Client Challenges

• Generate robust real-world data to evaluate the comparative effectiveness of select mammography screening modalities

• Collect patient-level data from diverse real-world sources and transform them into a rich, usable platform for analysis and dissemination

• Generate high quality evidence efficiently and quickly to inform national clinical guideline development timelines and support marketing claims

OM1 Solution

We developed a large-scale, automated patient registry to:

• Collect longitudinal data rapidly, and at scale

• Get deep data on cancer risk profiles and outcomes • Reduce enrollment burden on sites

• Publish results rapidly

• Enable additional studies as needed with minimal effort

Results at a Glance

Introduction

Digital breast tomosynthesis (DBT), also known as 3D mammography, is becoming the preferred imaging modality for breast cancer screening compared to the traditional X-ray mammography introduced in the 1960s. Despite recent advancements in mammography screening, a lack of robust research on the comparative effectiveness of mammography screening modalities has led to differences between clinical practice and screening guidelines. Hologic, a leading healthcare diagnostics company primarily focused on improving women’s health, sought to address evidence gaps and demonstrate the effectiveness of advances in mammography screening.

Hologic partnered with OM1 to develop a large-scale patient registry and reusable evidence generation platform to rapidly evaluate diagnostic patterns and associated outcomes for select mammography screening methods.

The Challenge

Understanding Real-World, Patient-Level Performance

While existing data on general screening performance, such as cancer detection rate and recall rate, provide valuable insights, these data often fall short in understanding the long-term performance of screening modalities. In order to better understand the real-world impact of 3D mammography, including impacts in key patient subgroups, Hologic needed to generate data from a large, representative population. However, traditional approaches such as randomized control trials pose cost, time, and occasionally ethical challenges, and standard patient registries and prospective studies prove impractical with respect to resource requirements or patient recruitment.

These traditional research methods are based on linear, rigid designs that introduce several challenges:

• Lengthy process to generate quality, research-ready data

• Steep cost of implementation for traditional study models

• Selection bias introduced when patients and physicians are required to opt-in

• Hard to find large, representative patient cohorts

• Hard to scale to multiple research sites or data sources

The Solution

Collecting Data at Scale in a Fraction of the Time and Cost

Leveraging our expertise with patient registries, proprietary data sourcing and processing platforms, and medical language processing (MLP) and machine learning capabilities, we designed a flexible, scalable patient registry to:

• Automate retrospective and prospective data collection, integration, and transformation of data from EMR, radiology information systems, and tumor registries from 5 large health systems and 63 imaging facilities into a common data model

• Collect and integrate data on more than 1.1 million women and 2.5 million breast screening exams

• Extract relevant attributes from structured and unstructured data, including patient-level demographics, calculated breast cancer risk scores, and clinical outcomes

• Link additional extrinsic data including claims and vital status

OM1’s clinical experts worked collaboratively with the Hologic team and principal investigators on the study design, program planning, execution, analysis, and publication.

Results

By adopting OM1’s automated, hybrid retrospective and prospective research approach, Hologic was able to achieve much greater efficiency, reduced burden and selection bias, and increased patient numbers and follow-up over traditional data collection for research studies.

OM1’s solution enabled Hologic to quickly and efficiently conduct unprecedented, large-scale assessments of the most important mammographic screening outcomes outside of clinical trial settings, in real-world practice. Since 2018, Hologic has published several high-impact studies, including 3 manuscripts and 10 conference abstracts, with additional analyses and publications planned.

In the largest study of 3D mammography exams to date, OM1 and Hologic’s research findings validate the results of previous smaller studies and add to the growing amount of evidence demonstrating 3D mammography is a superior screening method that has a higher breast cancer detection rate.¹ The latest draft guidelines from the United States Preventive Services Task force (USPSTF) now recommend 3D mammography.

Expanding Research Horizons to Drive Health Equity

As the team conducted initial analyses, new research questions started to emerge. OM1’s reusable, adaptable solution provided a seamless infrastructure that enabled Hologic to conduct additional studies as needed.

The power of the registry’s sample size and design enabled Hologic’s team to examine key subgroups of interest, including mammographic density and race/ethnicity. One key study investigated racial differences in the utilization and performance of screening modalities. Our analysis found that all racial groups had improved outcomes with 3D mammography screening, but disparities were observed in 3D mammography utilization. These data suggest that reducing inequities in 3D mammography utilization may improve the effectiveness of breast cancer screening. These findings help raise awareness of disparities and drive adoption of best practices in cancer screening to improve health outcomes.

References

1. 1 Conant E.F., et al. Radiology, vol. 307, no. 3, 2023, https://doi.org/10.1148/radiol.221571