Associate Section Chief, Breast Imaging and Intervention Section, Associate Program Director, Breast Imaging Fellowship, and Associate Program Director, Diagnostic Radiology Residency, Department of Radiology, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin.
Assistant Professor, Department of Biostatistics and Medical Informatics, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin.
J Am Coll Radiol. 2022 May;19(5):604-614. doi: 10.1016/j.jacr.2022.01.021. Epub 2022 Mar 28.
Data on utilization rate and cancer yield of BI-RADS® category 3 in routine clinical practice in diagnostic mammography are sparse. The aim of this study was to determine utilization rate and cancer yield of BI-RADS 3 in diagnostic mammography in the ACR National Mammography Database (NMD).
Retrospective analysis of NMD mammograms from January 1, 2009, to June 30, 2018, was performed. BI-RADS 3 utilization rate in diagnostic setting was calculated and stratified by patient, facility, and examination-level variables. Patient-level cancer yield was calculated among women with BI-RADS 3 assessment and adequate follow-up (imaging follow-up ≥24 months or biopsy). Logistic regression was performed to assess the odds of utilization of BI-RADS 3, with respect to facility, examination, and patient variables, and the odds of malignancy among patients with probably benign findings. Chi-square and t tests were used to determine significance (P < .05).
Data from 19,443,866 mammograms from 500 NMD facilities across 31 states were analyzed, of which 3,039,952 were diagnostic mammograms. Utilization rate of BI-RADS 3 was 15.5% (470,155 of 3,039,952) in the diagnostic setting. There was a statistically significant difference in BI-RADS 3 utilization rate across all collected variables (P < .001). Patient-level cancer yield at 2-year follow-up was 0.91% (2,009 of 220,672; 95% confidence interval [CI], 0.87%-0.95%) in the diagnostic setting. Patient and examination variables associated with significantly higher likelihood of malignancy included calcifications (odds ratio, 4.27; 95% CI, 2.43-7.51), patient age > 70 years (odds ratio, 3.77; 95% CI, 2.49-5.7), and presence of prior comparisons (odds ratio, 1.23; 95% CI, 1.07-1.42).
In the NMD, BI-RADS 3 assessment was common in diagnostic mammography (15.5%), with an overall cancer yield of 0.91%, less than the benchmark of 2%. Utilization trends in diagnostic mammography warrant further research for optimization of use.
常规临床诊断性乳腺 X 线摄影中 BI-RADS®第 3 类的利用率和癌症检出率数据较为匮乏。本研究旨在确定 ACR 国家乳腺 X 线摄影数据库（NMD）中诊断性乳腺 X 线摄影中 BI-RADS 第 3 类的利用率和癌症检出率。
对 2009 年 1 月 1 日至 2018 年 6 月 30 日的 NMD 乳腺 X 线摄影进行回顾性分析。计算诊断性设置中 BI-RADS 第 3 类的利用率，并按患者、医疗机构和检查水平变量进行分层。在接受 BI-RADS 第 3 类评估且随访充分（影像学随访≥24 个月或活检）的女性中计算患者水平的癌症检出率。采用 logistic 回归分析评估与医疗机构、检查和患者变量相关的 BI-RADS 第 3 类利用情况的比值比，以及可能良性发现患者的恶性肿瘤比值比。采用卡方检验和 t 检验确定统计学意义（P＜0.05）。
分析了来自 31 个州的 500 家 NMD 医疗机构的 19443866 份乳腺 X 线摄影，其中 3039952 份为诊断性乳腺 X 线摄影。在诊断性设置中，BI-RADS 第 3 类的利用率为 15.5%（470155/3039952）。在所有收集的变量中，BI-RADS 第 3 类的利用率存在统计学显著差异（P＜0.001）。在诊断性设置中，2 年随访时的患者水平癌症检出率为 0.91%（2009/220672；95%置信区间[CI]，0.87%-0.95%）。与恶性肿瘤发生风险显著增加相关的患者和检查变量包括钙化（比值比，4.27；95%CI，2.43-7.51）、患者年龄＞70 岁（比值比，3.77；95%CI，2.49-5.7）和存在既往比较（比值比，1.23；95%CI，1.07-1.42）。
在 NMD 中，BI-RADS 第 3 类评估在诊断性乳腺 X 线摄影中较为常见（15.5%），总体癌症检出率为 0.91%，低于 2%的基准值。诊断性乳腺 X 线摄影的利用趋势值得进一步研究，以优化其应用。
