Home Marketplace Taking the fear out of clinical probabilities

Taking the fear out of clinical probabilities

A hospital physician designs a medical app for diagnostic probabilities

September 2012

Published in the September 2012 issue of Today’s Hospitalist

A NEW iPHONE APP developed by a practicing hospitalist aims to help physicians sharpen their diagnostic skills while brushing up on statistics and Bayesian reasoning.

The Medicine Toolkit app bills itself as a tool to help teach and apply statistics to clinical decision-making. Elizabeth Farrell, MD, a hospitalist at Beth Israel Deaconess Hospital in Boston who created the app, says it is particularly helpful in deciding when to order tests.

Take a patient who presents with shortness of breath. If you are relatively certain he or she has heart failure, you would give that patient a high pretest probability of CHF. And if the patient is really struggling to breathe and is requiring a lot of oxygen, your treatment threshold to give Lasix will be very low. In this setting, would it help to get a proBNP?
The app’s Bayes at the Bedside module will help walk you through the decision-making process by illustrating how a positive or negative test result will affect pretest probability, and how this result may or may not change how you treat the patient.

“The app makes you really step back and ask what your pretest probability is,” Dr. Farrell says, “instead of thinking that the patient could have heart failure, PE, COPD, pneumonia or the flu. Instead of ordering tests for all of those conditions, you identify your leading diagnosis and how certain you need to be with that diagnosis to proceed.”

Because concepts like biostatistics and Bayesian reasoning are slightly foreign “and fairly intimidating “for many physicians, Medicine Toolkit features a tutorial that walks users through concepts like probability and common errors in diagnostic reasoning. The app also contains two components to guide physicians’ clinical decision-making.

The first is Bayes at the Bedside, which contains more than 150 likelihood ratios. The second component, Pocket Evidence, offers summaries of more than 300 review articles, landmark studies and consensus guidelines.

Dr. Farrell says that besides helping her know when to order tests, Medicine Toolkit is also useful when she’s working with specialists.

“If you have a consultant who wants to treat for endocarditis and you don’t think it’s warranted,” she explains, “you can use this process to determine where the disagreement is. Maybe their pretest probability is higher than yours, or maybe their treatment threshold is a lot lower. Identifying that discrepancy can make for a much more productive conversation.”

More information about Medicine Toolkit is online at the iTunes app store.