OBJECTIVE. The purpose of the present study was to evaluate associations of individual radiologist subspecialties with workforce attrition. 

METHODS. This retrospective study used a subset of subspecialist radiologists (analyzed by individual radiologist-year observations) from an earlier study of a national sample of Medicare-enrolled radiologists from 2014 to 2022. The prior analysis determined radiologist characteristics, practice characteristics, and designations of radiologist attrition (i.e., a clinically active radiologist's exit from the workforce based on a lack of subsequent claims in a Medicare fee-for-service dataset or a commercial, Medicare Advantage, and Medicaid dataset). The present analysis assigned radiologists to one of seven subspecialities (abdominal imaging, breast imaging, cardiothoracic imaging, musculoskeletal imaging, neuroradiology, nuclear medicine, and vascular and interventional radiology [VIR]) by use of Medicare fee-for-service claims data, work relative value unit weighting, and a classification based on the Neiman Imaging Types of Service. Independent associations with attrition were identified by multivariable logistic regression analysis. Nonlinear regression models were used to estimate attrition as a function of years of practice and thereby estimate mean career lengths. 

RESULTS. The analysis included 159,490 radiologist-years for 29,770 subspecialist radiologists. The attrition rate over the study period was 2.2% and varied across subspecialties from 1.0% (VIR) to 4.3% (cardiothoracic imaging). Year-by-year attrition rates increased from 1.4% in 2014 to 2.7% in 2022 and also increased over time for each subspecialty (percentage-point increase ranging from 0.4% [musculoskeletal imaging] to 2.1% [cardiothoracic imaging]). Adjusted odds of attrition, compared with abdominal imaging, were significantly higher for breast imaging (OR = 1.31) and cardiothoracic imaging (OR = 1.81), significantly lower for VIR (OR = 0.78), and not significantly different for other subspecialties. Mean estimated career length was 1.1 and 1.5 years greater for men than for women among academic and nonacademic radiologists, respectively; these gender differences were smaller within individual subspecialties (academic: 0.4-0.6 years; nonacademic: 0.2-0.5 years). 

CONCLUSION. Attrition increased over time for each subspecialty but varied among subspecialties. 

CLINICAL IMPACT. Recruitment and retention initiatives are needed broadly across the workforce to avoid departure cycles that potentially exacerbate individual subspecialty shortages.