NEW YORK (360Dx) – Misdiagnosis affects at least one in 20 US adults each year, but health systems don't have a systematic approach to measure and improve diagnostic performance, according to researchers at the Baylor College of Medicine.
In an attempt to remedy the situation, which is surprisingly pervasive and can have devastating consequences for patients, researchers from Baylor and Geisinger Health System are collaborating under a $3.5 million grant from the Gordon and Betty Moore Foundation to develop a program to address diagnostic errors. Within a few years, they expect to generate tools and best practices that could provide a basis for an advisory service to clinicians and others working within health systems to help reduce such errors and improve patient care.
"In the outpatient setting, we've found quite a few common conditions being missed," said Hardeep Singh, a principal investigator on the project who works at the Houston VA Center for Innovations in Quality, Effectiveness, and Safety and is an associate professor of medicine at Baylor. "These are not just unusual conditions," he said.
And although diagnostic tests are not the sole source of inaccuracies, they are among the leading participants in process errors, Singh said.
Diagnostic inaccuracies emerge when clinicians test for and treat rare and unusual conditions, but they also tend to show up frequently when clinicians prescribe routine tests, he noted.
Test performance has also emerged as a key concern. Not all tests are as accurate as they could be, and clinicians are not always able to properly interpret the results, Singh said. This is true of some common tests that certain clinicians are not familiar with or are not used to interpreting on a routine basis, as well as relatively new tests that are also not familiar to some clinicians, he noted.
"As you go into new categories of testing, including genetic and biomarker tests, for example, the likelihood that they will be misinterpreted is higher," Singh noted.
Still, the researchers have found misinterpretation of results in "bread and butter," processes, such as tests of hemoglobin, kidney function, or blood chemistries, he noted.
The point is that with the more sophisticated tests, clinicians may be challenged in integrating the results they are receiving with what they are seeing relative to a patient's condition, he said, "but the same thing is true of simpler tests that some clinicians don't normally deal with."
Overall, there are several reasons for diagnostic errors that can be described by categories, or dimensions, of process breakdowns, Singh said.
"We've created a conceptual approach to address diagnostic errors that includes several dimensions, and within each of these, the causes might be different," Singh said.
In the category of patient-physician interaction, rushed visits are an important challenge, he noted. Physicians, sometimes overwhelmed by the demand for their services, don't put adequate effort into listening to a patient's medical history, and they are besieged by administrative tasks, such as accurately recording information in the patient's electronic health record.
Healthcare providers are so rushed, in some cases, that they don't take action on laboratory test results, Singh said. These providers say that, in some cases, they have so many tests that they cannot handle them properly, he added.
"If there's an abnormal test result, we often see that there is a lack of follow up and some patients fall through the cracks," Singh said.
He noted that his research to this point, over a period of 10 years, has shown that ordering and interpreting test results is one of the most important sources of errors. Decisions around who is responsible for interpreting tests and making decisions based on the results are also a problem, he said. These kinds of errors tend to occur frequently when clinicians are responding to abnormal results, and confusion emerges among general practitioners, specialists, and others around who have responsibility to follow up on the results.
"We see a diffusion of responsibility around who should follow up on these abnormal results as one of the big things in our work," Singh said.
The researchers will use the grant from the Gordon and Betty Moore Foundation — which includes improving the experience and outcomes of patient care among its missions — to develop what they are calling a Safer Dx Learning Lab.
It would enable Baylor to work with health system leaders and clinical teams, such as those at Geisinger, to translate their research of diagnostic inaccuracies into improvements in care for patients, Singh said.
Ultimately, he hopes to provide a range of services stemming from the collaboration with Geisinger, which clinicians and other healthcare providers could use to improve the general standard of care and diagnosis.
Dennis Torretti, associate chief medical officer at Geisinger Medical Center in Danville, Pennsylvania, said that the healthcare system has formed a committee to improve clinical diagnosis and to target vulnerabilities in the diagnostic process. He said that he believes that the partnership with Baylor "will yield additional insight and approaches to the problem of diagnostic error."
The Baylor-Geisinger group has dedicated a researcher to the project who will connect the research team in Houston with the clinical operations team at Geisinger.
The project was partly inspired by a 2015 report published by the National Academies of Medicine entitled Improving Diagnosis in Health Care, which stated that most people will experience at least one diagnostic error in their lifetime, "sometimes with devastating consequences."
The report also recommended that health systems develop approaches to "identify, learn from, and reduce diagnostic errors and near misses in clinical practice." Singh said that the lab they are working on will build upon this recommendation.
In addition to analyzing data on missed diagnostic opportunities by leveraging electronic health record data warehouses, the team will evaluate how best to gather diagnostic safety concerns directly from patients and clinicians.
The research also will focus on estimating the cost of missed diagnostics opportunities, Singh said. Understanding the economic benefits associated with making safer, accurate, and timely diagnoses could provide further stimulus for efforts to reduce diagnostic errors across many health systems, something that has been slow to get off the ground, he said.
They will cast a "wide net" that consists of three categories of diagnostic dysfunction.
First, by retroactively looking at electronic health records, they may be able to trace patterns in diagnostic errors for patients that unexpectedly become sick a few days after they have seen a physician, Singh said.
They are also working to develop tools and services that are designed to elicit information from clinicians about their challenges related to the diagnostic process, and similarly from patients to better understand their challenges.
Their ultimate objective is to employ a collaborative approach with many stakeholders in the diagnostic process, he said. "We don't think people are intentionally generating diagnostic errors," he said.
Having developed tools and services, the key question as to how "we disseminate this information and make it useful to health systems" still remains, he said.
"We realize that not everything we develop is going to succeed and that's why we call this a lab. Some things will work, and some may not," he added. "If you are a healthcare system — a hospital, clinic, or an emergency room — how would you as an entity go about improving the diagnostic process in terms of these breakdowns," Singh said.
The researchers hope to focus on what proves successful at Geisinger, and roll it out to other health systems. One outcome could be the development of software including apps or other tools that they find are particularly useful in identifying diagnostic errors. The researchers would then advise the health systems on how to properly implement the tools.
After three years, the collaborators expect to have "a base of knowledge" about what could be applied to improve diagnostic processes in health systems, he said. Some people call it a "change package," where an array of choices would be available that can be applied in specific diagnostic circumstances, he noted.
Their solutions could include a toolkit of strategies that can be applied by clinicians following up on test results, Singh said.
The R&D work should also have a beneficial economic effect on health systems, he noted. The costs associated with patient malpractice suits alone are significant, he said, not to mention the economic and patient costs associated with the provision of unnecessary tests and treatments.
At present, while the national US healthcare debate is mostly about getting patients affordable access to testing and treatments that they need, and receiving reimbursement through insurance as well as having access to insurance, there is another level of conversation that needs to occur, he said, which has more to do with reducing costs to healthcare systems by reducing diagnostic errors.
Within a few years, the researchers believe that they will have a demonstration project ready for testing in multiple health systems, which will contribute to an overarching objective of health systems to reduce costs and provide value-based care, Singh said.