This article by Ambra Health CEO, Morris Panner, was originally published on Forbes on August 12, 2019.
According to Kaiser Health News, “Walmart Inc., the nation’s largest private employer, is worried that too many of its workers are having health conditions misdiagnosed, leading to unnecessary surgery and wasted health spending.” Company officials discovered that about half of its workers seeking treatment were advised to undergo back surgeries and did not need them.
Radiology was a major contributor to these high error rates. While the average real-time error rate in typical radiology examinations is about 3-5%, in the case of advanced images run by CT scans and MRIs, retrospective errors from false-positive readings can reach as high as 30% of diagnoses.
Diagnostic errors are an indication of poor patient care. According to research conducted by Johns Hopkins University, over 250,000 (or 10%) of all U.S. deaths are due to medical errors. Radiology plays a key role in the diagnostic process for many patients. According to a Coverys report (via Radiology Business), a staggering”80% of missed diagnosis claims in radiology result in either permanent injury or death.”
Misdiagnosis Stemming From Cognitive Errors
When Harvard Medical School famously conducted a test by superimposing an image of a tiny gorilla on a lung Xray, 83% of radiologists failed to notice the gorilla, even when it was 48 times the size of the nodule they were looking for. Apparently, these professionals were so busy looking for what they’re trained to find that they completely ignored the gorilla staring at them in the face.
Cognitive biases are usually “the result of psychological distortions in the human mind, which persistently lead to the same pattern of poor judgment, often triggered by a particular situation.”
Misdiagnosis Relating To System Failure
The second factor that affects diagnostic readings is system-related and includes not only equipment failure but problems caused by policies and procedures that bring about undue visual and mental fatigue. These factors can include heavy workloads, lighting conditions, shift length or timing, task repetitiveness, pace of reading images and other environmental distractions such as phone calls and other interruptions.
The constant staring at static radiographs and other images can lead to visual exhaustion, physical discomfort, eye strain and low motivation that, when combined, inevitably results in reduced detection accuracy. Radiologists and physicians need to make critical decisions constantly, which results in mental strain, giving way to decision fatigue. This can result in poor judgment and diagnostic errors.
How can we ensure that in the future we don’t miss the gorilla? Here are ways to lower radiology errors and misdiagnosis in radiology exams, according to best practices advocated by the American Journal of Roentgenology.
1. Instituting a radiologist peer-review practice. Putting in place continuous, structured and systematic procedures that evaluate work performance via peer review is one of the better ways to improve diagnostic accuracy. This peer-review process should be built upon a commitment toward diagnostic accuracy and mutual respect. Anonymous feedback and regular reviewer ratings can help identify opportunities for additional education and self-improvement for practicing radiologists. Several peer review software solutions are available in the market that can aid in managing this process efficiently.
2. Managing workloads that reduce burnout. There is a wealth of research and literature available on physician burnout. Excessive workloads, long work hours, electronic health records, documentation for insurance filings and more have been known to cause excessive burnout in physicians, leading to poor job satisfaction and increased medical errors.
Instituting double reads, limiting the length of work shifts, establishing structured breaks, and switching between modalities during the workday can all be part of an orchestrated strategy to improve productivity and reduce diagnostic errors.
3. Deploying AI-based decision support systems. Advances in artificial intelligence (AI) and machine learning technology are leading to the rise of remarkable new capabilities and tools that have the ability to read large datasets and discover meaningful insights to advance patient therapies.
In a recent example, a commercial artificial intelligence (AI) system “matched the accuracy of over 28,000 interpretations of breast cancer screening mammograms by 101 radiologists.”
Measuring the speed of detection in exposing specific lung nodules on chest CT images, Harvard Business Review reported how AI-driven automated analysis outperformed a panel of radiologists by as much as 62% to 97%. It estimated $3 billion in annual savings could be realized.
4. Consistent follow-up. One of the key reasons for diagnostic errors is the lack of follow-up on incidental findings. The solution needs to be one that puts a system in place to ensure a timely follow-up of imaging recommendations that can result in optimal patient care. Again, technology can come to the rescue, wherein a data visualization platform can be built and powered by an algorithm that specifically identifies cases with a follow-up imaging recommendation.
5. Structured reporting. One of the benefits of structured reporting is that it improves communication between the radiologist and referring physicians and can offer a viable solution for solving the cognitive bias problem. A structured report acts like a checklist that can be used to improve thinking, reduce reliance on memory and guard against a negative emotional state that is an outcome of fatigue. Structured reporting is already rewriting the rules and processes in major hospitals across the country.
6. Education. This is one of the lowest hanging fruits when it comes to solving the misdiagnosis problem. It ideally needs to be a multi-pronged approach that includes case-based learning, focuses on real-world diagnostic decisions, putting a premium on the convergence of intuitive and analytical thinking. Learners must be evaluated on diverse competencies that include the ability to use the right technology, knowledge to make the right diagnosis, awareness of cognitive biases and corrective measures and more. Learning management systems and e-learning platforms can help train budding radiologists or used for continuous education.
A multipronged strategy is required to ensure that diagnostic errors receive much-deserved resources and attention including the need to invest in information technology, tools, and processes that can aid in making accurate diagnoses, freeing up time for issues that require greater intervention.