Learning to Deliver Consistent Healthcare
Variation in medical practices and healthcare decisions accounts for a significant part of health system costs. Some of these variations may be due to “weak best practices” that arise from uncertainty and the vagaries of tacit learning — the informal knowledge that is difficult to transfer to another person. A study in a U.S. academic hospital by David Chan of Stanford School of Medicine found a large variation in spending on medical tests by residents in general medicine but little variation when the same residents worked in a specialized area such as oncology. Because of the difficulty in codifying general medical decisions, it is less likely that “a common way of doing things that is agreed to be superior to other ways” will take root in a healthcare setting, Chan says. Chan presented his findings at the Economics of Organizations workshop at Smith School of Business at Queen's University.
As in many industries, practice variation is a persistent and worrying phenomenon in healthcare. Variation is seen geographically — you are twice as likely to have cardiac surgery if you live in Kingston than in Toronto — and within healthcare organizations themselves.
Two explanations are generally used to explain practice variation within institutions. One relates to differences in the skills or preferences of healthcare practitioners, the other to differences in medical “schools of thought.”
A third explanation, however, may hit closer to the mark. According to David Chan of Stanford School of Medicine, the variations may be explained by “weak best practices” that arise from uncertainty and the vagaries of tacit learning — the informal knowledge that is difficult to transfer to another person.
To better understand practice variation in a healthcare setting, Chan studied the decisions that went into 3.2 million patient orders (spending on medical tests) made by interns and residents at a large U.S. academic hospital. He traced the spending effects of the same physicians as they progressed through training in different roles on teams and practice environments.
Chan presented his findings at the Economics of Organizations workshop at Smith School of Business.
In complex team-based environments such as hospitals, decisions are often based on informal evidence and the rationale behind decisions can be difficult to explain and learn from. “Despite advances in knowledge, indeed because advances in technology have relegated routine tasks to computers, decisions made by humans are increasingly difficult to codify,” says Chan.
This is especially true in the case of medical practices and technology. In such an environment, it is less likely that “a common way of doing things that is agreed to be superior to other ways” will take root, he says.
“Resident spending exhibits substantial convergence in specialist services, eliminating much of the variation by the end of the third year, while practices by the same residents showed no convergence in general medicine”
In his study, Chan found a large variation in spending on tests by interns and residents, identified by random assignment of patients. The variation was particularly pronounced among the more experienced residents; variation in spending tripled when physicians in training had more seniority on the medical teams and therefore more influence on whether or not to order medical tests.
Chan also compared the practice patterns of residents in specialist areas of cardiology and oncology against the practice patterns of the same residents in general medicine. The specialist areas showed less variation and stronger best practices.
“Resident spending exhibits substantial convergence in specialist services, eliminating much of the variation by the end of the third year, while practices by the same residents showed no convergence in general medicine,” he says.
This could be because best practices are stronger where knowledge is more specialized, Chan adds. As he points out, cardiology and oncology have perhaps highest volume of explicit knowledge in medicine, which reduces the reliance on tacit knowledge.
Economists have traditionally looked at differences in preferences and ability or in learned differences to explain practice variation, but Chan found no evidence of any of this in his large study. “This suggests that worker characteristics and formally learned differences have little role in explaining practice variation, at least within organizations.”
The findings should be of interest to healthcare administrators and policymakers alike. It has been estimated that eliminating practice variation could trim healthcare spending in the U.S. by 30 percent.
— Vanessa Santilli and Alan Morantz