Battling Radiology Burnout Without Sacrificing Productivity

In 2006, Comair flight 5191 crashed during takeoff and killed 49 people. Concern about the impact of controller fatigue on the accident led the National Transportation Safety Board to investigate four prior runway crashes, where they discovered that all four controllers reported excessive workload as an underlying cause of their mistakes. The Federal Aviation Administration responded to the findings by enacting comprehensive changes to the industry's shift regulations. Radiology and air traffic control are comparable in the sense that both fields demand prolonged focus on computer screens. The symptoms of visual and cognitive fatigue that contribute to controller error similarly lead to reduced diagnostic accuracy by radiologists. However, unlike aviation and most industries outside of medicine, radiology is impacted by declining reimbursement and high costs of training new talent, which make it difficult to counteract rising workloads with hiring adjustments. There is a limited number of radiology residency training positions in the US which ultimately limits potential additional hiring. As a result, simultaneously combating burnout and RVU expectations requires consideration of how to reduce workloads without sacrificing radiologists’ productivity.

How have radiology workloads changed in the past two decades?

A recent case study of a large general hospital found that radiology RVUs rose from 6,187 in 2006 to 24,584 in 2020 (297% increase over 15 years).  In the same time period, the number of radiology staff in the hospital remained stagnant, indicating that individual workloads increased significantly.   The results of this study build on a prior report which noted an increase of 70.3% annual work RVUs per full-time equivalent radiologist from 1991–92 to 2006–07.  Rapidly growing radiology workloads (defined as the product of the number and complexity of examinations performed per time unit) in the past 20 years can largely be attributed to technological developments including voice recognition software, picture archiving and communication systems (PACS), and advanced cross sectional imaging techniques.  Another factor has been the widespread adoption of defensive medicine practices and consequent medical imaging overutilization.  Global imaging volume is expected to double in the next eight years, so the trend of rising radiology RVU expectations will likely continue for the foreseeable future.  

What is burnout, and how does workload relate to radiologist burnout?

The Maslach Burnout Inventory is a highly established method of measuring burnout, and it considers three primary symptoms - emotional exhaustion, depersonalization, and low sense of personal accomplishment.  Radiologists experiencing burnout often feel unmotivated for work and lack compassion for their patients.  Therefore, it negatively impacts public health by adversely affecting physicians’ clinical performance. Burnout can also contribute to absences from work, unprofessional behavior, unhealthy lifestyle habits, and premature retirement.  

In 2016, the ACR Commission on Human Resources generated a list of ten risk factors for radiology burnout, which they ranked by an importance score determined by eight members on the commission.  Reducing these factors is a shared responsibility between radiologists and health care organizations.  Individual radiologists prevent burnout by striving to optimize their physical health, personal relationships, and pursuits outside of work.  Seeking guidance from career coaches or mental health professionals when needed has also been shown to effectively reduce burnout rates.  Organizations are responsible for acknowledging the problems contributing to burnout in their specific practices and encouraging open communication to cultivate a sense of collaboration and job control.  Although social factors such as workplace isolation can be solved without modifying workloads, the two greatest factors of burnout, inadequate staffing and prolonged stress, both stem from workload expectations.  Thus it is not surprising that increased workloads correlate with increased burnout rates.

Back in 2003, an ACR study indicated that 93% of radiologists enjoyed their work.  This was a higher level of satisfaction than most other physician specialties, but it was lower than the results of the equivalent 1995 ACR survey.  Of the radiologists who reported low satisfaction in the study, more than half cited workload as a primary reason.  Furthermore, radiologists who expressed dissatisfaction due to work overload felt the issue was serious enough that they were willing to accept proportionately lower income (i.e. take a 10% income cut for 10% reduction in workload).  As with physicians overall, radiology job satisfaction has continued to decrease over time.  Today more than 85% of US radiologists feel at or above full work capacity and approximately 70% of radiologists have experienced feelings of burnout. 

Why hasn’t AI already led to reduced radiology burnout rates?

Geoffrey Hinton is a Turing Award winning computer scientist known as one of the “Godfathers of AI.”  Five years ago he declared, “We should stop training radiologists now; it’s just completely obvious that within five years deep learning is going to do better than radiologists.”  Conversations about AI overtaking the work of radiologists have been so widespread throughout the past decade that it is now a factor for medical students moving away from radiology as their specialty choice. This discussion has only been further amplified by social media creating an inaccurate narrative around the speciality as easily replaceable with AI. However, current studies concede that previous speculations were overly optimistic and did not accurately consider regulatory processes and scarcity of training data.  

Since 2016, The FDA has approved 80+ AI algorithms for medical imaging, and ~30% of radiologists have adopted some form of AI into their practice.  Most of these AI solutions have not lived up to the expectation of reducing the burden on radiologists.  A group of Dutch researchers recently studied a random sample of 440 medical imaging studies published in 2019.  They found that rather than being integrated into existing workflows and taking over tasks for radiologists, AI applications generally led to additional post-processing and interpretation time.  As a result, 86% of AI-focused studies were actually associated with a higher workload for radiologists.  

How is Rad AI Omni increasing efficiency and reducing burnout rates?

Rad AI Omni uses artificial intelligence to generate report impressions from dictated findings.  Without adding a single click into existing workflows, Omni creates impressions which are customized to each radiologist’s language preferences in less than a couple seconds.  Radiologists using Rad AI achieve time savings of an hour a day, allowing them to heighten their focus on the most specialized aspects of their work - expert imaging interpretation and diagnosis.  Taking another look at the ACR’s list of risk factors for radiology burnout, the primary factor identified is inadequate staffing.  Hiring more radiologists decreases workload by reducing the necessary productivity levels of each radiologist in the practice.  Rad AI Omni instead minimizes the tedium and time required to read each study, thereby reducing workloads while maintaining or improving individual levels of productivity.  

The second greatest factor for radiology burnout identified by the ACR is prolonged stress in work environments.  In addition to decreasing stress by reducing the time pressure of hitting RVUs, Rad AI Omni decreases the mental strain of proofreading reports.  Automated impressions make typos in the findings more apparent.  Radiologists modify their findings in 16-23% of the reports where Omni has been utilized, and 5% of those edits correct significant clinical errors.  With Rad AI Omni, job satisfaction is improved because radiologists receive both time-saving support and improved confidence in their report consistency.  

Join the thousands of radiologists who trust Rad AI


Request a demo