What’s New in the Patient Safety World

August 2011

The Amazon.com Approach to Medication Reconciliation



We often lament that our supermarkets better use technology than our hospitals do. Our supermarket can tell us exactly how many boxes of cornflakes there are and exactly where they are in the store. Most of the hospitals we work with cannot tell us how many orthopedic widgets they have and where they are.


Online retailers like Amazon.com use populational databases to predict what someone might want to buy. How often have you bought something on an online site and saw a message that says something like “People who have purchased “X” have also often purchased “Y”? Great marketing tool!


But now healthcare researchers have taken that concept in attempt to improve the medication reconciliation process. Initial lists for medication reconciliation (sometimes called “best possible medication history” or similar names) very frequently omit important drugs that a patient has actually been taking. Hasan and colleagues (Hasan 2011) have used the above concept, which they refer to as collaborative filtering, to help identify medications omitted from patient medication lists at the time of medication reconciliation. They basically determine, based on large population databases, that patients who take drug “X” also often take drug “Y”. They established multiple different algorithms and applied them to sample patient data. In fact, their algorithms were able to guess correctly an omitted drug within 10 guesses about 50% of the time (they did even better guessing the therapeutic class of a missing drug). They found some of their algorithms might work better in certain settings or with certain populations.


While this is not yet ready-for-prime-time, and might be expected to produce some unintended consequences, it is a fascinating concept that we expect someday will prove to be very useful in improving the medication reconciliation process. Think about how this could also be used to improve another patient safety problem – diagnostic error.


Keep your eyes on this technology!





Hasan S, Duncan GT, Neill DB, Padman R. Automatic detection of omissions in medication lists. JAMIA 2011; 18: 449-458 Published Online First: 29 March 2011















Tip of the Week Archive


What’s New in the Patient Safety World Archive