Sometimes, a new idea is exactly what we need to tackle a longstanding, otherwise insurmountable problem. A little bit of innovation can go a long way.
On the other hand, some new technologies are sexy and flashy but they don't really make a difference for society, or they generate new problems worse than the old.
I'm always on the lookout for creative solutions to the problem of overdiagnosis. Patients, healthcare providers, and society as a whole need to make changes to help create a sustainable, high-quality health care system.
In Doctors create app to help diagnose, treat patients at point of care, Globe and Mail journalist Ivor Tossel describes the aims behind SnapDx.
Dr. Rahul Mehta, an internal medicine resident at the University of Calgary, partnered with colleague Dr. Aravind Ganesh to create the SnapDx app. The app uses evidence-based guidelines to help guide physicians delivering care.
SnapDx Clinical is an efficient bedside assessment tool designed for use by medical trainees and clinicians at the point of care.
We provide the best evidence-based questions and tests to be used as part of your history and physical examination to confidently sort through your differential diagnosis. (from iTunes App description)
The idea is that SnapDx can help aid decisions about diagnosis, giving clinical probabilities that might override the need for ordering laboratory or radiological tests. It does this by emphasizing thorough physical exams, filtering these findings through well-evidenced decision-making tools, and providing probabilities for diagnosis.
Try downloading the SnapDx App (iTunes) yourself, or see the screenshots below for an idea of what it looks like.
I applaud the effort, and I imagine it must have taken a heroic effort to tackle the statistical nightmare behind the scenes, converting everything into a standardized interface.
Despite recognizing this, I must admit I found it a bit cumbersome and hard to understand. Each section has an estimated pre-test probability which is often set to the prevalence rate from a major research study; then, you tick yes/no to various scoring criteria (which are helpfully described in the Info sections). With this, you see the probability for/against a diagnosis. I think. Though it doesn't explicitly say if you should order a test, or which test you should order.
I got a bit bogged down in the details. One big issue I had is that I was not clear on is how to set the positive pre-test probability accurately.
For example: when I tried the Pulmonary Embolus (PE) tool, I was thinking of a patient who had recent surgery, recent cancer, immobilization, chest pain, tachypnea, no fever, a normal blood gas, and a normal chest x-ray. There was almost no other diagnosis possible besides a clot in his lungs. Yet because he did not have signs of DVT, hemoptysis, a clotting history, or tachycardia he would not score very high on any of the scales. Of course my pre-test probability for him was high, but I don't know if it was 50% or 99%, and the possible harms of a CT-scan were outweighed by the benefit of ensuring the treatment (high-dose blood thinner; possibly quite harmful) was in fact necessary.
The ambition of Drs Mehta and Ganesh is admirable, and I will keep the app around, looking for future iterations of it. It has the potential to improve clinical accuracy and to decrease ordering of tests that would only confirm what we already know from physical exam.
Using our detective skills rather than requisitioning a test? It is a great idea.