Patients in emergency departments who are treated by physicians with a high propensity to admit those they see into the hospital are more likely to be discharged after only a short stay, suggesting a possible unnecessary admission, while they are no less likely to die, new research suggests.
The findings suggest that differences in physicians’ skill or risk aversion may come into play when they make admitting decisions, said Dr. Dan Ly, assistant professor of medicine in the division of general internal medicine and health services research at the David Geffen School of Medicine at UCLA.
“Physicians, just like professionals in other domains, differ in their abilities and decisions, which has consequences for how much care you might receive, including, in this case, your likelihood of being hospitalized,” said Ly, who co-authored the paper with Stephen Coussens, a health economist and senior data scientist at the Washington state-based data management company Abett. “Some physicians may order more tests and hospitalize more of their patients, but this doesn't appear to translate to better health.”
The findings will be published in the peer-reviewed JAMA Internal Medicine.
While previous studies have shown great variation in emergency department physicians’ likelihood of admitting a patient to the hospital, there has been little evidence whether and how this affects patient outcomes.
Using electronic health records from Veterans Affairs for January 2011 through December 2019, the researchers compared physicians practicing within the same emergency department. Their cross-sectional study captured data for 2,100 physicians in 105 emergency departments across the U.S. comprising more than 2.1 million patient visits. They focused on patients coming into the emergency department for chest pain, shortness of breath, or abdominal pain.
About 41% of the visits led to hospital admission, with 19% of those patients discharged in less than 24 hours. Variation in admission rates varied greatly. For example, patients who saw high-admitting physicians (those in the 90th percentile) were almost twice as likely to be admitted as patients who were treated by low-admitting physicians (those in the 10th percentile), even though their underlying health did not differ. Overall, about 2.5% of patients died within 30 days. The researchers found no relationship between a physician’s admission rate and the 30-day mortality rate of their patients. This lack of relationship was also true for 7-day, 14-day, 90-day, and one-year mortality.
The findings suggest that differences across physicians in admission rates are driven less by a patient’s underlying health than variation in physician decision-making, the researchers write. In addition, these additional hospitalizations from high-admitting physicians led neither to short-term protection from severe outcomes nor to a reduction in patients’ risk for dying for up to a year afterward. And patients who were treated by physicians with a lower propensity to admit spent less total time after their ED visit in the hospital, which suggests that low-propensity physicians were not simply deferring a necessary hospitalization into the future, they write.
Limitations of the study include the possibility that some potential clinical confounders were not captured in the electronic health records; the manner that the researchers used to measure patients’ health prior to the emergency department visit, called the Elixhauser Comorbidity Index score, does not measure the severity of a person’s presenting condition; variations in physician characteristics, such as medical training, could not be controlled for; the findings are specific to the VA, whose patients are largely male, and has emergency department staffing with more non-emergency medicine trained physicians, so the findings may not be applicable to other institutions; and other factors besides the physician also play a role in variation in admission rates.
While more research is needed to fully understand these differences in admission rates, “our study is able to better account for patient differences and make apples-to-apples comparisons between physicians to demonstrate true differences in ED physician practice patterns and show that these differences do not translate to better patient health outcomes,” Ly said.
The study was funded in part by the Veterans Affairs Health Systems Research Center for the Study of Healthcare Innovation Implementation & Policy Locally Initiated Project. (LIP 65-175)
Article: “Variation in Emergency Department Physician Admitting Practices and Subsequent Mortality,” DOI: 10.1001/jamainternmed.2024.6925
