Fax: (011) 45 35 25 77 31
Increased risk of severe depression in male partners of women with breast cancer†
Version of Record online: 27 SEP 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 23, pages 5527–5534, 1 December 2010
How to Cite
Nakaya, N., Saito-Nakaya, K., Bidstrup, P. E., Dalton, S. O., Frederiksen, K., Steding-Jessen, M., Uchitomi, Y. and Johansen, C. (2010), Increased risk of severe depression in male partners of women with breast cancer. Cancer, 116: 5527–5534. doi: 10.1002/cncr.25534
We thank Visti B. Larsen and Aslak H. Poulsen of the Institute of Cancer Epidemiology, Danish Cancer Society, for their assistance with the statistical analyses.
- Issue online: 23 NOV 2010
- Version of Record online: 27 SEP 2010
- Manuscript Accepted: 14 JUN 2010
- Manuscript Revised: 3 JUN 2010
- Manuscript Received: 19 FEB 2010
- affective disorder;
- breast cancer;
- cohort study;
A few small studies published to date have suggested that major psychosocial problems develop in the partners of cancer patients; however, to the authors' knowledge, no studies to date have addressed their risk for severe depression. In a retrospective cohort study, the risk for hospitalization with an affective disorder of the male partners of women with breast cancer was investigated, using unbiased, nationwide, population-based information.
Followed were 1,162,596 men born between 1925 and 1973 who were aged ≥30 years at study entry, resided in Denmark between 1994 and 2006, had no history of hospitalization for an affective disorder, and had lived continuously with the same partner for at least 5 years. A Cox regression analysis included detailed clinical information regarding the diagnosis and treatment of breast cancer and on annually updated socioeconomic and health-related indicators obtained from national administrative and disease registers.
During the 13 years of follow-up, breast cancer was diagnosed in the partners of 20,538 men. On multivariable analysis, men whose partner was diagnosed with breast cancer were found to be at an increased risk of being hospitalized with an affective disorder (hazards ratio, 1.39; 95%confidence interval, 1.20-1.61), with a dose-response pattern for the severity of breast cancer. Furthermore, men whose partner died after breast cancer had a significant, 3.6-fold increase in risk for an affective disorder when compared with men whose partner survived breast cancer.
The results of the current study supported the hypothesis that men whose partner had breast cancer were at an increased risk for hospitalization with an affective disorder. Cancer 2010. © 2010 American Cancer Society.
During the past few decades, our understanding of the consequences of cancer has changed, with recognition not only of physical but also of psychological and social effects.1, 2 In response, various supportive psychosocial intervention strategies have been designed, tailored to the problems that cancer patients face during the course of their disease.3-6 The extent to which cancer affects not only patients but also their closest relatives was first addressed in a seminal article published more than 20 years ago, in which House et al illustrated how several diseases can influence people close to the patient.7 The mechanism of this effect may involve several interacting pathways: the event may cause stress in the partner; it might deprive the partner of emotional, social, and economic support; and it can influence the daily life and behavior of the partner.7, 8
The effect of cancer on the psychological well-being of the partner could result in increased risks of several psychiatric disorders related to stressful life events, including neurotic, stress-related, somatoform, substance abuse-related, and affective disorders.
A recent population-based retrospective cohort study including 11,000 spouses of cancer patients indicated that risk of psychiatric diseases was increased among spouses of cancer patients.9 In keeping with these findings, some of the previous studies on depression after a diagnosis of cancer in a spouse support an association,10-13 whereas others have reported negative or null findings.14-16 The majority of the studies had several limitations, including small sample sizes of spouses with cancer (<200 subjects),10-16 and only a few small studies did include a relevant comparison group.12, 13 To the best of our knowledge, none of the previous studies regarding the risk of depression in partners of cancer patients used an objectively defined state of depression, diagnosed years before the study was conducted, and obtained from an administrative, population-based and nationwide database to reduce the risk for misclassification of the outcome under study as well as recall and information bias.10-16
In a retrospective study with a cohort design, based on a nationwide sample and register-based information, we carefully examined the risk for hospitalization for an affective disorder among the male partners of women in whom breast cancer had been diagnosed. We focused purely on the affective spectrum because these disorders are severe, are commonly associated with stressful life events, and more systematically lead to a psychiatric diagnosis or hospitalization compared with other possibly stress-related disorders.
MATERIALS AND METHODS
Since 1968, all Danish residents have been registered in the Central Population Register and assigned a unique personal identification number that contains their date of birth and sex. Individual information is kept under this identification number in all national registries, ensuring accurate linkage of information among these registries.17 We identified all 3.47 million persons who were born between 1925 and 1973, who resided in Denmark in the period between 1994 and 2006, and who entered the cohort at aged ≥30 years without a previous cancer. Of the total of 1,628,651 men identified, 1,170,582 had been either married or cohabiting with a partner continuously for at least 5 years at the time of study entry. Cohabitation was included because in Denmark, >20%of all couples living together in 2002 were unmarried.18 We defined cohabitation in our context as 2 persons of the opposite sex, aged >16 years, with a maximum age difference of 15 years, living at the same address in the absence of marriage, with no other adult in residence. Information regarding cohabitation status as well as death and emigration was obtained through linkage to the Central Population Register.
We identified women in whom breast cancer had been diagnosed through linkage with the Danish Breast Cancer Cooperative Group (DBCG). Since its establishment in 1977, nearly all new breast cancer patients have been registered in the DBCG database.19 The DBCG has prepared national guidelines for diagnostic and therapeutic procedures in patients with primary invasive breast cancer, which were later supplemented by guidelines for in situ carcinomas and hereditary breast cancer.19 The information on the breast cancers includes date of primary surgery, tumor size (in mm), number of tumor-positive lymph nodes, estrogen and progesterone receptor status, chemotherapy and/or hormone therapy, radiotherapy, and date of disease recurrence and/or death.
Information on hospitalization with affective disorders was obtained from the nationwide Danish Psychiatric Central Register,20 which contains information regarding all admissions to Danish psychiatric inpatient facilities, both psychiatric hospitals and psychiatric departments in general hospitals, and, since 1995, information from outpatient contacts. The admission record includes the hospital department, the date of admission, the date of discharge, 1 primary psychiatric diagnosis, and up to 3 auxiliary diagnoses. The affective disorders were classified according to the eighth edition of the International Classification of Diseases (ICD-8)21 between 1969 and 1993 and according to ICD-1022 since 1994 by the following codes: ICD-8 codes 296.09 to 296.99, 298.09 to 298.19, 300.19, and 300.49; and ICD-10 codes F30 to F39.
Socioeconomic and Demographic Indicators
Information regarding the socioeconomic characteristics was obtained from the population-based Integrated Database for Labour Market Research in Statistics Denmark, which contains yearly data since 1980. The core variables in the database are derived by linkage with the Central Population Register, the taxation authorities, the Register for Education Statistics, the Register Relating to Unemployment, and a register of all companies with >1 employee. For all persons in the study population, we obtained information at the individual level regarding several demographic and socioeconomic variables for each year of the study period.23, 24 Four indicators were defined: number of children living at home ages birth to 17 years (0, 1, or ≥2), highest attained educational level (basic or high school, vocational education, higher education, or unknown), disposable household income (lowest [1st quartile], middle [2nd-3rd quartile], or highest [4th quartile]), and affiliation to the work market (working, unemployed or other, early retirement, or unknown). We obtained information concerning the income of all family members to estimate total family income.
Using the Danish National Patient Register25 and the Psychiatric Case Register,20 we obtained full histories of any disease leading to hospitalization from 1978 and, beginning in 1995, also outpatient visits by each study participant through 2006. The information in these registers includes the dates of admission and discharge and diagnoses coded according to the Danish modified versions of the ICD-821 and, from 1994 onward, ICD-10.22 On the basis of information from these registers, we defined 3 health-related indicators to account for the presence of chronic somatic and psychiatric comorbidity: the Charlson index,17, 26 alcohol-related mental disorders (ICD-8 codes 291.09-39, 291.99, 303.09-29, 303.99, and 393.09; and ICD-10 code F10), and schizophrenia and other psychoses (ICD-8 code 295 and ICD-10 codes F20 and F25).
Of the 1,170,582 men who had been living with a partner for ≥5 years, we excluded 7951 men who had had an affective disorder (n = 7216), schizophrenia, or another psychosis (n = 735) before the beginning of follow-up and 35 men who were censored (because of death, emigration, or remarriage) before the beginning of follow-up, leaving 1,162,596 men (99.0%) for the analyses.
We used Cox regression analysis to assess the risk for hospitalization with an affective disorder after the diagnosis of breast cancer in a partner. The hazard ratio (HR) for hospitalization with an affective disorder after having experienced breast cancer in a partner compared with not having experienced breast cancer in a partner was estimated using the PHREG procedure in the SAS statistical software package (version 9.1; SAS Institute, Cary, NC). Follow-up time was counted from January 1 of the year in which the man had lived continuously with the same partner for 5 years, January 1 from the year the man turned 30 years old, or January 1, 1994, whichever came last and until the date of hospitalization with an affective disorder; death; death of partner; emigration; onset of schizophrenia; December 31, 2006; or new cohabiting partner, whichever came first. The exposure variable (ie, the diagnosis of breast cancer in the partner) was included as a time-dependent variable. Thus, person-time before the partner's breast cancer diagnosis was counted as unexposed, whereas person-time after the breast cancer diagnosis was counted as exposed.
The HRs were adjusted for number of children, highest attained educational level, disposable household income, affiliation to the work market, Charlson index, and history of alcohol-related mental disorders. Information regarding these factors (except for age) was extracted 2 years before the man's date of entry to avoid misclassification due to, for example, a change in socioeconomic or health status that might be related to the presence of an as-yet undiagnosed breast cancer. To examine the effect and potential effect modification of the factors included as confounders, we conducted separate analyses of affective disorders according to the adjustment factors. We determined whether the association between the partner's breast cancer and the risk of hospitalization with an affective disorder was affected by interactions with these variables. To assess whether the results were influenced by change in cohabiting partner, we also conducted analyses with no censoring at the date of new cohabiting partner. In a sensitivity analysis of HRs of hospitalization for an affective disorder according to time since entry, we examined HRs in 3 intervals of follow-up after the partner's breast cancer diagnosis: 1 year, 2 to 3 years, and 4 to 13 years of follow-up. To assess whether the risk for hospitalization with an affective disorder was affected by the severity of the breast cancer, we also estimated HRs for hospitalization with an affective disorder according to breast cancer tumor size, number of tumor-positive lymph nodes, estrogen and progesterone receptor status, chemotherapy and/or hormone therapy, and radiotherapy. In analyses of tumor size and number of tumor-positive lymph nodes, P values for the linear trends were calculated as a continuous variable in the regression model (tumor size in mm and number of tumor-positive lymph nodes). Among the men experiencing a diagnosis of breast cancer in their partner, we assessed the HRs for hospitalization with an affective disorder according to the partner's disease recurrence or death after the breast cancer diagnosis, respectively. In these analyses, follow-up time was counted from the date of the partner's breast cancer diagnosis until date of death, emigration, new cohabiting partner, onset of schizophrenia, hospitalization with an affective disorder, or December 31, 2006, whichever came first. When examining breast cancer recurrence as the exposure variable, person-time before disease recurrence was counted as unexposed, person-time after disease recurrence was counted as exposed, and censuring took place at death of the partner. When examing death after breast cancer in the partner as the exposure variable: person-time before date of death of the partner was counted as unexposed and person-time after date of death was counted as exposed.
In all analyses, time since entry into the study was used as the time scale and the baseline HR was stratified by age at entry (30-34, 35-39, 40-44, 45-49, 50-54, 55-59, 60-64, 65-69, and ≥70 years). All P values are 2-sided, with a 2-sided α error level of 5%.
Of the 1,162,596 men, a total of 11,855,300 person-years of follow-up were accrued, yielding a median follow-up of 13 years (range, 0-13 years). We identified 20,538 male partners of women diagnosed with breast cancer. Of the male partners, a total of 12,365 were hospitalized for an affective disorder and among these men, a total of 180 had a partner who was diagnosed with breast cancer.
The socioeconomic, demographic, and health-related indicators are listed in Table 1. A high percentage of the men had no children, attained vocational education, were working, and had no comorbidity.
|Mean/median age (SD) y||45/44 (11)|
|No. of children (%)a|
|Highest attained educational level, no. (%)a|
|Basic or high school||369,364 (32)|
|Vocational education||509,566 (44)|
|Higher education||260,257 (22)|
|Disposable income, no. (%)a|
|Lowest (1st quartile)||232,545 (20)|
|Middle (2nd-3rd quartile)||605,369 (52)|
|Highest (4th quartile)||324,682 (28)|
|Affiliation to work market, no. (%)a|
|Unemployed or other||104,665 (9)|
|Early retirement||29,150 (3)|
|Charlson index, no. (%)a|
|Alcohol-related mental disorders, no. (%)ab|
The clinical characteristics of the breast cancers are listed in Table 2. A high percentage of women with breast cancer had tumors that measured 11 to 20 mm, had no tumor-positive lymph nodes, and received chemotherapy and/or hormone therapy as well as radiotherapy.
|Clinical Characteristic||No. (%)|
|Tumor size, mm|
|No. of tumor-positive lymph nodes|
|ER and PR statusa|
|Chemotherapy and/or hormone therapy|
On multivariable Cox proportional regression analysis, men whose partner was diagnosed with breast cancer were found to be at an increased risk of being hospitalized with an affective disorder compared with men whose partner was not diagnosed with breast cancer (HR, 1.39; 95% confidence interval [95% CI], 1.20-1.61) (Table 3). In analyses without censoring at the date of new cohabiting partner, only small differences in results were observed (HR, 1.39; 95% CI, 1.20-1.61).
|Person-y of follow-up||11,724,718||130,582|
|No. of affective disorders||12,185||180|
|Multivariate HR (95% CI)||1.00 (reference)||1.39 (1.20-1.61)|
When we examined the risk of hospitalization for an affective disorder according to different levels of clinical variables among the women with breast cancer, we identified a consistently increased risk of affective disorders concurrently with an increase in the disease severity, especially for tumor size (P for linear trend = .003) and number of tumor-positive lymph nodes (P for linear trend = .002) (Table 4). Furthermore, we identified a consistently increased risk for hospitalization with affective disorders that did not appear to be influenced by receiving chemotherapy and/or hormone therapy and radiotherapy. In the subgroup of men whose partner was diagnosed with breast cancer, the death of the breast cancer partner was associated with a 3.6-fold significant increased risk of hospitalization for an affective disorder. Experiencing disease recurrence in the partner was also found to be significantly associated with an increased risk of an affective disorder (Table 5).
|Tumor Size, mm||0-10||11-20||21-50||≥51|
|Person-y of follow-up||11,724,718||22,129||56,818||46,712||4923|
|No. of affective disorders||12,185||27||79||66||8|
|Multivariate HR(95% CI)||1.00 (reference)||1.26 (0.86-1.84)||1.41 (1.13-1.76)||1.40 (1.10-1.79)||1.64 (0.82-3.27)|
|P for linear trendb||.003|
|No. of Tumor-Positive Lymph Nodes||0||1-3||≥4|
|Person-y of follow-up||11,724,718||72,723||37,633||20,226|
|No. of affective disorders||12,185||95||48||37|
|Multivariate HR(95% CI)||1.00 (reference)||1.30 (1.06-1.60)||1.32 (0.99-1.75)||1.85 (1.34-2.55)|
|P for linear trendb||.002|
|ER and PR Statusc||Positive||Negative||Unknownd|
|Person-y of follow-up||11,724,718||27,357||91,233||11,992|
|No. of affective disorders||12,185||45||116||19|
|Multivariate HR(95% CI)||1.00 (reference)||1.67 (1.24-2.23)||1.29 (1.07-1.54)||1.56 (0.99-2.44)|
|Chemotherapy and/or Hormone Therapy||None||Received||Unknownd|
|Person-y of follow-up||11,724,718||51,412||78,614||556|
|No. of affective disorders||12,185||70||109||1|
|Multivariate HR(95% CI)||1.00 (reference)||1.33 (1.05-1.68)||1.43 (1.18-1.73)||2.07 (0.29-14.66)|
|Person-y of follow-up||11,724,718||68,282||62,300|
|No. of affective disorders||12,185||96||84|
|Multivariate HR(95% CI)||1.00 (reference)||1.37 (1.12-1.68)||1.41 (1.14-1.75)|
|No recurrence||Recurrence||No death||Death|
|Person-y of follow-up||109,870||21,191||115,780||14,792|
|No. of affective disorders||139||41||122||58|
|Multivariate HR (95% CI)||1.00 (reference)||1.54 (1.09-2.19)||1.00 (reference)||3.63 (2.64-5.01)|
In the sensitivity analysis of hospitalization for an affective disorder according to time since entry, the HR was 1.36 (95 %CI, 0.85-2.17; N = 23) for the first year of follow-up, 1.40 (95% CI, 1.03-1.90; N = 48) for 2 to 3 years of follow-up, and 1.66 (95% CI, 1.38-1.99, N = 109) for 4 to 13 years of follow-up. The risk for hospitalization with an affective disorder was not found to be significantly affected by adjustment for covariates (data not shown).
In this cohort study based on a large, nationwide, population-based sample, with extensive control for confounders, we found that the male partners of women in whom breast cancer has been diagnosed were at a significantly increased risk for severe depression. The observed association might be explained by stress and a lack of social support.27 Breast cancer in the partner might also lead to changes in the lifestyle of the male partner, which could affect the risk for affective disorders.28 It has been very well established that family caregivers in general and those in oncology in particular have a high risk of depression because they experience chronic stress that affects their physical state.29, 30
In keeping with the results of the current study, 2 case-control studies of the risk for depression in the partners of cancer patients that had small samples but relevant comparison groups also found a significantly increased risk in the exposed group.12, 13 Although the outcomes of these 2 studies are less marked than in the current study, the results support the hypothesis that a partner's cancer diagnosis has an effect on the spouse's risk for depression.
We identified a consistently increasing risk of affective disorders based on breast cancer severity (tumor size, number of tumor-positive lymph nodes, and disease recurrence or death after the breast cancer diagnosis). The severity of the breast cancer influences the risk of the physical and psychosocial effects of treatment in the woman, such as lymphedema,31 cardiovascular disease,32 fatigue,33 depression,33 and anxiety.33 If the partner with breast cancer experiences late effects, develops disease recurrence, or dies, these events each may increase the psychological burden on the man. In keeping with the results of the current study, a community-based cohort study of 1046 elderly individuals (aged ≥65 years), of whom 139 had experienced widowhood, indicated that widowhood was associated with an increased risk of depression.34 Two follow-up studies also demonstrated a significantly increased risk of depression among individuals whose partner had cancer in the terminal stage.10, 15 Thus, the effect of breast cancer in a partner on the man's risk for hospitalization with depression might depend both on the burden of late effects of the breast cancer and treatment and the overall prognosis.15
Previously, we conducted a nationwide, population-based study to determine whether cancer is followed by an increased risk of divorce or separation in cohabiting partners, which might explain the increased risk of severe depression. We found an increased risk of divorce only among women with cervical cancer.35
We studied the risk of affective disorders severe enough to warrant hospitalization. Because such events are rare, cases in both the exposed (137.8 per 100,000 person-years) and the unexposed (103.9 per 100,000 person-years) groups were small. Nevertheless, the diagnosis of an affective disorder by a psychiatrist followed by hospitalization ensures that the risk of misclassification of the outcome under study is minimal. We speculate that the effect would have been even stronger if we had used outcomes such as antidepressant usage, referral to a psychologist, or reduced quality of life.
The current study has several advantages, including the design, which minimized selection bias. The data were obtained from public administrative registers established years before the study was initiated, thereby leaving little room for information bias. The register-based data regarding hospitalization for affective disorders and the clinical data concerning breast cancer provided precise information on the timing of both exposure and outcome. Cases of breast cancer are recorded by the DBCG only if a detailed pathology report is available, thus minimizing the risk of misclassification of the exposure. Access to nearly complete clinical data on the breast cancer cases made it possible to examine the dose-response relation by disease severity, and the findings further supported the results. In addition, we were able to adjust for important confounding factors and to include both officially married and cohabitating couples, which is relevant in Western countries such as Denmark, in which a substantial proportion of couples are not married.18 Men in this study were followed for a maximum of 13 years, and the results demonstrated that men whose partner had been diagnosed with breast cancer were at a similarly increased risk of hospitalization for an affective disorder throughout the follow-up period.
The current study also had some limitations. The men in this study were cancer free at baseline. Cancer patients' partners may be more likely to have cancer.36 Thus, exposed men in this study could potentially have been more likely to be diagnosed with cancer and affective disorders after a cancer diagnosis in a partner than unexposed men.37 The current study may have underestimated the risk of affective disorders. We had no information regarding lifestyle factors such as alcohol consumption, physical activity, or dietary habits that might be associated with the risk of depression as well as possibly breast cancer in the partner due to shared lifestyle.38, 39 The fact that these potential confounding factors were not taken into account might have resulted in an overestimation of the true effect. This study focused only on the partners of breast cancer patients, because we had access to detailed clinical information for this group of patients on a nationwide and population-based basis. We conducted a further analysis focusing on the partners of patients with cancer at any sites using the Danish Cancer Registry dataset with follow-up until December 31, 2003. We followed 1,116,829 men born between 1925 and 1973 who were aged ≥30 years at the time of study entry and resided in Denmark between 1994 and 2003. During the 10 years of follow-up, we identified 42,683 male partners of women in whom any cancers had been diagnosed. Of the male partners, a total of 8198 were hospitalized for an affective disorder. On multivariable analysis, men whose partners were diagnosed with any cancer were found to be at an increased risk of being hospitalized with an affective disorder (multivariate adjusted HR, 1.57; 95% CI, 1.38-1.79). Thus, we conclude that the results of the current study may be generalized to the partners of any cancer patient, and we hypothesize that the partners of patients with cardiovascular disease and other severe, chronic disorders might also be at an increased risk for affective disorders.40 A further potential limitation in the current study is lack of information regarding cause of death, which may have influenced our results. In some elderly people, death may occur from non-cancer-related, even in those in whom breast cancer was previously diagnosed. Lastly, the results of the current study may be due to chance. Our study was large but included few cases, and we conducted several statistical analyses; in addition, to our knowledge, no other studies to date have used hospitalization with depression as the outcome. Thus, further studies are needed to confirm the results of the current study.
We conclude that a diagnosis of breast cancer affects not only the life of the patient but may also seriously affect their partner. Such interpersonal effects on health challenge our understanding of cancer as a physical disease, because it can also have important social and psychological effects. We suggest that some type of screening for depressive symptoms41 in the partners of cancer patients in general and those of breast cancer patients in particular might be important for preventing this devastating consequence of cancer.
CONFLICT OF INTEREST DISCLOSURES
Supported by grants from the Japan Society for the Promotion of Science Postdoctoral Fellowships for Research Abroad (Japan), the Foundation for Promotion of Cancer Research (Japan) for the 3rd Term Comprehensive 10-Year Strategy for Control, Southern Danish University, and the Danish Cancer Society (Denmark).
- 8Social Epidemiology. New York: Oxford University Press; 2000.,
- 18Official Web site of Denmark. Available at: URL: www. denmark.dk. Accessed February 5, 2010.
- 21World Health Organization. International Classification of Diseases. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. 8th rev. Geneva: World Health Organization; 1967.
- 22World Health Organization. International Statistical Classification of Diseases and Health Related Problems. Geneva: World Health Organization; 1992.
- 23Eurostat/Statistics Denmark. Statistics on Persons in Denmark. A Register-Based Statistical System. Brussels, Belgium: Office for Official Publications of the European Communities; 1995.
- 31Breast cancer-related lymphedema: attention to a significant problem resulting from cancer diagnosis. J Clin Oncol. 2008; 26: 5666-5667.