Using an innovative method for measuring doctor turnover, Weill Cornell Medicine researchers determined that between 2010 and 2018, the annual rate at which physicians left their practices increased by 43 percent, from 5.3 percent to 7.6 percent a year.  The causes of this trend are not known, but warrant further investigation, according to the researchers.

The study, published July 11 in the Annals of Internal Medicine, also found that the first three quarters of 2020 (the start of the COVID-19 pandemic in the United States) were not associated with higher turnover. However, more data are needed to fully understand turnover trends related to COVID-19.

Whether doctors are moving to a new practice or retiring, “it is important to study turnover because it can hurt the continuity and quality of patients care,” said one of the study authors Dr. Lawrence Casalino, professor emeritus of population health sciences at Weill Cornell Medicine.  “There’s a lot of mutual trust that builds between a doctor and patient over time that’s difficult to replace.”

The study authors, led by Dr. Amelia Bond, assistant professor of health policy and economics in the department of population health sciences at Weill Cornell Medicine, developed a new method to estimate turnover primarily using the Medicare Data on Provider Practice and Specialty (MD-PPAS). This database contains information on clinician characteristics such as age and sex, physician specialty, and tax identification numbers of the practices in which physicians work.

Using this data, the researchers developed an algorithm allowing them to estimate physician turnover from 2010 through the first three quarters of 2020. “I think this method may become the gold standard to identify turnover in the years to come,” Dr. Casalino said, adding that prior estimates, using surveys to assess turnover, have been unreliable.

The researchers found that between 2010 and 2018, the annual rate of turnover, based on doctors moving or not practicing anymore, increased from 5.3 to 7.6 percent, a change of 43 percent. “While the percentage change from year to year seemed somewhat modest, the percentage change over time was quite large,” Dr. Casalino said. 

Most of the turnover occurred between 2010 and 2014 and resulted from doctors who stopped practicing. Why turnover increased during those years remains unknown. “As we speculate in the article, those were the years when the electronic medical record became a requirement,” said Dr. Casalino. “Some prior studies have suggested a link between electronic health record use and physician burnout, and it may be that burned out physicians are more likely to stop practicing or move to another practice.”

Turnover may also create further barriers to accessing care for people living in rural areas and for underserved communities. The researchers observed that doctors in rural areas were more likely to move or to stop practicing medicine than urban doctors. Female doctors were more prone to turnover than male doctors. Physicians in larger practices were more likely to experience turnover than those in single-doctor or two-physician practices. Doctors who took care of more patients who were eligible for Medicare and Medicaid were also more disposed to turnover.

Following 2014, turnover rates remained stable through 2017, and then slightly increased in 2018. In the second and third quarters of 2020, turnover, which was based on data about moving, was slightly lower than in the corresponding quarters of 2019. While a great deal of turnover did not seem to be occurring early in the COVID-19 pandemic, “researchers need to look at the next year of two of data to get a better understanding of any pandemic trends,” Dr. Casalino said. 

In addition to learning more about COVID-19 physician turnover, the method will also make it possible to determine how much turnover occurs after a private equity company, a health insurer, or a hospital acquires a physician practice.

“Once we know the reasons behind doctor turnover, individual practices or hospitals can better develop programs to retain their providers,” Dr. Casalino said. “That’s why we’re making our algorithm public for other researchers to use.”