Pneumonia is a common infectious disease, and antibiotic treatment in hospitalized patients must balance efficacy, safety, and resistance risk. However, antibiotic selection and dose adjustment still rely heavily on clinician experience. Although large language models (LLMs) are promising for clinical reasoning, their direct use for antibiotic selection and dose recommendation is limited by hallucinations and weak adherence to clinical constraints.

This study aimed to develop and externally validate a constrained LLM-based clinical decision support pipeline for antibiotic selection and dose recommendation in hospitalized patients with pneumonia.

We conducted a multicenter retrospective study using electronic health record narratives, antibiotic orders, and laboratory indicators of hepatic and renal function from 331 hospitalized patients with pneumonia from 2 hospitals in China. The development cohort included 233 patients, and the external validation cohort included 98 patients. The pipeline integrated dual-branch retrieval (similar-case vector retrieval plus guideline-based knowledge graph retrieval), clinician-defined rule constraints, and hybrid-context reasoning. DeepSeek-V3, GLM-4.6, and GPT-4o were evaluated using F1-score and Jaccard accuracy.

On the internal test set, the full pipeline using DeepSeek-V3 achieved the best performance, with an F1-score of 0.8110 (95% CI 0.7371-0.8762) and Jaccard accuracy of 0.7624 (95% CI 0.6810-0.8386) for antibiotic selection and an F1-score of 0.7538 (95% CI 0.6671-0.8329) and Jaccard accuracy of 0.7076 (95% CI 0.6145-0.7938) for joint antibiotic selection plus dosing recommendation. On the external validation set, performance remained high, with an F1-score of 0.8605 (95% CI 0.7891-0.9252) and Jaccard accuracy of 0.8571 (95% CI 0.7857-0.9184) for antibiotic selection, and an F1-score of 0.8503 (95% CI 0.7789-0.9150) and Jaccard accuracy of 0.8469 (95% CI 0.7755-0.9133) for antibiotic selection plus dosing recommendation. The system also provided traceable evidence and rule trigger information to support clinician review.

A constrained, retrieval-augmented LLM pipeline improved the consistency and interpretability of antibiotic selection and dose recommendation for hospitalized patients with pneumonia and provided preliminary evidence of cross-site generalizability.

Understanding changes in opioid overdose-related emergency department (ED) visits and ED-based opioid use disorder (OUD) treatment post-coronavirus disease 2019 (COVID-19) pandemic can inform ongoing efforts to address the opioid crisis. We aimed to examine trends in opioid overdose-related ED visits, ED-based medication for OUD (MOUD) treatment, and appendicitis-related ED visits (as a control) before and after the initial COVID-19 peak in April 2020.

We conducted an interrupted time series analysis of monthly ED visits from January 2017-December 2022 at three hospitals in California. We modeled pre- and post-COVID-19 visit trends and the change in trend from pre- to post-COVID-19 peak using linear regression controlling for study site. Our primary outcome included monthly rates of opioid overdose-related ED visits, with monthly rates of ED visits with MOUD treatment as a secondary outcome. Appendicitis-related visits served as a control for temporal trends.

Of the 781,488 ED visits across the entire study period, there were 2,536 (0.32%) opioid overdose-related visits, 9,755 (1.25%) MOUD treatment visits, and 1,123 (0.14%) appendicitis-related visits. Pre-pandemic, monthly increases were observed in opioid overdose-related visits (6.7 visits/10,000 per month, 95% confidence interval [CI] 2.6-10.8, P = .001), MOUD-positive visits (34.9 visits/10,000 per month, 95% CI, 26.0-43.7, P < .001), and appendicitis visits (2.5 visits/10,000 per month, 95% CI, 1.2-3.9, P < .001). After April 2020, only MOUD-positive visits showed an immediate (level change) increase (49 visits/10,000 or 34% of April 2020 projected pre-COVID-19 visit rates, 95% CI, 26.1-71.9, P < .001), with opioid overdose-related visits subsequently declining (-4.6 visits/10,000 per month or -11% of April 2020 projected pre-COVID-19 visit rates, 95% CI, -7.2 to -2.0, P = .001). Across the entire pre- to post-COVID-19 period, significant decreases in overall visit trends were observed across all visit types, greatest for MOUD-positive visits (-39.2 visits/10,000 per month, or -27% of April 2020 projected pre-COVID-19 visit rates, 95% CI, -51.5 to -26.8, P < .001), followed by opioid overdose visits (-11.3 visits/10,000 per month, or -27% of April 2020 projected pre-COVID-19 visit rates, 95% CI, -16.1 to -6.5, P < .001) and appendicitis visits (-4.2 visits/10,000 per month, or -23% of April 2020 projected pre-COVID-19 visit rates, 95% CI, -7.2 to -1.2, P = .01).

From pre- to post-initial COVID-19 peak, absolute ED-based MOUD treatment trends declined over three times faster than those of opioid overdose-related ED visits (although percentage changes relative to expected April 2020 rates were both -27%). These findings may reflect reduced perceived urgency as overdose presentations decreased and a shift away from crisis-driven implementation, underscoring the need for intentional integration of MOUD into routine ED practice to sustain treatment capacity.

The last few decades have witnessed a resurgence in pertussis notifications in a number of countries with high vaccine coverage, including Sweden. The underlying causes of the resurgence have been the subject of much scientific debate. To arbitrate among the putative drivers of the resurgence in Sweden, we formulated a mechanistic transmission model which we fit to age-structured time-series notifications data via likelihood maximisation. Given our model, we find the data are best explained by the combined effects of a low basic reproductive number, incomplete (leaky) DTaP-derived immunity, a much lower reporting probability of infections in older individuals and waning of vaccine-derived immunity. In addition, our modelling explains the post-2014 resurgence as a combination of two factors. First, a dynamical transient known as the honeymoon effect, in which a rebound in transmission follows after a rapid decrease in the average population susceptibility. Second, an increase in the infection reporting probability from 2014 onwards, likely due to the use of new laboratory testing methods. Additionally, we used our fitted transmission model to reconstruct indirect protective effects of vaccination. Our results suggest immunization prevents about 50% of potential infections in individuals too young to receive vaccination. However, our statistical inference demonstrates that pertussis elimination is not possible with routine immunization using acellular vaccines due to the combined effects of vaccine leakiness and waning immunity.

Hypoventilation and hypercapnia are the primary physiological indicators of opioid-induced respiratory depression (OIRD), while hypoxia is typically a later manifestation. However, use of pulse oximetry to measure oxygen saturation (SpO₂) level remains the most widely available and commonly used monitoring modality in both clinical and community settings, whereas capnography is not routinely accessible. Given this reality, it is important to evaluate hypoxia thresholds reported in the literature to inform practical detection and intervention strategies. Oxygen saturation monitoring may serve as a valuable support tool to help determine when individuals experiencing opioid toxicity require intervention; however, there is limited consensus on SpO₂ thresholds that indicate OIRD. Our objective in this study was to evaluate existing evidence on SpO₂ thresholds for OIRD as a physiological marker to incorporate with clinical assessment to potentially prevent opioid-related outcomes such as hypoxemia, hypoxic brain injury, cardiac arrhythmias, and mortality, and to inform monitoring and intervention strategies for future research.

We electronically searched Ovid MEDLINE, Ovid Embase, PubMed, and grey literature databases for qualitative and quantitative studies published from 2014-2024. We included studies published in English where participants with opioid use disorder (OUD) experienced opioid toxicity and continuous SpO₂ monitoring in medical settings. Quality was evaluated using the modified Downs and Black checklist. The primary outcome measure was the SpO₂ level used to quantify opioid-induced respiratory depression.

Of the 2,864 articles screened, 37 underwent full-text review, and 16 were included in data extraction and analysis (total patients, N = 12,887). Reported SpO₂ thresholds ranged from 85-95% (median 92, interquartile range 90-95). In three studies where patients experienced opioid-induced respiratory depression, SpO₂ levels dropped to a median value of 80%.

Although hypoxia is a late indicator of opioid-induced respiratory depression, SpO₂ monitoring remains a widely deployed tool for detecting clinically significant deterioration. Reported SpO₂ thresholds for defining OIRD ranged from 90-95% across studies of "excellent" and "good" quality, underscoring the need for standardized, patient-centered thresholds to guide monitoring and intervention in future research.

No-reflow phenomenon (NRP) frequently complicates ST-elevation myocardial infarction (STEMI). This study evaluated the CHA₂DS₂-VA score (CV score) as a predictor of NRP in STEMI patients undergoing primary PCI.

A total of 725 STEMI patients were retrospectively enrolled and stratified by median of CV score (≤ 2 vs. > 2). The predictive ability of the CV score for NRP was assessed using logistic regression, receiver operating characteristic analysis, and DeLong tests.

NRP incidence was higher in the high CV score group (15.2% vs. 8.7%, p = 0.007). The CV score predicted NRP overall (adjusted OR: 1.295, p = 0.002), but this effect was confined to females (AOR: 1.597, p = 0.002), with no significant association in males. Area under the curve value (AUC) of the CV score for predicting NRP was 0.613 in all patients (38.8% sensitivity and 79.1% specificity, p = 0.001) and 0.673 in females (78.1% sensitivity and 47.6% specificity, p = 0.002). A significant sex × CV score interaction (p = 0.017) confirmed sex-modified predictive value. DeLong tests combining the CV score with laboratory markers improved prediction in males but not females.

The CV score is a sex-specific predictor of NRP in STEMI, with predictive value in females but not males. These findings support a tailored risk stratification approach: using the CV score alone for female patients, while integrating laboratory and procedural markers for male patients.

Cardiovascular-kidney-metabolic (CKM) syndrome poses a major health risk. This study assessed the impact of a healthy lifestyle on all-cause and cardiovascular disease (CVD) mortality in CKM individuals.

We analyzed 306 831 and 9823 CKM individuals from UK Biobank and NHANES (2007-2018). A healthy lifestyle score was created based on seven factors: no current smoking, moderate drinking, healthy diet, regular physical activity, adequate sleep, low sedentary behavior and appropriate social connection. Primary outcomes were all-cause and CVD mortality from linked health records and death registries.

Cox regression models showed that higher lifestyle scores were associated with reduced risks of all-cause (hazard ratio [HR]: 0.81, 95% confidence interval [CI]: 0.80-0.82) and CVD mortality (HR: 0.82, 95% CI: 0.80-0.84) in UK Biobank. In NHANES, higher scores were associated with 15% (95% CI: 0.80-0.90) and 11% (95% CI: 0.81-0.99) lower mortality risks. Both regular physical activity (HRs: 0.85 and 0.89 for UK Biobank, 0.76 and 0.73 for NHANES) and low sedentary behavior (HRs: 0.85 and 0.91 for UK Biobank, 0.79 and 0.67 for NHANES) were related to reduced mortality risks (p < 0.05). Participants with non-advanced CKM syndrome and a favorable lifestyle had the lowest mortality risks compared to those with advanced CKM syndrome and an unfavorable lifestyle.

A healthy lifestyle, particularly regular physical activity and low sedentary behavior, was significantly associated with lower all-cause and CVD mortality in individuals with CKM syndrome. These associations were stronger in the early stages, highlighting the importance of timely and targeted lifestyle interventions.

This 58-minute webcast features a conversation on the "State of the Art in Pericardial Disease"-the focus of Issue 22.2 on the theme of pericardial disease. Led by the issue's editor, the discussion engages the authors on emerging themes and lessons learned while researching and writing the articles. View the video at https://vimeo.com/1204339129/42dd65d78b.

Device sales representatives (DSRs) play a key role in providing vascular surgeons with device-specific education, operational guidance, and troubleshooting support. However, industry relationships may introduce bias, raising ethical concerns. This study aimed to evaluate the clinical benefit of DSR presence in the operating room (OR) during standard, infrarenal endovascular aortic repair (EVAR). A retrospective, case-control study was conducted to assess the association between DSR presence and surgical outcomes in elective, infrarenal EVAR cases at an experienced endovascular institution. Preoperative (age, gender, ASA classification, and past medical history), perioperative (device company, OR and fluoroscopy time, and use of additional endografts or stents) and postoperative (length of hospital stay post-operation, endoleaks at six weeks and six months, and mortality at six months) variables were collected for cases between June 1st, 2019 to December 31st, 2022. Fisher's exact test, chi-squared test and student's t-test were used to compare DSR-present and control groups. Ninety standard infrarenal EVARs were included with a DSR present in 64 (71.1%) cases. Cook and Gore devices were used in roughly equal proportions with a single case using a Medtronic device. DSR presence was not associated with a statistically significant change in operative duration (p = 0.53) or fluoroscopy time (p = 0.95), or in rates of endograft usage (p = 0.35) or adjunctive stent use (p = 1.00). Length of hospital stay (p = 0.46), short- (p = 0.63) and long-term (p = 1.00) endoleaks, and mortality rates (p = 1.00) were similar between groups. In this study of standard, infrarenal EVAR procedures at a high volume endovascular centre, DSR presence was not associated with statistically significant differences in case efficiency, device utilization or clinical outcomes. These findings suggest that routine DSR involvement in such procedures may not provide a clear clinical advantage. In light of important ethical considerations, the study calls for a more selective use of DSR support and for larger, multicentre studies to better understand their role in EVAR cases.

Prevalence of heart failure is increasing, with an associated rise in emergency department (ED)-related care. Social determinants of health (SDoH) are associated with adverse health outcomes, but the extent to which they influence ED use for heart failure care is poorly understood. We sought to describe the relationship between community-level social vulnerability and ED revisits and hospital admissions for heart failure within a single health system.

We conducted a retrospective review of Social Deprivation Index (SDI) scores (higher score = more deprivation) by ZIP code paired with administrative clinical data. Zero-hurdle regression was used to model the relationship between SDI and ED revisits within 90 days of a prior ED visit for heart failure (logistic model for > 0 revisits; Poisson model for count of revisits). We used a mixed-effects logistic model-accounting for repeat visits-to test the association of SDI with hospital admission at any given ED visit.

From January 2022-December 2023, there were 3,569 ED visits from 2,406 patients. Each standard deviation increase in SDI (30.3) was associated with increased odds of at least one 90-day revisit (OR, 1.53; 95% CI, 1.08-2.16). Higher SDI was also associated with more 90-day revisits, with varying magnitude by hospital. After adjusting for characteristics of prior ED visit, SDI was not associated with hospital admission.

Our results suggest that area-level social vulnerability influences the decision to seek heart failure-related care in the ED. Patients from more deprived areas may not have more severe clinical presentations, however, as evidenced by lack of association of SDI with hospital admission.

Methamphetamine-associated heart failure with reduced ejection fraction is a serious consequence of methamphetamine use often underrecognized in the emergency department (ED). Point-of-care ultrasound (POCUS) offers rapid, non-invasive cardiac screening for high-risk populations. This study evaluated the diagnostic yield of POCUS for detecting methamphetamine-associated heart failure with reduced ejection fraction in ED patients who use methamphetamine.

We conducted this prospective cohort study between December 2020-February 2024 at an urban Level I trauma center ED. The primary outcome was diagnostic yield of cardiac POCUS for abnormal left ventricular ejection fraction (LVEF) and abnormal sex-specific left ventricular end-diastolic diameter in patients with methamphetamine use, with secondary analyses assessing associations with use duration and frequency. Diagnostic yield was calculated as the proportion of completed POCUS examinations identifying abnormalities. We used E-point septal separation to calculate LVEF; < 40% was abnormal. Left ventricular end-diastolic diameter abnormality (> 5.8 cm males, > 5.2 cm females) was categorized as mild, moderate, or severe (mild, 5.9-6.3/5.3-5.6; moderate, 6.4-6.8/5.7-6.1; severe, > 6.8/> 6.1 cm males/females). Physician-performed POCUS assessed LVEF and left ventricular end-diastolic diameter in patients with a methamphetamine use history and a comparison group of non-users.

Of the 136 enrolled patients, 84 (61.8%) reported methamphetamine use. Among methamphetamine users, diagnostic yield of cardiac POCUS was as follows: reduced LVEF in 22 of 70 with measurable LVEF (31.4% [20.9-43.6%]); any sex-specific left ventricular end-diastolic diameter abnormality in 30 of 84 (35.7% [25.6-46.9%]); and severe sex-specific left ventricular end-diastolic diameter abnormality in 15 of 84 (17.9% [10.4-27.7%]). Corresponding values in non-users were 3 of 52 (5.8% [1.2-16.0%]; P < .001), 10 of 52 (19.2% [9.6-32.5%]; P = .05), and 3 of 52 (5.8% [1.2 - 16.0%]; P = .07), respectively. Longer duration (odds ratio [OR] 6.05, 95% confidence interval [CI] 1.46-25.10) and higher frequency (OR 5.93, 95% CI, 1.52-23.06) of methamphetamine use were associated with reduced LVEF.

Cardiac point-of-care ultrasound demonstrated a clinically significant diagnostic yield in detecting methamphetamine-associated heart failure with reduced ejection fraction among methamphetamine users in the ED, suggesting POCUS could aid early detection in this population, potentially streamlining management before disease progression.