Pakistan journal of pharmaceutical sciencesYong Ni, Zhiqiang Tang, Ting Huang, Jian Gong
BACKGROUND: Acute alcohol intoxication (AAI) is a common emergency with no specific antidote. Metadoxine has shown potential but lacks sufficient real-world evidence, particularly in Chinese populations. OBJECTIVES: To evaluate the clinical efficacy and safety of metadoxine in patients with acute alcohol intoxication. METHODS: This single-center retrospective cohort study included 124 patients with AAI admitted to an emergency department between January and December 2024. Patients were divided into metadoxine plus conventional treatment (n=62) or conventional treatment alone (n=62) after propensity score matching. Primary outcomes were time to consciousness recovery (Glasgow Coma Scale ≥14) and treatment effectiveness rate (symptom improvement within 6 hours without treatment escalation). Secondary outcomes included time to symptom relief, emergency stay, time to vital sign stabilization, blood ethanol decline rate, adverse reactions, and liver function changes. RESULTS: Metadoxine significantly shortened consciousness recovery time (3.05±1.45 vs. 5.65±1.58 hours, mean difference -2.60 hours, 95% CI: -3.14 to -2.06, p < 0.001) and improved treatment effectiveness rate (93.5% vs. 77.4%, OR=4.04, 95% CI: 1.30-12.56, p = 0.011). All secondary outcomes were significantly better in the metadoxine group (p < 0.001 for time-related outcomes and blood ethanol decline). No significant differences were observed in adverse reactions (9.68% vs. 6.45%, p = 0.510) or post-treatment liver enzyme abnormalities (12.90% vs. 16.13%, p = 0.610). Subgroup analysis showed a significant interaction by initial Glasgow Coma Scale (p = 0.037), indicating greater efficacy in patients with moderate-to-severe impairment. CONCLUSION: Metadoxine, when added to conventional supportive therapy, significantly accelerates consciousness recovery and improves clinical outcomes in acute alcohol intoxication without increasing short-term adverse events. The benefit appears more pronounced in patients with lower Glasgow Coma Scale scores.
BMC medical informatics and decision makingYi Lian, Xiaoqian Jiang, Qi Long
BACKGROUND: Electronic health records (EHRs) collected from diverse healthcare institutions offer a rich and representative data source for clinical research. Federated learning enables analysis of these distributed data without sharing sensitive patient-level information, preserving privacy. However, missing data remain a major challenge and can introduce substantial bias if not properly addressed. Very few distributed imputation methods currently exist, and they fail to account for two critical aspects of EHR data: correlation within sites and variability across sites. We aim to fill this important methodological gap. METHODS: We propose Distributed Mixed Model-based Multiple Imputation (D3MI), a novel federated imputation method designed to reduce bias in distributed EHRs. D3MI integrates the strengths from federated learning techniques, statistical learning methods for correlated data, and multilevel imputation algorithms to explicitly account for both within-site correlation and between-site heterogeneity using site-specific random effects. It operates under the Missing at Random (MAR) assumption. It preserves privacy by avoiding sharing raw data and features communication and computational efficiency. RESULTS: Through extensive simulation studies, we demonstrate that D3MI outperforms SOTA distributed imputation methods in both accuracy and consistency. We further demonstrate the use of D3MI in a real-world EHR case study involving incomplete and clustered data from participating hospitals in the Georgia Coverdell Acute Stroke Registry. CONCLUSION: By explicitly modeling the complex structure of distributed EHR data, D3MI addresses key limitations of existing approaches. It provides a powerful and efficient solution for handling missing data in distributed and privacy-sensitive settings and enhances the rigor and reproducibility of collaborative clinical research.
JMIR research protocolsXiaoyi Zhang, Christopher R Wilson, Hannah Eyre, David E Reed Ii, Travis Y Hee Wai, Alexander Kloehn, Ethan W Rosser, Gang Luo, Steven B Zeliadt
BACKGROUND: Chronic pain is a leading cause of disability and requires multidimensional assessment of pain intensity and functioning, yet electronic health records rarely capture these measures systematically. By contrast, surveys collecting patient-reported outcomes can assess pain over multiple dimensions but remain resource-intensive and difficult to scale for continuous population-level monitoring. OBJECTIVE: The objective of this study is to develop and validate a domain-informed natural language processing framework to derive pain severity and functional interference outcomes from unstructured clinical narratives. We aim to demonstrate that natural language processing-derived outcomes can serve as a reliable, scalable surrogate for resource-intensive patient-reported surveys. METHODS: This study uses a retrospective cohort of 3725 Veterans with chronic musculoskeletal pain initiating complementary and integrative health therapies across 18 Veterans Health Administration Whole Health Flagship sites (2021-2023). The dataset encompasses longitudinal patient-reported outcome surveys serving as the benchmark, linked with unstructured clinical narratives from the Veterans Health Administration electronic health record. Guided by established psychometric instruments and subject matter expert (SME) input, we developed a seed lexicon and annotation guidelines to identify language distinguishing 3 pain domains: pain severity, interference with enjoyment of life, and interference with general activities. Preliminary large language model (LLM) prompting was used to identify 600 candidate encounters (200 per domain) from 6747 notes across 260 patients for SME annotation, forming a ground truth validation sample. Two candidate LLMs will be evaluated on this sample; the best-performing LLM will generate a large library of span-level annotations to train a scalable, lightweight language model. The study uses a 3-stage validation process: (1) documentation completeness of pain interference in clinical narratives against SME-annotated references; (2) inference accuracy of the LLM-as-annotator and the fine-tuned lightweight model against SME annotations across note-level classification and span-level localization; and (3) concordance between the lightweight model's output and patient-reported pain, enjoyment, and general activity scores across a range of temporal windows. RESULTS: As of July 2026, the cohort of 3725 Veterans has been identified and linked to clinical notes. The seed lexicon and annotation guidelines have been developed. Applying a developmental LLM to screen 6747 text notes in 6642 unique encounters over a 7-month period for 260 patients, at least 1 of the 3 pain domains was identified in 75% of notes and 99% of patients. SME validation at the encounter level is in progress. Final results from the subsequent knowledge distillation and validation stages are expected in the first half of 2027. CONCLUSIONS: This protocol outlines a framework for identifying severe pain intensity and interference from clinical narratives, addressing a critical gap in health care system surveillance. To our knowledge, this is the first study to validate clinical text-based pain outcome extraction against patient-reported outcomes in a nationwide longitudinal cohort. If successful, this approach will enable health care systems to continuously monitor reports of pain-related functional interference and support more holistic, patient-centered pain management at scale.
Health information management : journal of the Health Information Management Association of AustraliaMaria Schwarz, Elizabeth C Ward, Sarah Jeffery, Joshua Simmons, Kristy Perkins, Philip Juffs, Sara Burrett
BACKGROUND: Electronic medical records (EMRs) currently rely on standardised data fields to support secondary data use for clinical care, performance monitoring, and system-level reporting. However, utilisation of standardised data capture and reporting within allied health remains underdeveloped in practice. Greater understanding of how allied health clinicians and managers perceive the purpose, value, and impact of standardised data collection is needed to support advancement in data maturity and transformation of utilisation. OBJECTIVE: To explore the perceived benefits, challenges and future opportunities of standardised data collection within an EMR, from the allied health perspective. METHOD: A qualitative descriptive study using semi-structured interviews was conducted with 28 allied health clinicians and managers from 5 allied health disciplines: Dietetics, Occupational Therapy, Physiotherapy, Speech Pathology, and Social Work. RESULTS: Five overarching themes reflected a gap between the acknowledged value of standardised data collection and its current utilisation in allied health practice: (i) Recognised value but unrealised potential of EMR data, including its role in clinical insights, service planning, financial justification, and professional advocacy; (ii) System and workflow misalignment limits meaningful data use, including disconnect between data collection and use, workflow burden, inefficient data capture, system constraints; (iii) Standardisation as both enabler and constraint, supporting benchmarking and system alignment while limiting clinical flexibility and contextual relevance; (iv) Pathways to advancing data maturity, including training, improved data and digital literacy, and integration with emerging technologies; and the broader impact of (v) Digital workflows reshaping workforce and care delivery, with implications for team-based and person-centred care. CONCLUSION: While standardised EMR data were recognised as valuable, their potential remains limited when data collection is poorly aligned with allied health workflows. Advancing allied health data maturity requires clinically meaningful, workflow-integrated and feedback-driven approaches that improve usability while supporting high-quality data for secondary use.Implication for health information management practice:Standardisation efforts must balance data quality with clinical relevance, ensuring data collection aligns with real-world workflows and supports meaningful use.
BMJ openFerguson Saapiire, Charlotte Tawiah, Francis Agbokey, Kenneth Ayuurebobi Ae-Ngibise, Wisdom Kudzo Axame, Frank E Baiden
OBJECTIVE: To explore the readiness of health managers to integrate effective coverage into routine monitoring of maternal and child health (MCH) services in two rural districts of Ghana. METHODS: An exploratory qualitative study was conducted among purposively selected district and subdistrict health managers in Kintampo North and Kintampo South, Ghana. Two rounds of in-depth interviews explored participants' perceptions and implementation considerations regarding the use of effective coverage. Between the two rounds, participants took part in a 3-month pilot implementation phase. This phase included workshops, hands-on training and planning sessions. During this period, participants were trained to compute and interpret effective coverage using routine data and to explore its application to MCH monitoring and evaluation. Data were analysed thematically in NVivo V.17 using Braun and Clarke's six-phase approach. RESULTS: Following the learning sessions, participants reported an understanding of the components of effective coverage and how to compute and apply the indicator using routine district data. They also reported confidence in computing and interpreting the indicator. In the follow-up interviews, participants expressed willingness to integrate the indicator into routine monitoring processes. However, they identified technical and logistical challenges and inadequate data in the current national health information management system. CONCLUSION: Health managers in rural Ghana expressed willingness to integrate effective coverage into the routine monitoring and evaluation of MCH programmes. However, limited availability of routine data within the national health information management system remains a major barrier to implementation.
Neurosurgical reviewMelis Demirci, Süheyla Serin Senger, Sevim Selen Karabulut, Sevgül İşeri, Mahmut Çamlar, Abdullah Bozoklar, Çağlar Türk
External ventricular drainage (EVD) remains indispensable for managing intracranial pressure and cerebrospinal fluid (CSF) disorders. Despite advances in neurosurgical care and infection prevention, EVD-related infections still contribute substantially to morbidity and mortality. This study aimed to identify predictors of mortality and evaluate the clinical relevance of infection-related and metabolic parameters in patients with suspected EVD-associated infection undergoing EVD placement. We retrospectively analyzed all adult patients who underwent EVD insertion and were clinically suspected of having an EVD-associated infection at a tertiary hospital over an 18-month period. Data from electronic medical records included demographics, neurological diagnosis, comorbidities, catheter characteristics, laboratory results, and antimicrobial therapy. EVD-associated infection was defined according to Centers for Disease Control and Prevention (CDC) criteria. Statistical analyses were conducted using IBM SPSS Statistics 26. A total of 102 patients with clinically suspected EVD-associated infection were evaluated. CSF cultures were positive in 33 (32%) cases, most frequently yielding Staphylococcus epidermidis and Acinetobacter baumannii . Intensive care unit (ICU) admission was associated with higher crude mortality rates, likely reflecting greater baseline neurological severity. Initial Glasgow Coma Scale (GCS) score (aOR = 0.85, p = 0.005) was found to be an independent predictor of mortality. Neither sex nor comorbidities predicted death. Hyperglycemia (> 140 mg/dL) and elevated lactate (> 1.2 mmol/L) independently associated with 30-day mortality. Shorter catheterization duration (≤ 5 days) also predicted death, likely reflecting greater neurological compromise. In our selected cohort of patients with clinically suspected EVD-associated infection, CSF culture positivity was not independently associated with mortality, although patients with culture-confirmed infection exhibited higher CSF pleocytosis and Cell Index (CI) values. Among patients with clinically suspected EVD-associated infection, initial neurological severity, together with markers of systemic metabolic stress, were independently associated with 30-day mortality. Hyperglycemia and elevated lactate emerged as key prognostic markers. While elevated CI values may support the diagnosis of culture-confirmed infection, CI did not demonstrate independent prognostic value for mortality. These findings should be interpreted within the context of this selected cohort and should not be generalized to the overall EVD population. Larger prospective studies including a greater number of culture-confirmed infections are needed to confirm these results.
JMIR agingLiqin Wang, Rebecca E Amariglio, Sheril Varghese, Jiazi Tian, Diane Seger, Li Zhou, Gad A Marshall
BACKGROUND: Subjective cognitive decline (SCD) typically refers to self- or informant-reported decline in cognition despite the absence of objective impairment on standardized testing. Older adults with documented normal cognitive test performance provide a pragmatic anchor cohort for electronic health record (EHR)-based SCD phenotyping. However, cognitive concerns are primarily recorded in unstructured notes and are inconsistently documented, making it unclear how often and in whom concerns are captured in routine care. OBJECTIVE: This study aims to operationalize EHR-based SCD phenotyping in an objectively normal-testing cohort using a large language model (LLM)-based natural language processing approach and to examine the frequency and clinical and sociodemographic correlates of cognitive concern documentation. METHODS: We conducted an EHR-based observational study of patients aged 65 years or older with a first normal cognitive test recorded in EHR flowsheets between January 2019 and April 2024 in a large health care system. We developed and iteratively refined a 2-stage LLM-based natural language processing pipeline to identify documented cognitive concerns in unstructured notes during the 12 months prior to the index date, and evaluated performance against manual review. We quantified the frequency of documented concerns within this normal-testing cohort and used multivariable logistic regression to assess associations with sociodemographic factors (age, marital status, insurance, and neighborhood Area Deprivation Index), sequentially adjusting for clinical comorbidities and relevant medications. RESULTS: Among 15,750 older adults with normal cognitive test scores, 13.8% (n=2175) had at least 1 documented cognitive concern in the prior year, captured in 1.2% (7394/605,177) of notes. On manual validation, the 2-stage pipeline (Med42-v2-8B screening followed by GPT-4o confirmation) achieved a sensitivity of 0.957, positive predictive value of 0.935, specificity of 0.985, and F1-score of 0.945 for cognitive concern identification. Documentation was more likely in older individuals, those with commercial (vs Medicare) insurance, and those with neurological and psychiatric conditions. In fully adjusted models, Parkinson disease (adjusted odds ratio [aOR] 5.29, 95% CI 3.60-7.77), traumatic brain injury (aOR 4.63, 95% CI 3.15-6.81), stroke or transient ischemic attack (aOR 3.46, 95% CI 3.15-4.18), epilepsy (aOR 2.52, 95% CI 1.83-3.49), depression (aOR 1.54, 95% CI 1.36-1.75), and excessive alcohol use (aOR 1.48, 95% CI 1.06-2.06) were among the strongest correlates of cognitive concern documentation. In contrast, obesity (aOR 0.73, 95% CI 0.65-0.82), hyperlipidemia (aOR 0.59, 95% CI 0.51-0.67), and residence in more deprived neighborhoods (higher Area Deprivation Index) were associated with lower odds of documented cognitive concerns. CONCLUSIONS: A 2-stage LLM pipeline enabled accurate identification of documented cognitive concerns consistent with SCD among older adults with normal cognitive testing. Documentation was uncommon and selectively captured by clinical and sociodemographic factors, with implications for equity and the validity of EHR-based phenotypes.
Clinical nurse specialist CNSMichael Gnidovec, Deb L Lindell
PURPOSE/OBJECTIVES: The project aimed to improve the safety of hospitalized tracheostomy and laryngectomy adults by reducing the time to initial respiratory therapist assessment and the delivery of emergency equipment to the bedside. DESCRIPTION: Tracheostomy and laryngectomy patients are classified as a low-volume, high-risk patient population. The specialized care and equipment needed for them are essential, especially in an acute care setting like a hospital. The ability to ensure safety equipment compliance at the time of admission is a challenge as it relies heavily on communication, which can be subject to human error. Three months of retrospective data were analyzed, which led the team to identify a gap in the notification process. The project team created an automated pager notification of new patients needing assessment and delivery of emergency equipment, triggered by an EHR notification when the patient's status changed to admitted. OUTCOME: A Kruskal-Wallis test compared time to assessment between the 2 groups and established that the difference was not statistically significant (P=0.121). However, the lower mean and median times to assessment in the postintervention group are clinically significant for this population. CONCLUSIONS: Automated notifications were effective in reducing delays in care and delivery of emergency equipment for this vulnerable population. .
PloS oneMohammad Rasouli, Elham Rahimi, Arshia Ilaty
Effective monitoring of thyroid cancer treatment requires timely detection of clinically meaningful changes across complex, multistage care pathways. Traditional monitoring approaches often rely on static thresholds and periodic assessments, limiting their sensitivity to subtle yet important process deviations and potentially delaying clinical intervention. This study aimed to develop and evaluate an integrated, data-driven framework that combines deep learning-based risk prediction with genetic algorithm-optimized statistical process monitoring to enhance thyroid cancer treatment surveillance. Informative clinical variables were first identified using the Information Functions algorithm, and a Multilayer Perceptron (MLP) neural network was then employed to predict patient-specific treatment risk, capturing nonlinear relationships within longitudinal electronic medical record data. The resulting risk estimates were sequentially incorporated into Exponentially Weighted Moving Average (EWMA) control charts to enable continuous process monitoring. To further improve chart performance, a genetic algorithm was used to optimize EWMA parameters, enhancing accuracy and sensitivity while maintaining chart stability. The results demonstrate that the MLP model outperforms conventional machine learning approaches in risk prediction across multiple evaluation metrics. Genetic algorithm-optimized EWMA charts exhibit smoother trajectories, improved stability, and more reliable detection of clinically meaningful deviations compared with non-optimized charts. The consistency of these improvements across multiple features suggests that GA-based optimization enhances overall control chart performance rather than feature-specific behavior. This hybrid framework demonstrates the feasibility of integrating machine learning-derived risk prediction with optimized statistical process control for longitudinal treatment monitoring. The proposed approach supports proactive treatment surveillance, risk-stratified follow-up, timely clinical intervention, and data-driven decision-making in thyroid cancer care. Furthermore, the framework has the potential to enable continuous and adaptive monitoring across complex multi-stage treatment pathways if integrated with electronic medical record (EMR) systems in future implementations.
BACKGROUND: Health care organizations increasingly rely on business associates (BAs) to provide clinical, administrative, and technology services that require access to protected health information. While the Health Information Technology for Economic and Clinical Health (HITECH) Act and the Health Insurance Portability and Accountability Act (HIPAA) Omnibus Rule extended legal liability to BAs, the frequency and characteristics of data breaches involving BAs have not been systematically tracked across the entire post-HITECH reporting era. Understanding these trends is critical for health information managers and cybersecurity professionals who are directly responsible for managing third-party risk. OBJECTIVE: The author examined the longitudinal trends in BA involvement in health care data breaches reported to the US Department of Health and Human Services (HHS) Office for Civil Rights (OCR) from 2009 to 2025, including changes in frequency, breach mechanisms, breach locations, and severity profiles of BA-involved incidents across 3 regulatory periods. METHODS: The author conducted a retrospective longitudinal analysis of health care data breaches (N=6612) reported to the HHS OCR breach portal between October 2009 and December 2025. The author operationalized BA involvement as breaches reported by BA entities or flagged as BA-related. Using logistic regression models, the author estimated annual trends in BA involvement, breach mechanism, and breach location. Chi-square tests assessed associations between BA status and breach characteristics across 3 regulatory periods: pre-Omnibus (2009-2013), post-Omnibus (2014-2019), and 2020-2025. Proportion tests compared BA-involvement rates across periods. RESULTS: BA-involved breaches accounted for 1950 of 6612 (29.5%) incidents and 285,718,494 (48.8%) of all affected individuals. The annual BA-involvement rate increased from 22.1% in the pre-Omnibus period to 36.6% in the 2020-2025 period (z score=8.29, P<.001). Logistic regression confirmed an 8% annual increase in the odds of BA involvement (odds ratio [OR] 1.08, 95% CI 1.07-1.10; P<.001). Hacking/IT incidents shifted from a minority of incidents to the dominant breach mechanism (OR 1.41 per year, 95% CI 1.39-1.44; P<.001), and the odds of network server breaches increased by 29% per year (OR 1.29, 95% CI 1.26-1.31; P<.001). BA-involved breaches were significantly more concentrated in hacking (1282/1950, 65.7% vs 2351/4662, 50.4%) and network server locations (1084/1950, 55.6% vs 1435/4662, 30.8%) compared with non-BA breaches (P<.001). The proportion of mega breaches (≥100,000 individuals) also increased annually (OR 1.16, 95% CI 1.13-1.19; P<.001), with BA-involved breaches exhibiting a significantly higher rate of mega breaches (12.4% vs 8.2%; χ21=28.44; P<.001). CONCLUSIONS: Building on prior evidence linking BA involvement to breach severity, this study demonstrates that BA-involved health care data breaches accelerated substantially across the post-HITECH reporting era, with the steepest increase beginning in 2020. The concurrent growth of hacking and the concentration of breaches on network servers coincided with digital transformation, cloud migration, and the ransomware epidemic, which may have amplified third-party risk exposure. Health information managers and cybersecurity professionals should prioritize BA risk management strategies that account for the evolving threat landscape, including enhanced vendor security assessments and data compartmentalization requirements.
Journal of medical Internet researchBenjamin Michaels, Jessica Pourian
BACKGROUND: Assessing medication adherence is central to quality care, yet linking electronic health record (EHR) medication orders to outpatient pharmacy dispense data remains technically complex. OBJECTIVE: This study aimed to present a generalized, reproducible tutorial for linking EHR medication orders to pharmacy dispense data that can be used to assess medication dispense proportions. METHODS: We developed and validated a structured query approach to link EHR medication orders to external pharmacy dispense data using patient identifiers, medication-level identifiers, pharmacy identifiers, and temporal constraints. The tutorial emphasizes key design decisions, including handling multiple triggering events, deduplication across vendors, and managing formulation changes. A retrospective cohort of pediatric acute otitis media encounters (January 1, 2021, to January 1, 2024) was used as an illustrative example. RESULTS: Overall, 98.3% (302/307) of pharmacies in the cohort returned at least 1 dispense record during the study period and were therefore classified as reporting pharmacies. Among 3404 orders, 2616 (76.9%) had a recorded dispense. CONCLUSIONS: EHR-integrated pharmacy data provide a feasible, timely proxy for assessing medication adherence. This tutorial provides a scalable framework for linking EHR and pharmacy data for medication adherence studies, while highlighting key methodological considerations for SQL coding.
JMIR human factorsLouise Nørgaard Olsen, Philipp Harbig, Anna Bay Laurberg, Jacob Laurberg, Morten Haaning Charles
BACKGROUND: Administrative workload in general practice limits time for direct patient care. AI-assisted documentation has been proposed as a way to reduce the documentation burden, but evidence from routine primary care settings remains limited. OBJECTIVE: This study aimed to evaluate general practitioners' (GPs) acceptance of AI-assisted documentation and its association with documentation time and clinical note quality in routine Danish general practice. METHODS: We conducted a quantitative pragmatic pre-post quality improvement evaluation in Danish general practice. A total of 20 GPs documented 239 consultations before and 236 consultations after implementation of an AI-assisted documentation system. Documentation quality, structure, clinical clarity, and documentation time categories were self-assessed using standardized audit forms completed immediately after each consultation. Technology acceptance and usability were assessed using the technology acceptance model (TAM) and the System Usability Scale (SUS). RESULTS: Self-assessed documentation structure increased from 3.99 to 4.45, while self-reported documentation time categories decreased from 2.85 to 2.29. Technology acceptance and usability were high (TAM domain means 3.76-4.19; SUS mean 77.5). GP-level paired analyses showed moderate improvements in structure and clarity and a reduction in documentation time. Combined blinded external assessments showed higher postimplementation scores for quality, structure, and clinical clarity, although reviewer-specific ratings diverged, and interrater reliability was low. The association between documentation time and perceived quality was negligible. TAM and SUS indicated high clinician acceptance. CONCLUSIONS: AI-assisted documentation was associated with lower self-reported documentation time categories while maintaining or modestly improving perceived clinical note quality. These findings support the feasibility of AI-assisted documentation in primary care, while highlighting the need for controlled studies with objective time measurement and longer follow-up.
Journal of medical systemsHyeonhoon Lee, Seonhye Choi, Duyeon Kim, Kyunglan Hong, Hyeonsik Kim, Chang Wook Jeong, Hyung-Chul Lee
Systematized Nomenclature of Medicine-Clinical Terminology (SNOMED CT) is the principal international standard for semantic interoperability of clinical information, but mapping free-text clinical narratives to SNOMED CT concepts remains labor-intensive. We developed a large language model agent system for mapping bilingual clinical text to SNOMED CT concepts and evaluated its effect on mapping accuracy and efficiency within a human-AI collaborative workflow. We designed a three-module agent system comprising translation, abbreviation expansion, and vector-based retrieval components, integrated with a pre-embedded SNOMED CT vector database. Three health information managers independently mapped bilingual clinical text segments using three approaches: human-only, Agent-only, and Agent-assisted human mapping. Performance was evaluated by using hit rate, precision, recall, and F1 score at k = 1 and 5, and R-precision. Mapping time was compared between human-only and human-AI collaborative approaches. A total of 2,261 de-identified clinical text segments across nine clinical categories were collected at a tertiary academic hospital in South Korea. The human-AI collaborative workflow, which expanded the set of valid SNOMED CT candidates presented at each mapping decision, raised pooled hit rate@1 from 0.837 to 0.868 (difference 0.031, 95% confidence interval [CI] 0.021 to 0.042; p < 0.001), raised R-precision from 0.632 to 0.674 (difference 0.042, 95% CI 0.034 to 0.051; p < 0.001), and reduced total mapping time by 53.9% (from 1.57 to 0.72 min per segment, including agent processing). By expanding the space of valid SNOMED CT candidates available to expert mappers, the human-AI collaborative approach improved SNOMED CT mapping accuracy while reducing time by about half. Its modular architecture, supporting periodic vector database updates without retraining, offers a sustainable and efficient solution for bilingual clinical terminology standardization.
JMIR mental healthBo Wang, Tyne W Miller-Fleming, Dongmei Yu, Donald Hucks, Emily Gantz, Rebecca Johnston, Angela Maxwell-Horn, Nancy Cox, James Sutcliffe, Carol A Mathews, Evon…
BACKGROUND: Obsessive-compulsive disorder (OCD) is a common psychiatric disorder, with two-thirds of affected individuals reporting severe impairment. Despite its substantial burden and moderate heritability, the etiology of OCD remains poorly understood, and treatments are often suboptimal. Although recent genome-wide association studies (GWAS) have identified some risk loci, much of the genetic architecture of OCD remains undiscovered, underscoring the need for scalable approaches to identify large, well-defined patient cohorts. OBJECTIVE: This study aimed to develop and validate a scalable electronic health record (EHR)-based phenotyping algorithm for identifying OCD cases to support large-scale genetic and translational research. METHODS: We leveraged EHR-linked biobank data from 2 large hospital systems, namely Vanderbilt University Medical Center (VUMC) and Mass General Brigham (MGB), to develop a high-throughput phenotyping algorithm integrating diagnostic codes, medication records, and natural language processing (NLP) of clinical notes. Algorithm performance was evaluated through expert chart review, and genetic analyses were performed in individuals of European genetic ancestry using the polygenic scores (PGS) of OCD, major depressive disorder (MDD), and height derived from the most recent GWAS. RESULTS: Expert chart reviews demonstrated our algorithm combining both International Statistical Classification of Diseases (ICD) codes and NLP achieved the highest positive predictive values (PPV) for OCD case identification (0.84 at VUMC; 0.91 at MGB) compared to using either ICD codes or NLP alone, albeit with reduced case yield. At both sites, algorithm-defined OCD cases of European genetic ancestry showed significantly higher OCD PGS than controls. In sensitivity analyses adjusting for MDD status, OCD PGS associations were more robust than MDD PGS associations, while height PGS showed no association, supporting the genetic plausibility and relative specificity of the phenotype. CONCLUSIONS: This study presents a scalable and cost-efficient EHR-based approach for identifying OCD cases across health systems. The algorithm achieves high PPV, and among individuals of European genetic ancestry, algorithm-defined cases show significant OCD PGS enrichment, supporting its utility for large-scale genetic studies and advancing understanding of the disorder's complex etiology.
Severe trauma is a leading cause of mortality worldwide; however, contemporary data on mechanism-specific incidence trends and factors associated with in-hospital mortality remain limited. We conducted a nationwide cohort study using a South Korean severe trauma registry spanning 2016 to 2024. Severe trauma was defined as an Injury Severity Score (ISS) ≥ 16. Annual incidence trends were assessed using negative binomial regression with incidence rate ratios (IRRs) and 95% confidence intervals (CIs). Factors associated with in-hospital mortality were identified using multivariable logistic regression, performed separately for patients with and without prehospital cardiac arrest. A total of 73,232 patients were included, of whom 44.9% had prehospital cardiac arrest. Overall in-hospital mortality was 58.2% (26.0% vs 97.7% for patients without versus with prehospital cardiac arrest, respectively). Transport-related injuries were the most common (53.5%), followed by falls/slips (38.2%). Transport-related severe trauma declined by 3.6% annually (IRR, 0.964; 95% CI, 0.955-0.974; P < .001), whereas falls/slips increased by 3.7% annually (IRR, 1.037; 95% CI, 1.024-1.050; P < .001). Among patients without prehospital cardiac arrest, higher ISS (ISS 25-40: adjusted odds ratio [aOR], 3.45; 95% CI, 3.26-3.64; ISS 41-75: aOR, 6.12; 95% CI, 5.40-6.93) and older age (≥65 years: aOR, 2.87; 95% CI, 2.16-3.84) were the strongest independent predictors of in-hospital mortality. Male sex, unintentional injury, lower-level receiving emergency departments, and falls/slips as the injury mechanism (aOR, 1.19; 95% CI, 1.13-1.26) were also independently associated with increased mortality. Among patients with prehospital cardiac arrest, conventional trauma system factors lost prognostic significance; self-harm was the strongest independent predictor (aOR, 2.14; 95% CI, 1.53-3.08). Female sex and injury in nonmetropolitan areas were also independently associated with increased mortality. The epidemiology of severe trauma in South Korea is evolving, with transport-related injuries declining and falls/slips increasing. Preventive strategies for falls/slips should become a public health priority. Given the near-universal mortality among patients with prehospital cardiac arrest, in whom conventional trauma system factors lose prognostic significance, primary prevention strategies, including suicide prevention and injury awareness programs, may yield greater reductions in severe trauma mortality than trauma system development alone.
Journal of medical Internet researchLaura Swinckels, Katharina Alves Rabelo, Eduardo L Delamare, Bruno G Loos, Pierre Lahoud, Harmen Bijwaard, Ander de Keijzer, Jinman Kim, Josef Bruers
BACKGROUND: Periodontitis is one of the most prevalent yet preventable oral diseases, as indicated by multiple clinical and radiographic factors. As these factors are recorded in electronic health records (EHRs), their reuse offers opportunities for personalized risk assessment and targeted prevention. Predictive AI and traditional machine learning models support fragmented detection tasks but lack the integration of textual and imaging predictors. Emerging multimodal large language models (M-LLMs) show promise in combining these data sources for clinical assessment. Evaluating the capabilities of M-LLMs and comparing them against the current clinical standard are therefore essential to determine their potential as digital assistants. OBJECTIVE: This study aimed to evaluate the ability of M-LLMs to assess periodontitis risk and suggest prevention strategies, based on EHR data and radiographic findings. Each M-LLM was individually evaluated by periodontal experts, benchmarked against other models, and compared with a periodontist as a reference. METHODS: A vignette study was conducted following TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) guidelines for the evaluation of LLMs. Ten periodontal vignettes were created, each including a panoramic radiograph and textual EHR data. Three LLMs capable of reasoning and handling multimodal data were compared to a periodontist who generated outputs manually, based on the same prompts and input data. Periodontal experts rated all outputs across 6 predefined criteria on a 5-point Likert scale. Statistical analyses evaluated overall performance per model and tested whether performance varied per model, scenario complexity, or rater. RESULTS: GPT o1 Pro and Claude Sonnet 4 showed strong performance, with 86.7% and 85.6% of ratings deemed acceptable-comparable to the periodontist's output (87.8%). Gemini 2.5 Pro was rated significantly lower than both the periodontist and the other models (59.4% acceptable; P<.002). Radiographic interpretation consistently received lower scores than other abilities across all models and the periodontist, with Gemini rated below the acceptable threshold. The time required for completion ranged from approximately 10 seconds for Claude to 37 seconds for Gemini; 3 minutes, 22 seconds, for GPT; and 5 minutes, 57 seconds, for the periodontist. CONCLUSIONS: M-LLMs demonstrated strong reasoning abilities in periodontal assessment. Across all models, unacceptable elements were consistently related to errors in radiographic interpretation, though refined prompting or newer model versions may improve this. Notably, even when radiographic findings were incorrect and plaque-retentive factors were absent, outputs were still rated well, indicating that EHR data alone provide a substantial basis. For clinical applicability, M-LLMs must at least perform comparably to a periodontist and meet the quality standards set by periodontal experts-a bar that GPT and Claude appear to approach.
Pharmacoepidemiology and drug safetyChristopher T Rentsch, Vanessa A Palzes, Mingjian Shi, Michael R Setzer, Samantha G Malone, Andrea H Kline-Simon, Emma L Winterlind, Lorenzo Leggio, Vincent Lo…
PURPOSE: Alcohol use disorder (AUD) remains a major public health problem, with few effective treatments. L-type calcium channel blockers (LTCCBs) have genetic and preclinical support as potential treatments for AUD. METHODS: We evaluated whether brain penetrant (BP)-LTCCBs are associated with reduced alcohol consumption in two preregistered observational cohort studies using electronic health records from the US Department of Veterans Affairs (VA) and Kaiser Permanente Northern California (KPNC). New users of BP-LTCCBs (nifedipine or felodipine) were compared with new users of a non-BP-LTCCB (amlodipine) and with unexposed patients from the same clinics, requiring a 180-day washout and ≥ 60 days' supply. Propensity score matching was performed across all exposure contrasts. The primary outcome was change in drinks per week from pre-index to end of follow-up using difference-in-differences (DiD) models, with prespecified subgroup analyses by AUD diagnosis, drinking level, and sex. RESULTS: Across both health systems, BP-LTCCB initiation was not associated with greater reductions in drinks per week than either comparator, with broadly consistent findings across all subgroups; though some subgroups were small and imprecise. In the VA, the DiD comparing BP-LTCCB with non-BP-LTCCB was 0.14 drinks/week (95% CI -0.06, 0.34), and versus unexposed was 0.00 (95% CI -0.18, 0.19). Similar results were observed in KPNC (BP-LTCCB vs. non-BP-LTCCB: 0.31, 95% CI -0.48, 1.11; BP-LTCCB vs. unexposed: 0.16, 95% CI -0.67, 0.99). CONCLUSIONS: In two large, preregistered EHR-based cohorts, we found no evidence that BP-LTCCBs were associated with reduced drinking relative to comparators. Despite compelling genetic and preclinical evidence, these findings do not support repurposing BP-LTCCBs for AUD and instead prioritize alternative pharmacologic targets.
Health expectations : an international journal of public participation in health care and health policyKrystal Warmoth, Nicola Small, Vanessa Davey, Patrick Burch, Alex Thompson, Jo Butterworth, Jonathan Gibb, Easter Joury, Steve Callaghan, Felicity Dewhurst
BACKGROUND: Care coordination is essential for improving healthcare experiences and outcomes for people living with multiple long-term conditions (MLTC), yet it remains challenging to measure care coordination in ways that are both meaningful and feasible at scale. Routine electronic health record (EHR) data offer potential for system-wide measurement, but it is unclear whether such measures reflect what matters to patients, carers, and those delivering care. METHODS: We undertook a multi-stage, collaborative consultation process to examine whether care coordination in MLTC can be meaningfully measured using routine data. Public contributors (n = 18), professional contributors (n = 23), researchers (n = 15) and data experts (n = 10) participated in structured discussions, co-developed appraisal criteria (including relevance to lived experience and feasibility in EHR data), and assessed measures identified through a linked systematic review. A consensus survey evaluated the appropriateness of shortlisted measures. RESULTS: Participants identified priorities including teamwork, communication, continuity, care management, and accountability. Appraisal criteria balanced patient relevance with feasibility in routine data. While a feasible shortlist of measures was identified, findings showed only qualified support for routine data-derived indicators. Continuity measures were most strongly endorsed but widely recognised as proxies capturing observable activity rather than the relational, experiential, and cross-sector dimensions central to coordinated care. Concerns were raised regarding data quality, conceptual validity, and the risk of incomplete or misleading representations. CONCLUSIONS: Routine data-derived measures of care coordination in MLTC hold clear potential but are inherently limited. The findings support a conditional position: such measures can provide useful system-level signals, but only if interpreted cautiously, supported by robust data infrastructure, and complemented by patient-reported insights. Advancing measurement will require integrated approaches that better capture what matters to patients and carers.
Journal of viral hepatitisDavid Etoori, Annabel Powell, Paul Trembling, Jennifer Plunkett, Ashley Brown, Maud Lemoine, Stephen Ryder, Sarah Montague, Eleanor Pilsworth, Anthony Bains, L…
Viral hepatitis mortality is an impact target required to evidence the elimination of viral hepatitis with a combined mortality rate of less than or equal to six deaths per 100,000 population. Within England, routine healthcare data linked with death registrations are used; however, where these data are not available, the World Health Organization (WHO) recommends overlaying the attributable fraction (AF) over the mortality envelope for decompensated cirrhosis (DC) and hepatocellular carcinoma (HCC). We aimed to estimate viral hepatitis mortality using the attributable fraction and compared it to surveillance methods. Six acute NHS trusts participated across England and reviewed the case records of individuals identified retrospectively up to 31 December 2023 through hospital episode statistics to confirm the diagnosis and complete a proforma on possible exposures and likely cause of DC/HCC. Overall, 563 individuals were identified, 390 for DC review with 127 (32.6%) confirmed and 173 for HCC review with 157 (91.8%) confirmed. The AF for DC was 0.008 for HCV and 0.016 for HBV, equivalent to a crude mortality rate of 0.09 per 100,000 population and 0.17 per 100,000 population, respectively. For HCC 0.1806 for HCV and 0.0645 for HBV, equivalent to 0.85 per 100,000 population and 0.30 per 100,000 population respectively. These results are comparable to applying the attributable fraction methodology to linking healthcare data sets and demonstrate that England continues to meet the WHO impact target.
Clinical and experimental dental researchEmma Fetchko, Linda Sangalli, Ariadne Letra
OBJECTIVES: Sexual dimorphism has been shown to influence disease predisposition and/or progression, although population-based studies in dental, oral, and craniofacial (DOC) diseases and conditions are limited. This study aimed to identify sex-based differences in DOC diseases/conditions using two large health data repositories. METHODS: Retrospective cross-sectional study of health record data obtained from adult participants (> 18 years old) in the NIH All of Us Research Program (n = 254,700) and the BigMouth Data Repository (n ≈ 4.7 million). The number of males and females presenting each selected condition in each database was recorded. Sex-specific association analysis for each condition was performed using chi-square tests (α ≤ 0.0002). Female-to-male odds ratio (OR) and confidence intervals were also calculated. RESULTS: Significant sex-related differences were found for 61/87 concepts (70%), with 33 concepts (54%) showing female bias and 28 (46%) showing male bias (p ≤ 0.0002). Analysis of BigMouth data showed sex bias for 90/230 (39%) concepts investigated, of which 87 (97%) showed female bias and 3 (3%) showed male bias (p ≤ 0.0002). CONCLUSIONS: This study provides evidence of sex bias in numerous oral and dental conditions in the populations studied. Additional studies in other populations might provide further insight into the role of sexual dimorphism in these conditions.
Basic & clinical pharmacology & toxicologyAndreas Halgreen Eiset, Signe Hertz Hansen, Sara Buttrup Rosenquist, Kenneth Skov, Lene Heise Garvey, Peter Gaarsdal Uhrbrand, Eva Aggerholm Sædder
Accurate documentation of drug allergies is critical to prevent re-exposure while avoiding unnecessary restrictions on safe pharmacotherapy. Although true opioid hypersensitivity is rare, adverse effects are frequently misclassified as allergies. The aim of the present study was to evaluate opioid allergy registration across six Danish hospitals. All opioid allergy registrations between 2020 and 2023 were analysed using text-mining algorithms. Registrations were dichotomised as 'probable allergy' or 'not allergy' based on predefined keywords and validated by manual journal audit. Among 1 199 314 patients, there were 14 661 allergy registrations towards an opioid. Only 19% (n = 2710) of the allergy registrations were classified as probable allergies, predominantly driven by rash and/or pruritus (71%). The likelihood of an allergy registration being 'probable' was universally low, although slightly higher for intravenously administered opioids compared with orally administered morphine (OR = 0.26, 95% CI: 0.21; 0.32) and tramadol (OR = 0.36, 95% CI: 0.25; 0.46). The audit of 203 allergy registrations corroborated the text-mining results. Notably, 63 patients (33%) received the same opioid despite a registered allergy. Opioid allergy registrations frequently misrepresent expected adverse drug reactions as true hypersensitivity, particularly for oral formulations. Improved diagnostic differentiation and documentation accuracy are imperative to optimise patient pain management protocols in hospitals.
Journal of evaluation in clinical practiceYerin R Lee, Louise Harris, Peter Rossos, Lucas B Chartier, Maxim Ben-Yakov
RATIONALE: Physician burnout is a major concern, particularly during periods of large-scale institutional change such as electronic health record (EHR) implementation. AIMS AND OBJECTIVES: To examine associations of burnout with physician age, clinical experience, and after-hours documentation burden ('pyjama time') among acute and post-acute care physicians undergoing EHR implementation, and secondarily to evaluate changes in burnout across three phases of EHR implementation (Pre-Implementation, Post-Implementation, Post-Optimisation). METHODS: We conducted anonymous electronic surveys of acute and post-acute care attending physicians at a tertiary academic health system in Toronto, Canada across three phases of EHR implementation. Burnout was assessed using the Maslach Burnout Inventory Human Services Survey-Medical Personnel (MBI-HSS-MP). Univariable and multivariable logistic regression analyses identified factors associated with high emotional exhaustion, high depersonalisation, and low personal accomplishment. RESULTS: We collected 306 responses in Pre-Implementation, 248 in Post-Implementation, and 214 in Post-Optimisation. Overall burnout levels remained stable across implementation phases (p ≥ 0.05). Physician age ≥ 65 years was independently protective across all burnout domains (adjusted odds ratio [aOR] 0.33-0.49; p < 0.01). Physicians with > 20 years in practice were less likely to report high depersonalisation (OR 0.39; p < 0.01) or low personal accomplishment (OR 0.54; p = 0.01). Increasing pyjama time was associated with higher odds of emotional exhaustion on univariable analysis (OR 1.06 per hour; p < 0.03) and remained independently associated with depersonalisation after adjustment (aOR 1.07; p = 0.02). Documentation burden peaked in the Post-Implementation phase (mean pyjama time 381.6 min/day), with higher depersonalisation observed on univariable analysis (OR 1.51; p < 0.02), although this association attenuated after adjustment. CONCLUSION: In this three-phase study, after-hours documentation burden was strongly associated with depersonalisation, while physician age ≥ 65 years consistently protected against burnout. Although overall burnout levels remained stable across implementation phases, the immediate Post-Implementation period was characterised by substantial documentation burden and transient increases in depersonalisation, highlighting the importance of robust training, workflow optimisation, and targeted support during early EHR adoption.
PloS oneYounggoun Jo, Yunchul Park, Euisung Jeong, Hyunseok Jang, Hyo-Sin Kim
Trauma patients requiring massive transfusion have mortality exceeding 40%, yet early risk-stratification tools for this population remain limited. We evaluated readily available predictors of mortality in massively transfused trauma patients, excluding those with severe traumatic brain injury (TBI) to focus on a hemorrhage-predominant cohort. This single-center retrospective cohort study included trauma patients who received massive transfusion at Chonnam National University Hospital between January 2018 and December 2023. Patients with severe TBI (Head Abbreviated Injury Scale [AIS] ≥ 4) were excluded. Three sequential multivariate logistic regression models were constructed: a primary model using variables available at emergency department (ED) arrival, a secondary model adding laboratory values, and an exploratory model incorporating early transfusion course variables. Of 172 massively transfused trauma patients, 124 met inclusion criteria (Head AIS < 4). Overall mortality was 41.1% (51/124), with 24-hour mortality of 23.7% (28/118). The Glasgow Coma Scale (GCS) was the strongest independent predictor across all models (OR = 0.77-0.79 per point, p < 0.001). In the primary ED arrival model (AUC = 0.770), GCS was the sole significant predictor; GCS alone achieved a comparable AUC of 0.771. The exploratory model incorporating transfusion variables achieved the highest discrimination (AUC = 0.828). All models showed adequate calibration (Hosmer-Lemeshow p > 0.05). Mortality decreased with longer time-to-first-transfusion (54.5% for ≤15 min vs. 23.7% for >60 min, trend p = 0.007), reflecting confounding by indication. In massively transfused trauma patients without severe TBI, GCS assessed at ED arrival is the single most informative predictor of mortality, enabling risk stratification before laboratory results become available. Metabolic acidosis markers and coagulopathy were strongly associated with mortality on univariate analysis but added no independent predictive value beyond GCS. Time-to-first-transfusion appeared to reflect hemorrhage acuity rather than a modifiable prognostic factor.
PharmacotherapyJennifer Beavers, Mary Katherine Cella Shultz, Leanne Atchison, Cara Lwin, Dandan Liu, Silky Chotai, Andrew J Medvecz, Robel Beyene, Michael C Smith, Bradley M…
BACKGROUND: Use of nonsteroidal anti-inflammatory drugs (NSAIDs) after a traumatic brain injury (TBI) is controversial due to potential worsening of an intracranial hemorrhage (ICH). This study evaluated if NSAID use within 14 days of TBI was associated with a clinically significant progression of intracranial bleed (PIB). METHODS: A retrospective, propensity score-weighted study was conducted in adult patients with a TBI admitted to an American College of Surgeons Certified Level I Trauma Center. The NSAID group included patients who received at least one dose of an NSAID within 14 days of injury while admitted. The control group included patients who did not receive NSAIDs. The primary end point was incidence of a clinically significant PIB (decrease in the Glasgow Coma Score (GCS) > 2 and an increased or new ICH). Secondary outcomes included need for additional neurosurgical intervention and PIB on imaging regardless of GCS change. RESULTS: 978 patients were included: 284 in the NSAID group and 694 in the control group. Most patients had an Abbreviated Injury Score head > 2 (88%) and a subdural hematoma (57%). Fifty-two percent of patients had a mixed bleed. A clinically significant PIB occurred in 0.35% in the NSAID group and 1.2% in the control group (Hazards Ratio 0.1, 95% Confidence Interval 0.01-0.82, p = 0.032). There was no difference in additional neurosurgical operations or PIB regardless of GCS change. In the NSAID group, there was no significant difference in time from admission to NSAID administration between patients who experienced a PIB and those who did not have a PIB (1.9 days vs. 4.7 days, respectively, p = 0.06). CONCLUSION: In this single-center, retrospective, propensity score-weighted study, NSAIDs were not associated with an increased risk of clinically significant PIB in patients with acute TBI when administered within 14 days of injury. NSAIDs appear safe in the acute phase post-TBI. Prospective studies are needed to validate our results.
Scandinavian journal of medicine & science in sportsBradley Sprouse, Avinash Chandran, Philip Hennis, John Morris, Simon Cooper, Charlotte Cowie, Subhashis Basu, Ian Varley
Due to the lack of large-scale research investigating the epidemiology of injury in English men's professional football, the present study aimed to examine the incidence rate, severity, and burden of injury in English men's domestic football. Injury surveillance data was prospectively collected and retrospectively analyzed. Time-loss injuries, and match and training exposure, were collected by club medical staff across 12 seasons (2013-2014 to 2024-2025) from English men's domestic clubs playing in the English Premier League and English Football League (EFL-Championship, League One, League Two) (416 team seasons). A bootstrapped negative binomial approach was adopted to calculate incidence rate and burden estimates, temporal analysis and incidence rate (IRR) and burden ratios (IBR), all with accompanying 95% CI. 15 986 time-loss injuries were recorded, resulting in 498 108 days absent over 2 286 030 exposure hours. Injury incidence rate was substantially higher in matches than training (24.9 vs. 3.6 injuries/1000 h; IRR: 7.0 [6.6-7.4]), as was injury burden (835.5 vs. 104.1 days absent/1000 h; IBR: 8.0 [7.5-8.7]). Match injury incidence rate remained stable across the study period, training injury incidence rate decreased by 4% per season (p < 0.001), but match injury burden increased by 5% (p < 0.001). Muscle injury incidence rate increased by 3% per season in matches (p = 0.004), but decreased by 3% in training (p = 0.028). Muscle injury burden increased in both matches (10%, p < 0.001) and training (4%, p = 0.006). Ligament injury incidence rate decreased by 5% per season in training (p < 0.001). Hamstring injuries were the most common diagnosis (1.33 injuries/1000 h) and contributed the greatest overall burden (40.37 days absent/1000 h). Over 12 seasons in English men's professional football, match-play consistently accounts for a greater injury incidence rate and burden than training. While match injury incidence rate has remained stable, burden has increased, specifically for muscle-related injuries.
Pharmacoepidemiology and drug safetyVinicius Pinho, Lucas Tramujas, Willian Soares, Matvey B Palchuk, K Arnold Chan
PURPOSE: In 2014, TriNetX launched a U.S.-based federated network of health system electronic health records (EHR) that is now used in academic and industry-sponsored studies and includes more than 250 health care organizations (HCOs) globally. The network expanded to Latin America in 2018, where currently there are over 40 HCO partners in Brazil and Colombia. This manuscript describes the TriNetX LATAM Collaborative Network and outlines how it may support observational studies and clinical trials. METHODS: We queried the LATAM Collaborative Network to determine patient counts for selected disease conditions. We used diagnosis codes, plus drug prescription and/or biometric measures to estimate the frequency of diabetes, hypertension, heart failure, renal impairment, and chronic obstructive pulmonary disease, and estimated survival among patients with breast cancer and prostate cancer. RESULTS: Approximately 4.54 million patients had at least one healthcare encounter in the data system in 2025; 58.8% of them were female, and 30% were age 60 or older. About 1 million adult patients had a diagnosis of hypertension, prescription of an angiotensin-converting enzyme inhibitor or angiotensin II antagonist, or systolic blood pressure above 140 mmHg. Among adult patients who had at least one serum estimated glomerular filtration test, 6.7% (43 720 of 655 560) had 2 instances of value < 60 mL/min/{1.73_m2} at least 3 months apart. Among 166 490 women with breast cancer, 5-year survival was 90.7%. Among 116 900 men with prostate cancer, 5-year survival was 91.1%. CONCLUSION: TriNetX LATAM Collaborative Network represents an emerging, scalable EHR-based data source to support pharmacoepidemiology studies and clinical trials.
Journal of human nutrition and dietetics : the official journal of the British Dietetic AssociationJacqui Bailey, Jolie Baird, Olivia Dullard, Sophie Kane, Caroline J Tuck, Katherine J Desneves
INTRODUCTION: Malnutrition contributes to significant health and economic burden. A 2022 Malnutrition Point Prevalence Survey (PPS) at Austin Health, Melbourne, reported a 40% prevalence rate, consistent with international studies. However, despite robust screening and referral processes, 28% of at-risk or malnourished patients were not receiving dietetic care on the PPS day. This study aimed to evaluate the rate and timing of electronic malnutrition screening tool (eMST) completion in adult inpatients against local hospital guidelines, compare the accuracy of nursing completed eMST with dietitian completed MST and identify common sources of error in screening completion and potential contributors for those patients who were identified as at risk of or with malnutrition and not receiving dietetic care. METHODS: A retrospective case series of dietitian referrals, weights, time of eMST completion and MST results between admission and the PPS day were assessed for timeliness against hospital guidelines and accuracy based on available weight history, dietitian completed MST and nutrition impact symptoms documented in the electronic medical record (EMR). RESULTS: On the day of the PPS 48 patients at-risk of (n = 19) or malnourished (SGA B n = 27, SGA C n = 2) were not receiving dietetic care. Of these, 60% (n = 29) missed dietetic care due to inaccurate MST completion. No weight loss (n = 20) was incorrectly reported despite patient reports or EMR data indicating otherwise. 35% of patients (n = 17) were weighed within 24 h of admission as per hospital guideline. 65% of MSTs completed during the morning nursing shift were inaccurate, compared with 48% during the afternoon and 36% during the night shift. CONCLUSION: This study examining the alignment between real-world screening practices and institutional guidelines for patients at risk of malnutrition identified clear gaps in the timely and accurate completion of MST screening and weight documentation. A multitude of factors likely influence this including nursing staff capacity, flawed design of EMR systems, and suboptimal protocol sequencing. Future research applying systems-level thinking is required.
PloS oneRupesh Agrawal, Dursun Delen, Bruce Benjamin, Rani Misra
OBJECTIVE: To characterize age- and sex-stratified prevalence patterns of obstructive sleep apnea and co-morbid conditions corresponding to [Formula: see text] components using a large, population-based electronic health record dataset. STUDY DESIGN: The study cohort included 556,474 unique de-identified OSA patients (ICD-9 code 327.23), mined from HIPAA-compliant Cerner [Formula: see text] database (1999-2016). The patient records were mined using Structured Query Language (SQL), stratified by age, gender, and stroke risk comorbidities per [Formula: see text], to conduct a population-based cross-sectional epidemiological study. METHODS: Clinical and epidemiological research informatics methodologies are applied to characterize OSA prevalence patterns across age (5-year intervals), gender, and stroke risk comorbidities per [Formula: see text] criteria. Body-level organ system involvement was examined according to the Major Diagnostic Category (MDC) classification. RESULTS: Age-stratified cross-sectional analyses showed that prevalence rates for OSA (m: 10.43%, f: 9.28%), hypertension (m: 7.37%, f: 7.11%), along with clinical encounters, peaked at age 55 for both genders, whereas the highest prevalence of cardiovascular comorbidities was observed in older age groups (ages 65-70). Hypertension was most prevalent (57.83%), followed by diabetes (34.7%), congestive heart failure (17.7%), atrial fibrillation (13.31%), vascular disease (12.3%), and prior stroke/TIA (4.78%). MDC analysis revealed that prevalence is pronounced in ENT, cardiovascular, and musculoskeletal body systems, consistent with the pathophysiology of chronic intermittent hypoxia in OSA patients. These patterns reflect population-level age-specific prevalence differences rather than longitudinal disease progression. CONCLUSION: This large cross-sectional study (556,474 patients) provides detailed age- and sex-specific prevalence estimates of OSA-associated stroke-risk comorbidities. While causal or temporal inferences cannot be drawn, the findings highlight population-level prevalence patterns that may inform hypothesis-driven longitudinal studies and future evaluation of stroke-risk models.
F1000ResearchKori Puspita Ningsih, Hartono Hartono, Nur Hafidha Hikmayani
INTRODUCTION: In casemix-based payment systems, clinical coding quality plays a critical role in determining data validity, reimbursement accuracy, and overall performance. Financial inefficiencies, skewed morbidity data, and claim denials might result from incorrect coding. However, the majority of studies pay little attention to thorough and context-specific evaluation, instead concentrating mostly on coding accuracy. OBJECTIVE: The purpose of this study was to use the Indonesian Case-Based Groups (INA-CBG) as the empirical backdrop for the development and psychometric validation of a multidimensional instrument for evaluating clinical coding quality in casemix systems. METHODS: An instrument development strategy based on ICD-10, ICD-9-CM, WHO coding guidelines, national casemix requirements, and results from coding audits was used to conduct a methodological investigation. The Content Validity Ratio (CVR) and Content Validity Index (I-CVI, S-CVI) were used to evaluate the content validity. Cohen's Kappa was used to assess inter-rater reliability, while Cronbach's alpha with 95% confidence intervals was used to quantify internal consistency. JASP was used for data analysis. RESULTS: The Clinical Coding Quality Assessment Instrument for Casemix (CCQAI-CM) comprises seven conceptual domains operationalized through 11 assessment items: diagnosis documentation, procedure documentation, documentation support, coding compliance, coding accuracy, diagnosis reselection, and procedure reselection. The instrument demonstrated high content validity (S-CVI = 0.972), varying item-level inter-rater agreement, and very high overall internal consistency (Cronbach's α = 0.982). CONCLUSION: The findings provide initial evidence of the content validity and reliability of the CCQAI-CM as a multidimensional framework for assessing clinical coding quality in casemix systems. Further validation across healthcare settings and against relevant casemix outcomes is warranted.
Breast cancer research and treatmentLaura C Pinheiro, Caroline Zeng, Darima Dorzhieva, Lovely Joseph, Zori Hamilton, Kathryn Edick, Anne Marie Mercurio, Desiree Walker, Victoria Blinder, Shoshana…
PURPOSE: Adults living with metastatic breast cancer (mBC) often have unmet employed-related needs. To address this gap, we developed, implemented, and evaluated the Routine Assessment and Integration of Screening for Employment (RAISE) program for adults with mBC. METHODS: We designed a quality improvement project informed by the Specific, Measurable, Attainable, Relevant, and Timebound framework. In Phase 1 (development), we built and integrated an English/Spanish screener into the electronic health record (EHR) platform. In Phase 2 (implementation), 48 h before an oncology visit, patients received alerts through the patient portal that a screening assessment was available. Patients who reported needing support for employment-related concerns triggered an EHR alert to a patient navigator who reached out to offer educational resources (physical booklet and/or mobile app). Phase 3 (evaluation) was guided by the RE-AIM framework and included surveys and interviews with patients, providers, and staff to evaluate acceptability and feasibility of RAISE. RESULTS: Between March 2024 and August 2025, 179 patients with mBC completed the RAISE screener Of these, 22 (22/179, 12.2%) requested assistance/navigation and were provided with educational resources. Among 41 (41/179, 23%) patients surveyed (evaluation phase), 33 patients had completed the screener, and 8 had not. Of those who completed the screener and the evaluation survey, 91% (29/33) reported they were somewhat or very satisfied with RAISE and 78% (25/33) reported the EHR screening questions were clear to them. Key themes that emerged from 5 patient and 10 provider interviews included utility of the resources offered, screening preferences, and barriers to incorporation of screening into the clinical workflow. CONCLUSIONS: Screening for unmet employment needs and delivering resources and navigation is preliminarily feasible and acceptable to patients with mBC, providers, and staff at two NYC hospitals, though key barriers to implementation exist, including low pre-visit screening uptake and integration into the clinical workflow. CLINICAL TRIAL NUMBER: Not applicable.
JMIR medical informaticsJiayu Lu, Kevin W McConeghy, Andrew R Zullo, Jinying Chen
BACKGROUND: Misspellings in medication names can compromise patient safety, reduce data utility, and impede large-scale data initiatives that integrate medication information from electronic health records (EHRs). Existing methods for detecting misspelled medical terms are mostly dictionary-based and can lead to high false-positive rates when correctly spelled but previously unseen (out-of-vocabulary) terms are encountered. OBJECTIVE: We aimed to develop and validate domain-specific, transformer-based language models for detecting misspelled drug names, with an emphasis on performance for unseen terms. METHODS: Using RxNorm as a standardized drug vocabulary, we created an RxNorm-augmented training corpus and developed two BERT (Bidirectional Encoder Representations from Transformers)-based models-BERTDrug and CharBERTDrug-for misspelling detection. Specifically, we randomly split 69,824 RxNorm drug names into training, development, and test sets (3:1:1) and generated k misspellings per name using text-perturbation techniques (k optimized for training; fixed at 1 for development and test sets). The models were fine-tuned on the training and development sets and evaluated using the RxNorm test set and 3586 drug names from the Long-Term Care Data Cooperative (LTCDC) database (external validation). The RxNorm test set and out-of-vocabulary LTCDC dataset (1922 terms), neither overlapping with the RxNorm training data, were used to evaluate performance on unseen terms. SpellChecker served as a dictionary-based baseline, while fastTextML and BioWordVecML, which used different subword embeddings as inputs for machine learning, served as additional baselines. Additionally, we compared model performance with GPT-4o, a generative large language model (LLM), using 2200 randomly sampled test terms. RESULTS: On the RxNorm test set, BERTDrug and CharBERTDrug outperformed the baseline models across most metrics. BERTDrug achieved the best overall performance (F1-score=0.859; area under the receiver operating characteristic curve [ROC-AUC]=0.947), followed by CharBERTDrug (F1-score=0.833; ROC-AUC=0.906). Both models also outperformed the baseline models on the out-of-vocabulary LTCDC dataset across most metrics, with CharBERTDrug performing best (F1-score=0.696; ROC-AUC=0.788), followed by BERTDrug (F1-score=0.669; ROC-AUC=0.786). In the secondary analysis, both models exceeded GPT-4o on most metrics (except Recall) for 2000 RxNorm terms. BERTDrug performed best (ROC-AUC=0.951; F1-score=0.855), followed by CharBERTDrug (ROC-AUC=0.911; F1-score=0.831) and GPT-4o (ROC-AUC=0.856; F1-score=0.721). In contrast, among 200 randomly selected LTCDC terms, GPT-4o performed best on most metrics except precision and specificity. CONCLUSIONS: Domain-specific language models improved detection of misspellings in out-of-vocabulary drug names and outperformed baseline models in both internal and external evaluations. The comparison with a generative LLM suggests that domain shift may substantially reduce the advantages conferred by domain-specific training. With further fine-tuning on diverse data that capture the terminology, formatting conventions, and spelling patterns encountered across real-world clinical settings, these models could be adapted for use in other clinical databases and EHR systems to improve medication data quality for research and to support future safety-focused applications.
European journal of trauma and emergency surgery : official publication of the European Trauma SocietyRoope Mustasilta, Elias Kultanen, Tero Puolakkainen, Aleksi Haapanen, Hanna Thorén, Johanna Snäll
PURPOSE: Patients with facial fractures are a diverse group with varying injury mechanisms and extent of injury. These patients often suffer from concomitant injuries (CIs) of different severity outside the facial region, some of which are life-threatening and may require urgent care. Therefore, early identification of patients at risk for CI is important. This study aimed to investigate exploratory risk groups for concomitant injuries in patients with facial fractures, focusing on the impact of injury energy, sex and age on the prevalence and type of concomitant injury. METHODS: In this retrospective cohort study, a data-driven analysis was used to perform statistical analysis and identify risk groups for different CIs. The study included 4170 patients with facial fractures evaluated at tertiary trauma centres from January 2013 to October 2020. RESULTS: 70.9% of patients with facial fractures were male. Patients' mean age was 45.7 years. Of the injury mechanisms, 36.0% were classified as high-energy trauma. In all, 31.7% of facial fracture patients had some type of CI. Female patients had more CIs than male patients (34.0% vs. 30.8%). The occurrence of CIs increased progressively with age, peaking in the age group 90-94 years, where over 55% of patients experienced CIs. However, considerable variation was observed in CI types across age groups. CONCLUSION: Due to the high CI risk in facial fracture patients, in conjunction with assessment and imaging of the facial trauma, we encourage routinely ruling out CIs, with a special focus on high-risk CI groups.
JMIR formative researchEmmanuel Chevallier, Catherine Letord, Jean Charlet, Marie-Odile Krebs, Emmanuelle Advenier-Iakovlev, Stefan J Darmoni, Julien Grosjean, Romain Leguillon
BACKGROUND: Off-label drug prescribing is prevalent across medicine, including psychiatry, often due to unmet therapeutic needs and inadequate responses to standard treatments. The PSYHAMM (Psychotropes Hors Autorisation de Mise sur le Marché) project, funded by the French Research Agency, investigates these practices. To support this research, a clinical data warehouse (CDW) with advanced data analysis tools was developed and deployed. This system integrates both structured and unstructured data from electronic health records, facilitating comprehensive data analysis. The goal is to improve understanding, regulation, and safety of off-label drug use in psychiatry by providing insights into prescribing patterns and their impacts, ultimately contributing to better clinical guidelines and patient care. OBJECTIVE: This study aimed to evaluate the precision (positive predictive value) of a CDW in identifying candidate off-label prescriptions in psychiatry among the cases automatically flagged by the system, rather than its overall accuracy, sensitivity, or specificity. METHODS: The PSYHAMM data analysis involved a retrospective study of pathology-medication pairs to evaluate the precision of a computerized system among system-flagged cases. This system was compared with manual checks performed by a psychiatrist. The evaluation process included verifying if the condition identified by PSYHAMM was documented in the medical record, assessing diagnostic agreement with tolerance for schizoaffective disorders, and ensuring the identified treatment was current or prescribed in the past. Precision was measured as the number of relevant documents retrieved divided by the total number of documents proposed and was computed for the precise diagnosis, the broad diagnosis, and the identified treatment among the flagged cases. RESULTS: The study analyzed 197 records, identifying 14 unique drug-pathology combinations. Bipolar disorder treated with sodium valproate represented the most cases (108/197, 54.8%), followed by schizophrenia treated with sodium valproate (37/197, 18.8%). The overall precision for detecting off-label situations was 51.3% (101/197). The precise diagnosis achieved a precision of 75.6% (149/197), while the broad diagnosis showed a higher precision of 84.8% (167/197). The identified treatment had a precision of 61.4% (121/197). The primary challenge was temporal discrepancies, such as distinguishing between acute and chronic conditions, which accounted for most of the 48.7% (96/197) of cases that were incorrectly classified. CONCLUSIONS: As a single-center, proof-of-concept evaluation, the PSYHAMM project demonstrates the potential of automated systems to support the identification of off-label prescriptions in psychiatry as a sensitive prescreening step requiring expert validation. The relatively high false-positive rate was driven mainly by temporal discrepancies (drugs prescribed before the index stay, discontinued during the stay, or only hypothetically mentioned) rather than by semantic errors. Future research should focus on integrating real-time data analytics and expanding to multiple institutions to improve the utility of off-label detection systems.
Journal of medical Internet researchPhilipp Remus, Daan Westra, Rachel Gifford, Frank van de Baan, Mark Govers
BACKGROUND: Hospitals continue to invest heavily to increase their level of digitalization. While advanced digital maturity is assumed to improve hospital performance, empirical evidence remains mixed. This tension is mirrored by the productivity paradox of IT, whereby investments in digital technologies do not consistently translate into observable performance gains. OBJECTIVE: This study aims to examine the relationship between the Healthcare Information and Management Systems Society (HIMSS) Electronic Medical Record Adoption Model (EMRAM) score and hospitals' financial, operational, and workforce-related indicators. METHODS: This longitudinal observational study used routinely collected hospital-level annual report data from the Dutch National Annual Healthcare Reports Database (CIBG) for Dutch hospitals from 2017 to 2023. Hospital-year records were linked at the institutional level to publicly available HIMSS EMRAM stage 6 or 7 certification information and operationalized dichotomously (stage 6 or 7 vs ≤5). The sample included 66 to 74 hospitals per year (mean 70.4, SD 2.8), corresponding to up to 498 hospital-year observations. Outcome measures were financial (profit margins, return on assets, asset-turnover ratio, and personnel-expense ratio), operational (length of stay and number of patients treated), and workforce (absenteeism) performance indicators. Linear mixed-effects models were estimated while controlling for hospital size, teaching status, staff-to-patient ratio, time trends, and COVID-19 effects. RESULTS: Advanced digital maturity was not significantly associated with improved financial, operational, or workforce performance after adjustment for multiple testing. For financial outcomes, high digital maturity showed no significant association with profit margin, return on assets, personnel-expense ratio, or asset-turnover ratio. Digitally mature hospitals initially appeared to treat more patients annually (β=31,664.318; 95% CI 8392.095-54,936.540; P=.008), but this association was not statistically significant after Holm-Bonferroni correction (adjusted P=.44). No significant associations were observed for length of stay or absenteeism. CONCLUSIONS: In a highly digitalized health system with near-universal electronic health record adoption, advanced technical digital maturity alone was not associated with measurable improvements in aggregated hospital-level financial, operational, or workforce performance. These findings provide longitudinal empirical support for the IT productivity paradox in hospital digitalization, suggesting that technical maturity is a necessary but insufficient condition for performance gains.
BACKGROUND: AI-enabled voice electronic medical records (EMRs) are increasingly promoted as tools to reduce clinician documentation burden; however, empirical evidence from multilingual, resource-constrained health systems in sub-Saharan Africa remains limited. OBJECTIVE: This study examined health care professionals' perceptions, anticipated benefits, and concerns regarding AI voice-enabled EMRs in Ethiopia to inform context-sensitive implementation strategies. METHODS: We conducted a qualitative, multisite study using semistructured written responses from 43 Ethiopian health care professionals recruited via purposive and maximum variation sampling between April 11, 2026, and April 21, 2026. Data were analyzed in Taguette using a hybrid deductive-inductive approach integrating 3 complementary frameworks: the Consolidated Framework for Implementation Research (CFIR), the technology acceptance model (TAM), and Normalization Process Theory (NPT). Analytical rigor was strengthened through independent dual coding, structured reconciliation, reflexive memos, and a version-controlled audit trail. RESULTS: Four overarching themes were identified. First, participants anticipated clear clinical benefits, including reduced typing burden, improved documentation continuity, and enhanced patient interaction, yet expressed substantial concerns about automation errors, accent-related transcription failures, and persistent infrastructural instability. Second, usability barriers, including interface complexity, inadequate training, and digital anxiety, shaped technology acceptance across cadres. Third, participants anticipated shifts in workflow, task distribution, and clinical collaboration as documentation practices evolved. Finally, ethical and governance concerns, particularly regarding data confidentiality, unclear consent procedures, and fear of surveillance, emerged as major determinants of trust. Cross-framework synthesis revealed that adoption readiness was jointly shaped by organizational capacity, usability perceptions, emotional and cognitive responses, and evolving workflow expectations. CONCLUSIONS: Successful implementation of AI voice-enabled EMRs in Ethiopia requires coordinated investments in digital infrastructure, locally adapted language models, strengthened data governance, and iterative user onboarding. These findings underscore the urgency of context-sensitive and ethically grounded approaches when deploying speech-based AI in low-resource health systems.
Health information management : journal of the Health Information Management Association of AustraliaYicong Xu, Jingya Zhou, Zhenghua Liu, Naishi Li, Dong Cai, Shengdong Pan
BACKGROUND: The accuracy of clinical coding, which is essential for supporting managerial and policy decisions such as funding formulas, is closely tied to the work engagement of clinical coders. OBJECTIVE: This paper aimed to evaluate the level of work engagement and associated factors among clinical coders through a nationwide survey in China. METHOD: Thirty-one tertiary public hospitals in China were selected via convenience sampling, and 354 clinical coders were surveyed with Question-star software. Work engagement was measured using the 17-item Chinese version of the Utrecht Work Engagement Scale (UWES), with each item rated from 0 to 6. Twenty-two potentially related variables were collected. Descriptive statistics and one-sample t-tests (vs UWES norms) were used to assess work engagement levels. Cluster analysis was performed to identify engagement patterns. Independent-samples t-tests or one-way analysis of variance were used to examine engagement differences across the 22 variables. Variables with p < 0.05 in univariate analysis were entered into multiple linear regression to identify associated factors. RESULTS: The mean work engagement score among clinical coders was significantly lower than the UWES-17 norm (3.26 vs 3.82; t = -8.378, p < 0.001). The participants were classified into three engagement groups: high (29.5%), moderate (50.7%), and low (19.8%). Multiple linear regression analysis revealed seven variables that were significantly associated with work engagement (F (7, 331) = 11.493, p < 0.001, R2 = 0.196): region, continuing education, ideal clinical coding software, work position, perceived leadership support, physician-coder communication and sex. CONCLUSION: Clinical coders in China demonstrated moderate-to-low work engagement, shaped primarily by objective career development conditions, along with their perceived software support and organizational support.Implications for health information management practice:Enhancing work engagement requires managerial focus on continuing education, software usability, and perceived leadership support.
JMIR medical informaticsAlex H Lee, Devesh Narayanan, Kristan Staudenmayer, Aussama K Nassar, Syed Morad Hameed
BACKGROUND: Systematic strategies to harness electronic health record (EHR) workflows, reduce redundancy, and support decision-making remain limited in acute care surgery (ACS). Understanding how EHR systems and workflows intersect with time-sensitive settings is critical to improving decision-making and outcomes in ACS. OBJECTIVE: This study aimed to evaluate how ACS clinicians leverage the EHR for decision-making and to identify opportunities and challenges for EHR-enabled decision support. METHODS: We conducted a qualitative ethnographic study over a 6-month period, combining in-depth interviews with 15 ACS surgeons and providers and 100 hours of "paired fieldwork" observations spanning the entire perioperative arc by a surgical provider and an organizational sociologist. Using constructivist grounded theory, we identified the enablers, challenges, and opportunities for EHR-enabled decision-making in ACS. RESULTS: Surgeons fell into two groups: (1) those who accepted information overload as inherent to the EHR, relying on generic templates and standard attestations, and (2) others who viewed it as a problem to fix, actively correcting errors and composing individualized summaries. Ambiguity in billing requirements drove overdocumentation, resulting in "note bloat" that obscured high-yield information. EHR use during decision-making focused primarily on risk assessment, though navigation challenges hindered access to critical data. Forecasting key outcomes that alter management or facilitate shared decision-making was seen as valuable. Automated risk stratification, generated from live EHR data while minimizing alert fatigue, was seen as a potential solution. CONCLUSIONS: ACS clinicians use various tactics to navigate EHR challenges and focus on high-value tasks. Streamlined risk assessment using EHR data may strengthen decision-making in critical moments, but solutions must integrate seamlessly within existing workflows to provide rapid and accurate outputs that prioritize meaningful outcomes.
BMJ openAlice Broadbent, Hannah Woods, Matthew Broadbent, Robert Stewart, Mariana Pinto da Costa
BACKGROUND: Reporting diversity characteristics is required for more inclusive, equitable and policy-relevant research. OBJECTIVES: To evaluate the reporting of diversity characteristics in publications using a large mental healthcare electronic health record (EHR)-derived research data resource, and to compare reporting of these characteristics in publications with their availabilities in the underlying dataset. DESIGN: Cross-sectional review of Clinical Record Interactive Search (CRIS)-derived publications and assessment of diversity characteristic availability within the underlying EHR-derived database. SETTING: The South London and Maudsley (SLaM) National Health Service (NHS) Foundation Trust Biomedical Research Centre Case Register, accessed via the CRIS platform, representing secondary mental healthcare delivered to a geographic catchment area covering four boroughs in south London. METHODS AND ANALYSIS: All CRIS-derived publications were reviewed to ascertain reporting of protected characteristics, as defined in the UK Equality Act 2010, alongside additional diversity-related characteristics. The availability of each characteristic within CRIS was assessed from records available up to 15 April 2026. Descriptive statistics were used to summarise reporting and data availability. RESULTS: A total of 362 publications were evaluated. The mean number of diversity characteristics reported per publication was 4.0, and no publication reported more than 10 characteristics. Age (89.3%), sex (87.3%) and ethnicity (80.4%) were the most frequently reported characteristics. Socioeconomic status was reported in 42.5% of publications, while marriage and civil partnership (35.0%) and disability (27.0%) were reported in a smaller proportion of studies. All remaining characteristics were reported in less than 10% of publications.Data availability within CRIS was highest for sex (99.9%) and age (99.8%), followed by socioeconomic status, geographic location and homelessness (all 97.5%) and ethnicity (86.5%). However, several characteristics were reported far less frequently than they were available in the dataset, particularly geographic location (6.6% reported despite 97.5% availability) and homelessness (5.8% reported despite 97.5% availability). CONCLUSIONS: Reporting of diversity characteristics in this case study for EHR-based mental health research was uneven and did not fully reflect availability in the source data. While age, sex and ethnicity are commonly reported, several other protected and diversity-related characteristics are rarely used by researchers despite their availability, and other characteristics remain challenging to capture. As routine EHR-derived datasets increasingly inform mental health research and policy, greater attention to recording, accessibility and reporting of diversity characteristics is required to support more inclusive and representative research.
Revista da Associacao Medica Brasileira (1992)Sefa Gümrük Aslan, Şehri Bariş Saklica, Sinem Uyar Köylü, Kurtuluş Köklü
OBJECTIVE: The objective of this study was to describe the demographic, clinical, cognitive, and functional characteristics of acquired brain injury patients, analyze complications, and evaluate inpatient rehabilitation outcomes. METHODS: Retrospective review of 50 acquired brain injury patients undergoing multidisciplinary inpatient rehabilitation (2019-2024). Glasgow Coma Scale, post-traumatic amnesia, cognitive scales (Mini-Mental State Examination, Montreal Cognitive Assessment, Rancho Los Amigos), Functional Ambulation Scale, and complications were assessed. RESULTS: Traumatic etiology comprised 76% of cases; 72% were severe (Glasgow Coma Scale 3-8). Prolonged post-traumatic amnesia (>24 h) was significantly associated with severe injury (p<0.001). Spasticity (64%), contractures (42%), speech disorders (42%), and dysphagia (34%) were most prevalent. Functional Ambulation Scale score 0 decreased from 58 to 42% post-rehabilitation; Functional Ambulation Scale 3-5 increased from 24 to 46% (Wilcoxon signed-rank test, p<0.001). CONCLUSION: Structured multidisciplinary rehabilitation was associated with measurable functional gains in acquired brain injury patients despite severe deficits and frequent complications; however, the absence of a control group precludes causal conclusions.
BMJ openJessica K Bone, Feifei Bu, Daniel Hayes, Daisy Fancourt
OBJECTIVES: We aimed to describe the characteristics of children and young people referred to social prescribing across the UK and understand what social prescribing looks like for these young people. Additionally, we aimed to explore whether access to and experiences of social prescribing vary with age and have changed from 2017 to 2025. Overall, we aimed to identify whether social prescribing reduces or exacerbates health inequalities among children and young people, and whether this has changed over time. DESIGN: Analysis of social prescribing electronic health records. SETTING: Social prescribing hubs and services across the UK that use Access Elemental (a cloud-based social prescribing platform). PARTICIPANTS: 52 585 individuals referred to social prescribing in 2017-2025 aged 4-25 years (mean=20.04 years, SD=4.71), of whom 57% were female, 39% male, <2% were in other gender groups and 3% did not disclose their gender. PRIMARY AND SECONDARY OUTCOME MEASURES: We summarised young people's characteristics (age, gender, country, urban area, area deprivation, referral route, reason for referral) and the care pathway received (case status, number type and length of contacts, onward signposting and interventions) using descriptive statistics. We then used unadjusted linear, logistic, multinomial logistic and negative binomial regression models to describe whether these factors differed by age and over time. RESULTS: Most individuals were aged 18 and over, 91% lived in urban areas and 58% lived in the top three most deprived deciles of the UK. Most were referred by general practitioners or other allied health workers (79%) and mental health was the leading reason for referral (44%). The typical pathway included 4.64 social prescribing contacts (SD=7.70) totalling 66 min (SD=108), with 34% receiving an onward referral to community support. The average age of those referred to social prescribing increased over time. CONCLUSIONS: Our findings indicate that relatively few children under 18 were referred to social prescribing and this disparity may be increasing. It was promising that children and young people referred to social prescribing were more likely to live in deprived areas. However, given current findings, more work is needed to increase the reach of social prescribing for children and young people across the UK.