This prospective study contrasted health outcomes and healthcare utilization for a cohort of breast cancer patients over five years following their cancer diagnosis with those of controls matched by age, geographic location and total comorbidity burden. Non-breast cancer comorbidity at baseline and five-year survival were similar in the two cohorts at baseline. Although similar in all other years, non-breast cancer comorbidity was significantly higher for cases in the first year following breast cancer diagnosis. Among survivors in each cohort, hospital utilization exhibited the same pattern, with more days of hospitalization and Medicare payments in the first follow-up year, but similar amounts subsequently. In contrast, both numbers of physician visits and total Medicare payments for outpatient services (as measured by Medicare payments) were notably higher for cases—by over 25%—during most of the five follow-up years. Quality of care, as measured by guideline-adherent monitoring for those with baseline CVD and DM, and bone density testing for all patients, was also similar in the groups. However, colorectal cancer screening guideline adherence was better for cases.
Our estimates of utilization following breast cancer diagnosis are like those of Yabroff et al. in which Medicare payments in the first year following breast cancer diagnosis were $11,728 (1999–2003), and higher than for a cohort matched by age and geographic location, but not comorbidity.25
While the corresponding estimate from our study was $8,034, these figures are not directly comparable due to differences in control selection, study time-period, and geography. However, we estimated inpatient costs for our cases to be 27% of total Medicare payments for the cases, which is similar to the Yabroff et al. estimate of 25%. Note that Medicare payments serve as a good proxy for inpatient and outpatient health care services as Medicare is the primary payer, accounting for over 86% of such expenditures for the elderly.26
Some studies have found greater comorbidity among cancer patients than among age-similar non-cancer patients.5–7
Therefore, when breast cancer patients are compared with non-cancer patients matched only on age and location,25,27,28
the cancer patients could have higher healthcare utilization for this reason alone. In this study we matched by baseline comorbidity in addition to age, race and geography; this enabled us to ask if the breast cancer diagnosis and treatment itself have longer term effects on healthcare utilization? While inpatient care utilization was similar between the two cohorts, numbers of physician visits and Medicare payments for outpatient care were consistently greater among breast cancer survivors—by over 100% in the first year and at least 25% in all four subsequent years. Breast cancer survivors visited cancer specialists more often, but numbers of visits to mental health specialists and generalists were similar. Some of this difference may be due to a “volunteer effect”, in which those who agree to enroll in a study are more likely to engage in systematic care-seeking than the controls who were selected by retrospective matching; the control group contains both women who would have and those who would not have accepted study enrollment had it been offered.29,30
Greater outpatient care among breast cancer survivors could also reflect long-term and late complications of cancer treatment,13,31
since comorbid conditions and their treatment may interact with cancer and its treatment, leading to worse physical health and higher mortality risk.6,32,33
However, survival for at least five years following diagnosis was quite similar for cases and controls, as was measured comorbidity levels in most years. Hence, more physician visits and outpatient care among breast cancer survivors does not appear to be due to their being sicker. More likely is that increased outpatient care is associated with breast cancer follow-up care. Indeed, the Institute of Medicine report From Cancer Patient to Cancer Survivor: Lost in Transition
focuses on the complex follow-up needs of cancer survivors, including preventive care, monitoring for treatment side effects (e.g., adjuvant hormonal therapy) and recurrence surveillance.9
That is, breast cancer patients, after being drawn into the caregiving network, are likely to remain engaged. Several studies have examined whether breast cancer modifies care-seeking for comorbid conditions among older adults.11,16,17,34–36
We examined colorectal cancer screening and bone density testing for all patients, and monitoring for two important chronic condition subgroups—those with CVD and DM. We found no differences between cases and controls, except in the case of colorectal cancer screening. Similar or better quality of care among breast cancer survivors is consistent with our finding that they had more physician visits compared to the controls. Differing study design makes comparisons with previous reports difficult. Snyder et al. found that in each of the five years of follow-up, breast cancer survivors had less colorectal cancer screening, bone density and lipid testing than matched controls.36
However, their controls were chosen from women who had had a mammogram during the baseline year, making it likely that their overall quality of care was also above average. In contrast, Earle et al. found that breast cancer survivors with diabetes had higher rates of lipid testing than matched controls.17
Since comorbidity was not a matching criterion, this could have been due to differences in comorbid disease burden. Ultimately it has not been clear whether breast cancer survivors receive either more or less chronic disease care than similarly-ill women with the same morbidity burden. Keating et al., which also matched controls by comorbidity, found any-cancer survivors with diabetes receiving diabetes screening “of generally similar quality” as non-cancer diabetics.16
This is consistent with our findings.
Previous studies have examined breast cancer survivors retrospectively;17,37
in contrast, we have been able to examine survival prospectively. The estimated five-year survival for this breast cancer cohort (79.6%), is lower than that for National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) regions (89.6%; 1996–2003).38
This could partly be due to the difference in populations represented—our subjects were from four selected geographic areas, while SEER data is obtained from areas containing over 26% of the national population. Further, because we were also interested in studying utilization, we examined only fee-for-service Medicare beneficiaries.
This study has several limitations. The study population was clustered in four geographic areas and may not generalize nationally. The breast cancer cohort had volunteered for a study; they may be healthier, or more prone to positive health behaviors, than other survivors, or a matched cohort selected from the general population.29,30
Also, we used only Medicare administrative data, with limited clinical detail on comorbidity severity and no pharmacy data; since patient comorbidity was identified from diagnoses coded on claim forms, breast cancer survivors’ greater interaction with health care providers may partly explain their higher measured comorbidity. Both cases and controls participated in Medicare FFS; in this sense they were similar, but not fully representative of Medicare enrollees, upwards of 20% of whom are enrolled in managed care plans. Also, since studying survivorship requires waiting for the data to mature, our subjects’ care does not reflect recent advances in breast cancer management, such as sentinel lymph node biopsy and the use of aromatase inhibitors. We also recognize that the sub-cohorts examined for quality of care, involving those with baseline CVD and diabetes, were small. The study cohort has few minorities (4.3% black), a result of a) the population distribution of the study regions and b) older black women being at lower risk of developing breast cancer, but at higher risk of presenting with late stage disease (and therefore excluded from our study). Finally, due to lack of completeness of provider specialty field, the provider type of a sizable proportion of physician visits (about 25%) could not be determined.
In summary, this study clarifies the mixed picture related to longitudinal health care for older breast cancer survivors. We found that, beyond the first year after breast cancer diagnosis, older survivors have patterns of disease burden, inpatient care and quality of care for other health problems quite similar to those of women without breast cancer. There was one exception—they had notably more physician visits, especially for cancer and surgical specialists, than non-breast-cancer patients in each of five years of follow-up. Reasons for this should be pursued using more detailed clinical data. For example, were the excess visits for cancer surveillance, and, if so, did they follow accepted guidelines?