University of Maryland Medical System

In surgical research, statistical sophistication is too often mistaken for scientific rigor. Across a growing body of plastic surgery literature, adjusted odds ratios, hazard ratios, and regression coef- ficients are frequently presented without the crude event rates or absolute measures of effect that give findings clinical meaning. We describe this phenomenon as "runic statistics": results that are statistically valid yet clinically opaque. Through examples drawn from contemporary plastic surgery studies, we highlight three recurrent interpretive flaws: reliance on statistical significance without consideration of clinical relevance, reporting of relative measures without baseline risks or absolute differences, and conflation of association with causation. We further demonstrate how case-mix imbalances can create apparent contradictions in results (Simpson's paradox), and how identical odds ratios can translate into very different clinical implications depending on the base- line risk. To address these challenges, we propose a thirteen-step reporting framework designed to promote transparency, interpretability, and clinical applicability. Key elements include explicit definition of the estimand, presentation of both crude and adjusted data, translation of relative effects into absolute risks and patient-facing numbers, assessment of minimal clinically important differences, careful handling of confounding, and restraint in the use of causal language. By an- choring statistical reporting in clinical realities, surgical research can remain both methodologically rigorous and directly relevant to patient care. Our goal is not to simplify science, but to ensure that its communication is clear, transparent, and ultimately useful at the bedside.

BackgroundRetention in HIV care is essential to end the HIV epidemic and improve individual patient outcomes, yet 1/3 of people living with HIV in Maryland are not consistently retained in care. Predictive modeling using electronic health record (EHR) data is a promising strategy to identify patients at risk for lapses in care; however, existing efforts remain limited. Regional analyses using statewide health systems can uniquely inform local strategies by accounting for demographic, clinical, and structural variations. This study aimed to develop predictive models utilizing comprehensive EHR data from the University of Maryland Medical System (UMMS) to identify people living with HIV who are at risk of lapsing in care.Figure 1.Receiver Operating Characteristic (ROC) Curve demonstrating the performance of the Random Forest model in predicting lapses in HIV care. The model achieved an area under the curve (AUC) of 0.91, with high recall (98%) and precision (81%), indicating strong predictive capability for identifying patients at risk of lapsing in HIV care.Figure 2.SHAP summary illustrating the impact of key features on the Random Forest model's predictions of HIV care lapses. Each dot represents a patient encounter, with dot colors indicating feature value magnitude (high in red, low in blue).MethodsUtilizing EHR data from UMMS including HIV-related prescriptions, laboratory tests and clinical visits from adults receiving antiretroviral therapy from 1/2016-6/2024, we identified 8,518 patients with 205,633 encounters. Consolidating multiple encounters within 10 days as one encounter, and including only encounters preceding the last recorded encounter for both lapsed and non-lapsed groups, resulted in 4,735 patients. Of those, 2,667 (56%) had lapses - defined as not having an HIV-related clinical encounter within 12 months. We used a Random Forest classifier with extensive hyperparameter tuning via randomized search cross-validation. Model performance was assessed via the performance matrices and SHAP values for feature importance.ResultsThe Random Forest model demonstrated an AUC of 0.91 with 98% recall for lapses, and 81% precision (Figure 1). Feature importance analysis highlighted significant predictors, including time on treatment, age, BMI, alcohol use, employment status, racial demographics, and comorbidities such as hypertension and diabetes (Figure 2).ConclusionPredictive modeling using EHR data from a comprehensive statewide health system can effectively identify patients at risk for lapses in HIV care. Clinically actionable predictors, such as treatment duration, demographics, and specific health conditions, provide practical insights that can guide interventions to enhance patient retention and ultimately improve health outcomes for people living with HIV.DisclosuresAll Authors: No reported disclosures

BackgroundThe optimal tube size for managing traumatic hemothorax, pneumothorax, or hemopneumothorax remains debated. While large-bore chest tubes (LCTs—≥ 28 Ch) are traditionally favored, emerging evidence suggests that small-caliber tubes (SCTs—≤ 14 Ch), such as pigtail catheters and small straight tubes, may offer similar efficacy with fewer complications. This study aimed to evaluate the comparative effectiveness and safety of SCTs versus LCTs from Randomized Controlled Trials (RCTs) in adult trauma patients and to assess the conclusiveness of the current evidence using trial sequential analysis (TSA).MethodsThe study was conducted according to the Cochrane recommendations, searching the PubMed, Scopus, and EMBASE datasets up to 25th March 2025 without language restrictions (PROSPERO ID: CRD420251023165). The primary outcome was treatment failure; secondary outcomes included insertion-related complications, duration of drainage, and length of hospital stay. Random effects models based on restricted maximum likelihood and Hartung-Knapp correction were developed. Sensitivity analysis was conducted to detect sources of heterogeneity. The risk of bias was assessed using the Cochrane RoB 2 tool. TSA was used to evaluate the risk of random error and to determine whether the required information size (RIS) had been reached.ResultsFour RCTs (n = 676 patients) were included. Pooled analysis showed no significant difference in failure rates between SCTs and LCTs (RR 0.95, 95% CI 0.66–1.35, I2 = 0%). No significant differences were observed in complication rates or hospital stay. Duration of tube placement was significantly shorter in the SCT group (MD − 0.49 days, p = 0.02). TSA indicated that the cumulative evidence was underpowered, achieving only 22% of the RIS (3110 patients). The Z-curve did not cross thresholds for benefit, harm, or futility.ConclusionSCTs appear to be as effective and safe as LCTs for selected trauma patients with uncomplicated thoracic injuries. However, due to limited sample size and heterogeneity across trials, current evidence is inconclusive. Larger, high-quality RCTs are warranted to confirm these findings and guide clinical practice.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13017-025-00655-x.

Background: Recurrence of atrial fibrillation (AF) remains a concern even after catheter ablation. The impact of Sodium-Glucose Cotransporter 2 Inhibitor (SGLT2i) on Atrial tachyarrythmia free event remains unclear. Hypothesis: This analysis aims to analyze outcomes in patients treated with SGLT2i after catheter ablation. Methods: A meta-analysis of available studies comparing SGLT2i to non- SGLT2i group following ablation was conducted using electronic databases; PubMed and Embase until May 2025. A random-effects meta-analysis using the DerSimonian–Laird model was performed. Primary outcomes: tachyarrhythmia-free survival and secondary outcomes: all-cause hospitalization and all-cause mortality, and Relative risks (RR) with 95% confidence intervals (CI) were calculated with p <0.05 considered significant. Results: A total of 10 studies were included for meta-analysis with sample size of 10,985 (SGLT2i group: 5534 and non-SGLT2i group: 5451). Primary outcome: There was no significant difference in tachyarrhythmia-free survival (RR: 0.99, 95% CI: 0.76–1.29; p = 0.96), and high heterogeneity (I2 = 84.6%) was observed for this endpoint. Secondary outcome: SGLT2i therapy significantly reduced all-cause hospitalization (RR: 0.74, 95% CI: 0.59–0.94; p = 0.01; I2 = 36.4%) and all-cause mortality (RR: 0.64, 95% CI: 0.45–0.89; p = 0.01; I2 = 0%) Conclusion: SGLT2i therapy offers additive clinical benefits over non-SGLT2i group, notably in reducing hospitalizations and cardiovascular mortality. Its effect on arrhythmia prevention remains uncertain, highlighting the need for further investigation into patient selection and mechanistic interactions.

Thirty years ago, education leaders in emergency medicine (EM) developed a standardized letter of recommendation to address limitations of narrative letters of recommendation in the residency selection process. Since then, multiple iterations and improvements with specialty-wide adoption have led to this letter being cited as one of the most essential pieces of a residency application. Based on the experience and success in EM, many other specialties have also now adopted standardized letters of their own. In this paper, we detail the 30-year history of the EM standardized letter including form changes and technological innovations, research and validity evidence, and discussion of research and administrative priorities for the future.

BACKGROUNDPulmonary hypertension (PH) is a highly morbid disease, but underdiagnosis is common outside of expert referral centers. Consequentially, there may be opportunities to automate PH diagnosis using artificial intelligence (AI) clinical decision support tools. Analysis of patient-level right heart catheterization (RHC) data is required to optimize AI-based PH diagnosis but has not been reported previously.METHODSWe performed a retrospective cohort analysis of all RHC studies (January 1, 2016 to December 31, 2024) performed at the University of Maryland Medical System (UMMS), which is a Maryland statewide clinical network of 12 hospitals serving >2 million patients. We developed an automated large language model (LLM)-driven Pattern Repository (LDPR) method, featuring three task-specific LLM agents for extracting unstructured RHC data, which was manually cross-validated independently by two PH experts. To address data missingness, we used machine-learning to develop formulae to calculate mean pulmonary artery pressure (mPAP) from systolic (sPAP) and diastolic (dPAP) PAP, using an 80/20 train-test split.RESULTSThe study cohort included N=11,029 unique patients and 17,292 RHC reports (age 66±13.5 years; 43% female; 65% White, 30% Black or African American; mPAP, 28±11mmHg; 26% congestive heart failure). The precision for accurate mPAP, sPAP, and dPAP extraction by the LLM was 99.6%, 99.4%, and 99.4%, respectively, with a detection failure of 0.4%. A missing mPAP was noted in N=548 cases and N=507 unique patients (3.2% and 4.6%, respectively). When applying ML to the dataset, the simple, linear equation: mPAP=1.51+0.43*sPAP+0.45*dPAP returned the highest R2 of 0.94 and lowest mean square error of 8.3 mmHg, which outperformed linear equations used currently (all p<0.001). The ML-derived formula was then directed to patients with missing mPAP (N=507) and identified N=382 patients (75.3%) with mPAP >20mmHg, and therefore reclassifying patients from no diagnosis to a diagnosis of PH.CONCLUSIONIn this retrospective cohort analysis, combination LLM-ML-based extraction and interpretation of RHC was used to automate PH diagnosis in a large and heterogenous patient population. This approach is an efficient and scalable solution to preventing under-diagnosis of PH and demonstrates the feasibility of generative AI for advancing clinically-actionable tools that can improve cardiovascular disease phenotyping and diagnosis in real-world settings.