Radiology and Imaging Sciences, University of Utah/Huntsman Cancer Institute, 30 N 1900 E, Salt Lake City, UT 84132.
Radiology and Imaging Sciences, University of Utah/Huntsman Cancer Institute, 30 N 1900 E, Salt Lake City, UT 84132.
Acad Radiol. 2019 Feb;26(2):210-214. doi: 10.1016/j.acra.2018.04.012. Epub 2018 May 7.
This study aimed to compare performance metrics for immediate and delayed batch interpretation of screening tomosynthesis mammograms.
This HIPAA compliant study was approved by institutional review board with a waiver of consent. A retrospective analysis of screening performance metrics for tomosynthesis mammograms interpreted in 2015 when mammograms were read immediately was compared to historical controls from 2013 to 2014 when mammograms were batch interpreted after the patient had departed. A total of 5518 screening tomosynthesis mammograms (n = 1212 for batch interpretation and n = 4306 for immediate interpretation) were evaluated. The larger sample size for the latter group reflects a group practice shift to performing tomosynthesis for the majority of patients. Age, breast density, comparison examinations, and high-risk status were compared. An asymptotic proportion test and multivariable analysis were used to compare performance metrics.
There was no statistically significant difference in recall or cancer detection rates for the batch interpretation group compared to immediate interpretation group with respective recall rate of 6.5% vs 5.3% = +1.2% (95% confidence interval -0.3 to 2.7%; P = .101) and cancer detection rate of 6.6 vs 7.2 per thousand = -0.6 (95% confidence interval -5.9 to 4.6; P = .825). There was no statistically significant difference in positive predictive values (PPVs) including PPV1 (screening recall), PPV2 (biopsy recommendation), or PPV 3 (biopsy performed) with batch interpretation (10.1%, 42.1%, and 40.0%, respectively) and immediate interpretation (13.6%, 39.2%, and 39.7%, respectively). After adjusting for age, breast density, high-risk status, and comparison mammogram, there was no difference in the odds of being recalled or cancer detection between the two groups.
There is no statistically significant difference in interpretation performance metrics for screening tomosynthesis mammograms interpreted immediately compared to those interpreted in a delayed fashion.
本研究旨在比较即刻和延迟批量解读筛查断层合成乳腺 X 线摄影的性能指标。
本研究符合 HIPAA 规定，并经机构审查委员会批准豁免同意。回顾性分析了 2015 年即刻解读的断层合成乳腺 X 线摄影的筛查性能指标，并与 2013 年至 2014 年的历史对照数据进行比较，当时的断层合成乳腺 X 线摄影是在患者离开后进行批量解读的。共评估了 5518 例筛查断层合成乳腺 X 线摄影（批处理解读组 n=1212，即刻解读组 n=4306）。后者组的样本量较大，反映了小组实践的转变，即对大多数患者进行断层合成。比较了年龄、乳房密度、比较检查和高危状态。采用渐近比例检验和多变量分析比较性能指标。
与即刻解读组相比，批处理解读组的召回率或癌症检出率无统计学差异，批处理解读组的召回率分别为 6.5%和 5.3%（95%置信区间-0.3%至 2.7%；P=0.101），癌症检出率分别为 6.6 例和 7.2 例每千例（95%置信区间-5.9 例至 4.6 例；P=0.825）。批处理解读（分别为 10.1%、42.1%和 40.0%）和即刻解读（分别为 13.6%、39.2%和 39.7%）的阳性预测值（PPV），包括 PPV1（筛查召回）、PPV2（活检推荐）和 PPV3（活检进行）也无统计学差异。在调整年龄、乳房密度、高危状态和比较乳房 X 线摄影后，两组之间的召回或癌症检出率的比值无差异。
即刻解读与延迟解读的筛查断层合成乳腺 X 线摄影的解读性能指标无统计学差异。
