Union Against Cancer.

Marital status vs. married  
 Separated/divorced1.12 (0.99–1.28)0.082
 Widowed1.09 (0.97–1.22)0.142
 Single1.26 (1.11–1.43)< 0.001
Age in yrs vs. < 50  
 50–591.11 (0.90–1.37)0.315
 60–691.47 (1.22–1.78)< 0.001
 70–791.98 (1.63–2.39)< 0.001
 ≥ 803.22 (2.60–3.98)< 0.001
Female gender vs. male gender1.09 (1.00–1.19)0.049
Race/ethnicity vs. white  
 African American1.26 (1.07–1.49)0.007
 Hispanic0.95 (0.79–1.15)0.602
 Other0.76 (0.64–0.91)0.002
AJCC/UICC stage vs. Stage I  
 II1.13 (0.99–1.30)0.073
 III1.81 (1.58–2.08)< 0.001
 IV2.64 (2.29–3.05)< 0.001
Lymph node positivity1.30 (1.16–1.45)< 0.001

Several covariates were significantly associated with survival. Compared with subjects age < 50 years, those age ≥ 60 years had progressively worse survival rates, with those age ≥ 80 years having the worst survival rate (HR = 3.22, 95% CI, 2.60–3.98). Women were found to have significantly worse survival than men (HR = 1.09, 95% CI, 1.00–1.19). Compared with white subjects, those of other race/ethnicity (which includes Asian, Pacific Islander, and Native American patients) had better survival. African-American subjects had worse survival (HR = 1.26, 95% CI, 1.07–1.49). Disease stage and lymph node status also were found to significantly predict survival. Although mortality increased progressively with increasing stage, the difference in survival between subjects with Stage I and Stage II bladder carcinoma was not significant. Otherwise, the increased mortality for each successive stage was highly significant. Finally, subjects with metastases to regional lymph nodes had significantly worse survival than those with negative lymph nodes, independent of all other factors (HR = 1.30, 95% CI, 1.16–1.45).

Analysis of missing data revealed that exclusion of subjects missing information on marital status or stage may significantly bias our results. Compared with our study sample, subjects for whom data were missing were younger (P < 0.001), were more likely to be white (P < 0.001), and were more likely to be married (P < 0.001). No differences in gender were noted between these groups (P = 0.252). The incorporation of data from these subjects into our multivariate model was found to affect the association between marital status and survival after radical cystectomy significantly. Independent of our covariates, all nonmarried categories were found to have significantly worse survival compared with married subjects (HR = 1.15, 95% CI, 1.03–1.30 [P = 0.017] for separated/divorced subjects; HR = 1.13, 95% CI, 1.03–1.24; [P = 0.014] for widowed subjects; HR = 1.31, 95% CI, 1.16–1.46 [P < 0.001] for single subjects). Covariates significantly associated with mortality after radical cystectomy were consistent with our final model, including African-American race (P < 0.001), advancing age (P < 0.001), later stage of disease (P < 0.001), and metastases to regional lymph nodes (P < 0.001).


Marriages positively affects survival after radical cystectomy. In our cohort, married subjects tended to have lower stage disease at the time of cystectomy, which likely impacted the differential survival observed by univariate analysis. Furthermore, even after controlling disease stage, age, race/ethnicity, and lymph node status, married subjects demonstrated significantly better survival than single subjects. These results suggest that being married confers a beneficial effect that is independent of clinical factors and covariates predisposing patients to poorer disease outcomes.

The impact of marital status on survival has been extensively studied. Results from the National Longitudinal Mortality Study showed an increased relative risk of mortality from all diseases for all nonmarried categories, an effect that decreased with age.11 Similarly, being married conferred a survival advantage in patients with breast and prostate carcinoma.6, 8 Associations between marriage and mortality appear to be independent of health status.12 Given the protracted time course of an indolent malignancy such as prostate carcinoma, spousal support may be expected to influence survival. However, we have demonstrated this effect in bladder carcinoma, a much more aggressive malignancy.

In previous investigations, the beneficial effect of marital status on survival seemed to relate most strongly to lower disease stage at presentation, indicative of an earlier diagnosis of disease.13 Marriage has been associated with early diagnosis in breast, prostate, cervical, and other carcinomas.14–16 This may reflect better adherence to routine physician visits and screening examinations. For example, compared with unmarried subjects, married subjects are more likely to retain a primary care physician and maintain regular physician visits.17 Compliance issues are likely more important in diseases with generally accepted screening protocols, such as breast, prostate, and cervical carcinoma, than in bladder carcinoma, in which symptoms typically prompt evaluation. However, spousal support may encourage married patients to seek medical attention at the initiation of milder symptoms, leading to diagnosis at an earlier stage. Married patients with lung carcinoma are less likely to present with symptoms of advanced-stage disease than unmarried patients.18 Thus, marital status may affect survival through early detection, where compliance with health care screening and care-seeking behavior at the onset of less advanced symptoms lead to an earlier cancer diagnosis.

After a cancer diagnosis, patient outcomes may be impacted by treatment choices and a patient's ability to modify health-related risk behaviors. With regard to treatment, morbidity and mortality are reduced both in high-volume centers and by high-volume providers.19, 20 Iwashyna and Christakis21 defined high-quality hospitals as either university-based tertiary referral centers or those so labeled by US News and World Report and found that, compared with unmarried patients, married patients consistently receive care at higher-quality hospitals, have shorter lengths of stay, and have a decreased propensity to be discharged to a nursing home or transitional care facility. They postulated that the presence of a spouse contributing to decision-making may influence the choice of care facility and associated treatment. With respect to risk behaviors, bladder carcinoma has been intimately associated with cigarette smoking. A systematic review of 15 previous investigations revealed that smoking cessation represents a simple lifestyle change that can significantly alter patient prognosis after treatment for bladder carcinoma.22 Cigarette smoking may be a more modifiable health behavior in married patients and may contribute to the differential mortality rates. Spouses concurrently enrolled in smoking cessation programs are more successful than unmarried patients, and men seem to reap more benefit from spousal support on their chances of successful smoking cessation than women.23, 24

Tumor burden and the efficacy of treatment are most likely the most important prognostic indicators, yet evidence persists that psychosocial factors affect survival independent of tumor characteristics and treatment type. Depression exacerbates all-cause and cardiovascular mortality and may explain the decreased survival among widowed patients.25 Bereaved men and women have a higher relative risk of mortality, an effect that is more pronounced in healthy men than in those debilitated by chronic diseases.26 In our cohort, univariate analysis revealed markedly worse survival among widowed subjects compared with all other marital status categories. However, when marital status was incorporated in a multivariate model with age and disease stage, no difference was noted. Although our widowed subjects experienced increased mortality after radical cystectomy, they also tended to be older and have later disease stage at presentation, which attenuated the effect of the increased mortality in our multivariate model. Depression and negative affect impact other marital status categories beyond our widowed subjects. The dissolution of a marriage has pronounced effects on both mental and physical health. Separated and divorced patients experience increased depressive symptoms and are more likely to use mental health services independent of the quality of the marriage.27

Our emerging understanding of the interaction between the body's response to stress and immune function may explain psychosocial differences in survival. The field of psychoneuroimmunology evaluates immune and endocrine function changes in response to stressful stimuli. Cancer survival is most closely correlated with the nature of the cancer itself, yet the body's ability to react and combat the cancer must play a role. Although no causal relation between psychologic stressors and tumor development has been defined, stress has been positively associated with tumor progression and mortality.28, 29 Most investigations of psychoneuroimmunology in patients with cancer have focused on the role of natural killer (NK) cells involved in tumor surveillance and the innate immune response to infection and malignancy. In women with breast carcinoma, stress has been shown to promote a decrease in NK cell activity and a blunted cortisol response to stimulation.30 Furthermore, patients with breast carcinoma who are either married or perceive a high level of social support have higher levels of NK cell activity, independent of other clinical factors.31 Conversely, those who are separated or divorced have depressed immune function.32 Although speculative, psychoneuroimmunology may provide the link between psychosocial factors that affect cancer biology, tumor progression, and survival.

The current study has several limitations. First, although we included a large sample of patients, the SEER registry is an observational database. Thus, even though we can speculate as to the possible mechanisms behind the relations we found in the current study, we cannot definitively identify mechanisms responsible for the increased survival among married patients. Second, in the current study, we controlled the factors previously identified as independent predictors of survival in bladder carcinoma, such as the sociodemographic variables of age and race/ethnicity and the clinical variables of stage and lymph node status, but we did not include SES as a covariate in our analyses.33 Using the SEER public-use dataset, we would have to ascribe group characteristics to our individual subjects based on county or region of residence, as no personal socioeconomic information is available through the dataset. Although we found a significant association between marital status and survival, incorporating a measure of SES may affect that association. Third, we lack information regarding our subjects' marriage beyond the marital status classification. Within the married cohort, the quality of the marriage can significantly impact health-related quality of life, as patients in partnerships that adjust poorly to a new cancer diagnosis suffer greater depressive symptoms than those in emotionally healthy relationships.34 Among married patients, the negative effect of marital strain may overwhelm the positive effect of intimate social support. In this regard, our results may underestimate the effect of marriage on survival after radical cystectomy. Finally, 19.5% of our initial cohort was excluded because of missing information on marital status and disease stage. The demographics of this cohort was not consistent with our study sample, as the majority of these subjects had characteristics associated with improved outcome: younger age, white race, and married partner status. Multivariate models incorporating data from these subjects enhanced the differential survival for married subjects compared with all nonmarried categories.

Despite these limitations, we found a significant relation between marital status and survival in patients undergoing radical cystectomy for bladder carcinoma. This relation may affect future investigations of interventions directed at bladder carcinoma and mandates attention to balanced treatment groups with respect to marital status in the design of clinical trials. Although intuitively, it would seem that the benefits of spousal support would be more pronounced in more indolent malignancies such as prostate carcinoma, the survival benefit conferred by marriage was significant even in bladder carcinoma, a more aggressive malignancy. The mechanisms of that benefit are unclear. Possible sources include improved access to care manifested as an earlier diagnosis of disease and care in higher-quality facilities, as well as a reduction in risk behaviors that may affect survival. Furthermore, married patients have fewer depressive symptoms than unmarried patients, and depression has been established as an independent risk factor for survival. Finally, our understanding of the relation between psychosocial and physical health is evolving, but work in the field of psychoneuroimmunology may provide explanations for findings associating psychosocial relations with disease outcomes, especially survival.