Must Breast Care Delivery Models from Resource Rich Environments Be Applied to Resource Limited Environment: Reply

Abstract

In response to our invited editorial ‘‘Balancing resource constraints against quality of care’’ [1], Drs. Mouton, Apffelstaedt, and Baatjes ask whether breast care delivery models from resource-rich environments should necessarily be applied to resource-limited environments. We fully agree with our South African colleagues that health care delivery strategies must be adapted to existing resources within a given environment. Our multinational coalition, the Breast Health Global Initiative (BHGI), has published evidencebased, economically feasible, and culturally appropriate guidelines for breast care in lowand middle-income countries (LMCs), the purpose of which is to provide guidance to in-country health care leadership for strategic programmatic development [2]. A core principle developed through the collaborative efforts of breast cancer experts from more than 40 countries convened in four BHGI Global Summits between 2002 and 2010 is that novel breast care delivery models must be developed and applied in LMCs to improve outcomes on a global scale. So, regarding the challenge of health care resource allocation in LMCs, we are in alignment with our South African colleagues. We respectfully suggest that the authors have not understood our primary critique of their now published article [3]. Drs. Mouton, Apffelstaedt and Baatjes have interpreted their data to suggest that ‘‘dedicated breast surgeons are equally proficient at mammography interpretation as radiologists.’’ We reaffirm that this bold conclusion cannot be validated on the basis of their South African data because the surgical investigators have applied a previously abandoned scheme to mammographic interpretation and biopsy that blunts their ability to make a legitimate comparison between their experience and that of radiologists who follow well established and tested approaches for mammographic interpretation. We disagree with the authors’ assertion that their threepoint scale for classifying mammographic lesions as ‘‘negative/indeterminant/positive’’ allows them to make assertions about the reliability of the surgical interpretation in comparison to experienced, trained mammographic readers. Unlike BI-RADS, this three-point scale cannot directly measure individual reader reliability in a graded way. The five-point BI-RADS scale, now universally accepted as the standard method for interpreting mammograms, not only forces the reader to determine whether a biopsy should be performed (the three-point scale), it also provides a qualitative assessment of the likelihood of the lesion to proving to be malignant or benign on biopsy (the five-point scale). Because the BI-RADS categories 4 (suspicious) and 5 (malignant) were collapsed into a single group in the South African study, the relative accuracy of the surgical interpretations cannot be weighed with the level of granularity that permits a legitimate comparison of diagnostic accuracy between surgeons and radiologists. It is certainly encouraging that the cancer diagnosis rates based on clinical follow-up from this South African clinic seem similar to other reported experiences in the literature, but this analysis does not meet evidence-based criteria for establishing the surgeons’ mammographic interpreting skills in comparison to trained radiologists. At a minimum, the authors should acknowledge that they have overstated the findings of their study. B. O. Anderson (&) Department of Surgery, University of Washington, Box 356410, 1959 NE Pacific Street, Seattle, WA 98195, USA e-mail: banderso@u.washington.edu