Citations

Background. Solid organ transplant recipients are at risk of severe coccidioidomycosis and are given prophylaxis to mitigate the risk. Patients with seropositive testing typically receive lifelong prophylaxis; currently, this prophylaxis strategy includes patients who are positive only for IgM by enzyme immunoassay (EIA-IgM-only), although this result may be falsely positive. Methods. We conducted a retrospective study at a large-volume transplant center in an endemic coccidioidomycosis region to compare outcomes of non–lung transplant recipients who were seropositive for Coccidioides but discontinued prophylaxis (case patients) to outcomes of patients who continued prophylaxis (controls). Results. No case or control patients developed active coccidioidomycosis during the follow-up period. Before transplant, 62 of 77 case patients (80.5%) had a single positive serologic test, of whom 27 of 62 were EIA-IgM-only positive (43.5%). In contrast, 57 of 77 controls (74.0%) had a single seropositive result (16/57 [28.1%] were EIA-IgM-only). The single EIA-IgM-only result was classified as falsely positive by infectious disease consultants in 20 of 43 patients (46.5%) compared with all other Coccidioides serologic results (13/111 [11.7%], P < 0.001). Lifetime antifungal prophylaxis was felt to be unnecessary for 28 of 43 patients (65.1%) who were EIA-IgM-only versus 31 of 111 patients (27.9%) with other serologic results (P < 0.001). Conclusions. For patients repeatedly positive for EIA-IgM-only and no evidence of seroconversion, compatible clinical illness, or radiographic findings, discontinuing antifungal prophylaxis may be reasonable after the first posttransplant year.

Abstract Artificial intelligence (AI) models in medicine often fall short in real-world deployment due to inability to incorporate multiple data modalities in their decision-making process as clinicians do. Clinicians integrate evidence and signals from multiple data sources like radiology images, patient clinical status as recorded in electronic health records, consultations from fellow providers, and even subtle clues using the appearance of a patient, when making decisions about diagnosis or treatment. To bridge this gap, significant research effort has focused on building fusion models capable of harnessing multi-modal data for advanced decision making. We present a broad overview of the landscape of research in multimodal AI for radiology covering a wide variety of approaches from traditional fusion modelling to modern vision-language models. We provide analysis of comparative merits and drawbacks of each approach to assist future research and highlight ethical consideration in developing multimodal AI. In practice, the quality and quantity of available training data, availability of computational resources, and clinical application dictates which fusion method may be most suitable.

Background: The relative advantages and disadvantages of 2-stage versus 1-stage management of infection following total hip arthroplasty (THA) are the current subject of intense debate. To understand the merits of each approach, detailed information on the short and, importantly, longer-term outcomes of each must be known. The purpose of the present study was to assess the long-term results of 2-stage exchange arthroplasty following THA in one of the largest series to date. Methods: We identified 331 periprosthetic joint infections (PJIs) that had been treated with a 2-stage exchange arthroplasty between 1993 and 2021 at a single institution. Patients were excluded if they had had prior treatment for infection. The mean age at the time of reimplantation was 66 years, 38% of the patients were female, and the mean body mass index (BMI) was 30 kg/m 2 . The diagnosis of PJI was based on the 2011 Musculoskeletal Infection Society criteria. A competing-risk model accounting for death was utilized. The mean duration of follow-up was 8 years. Results: The cumulative incidence of reinfection was 7% at 1 year and 11% at 5 and 10 years. Factors predictive of reinfection included BMI ≥30 kg/m 2 (hazard ratio [HR] = 2; p = 0.049) and the need for a spacer exchange (HR = 3.2; p = 0.006). The cumulative incidence of any revision was 13% at 5 and 10 years. The cumulative incidence of aseptic revision was 3% at 1 year, 7% at 5 years, and 8% at 10 years. Dislocation occurred in 33 hips (11% at 10 years); 15 (45%) required revision. Factors predictive of dislocation were female sex (HR = 2; p = 0.047) and BMI <30 kg/m 2 (HR = 3; p = 0.02). The mean Harris hip score (HHS) improved from 54 to 75 at 10 years. Conclusions: In this series of 331 two-stage exchange arthroplasties that were performed for the treatment of infection, we found a low rate of aseptic revision (8%) and a low rate of reinfection (11%) at 10 years. These long-term mechanical and infection data must be kept in mind when considering a paradigm shift to 1-stage exchanges. Level of Evidence: Therapeutic Level III . See Instructions for Authors for a complete description of levels of evidence.

Abstract Background Inflammatory bowel disease (IBD) predisposes individuals to frailty, linked with adverse outcomes. While the Fried Frailty Index (FFI) is a well-established phenotypic tool to assess frailty, its administration is cumbersome. The FRAIL scale, simpler but not widely used in IBD patients, presents an alternative. We aimed to assess the prevalence of frailty and compare the FRAIL scale with the FFI. Methods Two hundred and nine confirmed IBD patients underwent assessment using both the FFI and the FRAIL scale. Patients were categorized as non-frail, pre-frail, or frail. Pearson coefficients evaluated correlation, and unadjusted analyses assessed frailty risk factors. ROC curve analysis evaluated the FRAIL scale. Results Among participants (53.5% female, median age 44), 37% were non-frail, 50% pre-frail, and 13% frail. The FRAIL scale exhibited strong correlation with the FFI for all three categories. Age showed no significant association with frailty. Frail individuals displayed higher inflammatory markers and more severe clinical disease, with frailty more prevalent in UC than CD patients. Frail individuals also exhibited lower hemoglobin, creatinine, and albumin levels. Conclusion Frailty and pre-frailty are prevalent in the IBD population, not necessarily linked with older age. The FRAIL scale demonstrated excellent correlation with the FFI, offering a practical tool for identifying frailty in IBD without physical measurements. Future studies should explore multivariable models incorporating frailty risk factors and interventions to mitigate adverse outcomes in the IBD population.

Importance Eligibility criteria for randomized clinical trials (RCTs) are designed to select clinically relevant patient populations. However, not all eligibility criteria are strongly justified, potentially excluding marginalized groups, and limiting the generalizability of trial findings. Objective To summarize and evaluate the justification of exclusion criteria in published RCTs in critical care medicine. Evidence Review A systematic sampling review of parallel-group RCTs published in the top 5 general internal medicine journals by impact factor ( The Lancet, New England Journal of Medicine, Journal of the American Medical Association, British Medical Journal, and Annals of Internal Medicine ) between January 1, 2018, and February 23, 2023, was conducted. RCTs enrolling adults in intensive care units (ICUs) and RCTs enrolling critically ill patients who required life-sustaining interventions typically initiated in the ICU were included. All study exclusion criteria were categorized as either poorly justified, potentially justified, or strongly justified, adapting previously established criteria, independently and in duplicate. Findings In total, 225 studies were identified, 75 of which were included. The median (IQR) number of exclusion criteria per trial was 19 (14-24), with 1455 total exclusion criteria. Common exclusion criteria were related to the risk of adverse reaction to interventions (302 criteria [20.8%]), followed by inability to obtain consent (120 criteria [8.2%]), and treatment limitation decisions (97 criteria [6.7%]). Most exclusion criteria were either strongly justified (1080 criteria [74.2%]) or potentially justified (297 criteria [20.4%]), whereas 5.4% (78 criteria) were poorly justified. Of the 78 poorly justified exclusion criteria, the most common were pregnancy (19 criteria [24.4%]), communication barriers (11 criteria [14.1%]), lactation (10 criteria [12.8%]), and lack of health insurance (10 criteria [12.8%]). Overall, 45 of 75 studies (60.0%) had at least 1 poorly justified exclusion criteria. Conclusions and Relevance Most exclusion criteria in critical care medicine RCTs were strongly justifiable. Across poorly justified criteria, the most common exclusions were pregnant or lactating persons, those with communication barriers, and individuals without health insurance. This highlights the need to carefully consider exclusion criteria when designing trials to minimize the inappropriate exclusion of participants and enhance generalizability.