False-Negative Screening and Diagnostic Mammograms in the National Mammography Database From 2010 to 2022

BACKGROUND. False-negative (FN) mammograms typically delay breast cancer diagnoses and may impact clinical outcomes. However, systematic evaluations of FN mammograms are challenging to conduct due to the low incidence of interval cancers. OBJECTIVE. The purpose of this study was to evaluate the rates of FN screening and diagnostic mammograms in the NMD and to assess associations of FN rates with patient- and facility-level characteristics. METHODS. This retrospective study included all screening and diagnostic mammograms in the National Mammography Database (NMD) obtained from January 1, 2010, to December 31, 2022. Patient- and facility-level factors were extracted from the NMD. FN mammograms were defined as those with a negative result in a patient with a tissue diagnosis of breast cancer within the subsequent year. FN rates per 1000 examinations were computed. Separate multivariable analyses were performed to identify associations with FN results for screening and diagnostic examinations. RESULTS. The analysis included 38,304,525 mammography examinations for 15,585,433 women (mean [SD] age, 64.4 ± 10.8 years). For 32,267,238 screening examinations, the FN rate was 1.9 (minimum, 0.7 in 2010; maximum, 2.5 in each year from 2020 to 2022). For 6,037,287 diagnostic examinations, the FN rate was 4.0 (minimum, 2.3 in 2010; maximum, 5.4 in 2020). In multivariable analysis, for screening examinations, the likelihood of an FN examination was lower for patients with race categories other than White (OR = 0.30-0.95), higher for patients with breast density categories other than almost entirely fatty breasts (OR = 1.60-2.00), higher for women with a personal (OR = 3.69) or family (OR = 1.29) history of breast cancer, and higher for academic or university-based facilities (OR = 1.37); for diagnostic examinations, the likelihood of an FN examination was lower for patients with race categories of Asian (OR = 0.91) and Hawaiian (OR = 0.77) and higher for patients with a race category of Black (OR = 1.12), lower for Hispanic patients (OR = 0.70), higher for patients with breast density categories other than almost entirely fatty breasts (OR = 1.50-2.78), higher for women with a personal (OR = 7.82) or family (OR = 1.31) history of breast cancer, and higher for academic or university-based facilities (OR = 1.37). CONCLUSION. Rates of FN screening and diagnostic mammograms increased over time and showed significant associations with patient and facility characteristics. CLINICAL IMPACT. Exploration of the causes of the observed associations could inform quality assurance efforts to reduce the risk of delayed breast cancer diagnoses.