Comparative effectiveness of breast imaging modalities: a natural experiment

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): We propose to conduct a comparative effectiveness study of screening and diagnostic breast imaging modalities and breast biopsies in accurately diagnosing breast cancer. Results will aid clinicians and policymakers in selecting among breast cancer detection methods while minimizing over-diagnoses and unnecessary biopsies. Several knowledge gaps in the area of breast cancer diagnosis require further study. Specifically, the evidence regarding whether digital mammography is more accurate than conventional film- screen mammography for screening certain patient subgroups is insufficient. Use of computer-aided detection to assist in interpreting images is gaining widespread use, but the evidence for improved accuracy with this technology is incomplete. There are also knowledge gaps in effectiveness for diagnostic ultrasound and magnetic resonance imaging. We propose a natural experiment to examine the comparative effectiveness of different imaging modalities and breast biopsy procedures in diagnosing breast cancer. We are partnering with Advocate Health Care (AHC), the largest integrated not-for-profit health care delivery system in the metropolitan Chicago region and one of the top 10 of its kind in the nation. Advocate patients are urban and suburban and have diverse racial/ethnic and socioeconomic backgrounds, and they reflect variation in breast cancer risk in the general population. The Illinois State Cancer Registry (ISCR) is one of the most highly recognized population-based cancer registries in the nation. After linking data from the AHC radiology database for years 2001 through 2011 with incident breast cancer cases from ISCR through 2012, we will model accuracy measures for different imaging and biopsy procedures stratified by important clinical characteristics. We will also evaluate the accuracy of imaging and biopsy procedures in detecting more aggressive breast cancer subtypes and more difficult to detect histologies. All models will adjust finely for measured facility, radiologist and patient characteristics, and calendar time. We will also examine how changes over time in clinical practices and introduction of new technologies may have affected imaging and biopsy accuracy. The sample size for this proposed study would far exceed any prior study on the topic of screening and diagnostic imaging accuracy in breast cancer. This is a highly innovative natural experiment that proposes to examine relations across the continuum of breast cancer screening, diagnostic imaging, and breast biopsy as practiced in a real-world setting. Given the large sample size and the proposed collection of detailed information on potential confounders, we believe that our study will make a substantial contribution to the literature and to future clinical, administrative, and policy decision-making.