Breast Cancer: Setting Priorities for Effectiveness Research (1990)

Before discussing high-priority patient management issues in breast cancer, the committee briefly reviewed its understanding of the major reasons that breast cancer had been selected for the Effectiveness Initiative research program. Breast cancer is prevalent among women generally and among those eligible for Medicare. We attach particular importance to the increasing incidence of breast cancer as women age and to its increasing prevalence among aging women. Evidence is mounting that breast cancer may be initially treated at a more advanced stage among older, Medicare-eligible women than among younger women.

The effectiveness of community-based treatment for breast cancer—both at any age and for women over age 65—may not be accurately predicted by efficacy studies from major research centers.⁹ Furthermore, the efficacy and cost-benefit of alternative treatment regimens remain uncertain for selected patient subsets. The costs associated with breast cancer and its treatment can be high, and it was noted that the (expected) advent of screening mammography as a covered benefit in 1990 would have unpredictable effects on total Medicare expenditures for this condition, particularly for the detection and management of intraductal carcinoma. Finally, the possibility that an approach to the study of breast cancer might be used as preparation for effectiveness studies of other forms of malignancy was acknowledged.

Thus, the committee agreed that the HCFA data banks should be used to develop preliminary information on the costs and effectiveness of prevention, diagnosis, treatment, and rehabilitation of breast cancer. The committee further endorsed the general concept that analysis of variations in patterns of care and outcomes by geographic area, by institution, by type of provider or practitioner, and by other factors should be carried out as part of any effectiveness research program. Finally, the committee believed that separate studies to determine the content and outcomes of care provided to the Medicare population will be of particular value, especially if those analyses can help to advance the understanding of the role and results of prevention strategies, to clarify whether treatments offered to older women differ systematically from those offered to younger women, and to identify preferred patient management regimens.

Several factors figure in selecting key patient management issues in breast cancer for effectiveness research, especially for elderly women. These factors are not equally well documented in the clinical, research, or health policy literature, but the committee believed that all are sufficiently important to be taken into account as specific study topics are developed. They include:

• Epidemiologic aspects of breast cancer among elderly women (e.g., higher or lower prevalence of more advanced disease in particular subgroups);

• Important health status and quality-of-life burdens of the illness on elderly women;

• Different treatment options characterized by, for instance, whether alternative therapies have different likelihoods of prolonging survival, producing major impairment and disability, or improving the woman’s physical functioning, emotional well-being, and independence;

• High degree of professional and clinical uncertainty or dispute about alternative strategies for managing the care of breast cancer in elderly patients, especially when different types of treatment modalities have been found to produce equivalent results (e.g., in terms of survival) or trade-offs among outcomes (e.g., in terms of treatment-induced and disease-related complications or costs);

• Substantial variation across geographic areas in the per-person use of services for breast cancer, including screening modalities; that is, variation beyond that apparently explained by differences in patient characteristics or health resources in the areas;

• Substantial variation across geographic areas or institutions in the

outcomes of care for breast cancer; that is, variation beyond that presumably explained by the differences in the severity or stage of illness or sociodemographic characteristics of patients;

• Relatively high costs to the Medicare program of reimbursing for the services provided to patients to screen for, diagnose, and treat breast cancer; and

• Relatively high out-of-pocket costs to the Medicare beneficiary for services to screen, diagnose, and treat breast cancer that are not covered or are only incompletely covered by the Medicare program.

In specifying more concretely the research activities we recommend for addressing the high-priority breast cancer topics, we raise three additional points concerning the content, conduct, and use of this research. First, at the 1988 Clinical Workshop, the committee had identified four generic areas of concern: (1) clarification of the implications of the difference between efficacy and effectiveness for the overall purposes of this research program; (2) screening and prevention of the illness; (3) generation and use of reliable and valid outcome measures that relate to functional status and quality of life; and (4) mental and emotional dimensions (cognitive functioning; anxiety and depression). Members of the breast cancer committee specifically endorsed these points, noting that several good measures of health-related quality of life (relating to functioning, emotional well-being, and so forth) are currently available or could be adapted for effectiveness research.

Second, in principle the research strategy should be the most appropriate one possible for the specific question being addressed. The choice of strategies include: monitoring through analysis of administrative data; observational (e.g., cross-sectional) studies; quasi-experimental (e.g., case-control or longitudinal-cohort) studies and demonstrations; and, potentially, RCTs. The choice is often determined by the research question, but if the question can be addressed by more than one approach, then the complexity, rigor, and expense of alternative research designs should be taken into consideration. Plans to use more than one strategy either simultaneously or sequentially should also be considered. In addition, projects based on newer, nonexperimental techniques such as meta-analysis and cost-effectiveness analysis should be included among the research options. The Medicare claims databases might also provide information that could be used to develop decision-analytic techniques, which incorporate data-based probabilities. The committee concurs that all these approaches should be considered

potential investigational methods for effectiveness research. By implication, we reject the notion that only one strategy is appropriate. We also emphasize the importance of evaluating beforehand the trade-offs implicit in selecting one approach over another and in mounting combined approaches.

Third, we stress the importance of contributions that effectiveness studies on particular kinds of illnesses can make as prototypes for ways to examine other problems in other conditions. Issues of prevention and screening, diagnosis and staging, treatment options, and rehabilitation as they relate to breast cancer, for instance, should be seen as opportunities to address conceptual and methods issues for other neoplastic disease prevalent in the Medicare population.

The nature and availability of data to examine key effectiveness questions through HCFA’s existing (or anticipated) administrative files had been questioned at the 1988 Clinical Workshop. That these remain critical issues was quite apparent to the breast cancer committee and occasioned considerable debate and discussion. Thus, the committee underlines here several points concerning the importance of acquiring additional data through special studies, surveys, and patient follow-up activities.

First, having adequate data on ambulatory management of breast cancer (e.g., drug therapy documented in physicians’ office records) is critical. The centrality of this requirement must be understood and acted on before any longitudinal studies of alternative therapies are undertaken with the Medicare files.

Second, data analysis and interpretation may be severely restricted to the extent that information about radiotherapy or chemotherapy in the treatment of breast cancer is not available in the administrative files. This limitation must also be acknowledged and overcome.

Third, the same points about extent and quality of data are true concerning staging or case-mix data. It is crucial for these analyses, for instance, not to combine (or confuse) patients with early disease with those with locally advanced disease.¹⁰ For instance, very ill patients may not have been hospitalized but rather were irradiated as outpatients. Patients in an “other than mastectomy” category can include two very different types of

women—those with early, favorable breast cancer who are having primary conservation and women with metastatic disease who are getting a biopsy only to establish the diagnosis and test for hormone receptors. Adequate staging data are needed not only for fully interpreting studies of practice variations but also for investigating outcomes, especially those with strong quality-of-life components.

Nevertheless, the administrative data sets even now can provide information on several important topics, such as the site of service, shifts in the proportions of procedures (e.g., from radical mastectomy to modified and/ or simple mastectomy), and geographic variations in broad patterns of care. They can also be used to look at in-hospital survival, time-based survival, readmissions, some complications, and other simple outcome measures when one is unable to measure functional status, mental health, or other health status outcomes directly. Thus, although the discussion emphasized the necessity of obtaining richer clinical and patient outcome data through Medicare PROs, linkages with local SEER registries, and (most important) primary data collection, the potential of the existing data sets was also acknowledged.