AHRQ’s contributions to diagnostic safety: past, present, and future

Introduction

The modern era of thinking about and researching diagnostic errors as a patient safety issue is barely two decades old. For centuries there have been efforts to improve diagnosis through clinician education and training, refinements of the physical exam, and advances in diagnostic tests, imaging, and other technologies. Merging these activities with the patient safety movement has illuminated major areas for improvement, especially in light of the increasingly recognized magnitude of diagnostic errors [1], [2], [3]. The Agency for Healthcare Research and Quality (AHRQ) has been at the center of this movement as a convenor, funder, and stimulus – supporting the ideas, data, and people that led us to the current state of diagnostic safety in 2025 [4], 5]. The tenth anniversary of the National Academies of Sciences, Engineering, and Medicine (NASEM) Improving Diagnosis in Health Care report [6], as well as recent turbulent events impacting the research community and the future of AHRQ, presents an opportunity to reflect on AHRQ’s contributions and the future of federal support and leadership to advance diagnostic error research [7], [8], [9].

We begin with an overview of AHRQ’s contributions to diagnostic excellence across several key domains using a publicly available list of AHRQ grants awarded over the last 25 years. The TreeMap below (Figure 1) provides a bird’s-eye view of AHRQ-funded diagnosis-related activities including the number of projects within each domain. These activities have been driven both by AHRQ’s priorities expressed in various Notice of Funding Opportunities (NOFOs) as well as key diagnostic areas proposed by investigators whose grants were selected for funding by AHRQ.

Figure 1:

Overview of AHRQ funded projects by domain. TreeMap summarizing the portfolio of 75 AHRQ-funded diagnostic research and conference activities from 2000 to 2022. The size of the boxes represents the proportion of total projects funded, not the dollars awarded.

Early AHRQ funding for diagnosis safety: The DEER Project

In the early 2000s AHRQ launched its major patient safety initiative by funding three large Centers for Patient Safety research and a dozen smaller Developmental Centers. Given the paucity of attention to diagnosis in early patient safety work, our Chicago-based team, the Diagnostic Error Evaluation and Research (DEER) Developmental Center for Patient Safety Research ((DCERPS) was the only one of the 93 early AHRQ-funded safety projects to focus exclusively on diagnosis safety [10]. The team, based at Cook County Hospital, Rush University, and the University of Illinois at Chicago (UIC), collected and reviewed hundreds of clinician-reported diagnostic error cases, as well as conducted weekly case-review conferences. Lacking clear definitions, standardized metrics, classification criteria, and agreement between reviewers as to whether an error had occurred, we waded in what seemed to be a “swamp” of diagnosis safety, understanding why others avoided this poorly mapped terrain. The DEER taxonomy tool was developed to classify where in the diagnostic process something potentially went wrong. This AHRQ-funded tool has been widely used in the ensuing 20 years [11], [12], [13], [14], [15]. Refinements to the DEER taxonomy continue with a newly developed cancer-specific version (CAN-DEER) to classify cancer cases as part of our current AHRQ-funded Improving Cancer Diagnosis (ICDx) Diagnostic Center of Excellence (DCE). The DEER project also included early IT surveillance work to link two previously siloed databases, one housing lab and one housing pharmacy data, to screen for potential diagnostic errors (e.g., patients with elevated TSH but who had not been prescribed levothyroxine, suggesting hypothyroidism may have been overlooked) [16].

AHRQ support for annual Diagnostic Error in Medicine (DEM) Conferences

The annual Diagnostic Error in Medicine (DEM) conferences have historically played a key role in bringing together the most concerned and expert people in working on diagnostic safety. Initially held in 2007 as a smaller gathering linked to the annual meetings of the American Informatics Association Meeting (AMIA) in 2007 and the Society for Medical Decision Making (SMDM) in 2008 and 2009, the DEM Conference evolved into an independent international annual event that brought together hundreds of clinicians, educators, researchers, and most importantly, patients and family members many of whom had personally experienced diagnostic errors [17], [18], [19], [20].

The DEM meetings spawned and were ultimately led by the Society for Improving Diagnosis in Medicine (SIDM), which was formed in 2011 which was active in leading diagnosis improvement for more than a decade. SIDM was supported in part by a series of diagnostic error projects funded by the Gordon and Betty Moore Foundation (see below). SIDM unfortunately ended in 2024 due to financial shortfalls when the Moore funding and other support was not sustained (the new Community to Improve Diagnoses in Medicine (CIDM) is currently attempting to pick up some of SIDM’s role [21]). AHRQ played a central role in these conferences by providing both modest conference grant support for many years and sending AHRQ staff to participate who in turn brought back learning and lessons that helped inform a series of subsequent funding calls prioritizing diagnostic safety.

AHRQ funded multiple diagnostic safety studies leading to and following the NAM report

Over the ensuing two decades AHRQ funded and helped disseminate a series of projects and tools related to studying the epidemiology of diagnostic errors, metrics and trigger tools for their detection, cognitive assessments and decision support, more reliable lab and radiology follow-up, and diagnosis improvement in special settings such as the Emergency Department and inpatient settings. Table 1 lists three dozen examples of some of these projects to illustrate their diversity and give a flavor of the issues that safety experts and patients have been collectively grappling with to improve diagnosis safety. Along with an investment by the Gordon and Betty Moore Foundation (roughly $100 million, similar in scale to the funding AHRQ committed to diagnosis safety projects over this period), and smaller funding grants from malpractice insurers (CRICO, Coverys, The Doctors Company Foundation) and recently the Hartford foundation, there has been genuine traction in advancing research and quality improvement efforts. A pivotal landmark in this journey was the 2015 report by the NASEM titled, Improving Diagnosis in Health Care, which was instrumental in defining foundational concepts and a framework of the diagnosis process, citing the extent of diagnostic errors, providing eight key recommendations, and challenging action on a national level [6].

Stimulated by the NASEM report, AHRQ issued two series of requests for proposals that focused partly or entirely on improving diagnosis – the Patient Safety Learning Laboratories (PSLL) and more recently the Diagnostic Centers of Excellence (DCE). Beginning in 2014 AHRQ funded several waves of Patient Safety Learning Laboratory (PSLL) grants with more recent NOFOs explicitly focused on diagnostic safety. These four-year projects were grounded in the understanding that systems engineering (SE) and human factors approaches were an underutilized resource in health care quality improvement. This led to a series of collaborations with systems engineers in a wide variety of diagnostic realms [22]. Table 1 includes several of these PSLLs (the full grantee list is available on the AHRQ website) [23]. These projects explored varied methods to detect and address failure modes in the diagnostic process with mixed success. For example, our team based at the Harvard Medical School Center for Primary Care and Beth Israel Deaconess Medical Center tackled the problem of “open loops” in action-requiring abnormal tests, referrals, and symptom follow-up. Open loop rates of 20 % or higher were repeatedly identified, depending on the diagnosis, process, and time frames we examined [24], [25], [26]. However, we (and several other PSLL projects) had only modest success in our improvement efforts due to IT challenges (including in our case impediments arising from a transition from homegrown EHR to Epic), staffing shortfalls and discontinuities (in large part COVID-related), and institutional distractions (including concurrent merger with another large health system), that resulted in low prioritization of the SE process improvement work by hospital leadership. Nonetheless, the PSLL’s are a work-in-progress, and lessons continue to be harvested and published [23]. The most recent of these projects are funded through fiscal year 2026.

AHRQ Diagnostic Centers of Excellence (DCEs)

In 2022 AHRQ funded ten four-year Diagnostic Safety Centers of Excellence (DCEs), each of which aimed to better characterize sources of diagnostic error and develop and test novel solutions to reduce harm [27]. As described below and detailed on AHRQ’s Diagnostic Safety Centers of Excellence web page, these projects, now in their third year, include a range of areas for improving diagnosis:

1. understand how, when, and why EHR-related diagnostic errors occur

2. apply systems science principles to eliminate patient harm from diagnostic error in the Emergency Department,

3. learn from patients to characterize and prevent missed opportunities for earlier cancer diagnosis,

4. decrease diagnostic error in medical imaging,

5. partner with patients and scientists to develop patient-centered diagnostic safety solutions and reduce the number of diagnostic safety events,

6. develop evidence-based approaches for specific diagnostic problems including unintentional weight loss, sepsis and hypertension,

7. enhance teamwork between patients and healthcare providers using Safety-II theory and practices,

8. study and improve inpatient diagnosis among a national consortium of hospitals,

9. characterize and enhance the diagnostic journey of children with multiple chronic conditions, and

10. develop a system using electronic triggers coupled with timely diagnostic feedback for clinicians and organizations using Safety-I and Safety-II approaches.

Not only do these DCEs cover a remarkably broad landscape, but the potential for cross fertilization of learning and collaboration across these centers is being promoted by AHRQ via regular meetings, an annual conference, and funding of a coordinating center [28].

Patient engagement in improving diagnosis

Support for patient engagement has been a major strategic stance and thread running through the priorities, projects, tools, and information disseminated by AHRQ related to diagnosis safety. AHRQ supports a series of diagnostic safety issue briefs. A recent issue titled, “The Patient’s Role in Diagnostic Safety and Excellence”, distills lessons learned from the literature and AHRQ projects [29]. Additional resources and recommendations include approaches to empower patients and families to ask key questions related to their diagnosis and ways they can help “co-produce” a better diagnostic experience by preparing critical information in advance of their visits, monitoring ongoing symptom evolution, and playing a role in reliable test result follow-up [30]. These diagnosis-centered activities have occurred within the context of the broader AHRQ portfolio that includes key pillars of patient engagement such as the CAHPS (Consumer Assessment of Healthcare Providers and Systems) which surveys patients regarding areas critical to good diagnosis such as access, timely scheduling, good communication and care coordination [31]. Equally important for evaluating and improving the staff and institutional “culture” patients encounter in their care are the Surveys on Patient Safety Culture (SOPS) that have been developed and validated; the results of which are collected and shared by AHRQ. A new supplement to the Medical Office Survey on Patient Safety Culture is available which includes 12 questions specifically related to diagnosis safety culture in the outpatient setting [32], 33]. AHRQ also supported development of the Toolkit for Engaging Patients to Improve Diagnostic Safety which seeks to improve communication during the clinician provider encounter (also discussed below).

Other AHRQ-supported diagnosis-related contributions

Some of the important work on diagnostic safety and followed from the general work in Patient Safety. The Patient Safety and Quality Improvement Act of 2005 authorized the creation of Patient Safety Organizations (PSOs) to facilitate voluntary reporting, aggregating, and analysis of patient safety events, and the development of a Common Formats for uniform reporting to standardize a reporting taxonomy. It wasn’t until 2022 that AHRQ developed and released a module specific to diagnostic events, the Common Formats for Event Reporting – Diagnostic Safety Version 1.0, to facilitate the standardized reporting of diagnostic safety events [34]. The ability to capture this data should facilitate more routine capture of diagnostic failures.

Similarly, the popular TeamSTEPPS program was originally developed to support teamwork and communication as part of AHRQ’s patient safety work [35]. The program has been quite successful, and as of 2023, it has been implemented in more than 1,500 hospitals across the nation, with more than 5,000 master trainers who have trained more than 300,000 health professionals. In 2022, a module specific to Diagnosis was introduced, emphasizing awareness of harm from diagnostic error, and introducing training modules on the diagnostic team structure, communication, leadership, situation monitoring, and reflective practice [36].

AHRQ also maintains several databases and manages a set of quality measures and performance indicators. In late 2024, they issued a Request for Information for public comments on the development of measures specific for diagnostic excellence; work towards diagnostic measures is ongoing.

PSNet

In 2001 AHRQ launched a novel and brave experiment – publishing Morbidity and Mortality (M&M) cases on its website. Although anonymized, medical legal fears obliged the editors to initially select only “near miss” cases. As these fears were overcome, the collection grew to include hundreds of patients (currently 650), with diagnostic error cases comprising the majority of medical errors identified when last analyzed [37]. To properly house the growing number of M&M cases a web-based resource known as the AHRQ Patient Safety Network (AHRQ PSNet) was created in 2005. In addition to the web M&Ms, the site features annotated reviews of the current literature and studies and practical guidance for clinicians to improve patient safety. In the ensuing two decades PSNet has become “the single most useful piece of infrastructure produced for the field of patient safety” [38]. Unfortunately, this valuable resource has been threatened by recent federal funding cuts that led to cancellation of the contract that funded the maintenance and updating of PSNet. Currently, the site is no longer being updated and absent a reversal, may cease to exist entirely. (There are efforts to have the Coordinating Center for Diagnostic Excellence, housed at UCSF, provide a similar resource in the future) [38].

Other recent activities

In recent years, AHRQ commissioned and published 27 diagnostic safety issue briefs and eight open access manuscripts published in peer review journals designed to succinctly summarize the state of science for organizations, clinicians, and patients to begin the work of actually reducing diagnostic errors [5], 39]. Examples include summaries of the current state of the science of diagnostic safety, measurement and reporting briefs, clinician reasoning educational materials, system design case studies, patient engagement activities, the role of (emerging) technology, and diagnostic issues for special populations such as older adults, rural communities, and pediatric patients. AHRQ will continue to develop new diagnostic safety issue briefs in the future.

AHRQ has also published three different diagnostic safety toolkits 1) Calibrate Dx, a self-evaluation tool for clinicians to assess and improve upon their own diagnostic decision making [40], 2) Measure Dx, a tool for healthcare organizations to measure and promote diagnostic excellence and reduce harm from diagnostic safety events [41], and 3) Toolkit for Engaging Patients To Improve Diagnostic Safety [30], to improve communication and information sharing within the patient-provider encounter. These tools are currently part of the Implementing Diagnostic Excellence Across Systems (IDEAS) project that is working with healthcare organizations to implement, sustain and evaluate the use of the tools [42].

Following the 2015 NAM report, the U.S. Congress requested that AHRQ convene a cross-agency work group to address the paucity of research dedicated to improving medical diagnosis, particularly diagnostic failures. The Federal Interagency Workgroup on Improving Diagnostic Safety and Quality in Healthcare was established and as of August 2025 met 19 times with meeting summaries and reports available on AHRQ’s website [43]. This workgroup included leaders from multiple Federal agencies including the CDC, the VA, Department of Defense, FDA, Indian Health Service, NIH, HRSA, ONC/ASTP, SAMSHA, and OASH, along with AHRQ, with a goal for “Federal agencies to develop a coordinated research agenda on the diagnostic process and diagnostic errors and to commit dedicated funding to implementing this research agenda” [6]. The workgroup has collaborated to produce issue briefs, and their meetings provide networking opportunities and facilitate coordination of activities focused on improving diagnosis across federal agencies.

More recent AHRQ work supported a NASEM workshop entitled Advancing Equity in Diagnostic Excellence to Reduce Health Disparities: A Workshop, and a special supplement in Academic Emergency Medicine devoted to research on improving diagnosis in emergency care [44], 45]. Both projects were instrumental in furthering diagnostic safety dialogue and knowledge.

Disease specific funded grants

AHRQ has funded a series of projects related to specific conditions including stroke (e.g. work on dizziness and stroke misdiagnosis), pulmonary embolism, cancer, sepsis, and maternal health [46], [47], [48], [49], [50]. Nonetheless, AHRQ’s ability to fund disease-specific research pales in comparison to the National Institutes of Health’s budget for these diseases. There has been a healthy tension and thread running through the diagnosis community and AHRQ-funded improvement activities between “diagnosis agnostic” cross-cutting processes, and disease-specific errors and pitfalls that are uniquely related to particular diagnoses and their specialized knowledge and tests [47], 51], 52].

Challenges and future uncertainties

From our perspective as safety researchers and recipients of AHRQ funding for diagnosis improvement work, we see many challenges and opportunities to improve the effectiveness, efficiency, equity, and reach of AHRQ’s work in diagnostic safety. The magnitude of diagnostic errors and quality improvement opportunities warrants sustaining and expanding the public funding of the type of work AHRQ has funded and coordinated over the past two decades. This includes longer term support for research centers and researchers as called for in chapter eight of the NASEM report [6].

We should be looking for ways to spread the reach and impact of AHRQ’s diagnostic safety tools and guides. The tools and information on the AHRQ and PSNet website are only as good as the quantity and quality of the people they reach and who adopt them. Unfortunately, for many reasons both internal and external to AHRQ, we know that the reach and impact needs to be much greater, especially considering how high quality and valuable many of these tools are. AHRQ’s recent commitment to the IDEAs implementation project is one example of their attempt to test and implement several of their diagnostic safety tools [42].

At present, AHRQ’s funding for diagnostic safety is uncertain as federal funding is threatened. To the chagrin of the health services research and patient safety community, AHRQ’s budget is proposed to be cut by 35 % (from $369 million in 2024 to $240 million in 2026) and its staff has been significantly reduced. As part of the March 2025 Executive Order, “Implementing the President’s ‘Department of Government Efficiency’ Workforce Optimization Initiative” AHRQ will be combined with the Office of the Assistant Secretary for Planning and Evaluation (ASPE) into the Office of Strategy “to conduct research that informs the Secretary’s policies and evaluates the effectiveness of the Department’s programs for a healthier America” [53], 54]. However, supporting AHRQ’s activities should be among the highest priorities for the stated goal of achieving a healthier, safer country, especially given what we have learned over the past two decades regarding the frequency of and harm from diagnostic errors [3], 55], 56]. Considering rapid changes in health care delivery, diagnostic and information technology, AHRQ should be expanded to help answer the many hard questions and overcome challenges in truly making diagnosis safer.

Imagining a more ambitious future

AHRQ has made a substantial impact on a host of measurement and improvement activities related to diagnostic safety and the reduction of preventable harm to patients. Instead of cutting back AHRQ and its diagnostic safety work, we should be reaching higher and further to more boldly and creatively imagine an agency with a far greater reach and impact. What if the motivated community that has come this far can receive additional AHRQ support to produce a 2025 follow-on NASEM-type report with recommendations to lead our country and the world to safer, more cost effective, and patient centered diagnosis?

We asked ChatGPT to contemplate what this might look like. Without hesitation, caution, or political inhibitions, it instantly produced a proposal that would fulfill many of our dreams for a brighter future (Supplementary Material). The AI program even suggested a name for a proposed new NIH Institute – the National Center for Diagnostic Excellence (NCDE), structured using elements from other successful NIH centers such as the National Cancer Institute (NCI), the National Center for Advancing Translational Sciences (NCATS), and the National Institute of Mental Health (NIMH). Rather than trying to edit or debate the feasibility and the likelihood of success of this new NCDE vision, here we share unedited this proposal that the ChatGPT outlined, and perhaps a future NASEM panel might someday recommend. Ideally, there should be debate on the merits of such an investment, carefully weighing the huge current human and financial costs of misdiagnosis, as well as the degree to which such an NCDE would be able to successfully reduce these harms and costs.

Conclusions

The journey to apply modern safety and quality improvement tools to achieve more timely, accurate, patient-centered, equitable, and efficient diagnosis has just begun. AHRQ has led the way in funding and evaluating tools addressing systems improvement and engineering, health IT and AI, patient engagement, teamwork and communication, and research aimed at understanding the epidemiology and causes of diagnostic errors. Despite the progress, we still have a long way to go. It would be a significant setback if there were significant reductions to federal efforts and leadership provided by AHRQ. With so many unanswered and new questions, coupled with limited other sources of funding from other sources (such as the Gordon and Betty Moore Foundation who had been providing a significant amount of funding support that has now been discontinued), now more than ever, we need AHRQ to fund and conduct diagnostic safety research to continue our progress to better diagnosis.