In Medicine, Rounding Works

Rounding is a time-honored tradition where doctors meet to talk about cases, either in a meeting room (the "rounding room" was named after a specific room at Hopkins) or at/near the bedside. Most often associated with inpatient medicine teams especially in training environments, the treating physician will present the case and discuss it with her colleagues. Not only is it thought that in this way, medical decision-making benefits from collective intelligence, but the anxiety provoked by immediate criticism (especially in trainees) sharpens one's thinking. A study in JAMA Network Open supports this. Teams here were internal medicine teams composed of multiple levels of training, from med students up to attendings. I don't think the findings would be too domain specific, but at a guess, I imagine the benefit would be even greater for psychiatry than for internal medicine, as psychiatry's diagnoses are fuzzier and more subjective.

Groups don't always arrive at better decisions than individuals - especially groups of non-expert individuals with no feedback - but teams with people who are experts, and who do get feedback benefit from collective intelligence, do better than individuals alone. So qualitatively this isn't surprising, but a problem in medicine is lack of quantitative thinking; especially in my specialty, psychiatry, where studies are constantly coming out showing that medical or psychiatric illness X increases the risk of psychiatric illness Y. No kidding! By how much is what we want to know. So what's the actual benefit of rounding?

For groups of 9, on average you need to treat about 4 people before you make a diagnosis that an individual would have missed (i.e. NNT is about 4.)

For groups of 5, NNT = 6.

For groups of 2, NNT = 8.

The simple plot below shows the % accuracy improvement per person based on group size, and again not surprisingly, there's a diminishing marginal return for adding more people. (Where does it go to zero? Nine is already on the big size for a rounding team.)


This of course doesn't take into account rounding time, which is a real consideration, and big teams are slow. Maybe the % improvement per minute drops at a certain point.

Therefore, don't hesitate to curbside-consult your colleague, because just by talking to one other person, every eight patients you're making a more accurate diagnosis.

Barnett ML, Boddupalli D, Nundy S, Bates DW, et al. Comparative Accuracy of Diagnosis by Collective Intelligence of Multiple Physicians vs Individual Physicians. JAMA Netw Open. 2019;2(3):e190096. doi:10.1001/jamanetworkopen.2019.0096