Mortality rates for breast cancer have declined since 1990. In 2006, there were about 2.5 million women in the United States who had survived breast cancer (diagnosed at any point prior to 1 January 2006) [1]. The survivorship of breast cancer patients depends on many factors, including age at diagnosis, tumor stage, tumor grade, estrogen receptor (ER) status, progesterone receptor status, socioeconomic factors and other non-cancer-related clinical conditions such as preexisting health status, functional status and social connections [2–4]. Research has shown that other preexisting chronic conditions, or comorbidities, also have a significant effect on the overall survival of breast cancer patients [5–7]. The current study assesses the effects of comorbidities on outcomes of breast cancer mortality and mortality from other causes among breast cancer survivors ages 66 and older.

The most commonly used outcome when conducting epidemiologic research on cancer survivorship is all-cause mortality, which, among individuals with breast cancer, is the sum of breast cancer-specific mortality and mortality from other causes. Although assigning a single cause of death is difficult and can be somewhat arbitrary, assessing outcomes for both breast cancer-specific mortality and mortality from other causes can assist healthcare providers and patients with breast cancer in assessing prognostic indicators and other decisions made in primary care. This study describes the effect of select characteristics, including comorbid conditions, on breast cancer-specific mortality compared to mortality from other causes and explores leading causes of death among breast cancer survivors.

Prevalence estimates at the time of diagnosis of breast cancer were 16.3% for previous cancer, 13.0% for diabetes, 12.8% for CVD and 8.8% for COPD, and the prevalence estimates were higher for older women. Women with no comorbidities were more likely than women with comorbidities to be alive at the end of the study period (Table 1). In the overall study population, 36.2% of women died as a result of causes other than breast cancer. Women with the following comorbidities were more likely to die as a result of other causes: CVD (59.2%), COPD (52.2%), diabetes (47.8%) and previous cancer (43.8%).

The distribution of causes of death varied by age and tumor stage (Figure 1). Women diagnosed with higher-stage cancers were much more likely to die as a result of breast cancer than from other causes. Other causes of death were more likely among women diagnosed with stages I and II cancers. At all ages, women diagnosed with stage I breast cancer were more likely to die as a result of CVD than from breast cancer. CVD was also the leading cause of death for women diagnosed with stage II breast cancer who were ages 85 years and older (42.2%) and 75 to 84 years of age (32.2%). Within all cancer stages, as women aged, there was a decrease in the proportion of all deaths that were due to breast cancer.

The proportional distribution of cumulative causes of death was dependent on follow-up time. In the first 5 to 10 years following breast cancer diagnosis, breast cancer was the cumulative primary cause of death. With longer follow-up, CVD became the cumulative primary cause of death (Figure 2). The proportion of deaths due to other causes also increased as follow-up increased. The primary causes of death in this category included deaths due to COPD (3.8% of all deaths), pneumonia (2.8%), Alzheimer's disease (2.2%) and diabetes (2.2%).

Older female Medicare enrollees were more than twice as likely to die as a result of diseases other than breast cancer (36.2% versus 15.1%). Characteristics associated with dying from other causes included older age, higher tumor stage and the presence of comorbidities, whereas characteristics associated with dying as a result of breast cancer included higher tumor-related stage and grade, ER-negative status and the presence of comorbidities. The finding that comorbidities had a significant effect on breast cancer-specific mortality is new and of particular interest, although relative hazards were significantly lower than those for other-cause mortality.

CVD was the leading cause of death in the study population (15.9%), closely followed by breast cancer (15.1%). CVD was the primary cause of death for women of all ages with stage I cancer diagnoses and for women in the two older age groups with stage II cancers. Within each cancer stage group, as women aged, they were more likely to die as a result of CVD and "other" causes and less likely to die as a result of breast cancer, other cancers or COPD. These findings are likely due to the facts that older women are more likely to have comorbidities [19] and that women with comorbidities are more likely to die as a result of causes other than breast cancer. This same age pattern is also seen in nationwide US vital statistics for the general population [20]. As expected, our study population was more likely to die as a result of cancer and less likely to die as a result of heart disease than older females in the general study population.

Our study comprised a long follow-up period (median 105 months). Over time, as women survive their breast cancer, they become less likely to die as a result of their cancer diagnosis. Still, even within the first five years of follow-up, CVD remains the leading cause of death for women of all ages with stage I breast cancers (data not shown).

Of those women who died as a result of CVD in our study, only 25.5% were also categorized as having CVD as a comorbid condition at the time of their breast cancer diagnosis. Conversely, of those women who had CVD as a comorbid condition at the time of their breast cancer diagnosis and who died during the study period, 41.9% died as a result of CVD. These percentages are low for several reasons. Most importantly, there was a long period of follow-up in this study, which allows for a substantial amount of time for women to develop CVD subsequent to our detection of the condition at the time of the breast cancer diagnosis. Chemotherapy has been shown to be associated with long-term cardiac toxicity in breast cancer patients, particularly in older patients [21]. In addition, there is likely to be underdiagnosis of CVD and/or lack of reporting in the Medicare record. Underdiagnosis and undertreatment of CVD may be important issues for women diagnosed with cancer, as their cancer diagnosis can be perceived by both the patient and her caregivers as the overriding medical priority.

We found four studies that have assessed the effect of patient characteristics apart from tumor characteristics on cause-specific breast cancer mortality. Two of these studies determined that patients over 65 or 70 years of age were more likely than younger patients to die as a result of other causes [22, 23]. Another study found that women with three or more comorbid conditions had a 20-fold higher rate of non-breast cancer mortality and a fourfold higher rate of all-cause mortality than did women diagnosed with breast cancer without comorbidities [24]. The third study assessed the role of comorbidity in explaining the shorter survival of black breast cancer patients and found that black women diagnosed with breast cancer were more likely to die as a result of causes other than breast cancer [25]. We also found that black women were slightly more likely to die as a result of other causes, but, after adjustment for cofactors, we found that black women had survival rates similar to those of white women.

In the current study, we detected a positive association between comorbidities and breast cancer-specific mortality. We are not aware of a previous study that has highlighted this relationship. There could be several potential explanations for these associations. Women with comorbidities are less likely to be treated or may be more likely to experience adverse effects of therapy and not complete their treatment regimens. Therefore, women with comorbidities could be more likely to die as a result of breast cancer. In addition, there may be misclassification in cause-of-death coding, and women with metastatic disease may be more likely to have their cancer listed as the underlying cause of death when another factor may have been the true primary cause. A surprising finding in our study was the significant association between tumor stage and mortality as a result of other causes. Again, it is not clear whether these findings indicate that women with higher-stage tumors are at higher risk of non-cancer-related mortality and women with comorbidities are at higher risk of breast cancer-specific mortality or whether these findings are due to misclassification in cause-of-death coding.

There can be considerable controversy surrounding the determination of a specific cause of death, so our study is limited by the validity of death coding. However, mortality data have been regulated for over 50 years by the World Health Organization's International Classification of Diseases [26]. Standard forms and model procedures are used for the collection of data and registration of events. One of the goals of this standardization is to have death certificates that are comparable both internationally and nationally. Nonetheless, the validity of cause-of-death determination is questionable and has been found to differ across cancer types, time periods and ages at death [27]. In addition, data from the Framingham Heart Study [28] have indicated that heart disease is overestimated as a cause of death in the general population. If this is true in our study population, then we cannot conclude that CVD is the primary cause of death, although the rates of CVD and breast cancer are likely very similar in this study population of older women diagnosed with breast cancer.

The current study is also limited by the fact that Medicare data were created for billing purposes and may not be optimal for clinical research. Recipients of Medicare may seek care outside their Medicare coverage, such as long-term care, oral prescriptions of tamoxifen or hormonal therapy. Treatments of both breast cancer and comorbid conditions have improved since the beginning of the period covered by our study. Information regarding medical history prior to enrolling in Medicare is also not available. The measurement of comorbidities does not consider the severity illness or the duration of diagnosis, and comorbidities that are present but not covered by Medicare would result in misclassification of comorbidity status. In addition, data on possible confounding factors, such as smoking, diet and functional status, are not available in the database.

A sizable portion (34%) of the eligible study participants were excluded from this study, mainly because they were enrolled in HMOs, for which Medicare files are not available. Differences do exist between these two Medicare healthcare delivery systems. Patients enrolled in HMOs are generally younger, healthier, diagnosed at an earlier stage and have better overall survival than FFS patients [9, 29]. There is no reason to believe that the differences between HMO and FFS patients would bias the findings observed in our study, but our results are most generalizable to older adults who are FFS patients.

In addition, there are a number of records with missing or unknown values for tumor stage and grade as well as ER status. These missing covariates are related to the study outcomes. Women who died during the study period were more likely to have unknown tumor stage and grade, ER status and surgery status. Missing tumor characteristics (tumor stage and grade as well as ER status) do not appear to be related to the specific cause of death (categorized as "Breast cancer" or "Other cause"). However, women with unknown surgery status were more likely to die as a result of breast cancer. Although the missing data are not random, they are an intrinsic characteristic of surveillance databases. Analyses done to account for missing data did not influence the study findings or conclusions.

Many older women diagnosed with breast cancer die as a result of comorbid conditions rather than from breast cancer. Other studies have found that comorbidities are associated with increased risk of mortality from other causes, and in our study we found that women with comorbidities were more likely to die as a result of breast cancer as well. We found that after 12 years of follow-up, older women diagnosed with breast cancer were almost equally likely to die as a result of breast cancer as they were to die as a result of CVD. Patient management rightfully focuses on cancer after diagnosis, but consideration of other existing comorbid conditions should also be integrated into patient management and recovery plans [30–32]. Especially among older breast cancer survivors, risk management of factors associated with CVD can help improve overall survival.