The data source for this study was the National Cancer Data Base (NCDB) maintained by the American College of Surgeons and the American Cancer Society. The NCDB data are derived from hospital registry data on more than 1500 CoC-accredited facilities and represents 70% of newly diagnosed cancers in the United States.4 Following IRB approval, the database was queried for all adults (men and women) between the ages of 18 and 90 years who were diagnosed with unilateral stage 0-III in situ or invasive breast cancer from 2004 to 2012. Patients who had bilateral breast cancers or unknown laterality were excluded. Information on epidermal growth factor Her-2 status was omitted due to limited data availability in the NCDB database. The consort diagram describing inclusion and exclusion criteria is included as Fig. 1 .
Collected covariates included age at diagnosis, gender, race, education, ethnicity, insurance status, income level, comorbidity score, stage at diagnosis, hormone receptor status, surgery type, hormone therapy after surgery, hospital type and geographic location, time from diagnosis to surgery (in days), use of radiation, use of chemotherapy, distance traveled to treating hospital, treatment received at more than 1 CoC facility, and tumor size. Race and ethnicity were combined to create 6 categories: non-Hispanic White, non-Hispanic Black, non-Hispanic other, Hispanic White, Hispanic Black, and Hispanic other. Due to the paucity of data on Asians, American Indians/Alaskan Natives, and Native Hawaiian/Pacific Islanders, these racial categories were collapsed into non-Hispanic other and Hispanic other. Insurance status was composed of 3 categories—private, government, and none. Income level was dichotomized into <$35,000 and ≥$35,000.
Hormone receptor status was based on estrogen (ER) and progesterone (PR) receptors and was coded as a combined ER/PR status with 4 possible values: ER+/PR+, ER+/PR−, ER−/PR+, or ER−/PR−. Surgery type included lumpectomy, unilateral mastectomy, and contralateral mastectomy. The use of radiation, chemotherapy, and hormone therapy was dichotomized as yes or no. Chemotherapy was further divided into neoadjuvant and adjuvant. NCDB designations of hospital type as academic, comprehensive, community, or integrated were applied. Twenty-five hospitals had multiple hospital type designations, and among these, the most representative designation was selected. Hospital locations were based on the 4 NCDB assigned geographic regions of Midwest, Northeast, South, and West. Comorbidity score was based on the Charlson-Deyo score and divided into 3 groups with scores of 0, 1, and ≥2. Age, distance travelled to treating hospital, and tumor size remained as continuous variables throughout the analysis.
The average annual hospital volume was calculated as the number of breast cancer cases treated at a given facility divided by the number of years the facility had participated in the NCDB. OS was defined as time from diagnosis to death or last follow-up. Patients who did not die were censored at the date of last follow-up. Mortality was not limited to breast cancer related mortality but rather defined as all-cause mortality.
Collected covariates included age at diagnosis, gender, race, education, ethnicity, insurance status, income level, comorbidity score, stage at diagnosis, hormone receptor status, surgery type, hormone therapy after surgery, hospital type and geographic location, time from diagnosis to surgery (in days), use of radiation, use of chemotherapy, distance traveled to treating hospital, treatment received at more than 1 CoC facility, and tumor size. Race and ethnicity were combined to create 6 categories: non-Hispanic White, non-Hispanic Black, non-Hispanic other, Hispanic White, Hispanic Black, and Hispanic other. Due to the paucity of data on Asians, American Indians/Alaskan Natives, and Native Hawaiian/Pacific Islanders, these racial categories were collapsed into non-Hispanic other and Hispanic other. Insurance status was composed of 3 categories—private, government, and none. Income level was dichotomized into <$35,000 and ≥$35,000.
Hormone receptor status was based on estrogen (ER) and progesterone (PR) receptors and was coded as a combined ER/PR status with 4 possible values: ER+/PR+, ER+/PR−, ER−/PR+, or ER−/PR−. Surgery type included lumpectomy, unilateral mastectomy, and contralateral mastectomy. The use of radiation, chemotherapy, and hormone therapy was dichotomized as yes or no. Chemotherapy was further divided into neoadjuvant and adjuvant. NCDB designations of hospital type as academic, comprehensive, community, or integrated were applied. Twenty-five hospitals had multiple hospital type designations, and among these, the most representative designation was selected. Hospital locations were based on the 4 NCDB assigned geographic regions of Midwest, Northeast, South, and West. Comorbidity score was based on the Charlson-Deyo score and divided into 3 groups with scores of 0, 1, and ≥2. Age, distance travelled to treating hospital, and tumor size remained as continuous variables throughout the analysis.
The average annual hospital volume was calculated as the number of breast cancer cases treated at a given facility divided by the number of years the facility had participated in the NCDB. OS was defined as time from diagnosis to death or last follow-up. Patients who did not die were censored at the date of last follow-up. Mortality was not limited to breast cancer related mortality but rather defined as all-cause mortality.
