HCI3 Update from the Field: Signal Detection

Submitted by francois.debrantes@hci3.org on Friday, December 7, 2012 - 12:18

Newtown, CT – December 7, 2012

A few decades ago, Signal Detection Theory came to life – Its purpose was simple (to detect a signal when there's a lot of noise) and has shown great relevance to the practice of medicine. A good summary of the theory and its application to healthcare can be found here, courtesy of Prof David Heeger, son of Nobel laureate Alan Heeger. The premise is straightforward: people make inferences and decisions based on information. Sometimes the information is clear and sometimes it isn't. Sometimes there's so much information that it's difficult to detect the real signal from the noise. Sometimes our own biases lead us to favoring a low signal (to avoid false negatives) at the expense of triggering too many false positives, or vice-versa. The trick, of course, is to find a signal that will minimize false positives and false negatives. Last week, in a rather quiet affair (low signal strength), a new web site was launched with a very powerful signal for all those who care about patient safety and transparency – www.hospitalsafetyscore.org.

What this means to you – To start with, everyone should immediately check their state ranking. While the page can use some work in clarity, it does show for each state the number of hospitals with an A and how many have an A as a percentage of all hospitals in the state. For example, my home state of Connecticut is 32nd on the list, with 6 hospitals (one out of 5) getting an A, and all others getting a lower rating. We're in good company with our neighboring state, NY, carrying a similar ratio, and put to shame by our neighbors to the north in MA and ME, where 4 out of 5 hospitals are getting an A. This is the signal we've been waiting for. There's far too much noisy information on hospitals, much of which is confusing to the average consumer. Oftentimes, whatever signal can be generated is drowned by the noise emanating from billboards, TV and radio. However, to make this signal the strongest it can be, we must now diffuse it. These scores should be plastered on the door of every hospital in the nation, but they won't and we know why. Short of that, let's act ourselves. For example, every employer should make these scores available to every employee and their family member. In addition, employers should band together in certain communities and take out full-page ads in their local papers to publicize the scores for each hospital in that community. Finally (for now), employers should insist that their health plan administrators use this information on all their websites and in designing P4P programs. And if you (or they) don't think hospital-based P4P works, then please read this month's Health Affairs paper on that topic. We've been waiting for a strong signal. Now that it's here, let's eliminate the false positives and reward the As. 


Francois de Brantes
Executive Director
Health Care Incentives Improvement Institute, Inc.
w: www.hci3.org