Conclusions
The committee came to four fundamental conclusions. First, we believe that the HCFA data sets, as presently constituted, can best be utilized to describe, track, and compare broad patterns of care for breast cancer according to subgroups of providers, practitioners, and patients. We fully support those goals of the Effectiveness Initiative. We offered three patient management areas—mammography (especially as a screening modality), therapeutic options (various surgical options, with and without radiation, chemotherapy, or hormonal therapy), and follow-up of patients treated for primary breast cancer—in which we believe effectiveness analyses would bear fruit. We suggested some of the variables that might have an important role in examinations of patterns (and variations) in care.
Second, we are less certain that at the present time the data sets lend themselves fully to the assessment of the important outcomes of therapeutic alternatives, believing that those outcomes presently proposed or available (e.g., death, proxy measures of complications of hospitalization) are insufficient to the task. We have taken seriously, however, the expressed intent of the framers of the Effectiveness Initiative that appropriate efforts for acquiring needed outcome data—either for the standard data sets themselves or for special projects—could and would be mounted. For this reason, and because ultimately we believe effectiveness research will falter without an ability to monitor or analyze outcomes appropriately, we have given strong emphasis in this report to outcome measurement; we hope that our suggestions here will strengthen that aspects of this work.
Third, we also wish to emphasize the need for appropriate case mix and severity adjustment or staging in effectiveness analyses and the need for ambulatory care data. The points raised concerning outcome measurement, such as mounting appropriate strategies for acquiring such data, apply as well.
Fourth, in our deliberations, many clinical questions were raised that are clearly in the purview of biomedical and clinical research, such as those relating to the appropriate physiologic variables in a staging index or to decisions to biopsy a patient following a positive screening mammogram. We do not believe that the program should try to tackle questions (such as these) that are best handled by prospective clinical RCTs. There are areas, however, where RCTs within the effectiveness arena might be considered. Among these might be alternative systems for delivering certain types of care for breast cancer, good mechanisms of feedback of practice-variation information to physicians, and ways to make information available to patients. We did not discuss these in any detail, but we encourage the agencies sponsoring effectiveness research to take them under advisement.