From the Department of Radiology, University of Wisconsin School of Medicine and Public Health, E3/311 Clinical Science Center, 600 Highland Ave, Madison, WI 53792-3252 (E.S.B.); Centre for Cancer Prevention, Queen Mary University of London, Wolfson Institute of Preventive Medicine, London, England (D.V., S.W.D.); and Nuffield Department of Population Health, University of Oxford, Oxford, England (R.G.B.).
Radiology. 2018 Jul;288(1):47-54. doi: 10.1148/radiol.2018171539. Epub 2018 Apr 3.
Purpose To determine whether low levels of recall lead to increased interval cancers and the magnitude of this effect. Materials and Methods The authors retrospectively analyzed prospectively collected data from the UK National Health Service Breast Screening Programme during a 36-month period (April 1, 2005 to March 31, 2008), with 3-year follow-up in women aged 50-70 years. Data on recall, cancers detected at screening, and interval cancers were available for each of the 84 breast screening units and for each year (n = 252). The association between interval cancers and recalls was modeled by using Poisson regression on aggregated data and according to age (5-year intervals) and screening type (prevalent vs incident). Results The authors analyzed 5 126 689 screening episodes, demonstrating an average recall to assessment rate (RAR) of 4.56% (range, 1.64%-8.42%; standard deviation, 1.15%), cancer detection rate of 8.1 per 1000 women screened, and interval cancer rate (ICR) of 3.1 per 1000 women screened. Overall, a significant negative association was found between RAR and ICR (Poisson regression coefficient: -0.039 [95% confidence interval: -0.062, -0.017]; P = .001), with approximately one fewer interval cancer for every additional 80-84 recalls. Subgroup analysis revealed similar negative correlations in women aged 50-54 years (P = .002), 60-64 years (P = .01), and 65-69 years (P = .008) as well as in incident screens (P = .001) and prevalent screens (P = .04). No significant relationship was found in women aged 55-59 years (P = .46). Conclusion There was a statistically significant negative correlation between RAR and ICR, which suggests the merit of a minimum threshold for RAR. RSNA, 2018 Online supplemental material is available for this article.
确定低召回率是否会导致间期癌的增加，以及这种影响的程度。
作者回顾性分析了英国国家卫生服务部乳房筛查计划在 36 个月期间（2005 年 4 月 1 日至 2008 年 3 月 31 日）收集的前瞻性数据，并对 50-70 岁女性进行了 3 年随访。每个乳房筛查单位和每年（n = 252）都有关于召回、筛查中检出的癌症和间期癌的数据。使用泊松回归对汇总数据进行分析，并根据年龄（5 年间隔）和筛查类型（现患 vs 新发病例），对间期癌与召回的关系进行建模。
作者分析了 5126689 个筛查期，平均召回率为 4.56%（范围，1.64%-8.42%；标准差，1.15%），每 1000 名筛查女性的癌症检出率为 8.1 例，间期癌发生率为 3.1 例。总体而言，RAR 与 ICR 之间存在显著的负相关关系（泊松回归系数：-0.039 [95%置信区间：-0.062，-0.017]；P =.001），每增加 80-84 次额外召回，就会减少约 1 例间期癌。亚组分析显示，在 50-54 岁（P =.002）、60-64 岁（P =.01）和 65-69 岁（P =.008）的女性以及新发病例（P =.001）和现患病例（P =.04）中也存在类似的负相关关系。在 55-59 岁的女性中，未发现显著的相关性（P =.46）。
RAR 与 ICR 之间存在显著的负相关关系，这表明 RAR 的最低阈值是有意义的。RSNA，2018 年在线补充材料可供本文参考。
