Computers Could Help Doctors Correctly Diagnose Skin Infections
LA BioMed Study Shows Technology Could Help Protect Patients
LOS ANGELES, CA, March 15, 2011 – A new study suggests technology can help prevent misdiagnoses and the healthcare costs associated with them. In the study, a computer-based software system designed to assist in the diagnosis of dermatologic and other visual medical conditions performed better than physicians in correctly identifying common skin infections.
The study, published today in the Dermatology Online Journal, found the physicians in two hospital emergency departments misdiagnosed 28 percent of patients with inflamed skin conditions as having cellulitis (deep skin infection). Based on symptoms input into the computer, the software called VisualDx, correctly diagnosed 64 percent of cases where the physicians had misdiagnosed cellulitis at one of the hospitals. The same system is also available on mobile devices and can be carried by physicians in their pockets to the bedside.
“Misdiagnosing non-infectious diseases as cellulitis is a common error and can result in unnecessary hospitalization and antibiotic use,” said Noah A. Craft, MD, PhD, the principle investigator at Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center (LA BioMed) and the corresponding author of the study. “This study shows that technology can help physicians make the correct diagnosis so that patients receive the best care possible and avoid unnecessary, expensive treatments.”
Cellulitis is a common bacterial skin infection that, if left untreated, can become life-threatening. It appears as a swollen, red area of skin that feels hot and tender, and other studies have found less serious skin infections are often misdiagnosed as cellulitis.
In the new study, which was conducted by LA BioMed researchers, specialists found emergency department physicians misdiagnosed cellulitis in 41 of 145 patients admitted to two hospitals. The researchers then entered symptoms and other health information from the 28 misdiagnosed patients into VisualDx, and in 18 (64 percent) of those patients, VisualDx suggested the correct alternative skin condition as a diagnosis. Asked to evaluate the same 28 patients, the physicians caring for the patients offered the correct alternative diagnosis in just four (14 percent).
“While this is a small sample, misdiagnosis of cellulitis can result in the unnecessary costly hospitalizations and potentially dangerous use of intravenous antibiotics that may further fuel the emergence of antibiotic resistant bacteria,” said Dr. Craft. “Conversely, if physicians fail to correctly diagnose patients who do have cellulitis, the patients could suffer worsening infection and serious complications, if they’re not treated promptly. Clearly, improved diagnostic accuracy is needed for cellulitis and a structured approach to diagnosing it could decrease misdiagnosis and improve the quality of care.”
The VisualDx technology used in the study takes physicians through a series of categories to fully describe their patients, the symptoms and circumstances surrounding the disease. Based on that information, VisualDx displays information about potential diagnoses and treatments, and it displays the relevant images from the world’s most comprehensive collection of digital medical images. Ultimately, this technology still depends on doctors to make the correct final diagnosis, but VisualDx can help physicians by suggesting a more accurate list of possibilities to consider.
LA BioMed researchers–Dr. Craft, Loren Miller, MD, MPH, and Carol Berkowitz, MD—contributed their expertise to the development of the VisualDx technology. Dr. Miller also participated in the study. Dr. Craft is a consultant for Logical Images, Inc., the makers of VisualDx, and another of the study’s authors, Art Papier, MD, is the company’s chief medical information officer. The study was supported in part by GCRC Grant M01-RR00425 National Center for Research Resources. To read the study, please visit: http://dermatology.cdlib.org/1703/index.html