Supervisor behavior tracked with physician turnover over two years
Physician retention appears to move with the behavior of the person at the top of the immediate chain of command. In a 10-year Stanford Medicine analysis published Oct. 14 in JAMA Network Open, stronger ratings of immediate supervisors’ leadership behavior were associated with a lower likelihood that physicians would leave their organizations in the next two years.
The finding is not a soft morale metric pulled from an employee-engagement brochure. Researchers linked survey responses to human resources records, allowing them to compare what physicians said about leadership and values with whether those same physicians actually stayed or exited. Across five survey periods from 2016 to 2024, the cohort included 7,351 survey responses from 3,124 physicians.
Leadership scores moved in step with retention
Each 1-point improvement in leadership behavior scores was associated with an 18% decrease in physician turnover over the following two years. In practical terms, the difference between an average manager and a better one was not abstract; it translated into fewer departures from the organization. For health systems struggling to stabilize medical staff, that kind of signal matters because turnover brings downstream disruption in clinic continuity, scheduling, recruitment, and team cohesion.
A 1-point improvement in alignment between physicians’ personal values and organizational values was associated with a 14% reduction in turnover as well. Alignment and leadership are not interchangeable, even though they can travel together in real workplaces. One speaks to the manager-physician relationship; the other speaks to whether clinicians can recognize themselves in the institution’s stated mission and operating norms.
The association between leadership and turnover remained after adjustment for age, sex, survey year, burnout, and professional fulfillment in a separate analysis comparing ratings at a single point in time. Values alignment did not retain that association after the same full adjustment. That pattern suggests leadership behavior may be the more durable operational lever, while values language may be partly entangled with other factors already captured by burnout and fulfillment.
Intent to leave was an imperfect predictor of actual exit
The study also separated physician intentions from physician behavior, and the gap was wide. The overall two-year voluntary turnover rate, excluding retirements, was 5.0%. Physicians reporting moderate intent to leave were more likely to depart, yet only 8% actually did so. Among those reporting likely or definite intent to leave, 17% left within two years.
That mismatch has real managerial implications. HR departments often rely on stated intention as a warning signal, but the data show that intention alone does not map neatly onto final behavior. Some physicians who express dissatisfaction remain in place, while others who do not voice a strong intent to leave still eventually depart. In a workforce already under strain, relying only on a survey response risks missing the quieter forms of attrition pressure.
Retention is often decided long before resignation paperwork appears. The study’s structure supports that reality by showing that supervisory behavior measured years earlier tracked with later turnover, not just with contemporaneous dissatisfaction.
Why leadership behavior may matter at the bedside
The mechanism is easy to understand even without overreading the statistics. Physician work is collaborative, high-stakes, and unforgiving of friction. Poor supervision can intensify administrative burden, erode trust, and make daily clinical work feel more fragmented than it already is. Better leadership can do the opposite: reduce avoidable conflict, clarify priorities, and give physicians enough support to stay engaged when patient volume, documentation, and institutional demands all press at once.
Values alignment likely operates through a different route. When clinicians believe the organization is acting in a way that matches its stated commitments, the workplace feels less transactional. When those values feel performative, disengagement can accelerate. The disappearance of the values signal in the fully adjusted model suggests that some of what looks like values mismatch may actually be explained by burnout or professional dissatisfaction already present in the survey data.
Still, the leadership association held up after adjustment, and that is the more actionable signal for hospital executives. Supervisor training, feedback loops, and promotion decisions that prioritize people management may have consequences beyond employee satisfaction scores. They may influence whether physicians stay long enough to build durable patient panels, maintain collegial teams, and preserve institutional memory.
Single-institution data point to a broader retention problem
The study does not establish causation, and it comes from a single institution. Those boundaries matter because organizational culture varies widely, and Stanford’s setting may not mirror every health system, academic department, or community practice. Even so, the sample was large enough to show a measurable relationship across multiple survey waves, and the linkage to actual human resources records strengthens the inference beyond a standard opinion survey.
The findings also land at a moment when physician turnover carries high operational cost. Replacing physicians is expensive, slow, and disruptive. Each departure can ripple through call schedules, referral patterns, and access to care. For leaders, the message is not that management style alone solves retention. It is that leadership behavior sits close enough to the center of the problem that it deserves the same seriousness given to compensation, workload, and staffing ratios.
A 5.0% voluntary turnover rate may sound modest, but even that level becomes costly when it is concentrated in specific services or when departures cluster around managers who repeatedly earn weak leadership ratings. The data argue for looking upstream, before the exit interview.
What health systems can infer from the cohort
The cleanest takeaway is narrow and practical: physicians who rate their immediate supervisors more favorably are less likely to leave. Health systems that want to reduce churn will likely need to do more than issue broad statements about culture. They will need to examine the quality of local leadership where physicians actually work day to day.
This study also shows why a simple pulse survey is not enough on its own. The most useful workforce metric is one that connects attitudes to actual behavior over time. In this cohort, the leadership signal did exactly that, while values alignment was more fragile once other clinician-level factors entered the model.
For organizations trying to hold on to experienced physicians, the next question is not whether leadership matters. It is how often leaders are assessed, coached, and held accountable for the turnover patterns that follow from their behavior. The answer will shape whether the next two-year interval looks like stabilization or another round of preventable departures.