Faculty Bibliography

PURPOSE/OBJECTIVE:As differences in imaging patterns may indicate unnecessary care, this study examined differences in repeat imaging rates between imaging studies interpreted by a nonphysician practitioner (NPP) versus a radiologist. METHODS:This multiyear (2013-2022) retrospective study evaluated imaging performed on Medicare fee-for-service beneficiaries using a CMS Research Identifiable File. Imaging studies, grouped by anatomic region and modality (eg, shoulder radiography [XR]) with ≥30 repeat studies within 90 days for both NPP-interpreted and radiologist-interpreted index studies, were included. Logistic regression was used to assess the likelihood of repeat imaging within 90 days for NPP-interpreted versus radiologist-interpreted index studies, adjusted for patient gender, age, race or ethnicity, comorbidities, urbanicity, and community income. RESULTS:There were 1,397,002 imaging studies that met the selection criteria. Of these, repeat imaging occurred for 12.5%. Unadjusted repeat imaging rates were higher for NPP-interpreted versus radiologist-interpreted imaging for XR (20.4% versus 14.6%), ultrasound (11.6% versus 4.5%), and MR (8.8% versus 3.8%). Adjusted for covariates, the odds ratio (OR) for repeat imaging was higher for NPP-interpreted versus radiologist-interpreted imaging: 1.35 (95% confidence interval [CI]: 1.33-1.37) for XR, 2.41 (95% CI: 2.21-2.63) for ultrasound, and 2.56 (95% CI: 1.81-3.64) for MR. By anatomic region-modality, these ORs ranged from 1.39 (95% CI: 1.34-1.44) for shoulder XR to 3.40 (95% CI: 2.80-4.14) for abdominal ultrasound, but was not significantly different for knee XR (OR: 1.01, 95% CI: 0.99-1.04). CONCLUSION/CONCLUSIONS:Among Medicare beneficiaries, imaging studies are more likely to be repeated when interpreted by a NPP than when interpreted by a radiologist. Potential excess reimaging has implications for unnecessary care.

UNLABELLED: HIGHLIGHTS/UNASSIGNED:A new oral cancer simulation model with risk factors including degrees of smoking and alcohol exposure, oral lesion features, and sex incorporates more accurate and precise representation of patient risk categories.We evaluated screening strategies for oral potentially malignant disorders with or without risk-stratified biopsy referral in both the general population and subpopulations defined by degrees of smoking and alcohol exposure.Men with a high degree of both smoking and alcohol exposure exhibited a significant reduction in cancer-specific deaths and cancer incidence from screening programs for oral potentially malignant disorders.Screening with risk-stratified biopsy, using a surgical treatment threshold of moderate dysplasia or worse, yielded the greatest efficiency in term of biopsies needed to detect 1 treatable case.

RATIONALE AND OBJECTIVES/OBJECTIVE:To evaluate the performance, stability, and decision-making behavior of large language models (LLMs) for title and abstract screening for radiology systematic reviews, with attention to prompt framing, confidence calibration, and model robustness under disagreement. MATERIALS AND METHODS/METHODS:We compared five LLMs (GPT-4o, GPT-4o mini, Gemini 1.5 Pro, Gemini 2.0 Flash, Llama 3.3 70B) on two imaging-focused systematic reviews (n = 5438 and n = 267 abstracts) using binary and ternary classification tasks, confidence scoring, and reclassification of true and synthetic disagreements. Disagreements were framed as either "LLM vs human" or "human vs human." We also piloted autonomous PubMed retrieval using OpenAI and Gemini Deep Research tools. RESULTS:LLMs achieved high specificity and variable sensitivity across reviews and tasks, with F1 scores ranging from 0.389 to 0.854. Ternary classification showed low abstention rates (<5%) and modest sensitivity gains. Confidence scores were significantly higher for correct predictions. In disagreement tasks, models more often selected the human label when disagreements were framed as "LLM vs human," consistent with authority bias. GPT-4o showed greater resistance to this effect, while others were more prone to defer to perceived human input. In the autonomous search task, OpenAI achieved moderate recall and high precision; Gemini's recall was poor but precision remained high. CONCLUSION/CONCLUSIONS:LLMs hold promise for systematic review screening tasks but require careful prompt design and circumspect human-in-the-loop oversight to ensure robust performance.

Ovarian cancer remains low in prevalence but has the highest mortality of all gynecologic malignancies. Population-based screening for ovarian cancer remains a topic of interest in contemporary practice, given that the majority of cancers encountered are high-grade aggressive malignancies, for which favorable survival is encountered in the setting of early-stage disease. This document summarizes a review of the available data from randomized and observational trials that have evaluated the role of imaging for ovarian cancer screening in average-risk and high-risk patients. When considering screening using pelvic ultrasound in average-risk patients, we found insufficient published evidence to recommend ovarian cancer screening. Randomized controlled trials have not demonstrated a mortality benefit in this setting. Screening with pelvic ultrasound may be appropriate for select patients at high risk, although the existing data remain limited as large, randomized trials have not been performed in this setting. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision process support the systematic analysis of the medical literature from peer reviewed journals. Established methodology principles such as Grading of Recommendations Assessment, Development, and Evaluation or GRADE are adapted to evaluate the evidence. The RAND/UCLA Appropriateness Method User Manual provides the methodology to determine the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where peer reviewed literature is lacking or equivocal, experts may be the primary evidentiary source available to formulate a recommendation.

BACKGROUND:IPMN consensus guidelines make implicit judgments on what cancer risk level should prompt surgery. We used decision modeling to estimate this cancer risk threshold (CRT) for BD-IPMN patients. METHODS:We created a decision model to compare quality-adjusted life years (QALYs) following surgery or surveillance for BD-IPMNs. We simulated treatment decisions for hypothetical patients, varying age, comorbidities and lesion location (pancreatic head/tail). The base case was a 60-year-old patient with mild comorbidities and pancreatic head IPMN. Probabilities, life expectancies, and utilities were incorporated from literature/public datasets. CRT was defined as the level of cancer risk at which the expected value of QALYs for surgery first exceeded that of surveillance. RESULTS:In the base case, surgery was preferred over surveillance, yielding 21.90 vs. 21.88 QALYs. The optimal CRT for a BD-IPMN patient depended on age, comorbidities, and location. CRT in the base case was 20 % and 3 % for an IPMN in the head and tail of the pancreas, respectively. Other drivers of preferred treatment were age and likelihood of postoperative mortality. CONCLUSION/CONCLUSIONS:For BD-IPMNs, the optimal CRT varies depending on patient age and risk of surgical complications. Personalized risk threshold values could guide treatment decisions and inform future treatment consensus guidelines.

Multi-cancer early detection (MCED) tests are already being marketed as noninvasive, convenient opportunities to test for multiple cancer types with a single blood sample. The technology varies-involving detection of circulating tumor DNA, fragments of DNA, RNA, or proteins unique to each targeted cancer. The priorities and tradeoffs of reaching diagnostic resolution in the setting of possible false positives and negatives remain under active study. Given the well-established role of imaging in lesion detection and characterization for most cancers, radiologists have an essential role to play in selecting diagnostic pathways, determining the validity of test results, resolving false-positive MCED test results, and evaluating tradeoffs for clinical policy. Appropriate access to and use of imaging tests will also factor into clinical guidelines. Thus, all clinicians potentially involved with MCED tests for cancer screening will need to weigh the benefits and harms of MCED testing, including consideration of how the tests will be used alongside or in place of other screening options, how diagnostic confirmation tests should be selected, and what the implications are for policy and reimbursement decisions. Further, patients will need regular support to make informed decisions about screening using MCED tests in the context of their personal cancer risks, health-related values, and access to care.

Endometriosis is a common condition impacting individuals assigned female at birth. Though incompletely understood, the disorder is caused by endometrial-like tissue located outside of the endometrial cavity, associated with inflammation and fibrosis. Clinical presentation is variable, ranging from asymptomatic to severe pelvic pain and infertility. Treatment is determined by the patient's individualized goals and can include medical therapies to temporize symptoms or definitive surgical excision. Imaging is used to help diagnose endometriosis and for treatment planning. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision process support the systematic analysis of the medical literature from peer reviewed journals. Established methodology principles such as Grading of Recommendations Assessment, Development, and Evaluation or GRADE are adapted to evaluate the evidence. The RAND/UCLA Appropriateness Method User Manual provides the methodology to determine the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where peer reviewed literature is lacking or equivocal, experts may be the primary evidentiary source available to formulate a recommendation.

The incidence of twin pregnancies has been rising, largely attributable to increasing use of artificial reproductive techniques. Ultrasound plays a critical role in establishing the chorionicity and amnionicity of multiple gestations, a key predictor of the expected risk and complications, along with guiding future clinical and imaging follow-up examinations and intervals. People carrying multiple gestations will typically undergo more ultrasound examinations (and occasionally fetal MRI) than those carrying singletons, at minimum including a first trimester dating scan, nuchal translucency scan at 11 to 14 weeks, an anatomy scan at 18 to 22 weeks, and other scans in the second and third trimesters for growth and surveillance. This document clarifies the most appropriate imaging guidelines for multiple gestations for seven clinical scenarios/variants, which range from initial imaging, follow-up imaging, growth and surveillance for uncomplicated multiple gestations, and those complicated by a known abnormality or discordance between fetuses. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision process support the systematic analysis of the medical literature from peer reviewed journals. Established methodology principles such as Grading of Recommendations Assessment, Development, and Evaluation or GRADE are adapted to evaluate the evidence. The RAND/UCLA Appropriateness Method User Manual provides the methodology to determine the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where peer reviewed literature is lacking or equivocal, experts may be the primary evidentiary source available to formulate a recommendation.

Burnout among radiologists is increasingly prevalent, with the potential for having a substantial negative impact on physician well-being, delivery of care, and health outcomes. To evaluate this phenomenon using reliable and accurate means, validated quantitative instruments are essential. Variation in measurement can contribute to wide-ranging findings. This article evaluates radiologist burnout rates globally and dimensions of burnout as reported using different validated instruments; it also provides guidance on best practices to characterize burnout. Fifty-seven studies published between 1990 and 2023 were included in a systematic review, and 43 studies were included in a meta-analysis of burnout prevalence using random-effects models. The reported burnout prevalence ranged from 5% to 85%. With the Maslach Burnout Inventory (MBI), burnout prevalence varied significantly depending on the instrument version used. Among MBI subcategories, the pooled prevalence of emotional exhaustion was 54% (95% CI, 45-63%), depersonalization was 52% (95% CI, 41-63%), and low personal accomplishment was 36% (95% CI, 27-47%). Other validated burnout instruments showed less heterogeneous results; studies using the Stanford Professional Fulfillment Index yielded a burnout prevalence of 39% (95% CI, 34-45%), whereas the validated single-item instrument yielded a burnout prevalence of 34% (95% CI, 29-39%). Standardized instruments for assessing prevalence alongside multidimensional profiles capturing experiences may better characterize radiologist burnout, including change occurring over time.