Dr. Anthony Paravati
According to the results of a large observational study of Medicare beneficiaries with early-stage breast cancer, the individual surgeon and the institution exert greater influence on whether patients undergo mastectomy compared with all clinical and demographic variables except for tumor size and nodal status (Abstract 1008).
Patients with early-stage breast cancer face the critical choice of whether to undergo breast-conserving surgery or mastectomy—approaches that yield equivalent cancer-specific outcomes. Ideally, physicians should objectively present all appropriate treatment options to patients. However, as suggested by the current findings, “the variation observed in [the mastectomy] decision suggests the presence of physician and institutional biases that should be addressed for the goal of reducing unwarranted health care system–related variability. This will take a systematic, concerted effort,” Anthony Paravati, MD, MBA, of the University of California, San Diego, said during his presentation of the study findings.
For the analysis, the investigators identified women with localized breast cancer diagnosed between 2000 and 2009 who underwent either breast-conserving surgery or mastectomy from the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, capitalizing on the wealth of patient- and treatment-related data provided. The final study cohort included 29,358 patients with a full complement of evaluable data, 22% of whom underwent mastectomy.
To identify the factors associated with the likelihood of mastectomy, the researchers conducted multivariate analyses and used odds ratios (ORs) to describe the impact of fixed demographic or clinical covariates, along with median ORs (MORs) to describe the relative impact of the institution and the individual surgeon. The latter metric helps translate variance into the OR scale and allows MORs for relative covariates to be directly compared with ORs for fixed covariates.
Multivariate analysis revealed that the strongest predictors of mastectomy included a primary tumor size of 2 to 5 cm (OR 3.06, 95% CI [2.85, 3.28]) and the presence of four or more positive nodes (OR 2.95, 95% CI [2.61, 3.33]), followed by the institution and surgeon combined (MOR 2.38, 95% CI [2.09, 2.70]) and the surgeon alone (MOR 1.97, 95% CI [1.71, 2.27]).
When the variation in mastectomy use was further explored, individual surgeons accounted for 23.7% of the variance and institutions for 14.9%, together totaling 38.6%, with fixed demographic and clinical factors accounting for the remaining 61.4%.
Dr. Paravati was careful to point out the caveats of this study: the inability to address potentially confounding covariates, such as multicentric disease or involved margins, and the lack of other relevant factors, such as patient choice, psychosocial issues, health behaviors, and patient-physician communication, in the SEER-Medicare data capture.
Still, discussant Elizabeth Ann Mittendorf, MD, PhD, of The University of Texas MD Anderson Cancer Center, believes the study findings—and the call for change—are valid. “Culture dominates. And it’s going to be difficult for us to change our culture,” she said.
However, change is necessary. “Patients do want an equal say in their care and decisions, particularly if they can be provided clear information about their treatment options,” Dr. Mittendorf stated, referring to the outcomes of a 2013 meeting convened by the Institute of Medicine focused on how to better deliver high-quality cancer care.
Dr. Mittendorf suggested that the best way to obtain robust, objective data to inform patients about their treatment options is to conduct clinical trials, as opposed to observational studies, which allows all physician biases to be put aside.
“Clinical trials are critically important in guiding treatment recommendations and informing those conversations with our patients,” she said.
– Kara Nyberg, PhD