The Gordon and Betty Moore Foundation is providing funding for nine novel ideas and approaches for new clinical quality measures targeting improved diagnosis, specifically addressing three major categories of disease responsible for the greatest harm from diagnostic errors: acute vascular events, infections and cancer. Earlier this year, the foundation announced this new funding opportunity as part of its Diagnostic Excellence Initiative.
Twelve million Americans experience a diagnostic error each year
It is likely that each of us will experience a diagnostic error in our lifetime. Delayed or missed diagnoses result in delays in treatment, allow undiagnosed conditions to persist or even progress, and worsen outcomes. The development of clinical quality measures for diagnosis will provide clinicians and medical institutions the ability to track and measure their success and failure rates and provide them with an opportunity to solve for deficiencies and implement improvements.
The foundation’s Diagnostic Excellence Initiative aims to reduce harm from erroneous or delayed diagnoses. The Initiative also addresses a pressing need to improve diagnostic performance, reduce costs and redundancy in the diagnostic process, improve health outcomes and save lives. Over the last few decades, significant progress has been made in medication safety and therapeutics, but work has been slow to address diagnosis.
“The ideas that we’ve chosen to fund through our request for ideas are varied in their approaches, but they all address the fundamental challenge of how to measure diagnostic performance,” says Karen Cosby, program officer for the Diagnostic Excellence Initiative. “We look forward to seeing the results of these approaches in the future.”
Previous work by the Moore Foundation, in addition to this new project, addresses important challenges in health care to improve patient experiences and outcomes.
A full list of awardees can be found below.
John Sather, MD
Yale University School of Medicine
Diagnostic excellence index for pulmonary embolism
Pulmonary embolism (PE) is a common and often deadly disease affecting up to one million patients each year in the United States. Despite 20 years of diagnostic safety research and established clinical guidelines, performance and measurement gaps persist for PE diagnosis. This project seeks to develop a composite measure of diagnostic excellence for PE which incorporates diagnostic accuracy, resource utilization and diagnostic yield to promote a balanced diagnostic approach that maximizes patient benefit while minimizing harm and optimizing resource utilization.
Hardeep Singh, MD, MPH
Baylor College of Medicine & Houston VA
Collaborating with University College London, University of Exeter and Geisinger
Safer Dx e-measures to reduce preventable delays in cancer diagnosis
The project will use electronic health record data to quantify how promptly clinicians recognize and act on concerning symptoms or test results that may indicate an undiagnosed cancer; outcome measures will quantify the percentage of cancer diagnoses that are made at an advanced stage or in emergency settings.
Matthew Thompson, MBChB, MPH, DPhil
University of Washington, Seattle
Collaborating with University of Cambridge, University of Leeds, and the National Cancer Institute
Development and pilot testing of a measurement tool for the diagnosis of lung cancer
Lung cancer is the most common cause of cancer-related death in the United States, and many patients have symptoms and repeated visits to health care providers in the months before diagnosis. This project will develop and test a new method for measuring and tracking the timeliness of the patient’s cancer trajectory from first onset of symptoms to final diagnosis of lung cancer. This measure will draw information from the electronic medical record as well as from patients and their caregivers. The goal of this measure is to identify places in the diagnostic pathway where improvements are needed to improve the early diagnosis and eventual outcomes for lung cancer.
David Seidenwurm, MD
American College of Radiology
Closing the loop: Improving care coordination towards early disease detection
Evidence-based recommendations for follow-up imaging and care for incidental imaging findings (unexpected abnormalities detected in images done for other purposes) provide opportunity for earlier detection of cancer, but systems often lack rigorous processes to assure patients and providers adhere to them. This project aims to develop a quality measure to determine if evidence based recommendations are met for incidental radiology findings.
Valerie Vaughn MD, MSc
Ashwin Gupta, M.D.
University of Michigan, VA Ann Arbor Healthcare System
Measures to improve the diagnosis of infection
The two most common misdiagnosed bacterial infections in the hospital are pneumonia and urinary tract infection. To reduce harm from misdiagnosis, this project will develop measures of misdiagnosis for these infections using a unique dataset from the Michigan Hospital Medicine Safety Consortium.
Shamim Nemati, PhD
University of California, San Diego
SEP1+: A composite measure to accurately assess early sepsis management
SEP1 measure has been criticized for its oversimplifying assumptions and its 'all-or-nothing' approach to bundle compliance. In this work, Nemati et al propose an alternative machine learning-based composite measure to provide a data-driven and non-binary assessment of a hospital’s compliance with bundled care.
Makoto Jones, MD
University of Utah, Salt Lake City
Measuring misdiagnoses of infections using machine learning in a flexible framework
This project will develop a novel measure of diagnostic error in infections using Big Data and machine learning to account for patient-level context. It will be implementable in today’s electronic health record systems, facilitating both rapid and comprehensive assessments of diagnostic quality.
Rajesh Keswani, MD
Northwestern University, Chicago
Co-investigators: Dr. Mozziyar Etemadi, Dr. Anthony Yang
Reducing cancer due to colonoscopy diagnostic errors using machine learning
Colorectal cancer is largely preventable through high quality colonoscopy but measuring colonoscopy quality is difficult in practice and thus is not reliably performed. This project will develop an automatically abstracted measure of colonoscopy skill, generated by machine learning analysis of video-recorded colonoscopies, that correlates with the ability of a colonoscopist to prevent colorectal cancer. This novel measure of colonoscopy skill can then be used to rapidly measure colonoscopy diagnostic quality, drive quality improvement efforts, and reduce the disease burden of colon cancer through improved prevention and earlier diagnosis.
Afshan Hameed, MD, FACC, FACOG
University of California, Irvine
Collaborating with the University of California, San Diego and the University of Tennessee, St. Thomas Health System
Diagnosis of cardiovascular disease in pregnant and postpartum women
Cardiovascular disease is the leading cause of death in pregnant and postpartum women, and symptoms of cardiovascular disease are often mistaken for normal symptoms of pregnancy. This project will measure the proportion of women screened for cardiovascular disease in pregnancy and the postpartum period, and the proportion of women who screen positive who complete the recommended evaluation. With improved detection of disease, appropriate treatment may improve the outcomes of women and their pregnancies.