To evaluate the association between preoperative breast magnetic resonance imaging (MRI) utilization and the rate of multiple surgeries, and to investigate the extent of any variation of rates of multiple surgeries among physicians. We identified patients with stage 0, I, or II breast cancer diagnosed between 2002 and 2007 in the Surveillance, Epidemiology, and End Results-Medicare database. Using diagnosis and procedure codes, we defined that the initial treatment episode had ended when a gap in surgery occurred at least 90 days after primary surgery. Surgical procedures of partial mastectomy or mastectomy during the initial treatment period were calculated to identify patients who received multiple surgeries. Multilevel logistic regression models were used to identify patient- and physician-level predictors of multiple surgeries. Of 45,453 women with early stage breast cancer who were treated by 2,595 surgeons during the study period, 9,462 patients (20.8 %) received multiple breast surgeries; of these patients, 8,318 (87.9 %) underwent one additional surgery, 988 (10.4 %) received two additional surgeries, and 156 (1.6 %) received three or more additional surgeries. Among 2,997 (6.6 % of the entire cohort) women who underwent preoperative breast MRI evaluation, 770 received multiple breast surgeries. After we adjusted for patient and tumor characteristics associated with multiple surgeries, we found that the rate of multiple surgeries was not significantly different between the two groups with or without preoperative breast MRI. Furthermore, the median odds ratio of 2.0, corresponding with the median value of the relative odds of receiving multiple surgeries between two randomly chosen physicians after controlling for other confounders, indicated a large individual surgeon effect. Substantial variation was observed in the rates of multiple surgeries in women aged 66 and older with early stage breast cancer. Evidence does not support that preoperative breast MRI reduces the incidence of multiple surgeries.
- Magnetic resonance imaging
- Multilevel logistic regression
- Practice variation