Scientific Session 28 — Breast - Screening/PathologyFriday, May 5, 2017
2225. Comparison of Screening Metrics: Online vs Batch Interpretation of Tomosynthesis Mammograms
Kim M*, Raghu M, Sheiman L, Hooley R, Philpotts L. Yale School of Medicine, New Haven, CT
Address correspondence to M. Kim (firstname.lastname@example.org)
Objective: Traditionally, screening mammograms are batch-read with results communicated to the patient via a letter. In recent years, communicating results immediately and directly to patients has become more feasible due to changing breast imaging practices including the implementation of PACS systems and digital mammography. Unfortunately, recall rates have been shown to be higher with online interpretation, which could render the practice less valuable. However, this has not been studied in the setting of screening mammograms performed with tomosynthesis. The objective of this study was to compare the tomosynthesis screening mammography metrics for online and offline reading. In addition, the time to biopsy was also compared for both groups.
Materials and Methods: A retrospective analysis of the breast imaging database was conducted to include all screening mammograms performed with tomosynthesis over a 1-year period from January 1, 2015, to December 31, 2015, at four outpatient sites. The recall rate (RR), positive predictive value of screening mammograms with abnormal findings (PPV1), and cancer detection rate (CDR) were compared between the online and offline groups. We included 22,149 screening mammograms that were performed with tomosynthesis (10,943 online and 11,206 offline). Online cases requiring additional imaging were generally performed at the patient’s convenience. In addition, for those cases where biopsy was recommended, the time from the screening mammogram to biopsy was calculated for one of our sites (5389 screening mammograms and 76 biopsies). The chi-square test was used to compare categorical data. The t test was used to compare days to biopsy. A two-tailed p-value of less than 0.05 was considered statistically significant.
Results: There was no statistically significant difference in RR (8.9% vs 8.7%, p = 0.6985), PPV1 (7.2% vs 8.1%, p = 0.5023), or CDR (6.7 vs 6.4 per 1000, p = 0.8625) when comparing screening tomosynthesis mammograms read online versus offline. There was, however, a statistically significant decrease in the average time to biopsy for those read online versus offline (9.2 vs 31.7 days, p < 0.001).
Conclusion: Online interpretation of screening mammograms in the era of tomosynthesis has great benefits by maintaining similar RR and CDRs, while significantly decreasing time to diagnosis, which may result in an overall improved experience for patients.