Physician Turnover and Productivity in the Veterans Health Administration

Wednesday, June 26, 2019: 9:00 AM
Taylor - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Aigerim Kabdiyeva

Discussant: Michael R. Richards


Clinical staff turnover affects the supply and cost of healthcare services at the Veterans Health Administration (VHA). An increase in clinical staff turnover results in higher costs from recruiting and training new hires, and losses due to lower productivity among new hires compared to quitting staff. Previous studies have also shown that higher turnover is associated with lower patient satisfaction and higher rates of medical errors and patient adverse outcomes. A VHA employee’s decision to quit is affected by non-pecuniary factors such as organizational culture and availability of clinical support staff as well as financial incentives such as salary, recruitment awards, retention awards and education debt repayment awards. In this study, we analyze the relationship between turnover and clinician demographic factors, tenure, salary, incentive awards and productivity. We hypothesize that increases in employee tenure, salary and incentive awards are associated with lower probability of quitting and that higher productivity employees are more likely to quit because they have more alternatives in the local healthcare labor market.


We focus on four clinical specialties for our analysis: nurse anesthetists, primary care physicians (PCPs), psychiatrists and psychologists. We obtain data on outpatient procedures performed by employees in these specialties between 2006 and 2017 from the VHA Corporate Data Warehouse (CDW). An employee’s productivity for each year is defined as the total Relative Value Units (RVUs) associated with the employee’s outpatient procedures divided by the employee’s total clinical time. Productivity data is merged with employee demographic, tenure, turnover, salary and incentive award data that we obtain from VHA CDW and the VHA Workforce Management and Consulting office. Our regression model estimates the relationship between the indicator for quitting and the employee’s demographic characteristics, salary, incentive awards and productivity. Since employees who are more likely to quit are also more likely to receive incentive awards, we face selection bias due to the unobserved propensity to quit. We attempt to control for the unobserved propensity to quit by including in the regression the medical center-year average quit rate for other specialties, which reflects unmeasured local conditions that affect quitting (like employee morale).


We find that higher tenure, age, and salary are associated with lower probability of quitting. Receipt of education debt repayment funds is associated with lower quitting for all four specialties while the receipt of recruitment, retention and relocation incentives is associated with lower turnover for different subsets of the specialties. The association between quitting and productivity is negative, contrary to our expectations.


To the best of our knowledge, this is the first study that examines the relationship between turnover at the VHA and key employee labor market characteristics including demographic variables, tenure, salary, receipt of incentive awards, and productivity. Managers need estimates like these to determine the optimal scale of retention programs and target them most effectively.