September 13, 2023 | Researchers from Johns Hopkins University and the Risk Management Foundation of the Harvard Medical Institutions have come up with a national estimate of serious harms from diagnostic error. An estimated 795,000 Americans become permanently disabled or die each year because dangerous diseases are misdiagnosed, according to findings published in BMJ Quality & Safety (DOI: 10.1136/bmjqs-2021-014130), and stroke was found to be the top cause of serious harm from misdiagnosis.
“The reason why stroke is on top is because it’s a combination of a relatively common dangerous disease that’s diagnosed at a relatively high rate,” David Newman-Toker, lead investigator of the study and director of the Johns Hopkins Armstrong Institute Center for Diagnostic Excellence, told Diagnostics World.
Diagnosing a stroke in progress is critically important because appropriate treatment relies upon knowledge of the type of stroke and sometimes the location of the brain injury, according to an American Stroke Association fact sheet. Part of the challenge also involves ruling out other conditions that may resemble a stroke.
Quickly identifying the signs and symptoms and seeking immediate medical attention can help reduce long-term disability and save lives, according to Patrick Brouwer, a neurointerventionalist and head of worldwide medical affairs at CERENOVUS, part of Johnson & Johnson MedTech. The sooner this disease can be diagnosed and treated, the sooner blood flow can be restored, which correlates to improved rates of survival and repair of functional abilities.
On the other hand, if a stroke is not diagnosed or treated promptly, there can be severe fallout. Brouwer, who was not involved with the BMJ study, explained that brain tissue not receiving blood flow will eventually die, and the result is that nerve cells cannot send signals to parts of the body responsible for crucial functions such as speech, thought, memory, or motor function.
To highlight the dramatic differences that exist among different diseases, Newman-Toker drew a comparison between stroke and heart attack. He pointed out that both conditions impact roughly a million people every year, depending on which patient groups are counted. “For diseases that are roughly equally common, you’d hope or think that we’d be about equally good at diagnosing them,” he said.
But that’s not the reality, according to his group’s recent findings. He explained that stroke is missed in roughly 17.5% of cases, whereas the heart attack error rate is down around 1.5%. The reason stroke is missed more often is twofold, according to Newman-Toker, who is also a professor of neurology, ophthalmology, & otolaryngology at Johns Hopkins. “One is that it’s intrinsically a little bit harder to diagnose stroke,” he said. “And two is that we’ve devoted a lot less attention to the problem than we have to diagnosing heart attack.”
A significant risk factor for ending up with misdiagnosis is the absence of an obvious clinical presentation. Strokes aren’t commonly missed when patients present with an inability to talk and weakness on one side, for instance, because it is relatively clear what is happening in these cases. Strokes tend to be missed when patients present with dizziness that resembles inner ear disease or headaches that resemble migraines, Newman-Toker explained
It may sound a bit trivial, he continued, but it’s an important point: “We miss things when they’re subtle—not when they’re obvious.” And one notable difference between these two diseases is that the symptomatic manifestations are broader and more complex for stroke than heart attack.
Beyond that, he pointed out the difference in time and energy devoted to these diseases over the course of decades. “Heart attack, in many ways, is kind of the gold star example of what we could be doing,” he said. “We’ve spent a lot of time over the last 50 years developing electrocardiograms and bioassays for heart enzymes of increasing accuracy and developing chest pain pathways and creating quality measures to improve diagnosis of patients with heart attack.” When it comes to stroke, however, he sees plenty of room for improvement.
Four-Pronged Approach to Stroke Solutions
Several stroke-centered efforts are underway at Johns Hopkins using a “four T” paradigm (teamwork, training, technology, and tuning) as an approach for considering solutions that address diagnostic errors. Teamwork involves getting access to expertise, training involves building up expertise, technology involves “bottling” that expertise, and tuning involves feedback and calibration.
“We don’t believe that there are any magic bullet solutions to any of these problems because it’s actually quite complex,” Newman-Toker said. Some are healthcare delivery issues, some are training issues, and some are technology issues. “So we have to actually develop all of these things in concert to have multiple safety nets to prevent misdiagnosis.”
On the teamwork front, he and colleagues have developed a remote specialty consultation program. Conceptually, it can be considered something along the lines of a “phone-a-friend” program. “We’ve shown that for stroke, the single biggest risk factor for being missed is having dizziness,” he said, noting that it dramatically increases the odds of being missed over the baseline rate of being overlooked. So that is where they’ve focused their attention initially.
According to Newman-Toker, if experts (sub-specialists who see and evaluate dizzy patients for a living) analyze people’s eye movements, they can discriminate between inner ear disease and stroke with greater accuracy than even an MRI scan in the early stages of the disease. “The first thing that we can do is make that expertise available remotely via this kind of phone-a-friend second consult idea,” he said. “We put eye movement recording technology in the emergency department, and then people call us, and we help them out.”
At the same time, there is a realization that there aren’t enough sub-specialists to go around, so there is a need to train people in scalable ways. One effort involves taking clinical trial data and “Tele-Dizzy” consult work and putting them into a virtual patient simulator software program. This makes it possible to use real cases to train people who are unfamiliar with evaluating stroke using eye movements.
In 2021, Newman-Toker and colleagues published a study (DOI: 10.1515/dx-2020-0127) showing that just nine hours of simulation training enabled medical interns to be more than twice as accurate as residents in diagnosing dizziness. “In other words, we’re trying to encapsulate and make more efficient the learning around the diagnostic process for this problem,” he said.
There are also efforts underway to use machine learning and artificial intelligence to help with the interpretation of eye movements using computer-based algorithms. The researchers are taking recordings from clinical trials and clinical care pathways and training algorithms in hopes that they will eventually be as effective as experts.
While highlighting “tuning” efforts, Newman-Toker explained that “right now, we don’t have good mechanisms for giving people feedback on their clinical performance at either the individual or institutional level.” He and colleagues have developed methods to offer that feedback, enabling professionals to “tune” their diagnostic skills and performance over time.
There’s another aspect involved. “If we’re trying to move the needle, we’ve built the needle to be moved,” he said. And that needle is a quality measure that reveals the rate at which patients are discharged from hospitals (without a stroke diagnosis) and then return and become hospitalized with stroke in a short timeframe.
“We can statistically determine how many are occurring above the expected rate based on the epidemiology in those same populations,” he said. Essentially, they are measuring the rate at which patients are experiencing adverse events or being harmed as a result of a missed stroke.
These solutions are in various stages of development, but all of them are designed with the potential to scale. The Tele-Dizzy consultation service is already deployed locally at Johns Hopkins, and there are plans to deploy it elsewhere with external partners. Meanwhile, the quality measures are under review nationally. If approved, they could potentially start to be deployed over the next couple of years, he said.
The technology program, which combines algorithm development and mobile phones to measure eye movements, is in an earlier research stage. “We’ll probably need to go through some clinical trial phases before it’s ready for FDA-approval-type of status and clinical dissemination,” he explained.
In Pursuit of a Saliva-Based Stroke Test
In the UK, meanwhile, University of Birmingham researchers and industry partner Marker Diagnostics are working to develop a rapid, saliva-based diagnostic test to identify patients in need of timely treatment to help ward off irreversible brain damage.
The researchers hope to pinpoint salivary small non-coding RNAs (sncRNAs) that will identify stroke and distinguish it from conditions that can mimic stroke, like seizures or migraines. They will also explore whether biomarkers can differentiate between the two main types of stroke, which call for different treatment approaches.
The Golden HOur for STroke (GHOST) study will include the West Midlands Ambulance Service University NHS Foundation Trust, Midlands Air Ambulance Charity, and University Hospitals Birmingham NHS Trust. Funding from the Stroke Association will include support from industry partner Marker Diagnostics UK to weave their Birmingham Biohub lab’s sncRNA discovery into the study.
The GHOST study is expected to last three years, with results anticipated in late 2026. It will be led by Antonio Belli, professor of trauma neurosurgery at the University of Birmingham’s Institute of Inflammation & Ageing.
Marker has previously developed a concussion test stemming from research that emerged from the University of Birmingham, and the company’s scientists plan to build a commercial pathway for any tests that might arise from the forthcoming research.
“Our previous studies detected ultra-early biomarkers and cellular responses that had never been reported in human studies before and resulted in a non-invasive diagnostic test that could change the way concussion is managed,” Belli said in a news release. Now, he and colleagues hope “to repeat this success with stroke.”
The GHOST study will rely upon the same methodology as these previous trials. Paramedics will gather blood, urine, and saliva samples from patients suspected of experiencing a stroke within the first hour following symptom onset. Additional sampling will continue as patients receive standard care within a hospital setting.
Richard Francis, head of research at the Stroke Association, expressed enthusiasm for the study because he sees its potential to lessen the amount of death and disability caused by strokes, both in the UK and globally.
Paramedics in the UK currently rely on the FAST (Face, Arms, Speech, and Time) test to determine whether someone is experiencing a stroke. The acronym can help with the recognition of signs and symptoms, but this method has its shortcomings. Paramedics can misdiagnose other conditions as stroke, for example, and not all those who suffer a stroke experience FAST symptoms.
“Having a saliva test would be a massive step forward in pre-hospital diagnosis for stroke and really help people to get the right diagnosis, to get to the right hospital for the right treatment and in the quickest time,” Francis added.
Paul Nicolaus is a freelance writer specializing in science, nature, and health. Learn more at www.nicolauswriting.com.