Disparities in the recommendation, acceptance, and performance of surgery as treatment modality for patients with stage 0-III breast cancer

e18509 Background: National data to identify disparities in breast cancer focus on the receipt of standard therapies, although receipt of those treatments is predicated upon three factors: provider recommendation, patient acceptance, and administration or performance of those therapies. We aim to determine whether surgical disparities exist for each of these three factors in patients diagnosed with stage 0-III breast cancer patients. Methods: The National Cancer Database (NCDB) was queried for females 40 to 70 years of age treated from 2010 to 2019 for stage 0-III breast cancer. Covariates of interest included race/ethnicity, socioeconomic status, and patient, facility, and tumor characteristics. Exclusion criteria were applied to mitigate biases in treatment variabilities. Outcomes of interest included whether surgery was recommended, whether the recommended surgery was accepted, and ultimately performed. Multivariable logistic regression models were fit to assess the outcomes of interest controlling for patient and tumor characteristics. Predictive margins were calculated, and chi-square tests were used to assess the differences between subgroups of interest. Results: Among 440,828 patients, unadjusted rates of recommendation, acceptance, and performance of surgery were 98.5%, 99.7%, and 99.6% respectively. After confounder adjustment, compared to White patients, the probabilities of having surgery recommended, accepted, and performed for Black patients were 0.6, 0.46, and 0.49 percentage points (ppt) lower respectively (all p < 0.0001). The probabilities of accepting and having surgery performed were both 0.14 ppt lower (p = 0.0012 and p = 0.0041) for Asian patients compared to White patients. Insurance status was a predictor for all outcomes, with the probabilities being 1.19, 0.78, and 0.78 ppt lower respectively when uninsured patients were compared to patients with private insurance (all p < 0.0001). Similarly, probabilities for all three factors were 0.92, 0.11, and 0.17 ppt lower respectively for patients treated at Academic/Research Programs compared to Comprehensive Community Cancer Programs (p < 0.0001, p = 0.001, and p = 0.0002). Income and Hispanic ethnicity had no significant association with the studied outcomes. Conclusions: Concern is justified about recommendation and differences in the performance of surgery as cause of breast cancer disparities by race and socioeconomic status, but acceptance of recommendations is also lower. Differences in understanding and a sense of trust may contribute to the lower likelihood of accepting surgical recommendations seen here. Further exploration is needed to ensure that patients of all backgrounds are equally clear regarding the reasons for surgical recommendations, and to foster trust. Such a focus on acceptance may help narrow disparities that have proven resistant to change so far.