He fell in love with the work environment and started contemplating medical school, even while noting that the devices and systems used in the hospital weren't engineered for optimal safety. He finished medical school in 2000, just as the Institute of Medicine released its landmark study "To Err is Human." He realized he was uniquely suited to bridge the gap between clinicians and engineers by combining his two areas of study, and he's spent the past decade researching how to reengineer health care for maximum safety and effectiveness. He also works as an emergency physician. "I will always practice," he says. "I enjoy it a lot, and I can't be good at what I do on the research side unless I work in a clinical domain."
On visual cues
When you approach a door that has a pull handle, your brain will tell you to pull it, and you don't need a sign that says, "Pull." But if people need to push that same door, it won't do any good to put a "Push" sign on it, because the perception of the pull handle is more basic than the perception of the sign. Labeling will not ensure that a system is used right if it's not designed properly otherwise, and if it is designed properly, there's no need for labels.
Everyone makes mistakes. The way to design the safest system is to understand where performance errors will occur and buffer the system against them. Both pilots and air traffic controllers make an average of two mistakes per hour, and yet there are very few accidents because the system is engineered to anticipate and correct them. If hospitals are basing patient safety efforts on the assumption that they can get people not to make mistakes, they're misdirecting their resources. They should study where errors occur, identify which ones put patients at higher risk, then redesign systems.
On screen design
Many EHRs allow physicians to open multiple tabs with different patients, which is important for workflow because they may need to deal with many patients at once, but there's very little differentiation among the screens. One vendor uses different color tabs for each patient, thinking the colors will alert the users, but they're not looking at the top of the screen-they're looking at where they're entering the morphine order. Human factors engineering uses eye tracking to know where the physician looks when placing an order.