The next revolution in radiology is not about a new imaging modality or machine learning, but about reducing reporting variation.
Recently, I attended the opening ceremony of the 4th World Congress of Thoracic Imaging, in Boston, and who was on the dais? None other than James A. Brink, MD, Chair of the ACR Board of Chancellors. He had been invited to give the Distinguished Congress Lecture. His presentation was entitled "Is the Art of Medicine Dead in the Era of Population Health Management?"
Aptly, Dr. Brink segued from his opening remarks to art, but not to the art of medicine. Instead, the next slide featured El Jaleo (1882) by John Singer Sargent, a 7'10" x 11'5" painting that hangs in Boston's Isabella Stewart Gardner Museum. Coincidentally, before travelling to Boston, I had already acquired a library discount pass to the museum for that afternoon; so I got the opportunity to see (and photograph) this artwork in situ:
Dr. Brink's inspiration was an article entitled "Art and the Uncertainty of Medicine," in which a group of medical students learning observation skills were pondering, among other things, whether the position of the contorted right arm of the dancer in the painting's foreground (see painting detail below) is natural or unnatural. As a chest radiologist, I noticed that the dancer's right arm position is similar to that when positioned for a PA chest radiograph, ie wrist dorsum touching the hip and elbow pointing forward (as much as possible), resulting in the scapula rotated away from the ribcage. Thus, I have a high degree of certainty about the dancer's arm position being exaggerated, but natural. On the other hand, I have only moderate certainty that her right thumb and forefinger are holding up a fold of the fabric of her long skirt, to prevent her from tripping on it while dancing. But perhaps she is exuberantly swishing the fabric in rhythm with the music. Greater uncertainty increases the range of possible alternate answers ie differential diagnoses.
The author of the article says physicians must develop "a tolerance and healthy respect for uncertainty, as well as recognition of the prominent role it plays in a patient's experience of illness." Uncertainty in medicine is a given. As diagnostic radiologists, we review imaging studies and provide opinions, impressions, interpretations —however you want to say it—in our reports, as well as make recommendations. Whenever possible, we try to reduce (or even eliminate) uncertainty.
So uncertainty is an enemy that needs vanquishing, right? No, says Dr. Brink. Rather, what needs to change is variation in reporting and making recommendations. Variation control is how radiologists can positively impact population health management by supporting uniform and appropriate follow-up care. Radiologists can achieve variation control by using algorithms that provide guidance on how to report and manage specific imaging findings.
We've all heard about decision support for providers, to order appropriate imaging studies when needed and avoid unnecessary imaging studies. Computer-assisted reporting represents decision support for radiologists, with care algorithms integrated directly into the dictation software platform. If all radiologists apply the same algorithms to the same imaging scenarios, then variation can be reduced for some report elements, such as classification (eg splenic laceration grading) or recommendation (eg pulmonary nodule management), no matter where a patient gets imaged or who reports the imaging.
The day before Dr. Brink's presentation, I walked the Freedom Trail and took a photograph of the Old South Meeting House (see the top of this blogpost), where Samuel Adams gave the signal to more than 5000 gathered colonists to start the Boston Tea Party on December 16, 1773. Thinking of this iconic historical call-to-action, as I listened to Dr. Brink, I had to wonder: Are we radiologists on the verge of a revolution in our profession? By revolution, I don't mean uprising, resistance, or rebellion, but I do mean transformation.
Any reports of the demise of the art of medicine are greatly exaggerated. Our artistry is evidenced in how we describe our image observations and synthesize those descriptions into an elegant whole. By transforming some elements of our reports, when appropriate, image-directed actions may become more uniform, whether in Boston or Topeka or Anchorage. If there are choices among those actions, the art of medicine is again used to assess patient-specific factors and preferences to select the best one.
T-shirts being sold by a street vendor near the start of the Freedom Trail proclaim: "Wicked smaht." I should have bought one for Dr. Brink. Next time I go to Boston, I will have to visit the Gardner Museum again, as well as a certain t-shirt vendor. No uncertainty there.
Campbell JL. Art and the Uncertainty of Medicine. JAMA 2014; 312:2337-2339. http://jamanetwork.com/journals/jama/fullarticle/2020377
Boland GWL, et al. Decision Support for Radiologist Report Recommendations. JACR 2011; 8: 819-823. http://www.jacr.org/article/S1546-1440(11)00442-X/fulltext