Author Art Papier, M.D., says VDDSS solutions can bring decision support to new levels of efficacy.

Medical errors and startlingly high levels of misdiagnoses have taken a toll on the public’s confidence in our healthcare system. The human costs, mistrust in the system and economic impact are considerable. According to recent studies, nearly 20 percent of diagnoses are incorrect. Why such levels? The nature of medical school training, financially driven time pressures put on doctors, a fragmented system and ever-increasing medical data are part of the problem.

There is simply too much to know and certainly too much for any primary care doctor to possibly memorize. Why is it, then, that diagnostic decision support is rarely used? For all the hand-wringing and conversations about improving quality of care, this technology is often discussed but rarely implemented.   

Doctors cite complexity, cost, workflow difficulties and outright skepticism that clinical decision support works.   

What can change this? The missing element in clinical decision support that counters all of these claims comes down to pictures. This may seem simplistic, but visual diagnostic decision support addresses the non-analytical, instantaneous perceptual moment occurring when the doctor examines the patient and then recognizes or does not recognize symptoms.

Visual diagnostic decision support can help tremendously, considering that half of all diseases have a skin or pattern clue, and 10 to 20 percent of a general practitioner’s diagnosis is visually based. Yet generalist practitioners receive minimal training in how to recognize these clues.

The Challenges of Diagnosis
Medical errors around physician orders and drug reactions grab headlines, but the more serious and underlying concern is misdiagnosis. The attention away from diagnosis-as-error is due to the systemic and cultural complexities of reporting and collecting data on diagnostic errors. Yet diagnostic error leads to incorrect treatment, failure to use an indicated diagnostic test, misinterpretation of test results and failure to act on abnormal results. Cognitive errors may result in poor quality of care, patient safety risks, increased costs and, in some cases, malpractice litigation.    

In their medical education, physicians are provided a highly structured method in which to think, and they habituate this methodology throughout their training and subsequent practice. Medical education stresses memorization of basic science and clinical facts; this means that students must focus on prototypical “classic cases” rather than learning all the variants.

As the medical student moves from the classroom to residency apprenticeship years, training shifts to “practice-based” learning from unique clinic or hospital patient cases. These residents soon realize that most patients do not neatly present as the textbook suggests. Thus, the life-long learning of the physician begins.    

Expertise is bred from experience; expertise evolves from interaction with thousands of patients, learning from the twists and turns of each individual “case,” synthesizing and remembering the vast array of symptom and examination patterns over a career. Our patients hope we have this expertise and assume if we are early in our careers, we have developed a methodology to think about, recognize and diagnose their problem regardless of how much experience we actually have. Yet a great number of patients will appear in our offices with patterns we have never seen in practice or read in texts.

If you practice family, internal, pediatric or emergency medicine, you are expected to recognize the pattern and make diagnoses that span all of the medical specialties. Given the immense variations of disease, it is clear that the frequent variants of the common as well as the rare diagnoses might be difficult for these non-specialty physicians to recognize, even after 20 years of practice, never mind the first five or 10.      

There are hundreds – if not thousands – of skin, eye and oral clues of diseases. When physicians are faced with making a diagnosis from skin or pattern-recognition-based clues, they readily admit they are insufficiently trained to recognize visual clues. This challenge is compounded every day in fast-paced clinics, emergency rooms and hospitals, where generalists are forced to make quick decisions, often with incomplete data and a dearth of experience in evaluating the subtleties of disease characteristics.     

Many practitioners falsely believe that search engines are an answer. But you cannot search by a diagnosis if you do not know the diagnosis, and there is the matter of accuracy. In a study by Tang, et al., Google-aided diagnoses were accurate 58 percent of the time. This rate is poor and should not be acceptable for medicine.    

Diagnostic decision support systems, though rarely used, allow physicians to enter their patient symptoms and other medical factors, such as laboratory results, to build a text-based listing of diagnostic possibilities. All of the diagnostic systems to date limit the dynamic nature of medical diagnosis and do not allow for the incorporation of perceptual and visual data into clinical thinking. Many physicians do not know how to verbally describe the visual clues and patterns they observe, and extensive words on a page or screen make it difficult to recognize patterns of disease. Furthermore, the findings in the physical examination are essential elements for diagnostic acuity.