Lisa Rosenbaum, writing in the New Yorker, looks at how changes in policy in complex systems attempt to do one thing but often do another.
In the article she explores how limits on shift lengths for residents, which were intended to increase quality of care by reducing mistakes presumably caused by fatigue, resulted in unintended consequences: the number of times a patient was passed from one doctor to another increased; residents started to focus on the clock; learning opportunities decreased; and quality of care, despite intentions, decreased.
For a young doctor, the right course of action isn’t always clear. Acquiring the necessary knowledge and experience requires feedback, which strengthens one’s ability to anticipate how the many variables and small decisions might affect the patient. What’s more, learning how to manage illness demands infinite tweaking; each patient is unique.
But now, residents spend less time directly caring for patients than they once did, and the feedback inherent in the hours once spent with more seasoned physicians has also diminished. In the earlier years of my training, morning rounds were a sacred time. The whole team would gather to learn about the patients who came in overnight, and discuss patients already in our care. We would hear their stories, examine them, review data, and then, together, make decisions about their care for the day. Now, however, the scheduling is such that overnight residents often have to leave before rounds, and the daily ritual has morphed into a race against the clock. Instead of beginning by asking who the sickest patient is, we now ask which resident needs to leave.
Manipulating one variable in a complex system, such as medical care, inevitably affects another. It’s increasingly difficult to do merely one thing, no matter how good our intentions.
We’re measuring everything we can but often the most important things, like patient outcomes, are the result of a large complex system. The impact of one variable is hard to manage and measure.
You can’t just alter one variable in that system and not expect to affect others. As we move towards optimizing the things that we can measure, we focus on what we can see, count, and report. However, what we can see and count is only a narrow subset of that complex system.