Background: Cardiovascular disease (CVD) remains a leading contributor to the global burden of disease. While pharmacological therapy is central to its prevention and management, medication adherence is often suboptimal, limiting treatment effectiveness.

Methods: This qualitative evidence synthesis followed the Joanna Briggs Institute Manual for Evidence Synthesis. A search strategy was conducted in September 2024 and updated in January 2025, with studies published between February 2004 to June 2024, were identified across four databases (Medline, CINAHL, Embase, PsycINFO) and analysed using framework analysis guided by the World Health Organization’s conceptual framework for medication adherence. A purposive maximum variation sampling approach was applied across five strata: low-income countries, underserved communities, rural populations, and primary and secondary CVD prevention settings. Sub-framework analysis was conducted to compare adherence behaviours and contextual influences across these groups.

Results: Of 145 eligible studies, 27 were sampled for synthesis. Barriers to adherence included fear of side effects, limited symptom awareness, negative beliefs about medications, and poor patient-provider communication. Enablers included trust-based relationships with healthcare providers, understanding of treatment purpose, simplified medication regimens, and strong social and family support. Participants in secondary prevention settings expressed heightened risk perception, while those in primary prevention often viewed CVD as asymptomatic and less threatening. Rural populations showed a preference for natural remedies and more fatalistic views. Low-income participants faced financial constraints and limited access to care. Structurally disadvantaged communities reported a lack of personalised care and continuity.

Conclusion: Medication adherence is shaped by a dynamic interplay of individual, clinical, and contextual factors. Strengthening patient-provider communication, addressing perceptual barriers, and tailoring strategies to population-specific needs are critical to improving adherence. Future research should prioritise diverse representation and integrate qualitative insights into medication adherence prediction tools and intervention design.

Introduction: Over half of the U.S. population aged 12 and older has used illicit hard drugs at least once in their lifetime. Social trust has been linked with alcohol use, smoking, and other risk behaviors. Although social trust may also share a similar pathway with hard drug use, the relationship has not been investigated in the general adult population.

Aims: This study examined the prevalence of hard drug use among U.S. adults aged 18 years and older. We also examined the associations between hard drug use with institutional social trust and interpersonal social trust.

Method: This serial cross-sectional study analyzed nationally representative data of community-living U.S. adults from the 2012–2022 U.S. General Social Survey. Participants reported ever misusing injection drugs or crack cocaine. We adopted validated measures of institutional trust (13-item) and generalized interpersonal trust (3-item). The study conducted binary logistic regression analysis and adjusted for socio-demographic factors.

Results: In our sample (n = 6,879), 7.3% reported hard drug use, and no statistically significant time trends were observed from 2012 to 2022. In unadjusted analyses, both lower levels of institutional and interpersonal trust were associated with hard drug use (OR = 1.42 to 1.88). However, only institutional trust (medium-level OR = 1.37; low-level OR = 1.50) remained independently associated with hard drug use in adjusted analyses. Among young adults, white people were more likely to use hard drugs compared to their black counterparts; blacks were more likely to do so among older adults.

Conclusion: Our findings identified institutional trust as independently related to hard drug use among U.S. adults and corroborated the crossover effect between age and race. Interventions to strengthen social trust in key institutions may reduce the likelihood of lifetime hard drug use and consequently prevent dependence.

Keywords: Social trust; Hard drug use; Population-based

Introduction: Artificial Intelligence (AI)-driven Clinical Decision Support Systems (CDSS) are increasingly being integrated into healthcare to improve decision making, patient safety, and workflow efficiency.  Currently, research on understanding and comparing the human and organizational factors influencing the transition from rule-based to AI-driven CDSS among healthcare workers remains limited, particularly in among hospital pharmacists. 
Aims: To compare the characteristics and experiences of rule-based and AI-driven CDSS among hospital pharmacists and examine their perceptions, attitudes, and readiness for AI implementation compared with physicians in two Indonesian hospitals.
Methods : A qualitative study was conducted in two private hospitals in Indonesia between April and June 2026. Semi structured interviews were conducted with pharmacists using rule-based CDSS and physicians using AI-driven CDSS. Data were analyzed using thematic analysis guided by the Socio-technical Systems framework.
Results: This study identified five major themes on informants’ experience, perceptions, attitudes, and readiness on AI technology in healthcare: individual factors towards AI technology, experience with CDSS, enablers of AI implementation, barriers of AI implementation, and governance and implementation readiness, along with 11 subthemes. Pharmacists and physicians emphasized that, despite the use of AI, responsibility for clinical decision making remains with healthcare practitioners. However, hospital pharmacists expressed concerns that overreliance on AI could reduce critical thinking and contribute to perception of professional replacement by AI technologies. Notably, this concern was not reported among physicians.
Conclusions: Using a Socio-technical Systems perspective, this pilot study identified social subsistem and external factors that may facilitate or hinder the transition from rule-based to AI-driven CDSS in Indonesian hospitals among hospital pharmacists and physicians. Although AI was perceived to improve analytical support and clinical vigilance, successful implementation may depend on addressing workflow integration, organizational culture, interprofessional dynamics, and also concerns regarding AI reliance. 
Keywords: Artificial Intelligence; Clinical Decision Support Systems; Socio-technical Systems

Introduction: Artificial intelligence (AI) technologies are rapidly advancing in medication assistance, yet their specific clinical implementations remain poorly mapped. Traditional systematic reviews struggle to keep pace with this exponentially growing literature.

Aims: This scoping review aims to map current developmental trends in AI-assisted medication. Additionally, it evaluates the efficacy of Large Language Models (LLMs) in accelerating literature screening and data extraction.

Methods: We systematically searched PubMed, Embase, and Cochrane database for relevant literature. Following deduplication, we designed a three-stage LLM-based automatic workflow for study screening and extraction: two sequential rounds of screening using the DeepSeek-V4-Flash model to determine study eligibility, followed by structured data extraction from the abstracts of included studies using the DeepSeek-V4-Pro model. Screening performance was evaluated on a balanced random sample of 200 records.

Results: As of April 1, 2026, we collected 91,249 initial records, leaving 74,259 after deduplication and ultimately including 2,024 studies following LLM screening. Validation of the screening process showed 93.5% agreement between DeepSeek-V4-Flash and GPT-5.5 Pro, with a Cohen’s kappa of 0.87. Cardiovascular (22.01%), endocrine (16.09%), and nutritional/metabolic diseases (11.30%) emerged as the primary targets for AI intervention. Core algorithms were predominantly rule-based and clinical decision support systems (CDSS) (17.84%), followed by closed-loop control (6.62%) and machine/deep learning (4.10%). Applications concentrated heavily on dose calculation and titration (54.84%), alongside drug selection (36.17%), and monitoring-based adjustment (32.95%). Notably, AI outperformed conventional care in 30.14% of studies, 9.63% reported comparable outcomes or no significant difference, and only 0.20% found AI inferior.

Conclusions: AI-assisted medication is advancing rapidly, particularly in cardiovascular care, yielding extensive research on optimizing critical tasks such as dose calculation and drug selection. Current evidence demonstrates the robust efficacy and immense potential of data-driven precision medicine.

Keywords: AI-assisted medication, Large Language Models, Scoping review

Introduction: Methylphenidate (MPH) is widely used to treat attention-deficit/hyperactivity disorder (ADHD), in patients with depression. However, concerns remain that MPH may increase the risk of manic episodes and transition to bipolar disorder (BD) in specific individuals.

Aims: We aim to estimate the treatment effect of administrating MPH on BD occurrence in patients with ADHD and depression using causal forest model with large-scale propensity scores.

Methods: We analyzed nationwide claims data from adults with ADHD and depression recorded in the Health Insurance Review and Assessment Service–ADHD database (2017–2020), converted to the OMOP Common Data Model. Patients were classified into MPH-treated and untreated groups and divided into training, validation, and test datasets. Propensity scores were estimated using random forests after large-scale covariate screening, and a causal forest model with inverse-propensity weighting was used to estimate average and conditional treatment effects (ATE and CATE). External validation was performed in a cohort from A tertiary hospital by applying the median CATE cutoff to classify patients into high- and low-risk groups and evaluating prediction of bipolar transition.

Results: Among 19,896 patients, 13,417 received MPH and 1,445 developed BD. After screening, 4,153 of 4,608 baseline covariates were retained, and the model was fitted using the top 30 variables by importance. The estimated ATE was 0.035 [0.025–0.102] in the validation set and 0.034 [0.025–0.102] in the test set. Significant heterogeneity in treatment effects was observed across CATE quintiles in the test set (χ²₄=12.40, p=0.0146). In the external cohort (n=34), median-based CATE stratification showed moderate predictive performance (accuracy 0.647, sensitivity 0.800, specificity 0.621).

Conclusions: Causal machine learning identified that MPH treatment may elevate BD risk in subgroups. The CATE-derived cutoff demonstrated external validity in real-world data, supporting the potential for individualized treatment-effect–based prescribing strategies in adults with ADHD and depression.

Keywords: Causal machine learning, treatment effect, bipolar disorder

Background
The AI-driven P-CARDIAC model family for cardiovascular disease, comprising 1°P-CARDIAC for primary prevention and P-CARDIAC for secondary prevention, has demonstrated high accuracy. However, performance can vary across distinct sub-cohorts.
Aim:
To externally validate P-CARDIAC models across diverse sociodemographic and clinical subgroups and evaluate its performance against traditional risk scores.
Methods
Utilizing a large electronic health record cohort, validation was stratified across subgroups based on sex, age (< 50, 50-64, >=75 years), and major comorbidities including diabetes, hypertension, chronic kidney disease, dementia, Atrial Fibrillation (AF), dyslipidemia, and smoking. Both basic and full 1°P-CARDIAC and P-CARDIAC models were evaluated. Discrimination was assessed using Harrell's C-index and compared against benchmark models (PCE PREDICT, Chiana-PAR, and Framingham score for 1°P-CARDIAC; TRS-2P and SMART2 for 2°P-CARDIAC). Decision curve analysis (DCA) and Net Reclassification Improvement (NRI) were conducted as additional analyses.
Results
The validation included 1,085,148 primary and 260,240 secondary prevention individuals. Both the basic models and the feature-rich full models demonstrated stable discrimination across most subpopulations. And the full model has outperformed the basic model in the subpopulation: 1°P-CARDIAC yielded a strong overall primary prevention C-index of 0.819. Despite its performance declining in complex cohorts like AF (0.645) and dementia (0.658), it consistently outperformed other risk tools across all subgroups (C-index improvement over PCE ranged from +0.022 [AF] to +0.178 [age 65–74]). Secondary prevention followed similar trends. 1°P-CARDIAC full provided a higher DCA net benefit than benchmarks overall and across key high-risk subgroups.
Conclusion
Both basic and full P-CARDIAC models demonstrate a consistent discrimination advantage over traditional risk scores across diverse subgroups in Hong Kong. However, absolute predictive performance attenuates in highly complex, older, and multimorbid clinical settings. Clinical deployment in these specific settings may require targeted calibration adjustments.

Introduction: Medication-related harm in older adults is difficult to identify and prevent due to multimorbidity, polypharmacy, and age-related changes. Artificial intelligence (AI) may support medication safety by detecting, predicting, and prioritising medication-related risk. However, its application across the medication-safety pathway in older adults has not been comprehensively mapped.

Aims: To examine how AI has been applied across the medication-safety pathway in older adults and identify evidence gaps relevant to clinical translation. Specific objectives were to describe AI methods, medication-safety applications, validation approaches, implementation status, and barriers to clinical translation.

Methods: We conducted a scoping review of studies published from January 2016 to 16 February 2026 in MEDLINE, Embase, and Scopus. Eligible studies included populations aged ≥60 years, or those in which older adults predominated, and investigated AI-based approaches addressing medication-related safety. Two reviewers independently screened studies, with disagreements resolved by consensus or a third reviewer.

Results: Of 4,436 records identified, 2,955 unique records were screened, 273 full texts assessed, and 46 studies included. Most studies focused on model development and validation (n=33), with electronic health records as the most common data source (n=35). Four application areas were identified: harm identification and prediction (n=22); prescribing appropriateness review (n=17); dose optimisation and monitoring (n=5); and medication review prioritisation (n=9). Machine learning, particularly gradient boosting methods, was the most common AI approach (n=31). Most studies reported internal validation only (n=28), while external validation (n=11) and real-world implementation (n=3) were limited. Outcomes focused mainly on model performance and process measures, with minimal evidence that AI reduced patient-level medication-related harm.

Conclusions: AI shows promise for improving the visibility and prioritisation of medication-related risk in older adults; however, evidence of clinical effectiveness remains limited. Future research should prioritise externally validated, workflow-integrated, and human-in-the-loop evaluations using clinically meaningful medication-safety outcomes.

Keywords: artificial intelligence; medication safety; older adults

Background: Medication adherence remains a major challenge in chronic disease management, yet conventional adherence measures may miss spontaneous patient narratives in digital spaces. This study aimed to examine Indonesian public discourse on chronic disease medication adherence on X by mapping adherence facilitators and barriers using the WHO five-dimension framework and assessing whether automated affective sentiment aligns with behaviourally meaningful adherence interpretation.
Methods: We conducted a cross-sectional infodemiology study of publicly available X posts from 1 January to 30 June 2025. Data were retrieved using seven disease-specific queries combining the Indonesian medication-taking phrase with chronic disease terms. After duplicate removal and screening, 611 tweets were analysed. IndoBERT was used to classify automated affective sentiment classification, whereas researcher-annotated behavioural sentiment captured adherence orientation of each tweet. Adherence facilitators and barriers were independently coded by two reviewers using the WHO five-dimension framework, with disagreements resolved through adjudication. BERTopic was used for topic modelling to identify semantic clusters.
Results: IndoBERT classified most tweets as negative (326/611, 53.4%), whereas researcher-annotated behavioural sentiment identified positive adherence-oriented content as the largest category (260/611, 42.6%). Agreement between automated affective and behavioural sentiment was low (κ=0.174), indicating divergence between emotional polarity and medication-taking behaviour. After adjudication, 402 tweets (65.8%) contained explicit adherence determinants: 230 facilitators (37.6%), 145 barriers (23.7%), and 27 mixed (4.4%). Patient-related factors dominated the WHO classification (335/402, 83.3%). Three emotion–behaviour divergence patterns were identified, accounting for 28.5% of the corpus. BERTopic identified clusters related to medication routines, patient-provider interaction, and side effects.
Conclusion: Indonesian public X discourse on chronic disease medication-taking revealed adherence facilitators and barriers, predominantly within patient-related domains. Automated affective sentiment did not align well with behavioural adherence interpretation, indicating that emotional polarity alone is insufficient for identifying adherence-relevant signals. Domain-informed interpretation is needed for social media-based medication adherence research and intervention planning.

Introduction: Machine learning prediction models increasingly inform immune checkpoint inhibitor (ICI) treatment in solid tumors, yet their clinical implementation depends on complete and transparent reporting. The TRIPOD+AI guideline was published in April 2024 to harmonize reporting of such models. Whether reporting quality has improved since its release is unknown.

Aims: To compare adherence to TRIPOD+AI reporting standards between ICI prediction-model studies published before versus after the guideline. We assessed both overall and item-level reporting quality.

Methods: We systematically reviewed prediction-model studies in ICI-treated patients across PubMed, Embase, Web of Science, and Cochrane Library. Two cohorts of 27 studies each were defined by first publication date (April 2019 to March 2021, pre-guideline; April 2024 to March 2026, post-guideline). Each study was scored against the 52-item TRIPOD+AI checklist (Yes/No/NA); adherence was Yes divided by applicable items. Cohorts were compared at study level (Mann-Whitney U test) and item level (Fisher exact test with Benjamini-Hochberg correction).

Results: Mean study-level adherence was 59.7% (SD 9.1) before and 66.0% (SD 8.2) after the guideline (Mann-Whitney U p=0.009; Cohen's d=0.73). Five items reached nominal significance (Fisher p<0.05), led by performance evaluation including calibration (12e, 18.5% to 55.6%) and predictor-distribution comparison between development and evaluation data (20c). However, no item remained significant after Benjamini-Hochberg correction (all q approximately 0.28). The least-reported items were sample-size justification (10), protocol availability (18c), health inequalities (3c), model fairness (14), and patient and public involvement (PPI; item 19). Items 14 and 19 were reported in no study in either cohort.

Conclusions: In this pilot study, overall reporting adherence improved significantly after TRIPOD+AI, but item-level gains did not survive correction. Fairness and PPI remained entirely unaddressed even after the guideline, representing the most urgent transparency gaps. These preliminary findings justify a larger confirmatory study.

Keywords: TRIPOD+AI; ML prediction model reporting; immune checkpoint inhibitors

Introduction: Emergency departments (EDs) are a critical point of contact for opioid overdose-related presentations. However, reliance on structured coding for surveillance can lead to underreporting or misclassification; prior work shows that only 42% of drug overdoses were coded properly, emphasising the need to leverage unstructured free-text fields.

Aims: To develop and internally validate a pipeline that applies natural language processing (NLP) and Machine Learning (ML) classifiers to identify opioid overdoses in routinely collected ED data, incorporating both structured and unstructured data.

Methods: An NLP-enhanced ML pipeline was developed and trained on a stratified sample of Victorian Emergency Minimum Dataset (VEMD) records. Medical concepts were extracted from free text using the Medical Concept Annotation Tool (MedCAT) to generate a Bag-of-Words (BOW) representation. Clinical ontologies SNOMED-CT and UMLS were tested for Named Entity Recognition (NER), with adjustments to concept-filtering parameters and model configurations. Subsequently, accounting for class imbalance, several ML models were trained and evaluated against a manually coded gold standard using standard performance metrics.

Results: The pretrained UMLS model and XGBoost classifier performed best, identifying 1010 opioid overdoses (precision = 0.86, F1-score = 0.77). It outperformed the customised SNOMED-CT model (precision = 0.16, F1 = 0.26) and the ICD-10 diagnosis codes (precision = 0.12, F1 = 0.20). The performance gap reflects the fragmented nature of ED texts; UMLS excels at capturing short, granular concepts, whereas SNOMED-CT captures specific, detailed concepts. Currently, the pipeline struggles with short, ambiguous texts, brand names and clinical abbreviations. Future work will address this by fine-tuning the model on customised concepts and contextual tagging.

Conclusions: This project demonstrates the potential of NLP-enhanced ML pipelines scaled across hospital settings to improve the accuracy, robustness, and generalisability of opioid overdose identification, ultimately providing a more reliable framework for surveillance.

Keywords: Natural Language Processing, Opioid Poisoning, Emergency Department Data

Introduction: Medication non-adherence remains a major challenge in the management of chronic diseases, leading to poor clinical outcomes, increased healthcare utilization, and higher costs. Conventional adherence-monitoring approaches, such as directly observed therapy and reminder-based mobile applications, are often resource-intensive, prone to inaccuracies, and lack real-time remote monitoring capabilities.

Objective: To develop, validate, and evaluate an artificial intelligence (AI)-based mobile application (Ai-Adhere) for real-time monitoring and improvement of medication adherence among patients with chronic conditions.

Methods: A prospective interventional study was conducted over six months at JSS Hospital, Mysuru. An AI-enabled mobile application was developed to assess medication adherence through video-based recordings. The prototype was initially validated using pre-recorded videos to determine model performance. Subsequently, 180 patients with chronic diseases were enrolled and instructed to use the application for two weeks by recording medication-taking events. Adherence was monitored in real time, and usability was assessed using the System Usability Scale (SUS).

Results: The AI model demonstrated robust performance, achieving a precision of 90.7%, recall of 93.2%, F1-score of 0.918, and overall accuracy of 90.5%. The application effectively monitored medication-taking behavior in real time and helped in identifying potential reasons for non-adherence. Medication adherence improved substantially among enrolled participants, increasing from 55% at baseline to 90% following application use. Usability assessment revealed a mean SUS score of 86.5 ± 8.2, indicating excellent usability and high user acceptability.

Conclusion: Ai-Adhere demonstrated high accuracy in monitoring medication adherence and showed potential to significantly improve adherence among patients with chronic conditions. The application offers a scalable, user-friendly, and real-time digital solution for adherence monitoring and may support improved long-term disease management.

Keywords: Medication Adherence; Artificial Intelligence; Chronic Disease Management; Digital Health; Real-Time Monitoring.

Introduction: Asthma and chronic obstructive pulmonary disease (COPD) are major causes of global morbidity and mortality. Inhalers remain the cornerstone of treatment; however, their effectiveness depends on correct inhalation technique and adherence. Improper inhaler use contributes to poor disease control, increased exacerbations, hospitalizations, and healthcare costs. Therefore, innovative approaches are needed to support and monitor inhaler use in routine care.
Aim: To develop and validate an artificial intelligence (AI)-enabled mobile application for real-time assessment of inhaler usage technique and evaluate its impact on inhaler adherence.
Methods: A six-month prospective interventional study was conducted among adults (>18 years) with asthma or COPD using controller inhalers. An AI-enabled mobile application was developed and validated using inhaler-use video recordings in real time. Performance was assessed using accuracy, precision, recall, and F1-score. Changes in inhaler usage technique and improved adherence following application use were evaluated, while usability and acceptability were measured using the System Usability Scale (SUS).
Results: The prototype achieved 71.1% accuracy, 83.3% precision, 47.6% recall, and an F1-score of 0.60. Following refinement, application real time performance validated on 130 recordings improved to 90.77% accuracy, 84.4% precision, 88.37% recall, and an F1-score of 0.86. AI-enabled monitoring significantly improved inhaler technique adherence, with confidence scores increasing from 38% to 82% in the Metered Dose Inhaler group, 0% to 91% in the Metered Dose Inhaler with Spacer group, and 22% to 78% in the Rotahaler group following real-time AI-guided interventions. The application demonstrated excellent usability and acceptability, with a mean SUS score of 84.06 ± 5.08.
Conclusion: The AI-enabled mobile application accurately assessed inhaler usage technique and improved adherence among patients with asthma and COPD. Its high usability and acceptability support integration into routine respiratory care.
Keywords: Artificial Intelligence; Inhaler Technique; Asthma; COPD.

Introduction:
Manually checking prescriptions for potential errors is time‑consuming and error‑prone in busy wards. Optical character recognition (OCR) and machine learning (ML) can support healthcare professionals by digitising prescriptions and flagging medication issues in real time.

Aims:
To design, implement, and validate an artificial intelligence (AI) based mobile application the Intelligent Medication Assistant (IMA) for active surveillance of medication incidents, and to measure its usability among clinicians and pharmacy staff.

Methods:
The IMA was developed as a progressive web application using the MongoDB, Express.js, React.js, and Node.js (MERN) stack, integrating Google Research's DocLens OCR‑based ML interface to read handwritten and printed prescriptions. De‑identified data were used to train and test the model, and the app was linked to a structured incident reporting workflow in inpatient wards. Validation measured detection rate and accuracy against pharmacist‑validated reference data. Usability was assessed over 10 days using the System Usability Scale (SUS) by 25 doctors, pharmacists, and pharmacy students.

Results:
The app successfully digitised complex prescriptions and matched drug names via a pharmacopoeia‑linked database. Of 620 medications validated, 617 were correctly detected (detection rate 99.52%), with an overall accuracy of 73.39% for medication recognition and incident identification. The mean SUS score was 82.30 ("excellent"); scores were 83.81 for doctors, 82.08 for pharmacists, and 81.14 for pharmacy students. Users reported reduced manual checking time and faster incident reporting.

Conclusions:
The IMA demonstrated high detection performance and excellent usability across all user groups, supporting the feasibility of AI‑enabled tools for medication safety surveillance in real‑world hospital settings. Further refinement may enable scaling to other wards and institutions.

Keywords:
Artificial intelligence; Mobile application; Optical character recognition; Medication safety; Usability

Introduction: Early identification of trauma patients at high risk of death is essential for triage and resource allocation. Traditional trauma scoring systems often perform suboptimally in contemporary, multicenter settings.

Aims: To develop and evaluate predictive models for emergency department (ED) and in-hospital mortality using prehospital and triage data.

Methods: We analysed Pan-Asian Trauma Outcomes Study (PATOS) registry (2015–2022) of trauma patients retrospectively, covering 14 Asian centers. Two temporally distinct outcomes were analyzed: ED mortality and in-hospital mortality among ED survivors. Predictors included demographics, prehospital interventions, triage vital signs, neurological status, injury severity scores, injury mechanism, and comorbidities. Machine learning (ML) models (Random Forest, XGBoost, Elastic-Net logistic regression, Naive Bayes), a point-based AutoScore system, and traditional trauma scores (TRISS, MGAP, GIFT) were compared using AUROC, accuracy, sensitivity, specificity, and F1 score.

Results: Of 208,184 patients, 1.17% died in-hospital. Prehospital CPR, ED CPR, total Glasgow Coma Scale, triage vital signs, and injury severity scores were consistently top predictors. For ED mortality, Random Forest and XGBoost using all variables achieved the highest discrimination (AUC 0.992), with clinical-variable-restricted models performing similarly (AUC 0.976-0.981). AutoScore yielded AUC 0.964 with high specificity (0.998) and moderate sensitivity (0.596). Traditional scores showed lower discrimination (TRISS AUC 0.958, GIFT 0.935, MGAP 0.890). For in-hospital mortality, Random Forest achieved AUC 0.989, XGBoost 0.986, and AutoScore 0.943. Traditional scores showed similar limitations (TRISS 0.930, GIFT 0.894, MGAP 0.891) with poor specificity (<0.01). An R Shiny application was developed for real-time risk assessment using key variables.

Conclusions: Neurological status, injury severity, and physiological compromise were key drivers of early trauma mortality. Tree-based ML models offered superior predictive accuracy (AUC 0.986-0.992), while AutoScore provided a practical, interpretable bedside tool for rapid risk stratification. The systematic underperformance of Western-derived traditional scores highlighted the need for region-specific prediction tools.

Keywords: trauma; mortality prediction; machine learning

Introduction: Infection management in multiple myeloma (MM) is critical, as patients are highly immunocompromised. However, identifying infection events in claims databases is challenging due to the absence of detailed clinical information. This study developed a methodological approach using a Japanese claims database to address this limitation.

Aims: To develop a claims-based algorithm to identify and characterize incident infections among MM patients.

Methods: We used a Japanese claims database (Medical Data Vision) to identify adult MM patients who initiated elranatamab (first dose as index date). Leveraging clinical expert input, we assessed patterns of infection diagnosis codes and anti-infective prescriptions to develop rules that identify infection episodes and classify them as bacterial, viral, or fungal. Incident infections were defined by mapping each episode to an anatomical site–pathogen combination, differentiating new events from pre-existing or ongoing conditions.

Results: The algorithm comprises three domains. For episode identification, an infection diagnosis requires a concurrent anti-infective prescription to reduce inclusion of rule-out diagnoses. For severe diagnoses like sepsis (<10% of episodes), distinguishing therapeutic versus prophylactic intent required either a ≥10-day continuous single-agent course or an escalation of therapy within 10 days. We excluded likely prophylaxis events (low-dose sulfamethoxazole/trimethoprim or events within 8 days post-index if the same agent was prescribed within 14 days pre-index), removing ~17% of initial episodes (mostly low-dose S/T). For type classification, infections were categorized by anti-infective class, recognizing that broad-spectrum therapy may overestimate bacterial events. Mapping episodes by anatomical site–pathogen combination distinguished new infections from pre-existing or ongoing conditions.

Conclusion: This structured approach to claims data supports improving infection identification. By focusing on anti-infective therapy patterns over diagnosis codes and incorporating site–pathogen mapping, the algorithm may help reduce misclassification. This provides a framework for evaluating infection patterns in claims-based pharmacoepidemiologic studies.

Keywords: multiple myeloma; infection algorithm; claims database

Introduction: Reliable background rates for conditions classified as adverse events of special interest (AESIs) are essential for interpreting vaccine safety signals in observed-versus-expected analyses, particularly for new vaccines, such as Lassa fever. In many low- and middle-income countries, pharmacovigilance relies predominantly on hospital-based data sources, which capture severe or clinically diagnosed events, potentially biasing background rate estimates for AESIs that are transient, self-managed, or treated in primary care. Yet such events can impose a significant burden and raise community concerns about immunisation programmes. Health and Demographic Surveillance Systems (HDSS) provide established population denominators and longitudinal follow-up but are not routinely configured for active vaccine safety surveillance. The Global Vaccine Data Network’s Background Rates of Adverse Events for Vaccine Evaluation in Africa (BRAVE) project addresses these gaps.

Aims: To adapt existing HDSS platforms to actively ascertain population-based AESI background rates.

Methods: A harmonised screening and follow-up module will be integrated into routine HDSS data collection in Ghana, Kenya, and the Democratic Republic of Congo. A screening question asks whether health events require medical attention or disrupt usual activities. Positive screens trigger structured, community-based follow-up to characterise symptom type, severity, timing, and care-seeking pathways. Standard operating procedures guide training, data quality assurance, ethics, governance, and incidence estimation using HDSS person-time denominators, with site-specific adjustments aligned with local care access and referral pathways.

Conclusions: Integrating active assessment of health events into HDSS enables population-based estimation of background rates beyond hospital-only surveillance, thereby supporting scalable, equity-focused strengthening of community-based vaccine safety monitoring across Africa.

Keywords: Health and Demographic Surveillance Systems, Adverse Events of Special Interest, Background Rates

Introduction: The estimand framework provides an approach to defining treatment effects by specifying the target population, treatment conditions, endpoint, summary measure, and handling of intercurrent events. Clinically relevant estimands may arise from a single research question, with intention-to-treat, modified intention-to-treat, and as-treated analyses reflecting distinct treatment strategies and causal contrasts. Although estimand specification is increasingly recognised in pharmacoepidemiologic study design, its application in the prevalent new user design (PNUD) remains challenging.

Aims: To examine how the estimand framework in PNUD studies and provide practical guidance for selecting estimands aligned with specific research questions.

Methods: We review core components of the estimand framework and discuss considerations specific to real-world evidence studies using PNUD. Using a case study of patients initiating sodium-glucose cotransporter-2 inhibitors versus dipeptidyl peptidase-4 inhibitors and cardiovascular outcomes, we illustrate alternative estimands and approaches to handling intercurrent events, including treatment discontinuation and switching. We further consider how study design and confounding adjustment methods influence the target population represented by the estimated treatment effect.

Results: Ambiguity in PNUD studies arises not only from handling intercurrent events but also from how the target population is implicitly defined through design and analysis choices. The time-conditional propensity score approach commonly used in PNUD studies targets the average treatment effect among the treated, though this is rarely stated explicitly. We propose a framework linking common design and analytic choices in PNUD studies to corresponding estimands and outline the role of sensitivity analyses in assessing robustness of findings.

Conclusions: Explicit estimand specification should be integral to PNUDs. Clearly defining the treatment strategy and target population improves transparency, aligns analysis with the research question, and strengthens interpretation of comparative effectiveness and safety findings. The proposed framework provides practical guidance for researchers implementing PNUD and evaluating study results.

Keywords: Estimands, prevalent new user design, real-world evidence

Introduction: Real-world evidence regarding the impact of sodium glucose cotransporter-2 inhibitor (SGLT-2i) on venous thromboembolism (VTE) occurrence remains scarce.

Aims: To examine whether initiation of SGLT-2i is associated with a reduced risk of VTE in type 2 diabetes mellitus (T2DM) patients.

Methods: We conducted new-user, active-comparator target trials emulation within the Shenzhen Public Health Data Platform to contrast SGLT-2i with dipeptidyl peptidase-4 inhibitors (DPP-4i) and glucagon-like peptide-1 receptor agonists (GLP-1RA), respectively, covering the period from 1 January 2019 to 1 October 2024. Overlap-weighted Cox proportional-hazard models were employed to quantify the association between SGLT-2i and outcomes.

Results: After a median follow-up of 1.86 years, VTE occurred in 423 (0.48%) of 89,141 SGLT-2i users and 478 (0.85%) of 56,815 DPP-4i users. SGLT-2i was associated with a lower risk of VTE occurrence versus DPP-4i (HR: 0.85; 95% CI: 0.74-0.97), with a more pronounced association in patients with advanced age, diabetic complications, or receiving concomitant antihyperglycemic agents (Pfor interaction<0.05). However, the significant association was observed only for deep venous thrombosis (DVT) (HR: 0.81; 95% CI: 0.69-0.94) but not for pulmonary thromboembolism (PTE) (HR: 0.98; 95% CI: 0.75-1.28). In primary analysis and sensitivity analyses, we observed no significant differences in the risk of VTE, DVT, or PTE occurrence between SGLT-2i users and GLP-1RA users.

Conclusions: SGLT-2i was associated with a lower risk of VTE compared with DPP-4i in T2DM patients, particularly among the old and those with advanced disease, while showing comparable safety to GLP-1RA.

Keywords: type 2 diabetes mellitus, venous thromboembolism, sodium-glucose cotransporter-2 inhibitors, glucagon-like peptide-1 receptor agonists, target trial emulation.

Introduction: Despite growing recognition of psychiatric safety in epilepsy care, the psychiatric consequences of antiepileptic drugs (AEDs) remain incompletely characterized. Prior studies have focused on individual agents or isolated outcomes such as suicidality, without systematic evaluation across pharmacological mechanisms or risk stratification for vulnerable subgroups.
Aims: This study evaluated mechanism-specific disproportionality signals for psychiatric adverse events (AEs) associated with AEDs and developed an interpretable machine learning model to support risk stratification.
Methods: This pharmacovigilance study utilized the FDA Adverse Event Reporting System (FAERS; 2004Q1–2025Q2), with the Canada Vigilance Adverse Reaction Database (CVAR) for external validation. AEDs were classified into GABA receptor agents, GABA transaminase inhibitors, and carbonic anhydrase inhibitors. Disproportionality analysis employed the information component (IC) and reporting odds ratio (ROR). For machine learning analysis, age, sex, and concomitant medications (ATC-based) served as predictors across a balanced dataset (96,207 cases with psychiatric AEs; 95,132 without). Seven algorithms were compared using stratified five-fold cross-validation, and SHapley Additive exPlanations was applied for model interpretability.
Results: Among 55,366 psychiatric disorder reports in FAERS, GABA receptor agents demonstrated the highest disproportionality signal (ROR, 5.58; 95% CI, 5.51–5.64), with signals most prominent in older adults and females. Findings were corroborated by CVAR. LightGBM was selected as the final model, achieving an AUROC of 0.704 in the FAERS test set and 0.771 in external CVAR validation. SHAP analyses identified age as the most influential predictor, followed by sex and concomitant nervous system medications.
Conclusions: GABA receptor agents exhibited the highest disproportionality signals. Routinely available clinical variables identified patients at elevated psychiatric risk via machine learning, supporting mechanism-based safety monitoring in epilepsy care. These hypothesis-generating findings warrant confirmation in future cohort studies.
Keywords: Antiepileptic drugs; Psychiatric disorders; Pharmacovigilance.

Introduction: In pharmacoepidemiologic studies, because treatment assignment is nonrandomized and comparisons between treatment groups are subject to confounding, causal effect estimation requires careful consideration. Balancing-score methods, including propensity score (PS) and disease risk score (DRS) methods, have been proposed for confounder adjustment. Existing DRS weighting methods primarily target the average treatment effect on the treated (ATT), reflecting that the DRS summarizes potential outcomes under control. Although the ATT is a natural target for DRS weighting when effect modification is considered, this does not preclude applying DRS weighting to average treatment effect (ATE) estimation.

Aims: We formulated DRS weighting estimators for the ATE under no effect modification. We derived asymptotic variance estimators.

Methods: We proposed three weighting strategies for ATE estimation targeting the DRS distribution in the overall population: target distribution weighting (TDW), inverse probability weighting (IPW), and stabilized inverse probability weighting (SIPW). We clarified algebraic relationships among the three weights. For each strategy, we formulated Hájek-type ATE estimators. We derived sandwich variance estimators using M-estimation.

Results: In the absence of effect modification, the proposed estimators targeted the ATE standardized to the DRS distribution in the overall population. The TDW weights were algebraically equivalent to the SIPW weights. Approximate variance estimators for the ATE estimators were obtained. In simulations, we compared the proposed DRS weighting estimators with naïve estimation and PS weighting. DRS weighting showed more favorable performance than PS weighting and naïve estimation.

Conclusions: The proposed DRS weighting estimators showed promise for ATE estimation in simulations. Although the ATE is mathematically definable under effect modification, its clinical interpretation may be challenging because is averages heterogeneous effects across subpopulation. Under no-effect-modification assumption, the proposed DRS weighting estimators with asymptotic variance estimators provided a viable alternative to ATE weighting in simulations.

Keywords: Prognostic score; Effect modification; Causal inference

Introduction: Medication safety studies using healthcare data are often limited by heterogeneous structures, local terminologies, and poor reproducibility. Standardised data models such as the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) improve transparency and scalability of pharmacoepidemiologic research. Intravenous aciclovir is widely used for suspected or confirmed herpes simplex virus central nervous system infection in children, yet uncertainty remains regarding aciclovir-associated nephrotoxicity in clinical practice.

Aims: To develop an OMOP-based phenotyping framework for paediatric medication safety studies using aciclovir-associated nephrotoxicity as a use case. To demonstrate how standardised phenotyping can support reproducible medication safety research.

Methods: A retrospective observational audit study was designed using electronic medical record data transformed into OMOP CDM. Eligible patients were aged under 18 years and received intravenous aciclovir between 2016 and 2023. Standardised concept sets were developed for aciclovir exposure, herpes simplex virus central nervous system disease, acute kidney injury, nephrotoxic exposures, and relevant subgroups. Cohort construction and phenotyping were implemented using methodologies compatible with the Observational Health Data Sciences and Informatics framework. Framework performance will be assessed through cohort ascertainment, phenotype validity, data completeness, and reproducibility.

Results: Phenotype development identified three key methodological challenges: defining suspected herpes simplex virus central nervous system disease using heterogeneous clinical proxies, identifying acute kidney injury using longitudinal creatinine measurements, and distinguishing aciclovir-associated nephrotoxicity from competing causes such as sepsis, dehydration, and concomitant nephrotoxic medications. These challenges informed standardised cohort definitions, temporal exposure windows, and confounder flags. Cohort extraction identified more than 3,000 eligible aciclovir exposures, with analysis underway.

Conclusions: Standardised phenotyping within OMOP CDM is feasible for complex paediatric medication safety studies and provides a transparent approach to exposure, outcome, and confounder definition. This methodology may support reproducible medication safety research and future multicentre analyses.

Keywords: OMOP Common Data Model; medication safety; aciclovir nephrotoxicity

Introduction: Diabetic foot infections (DFIs) are a major cause of non-traumatic lower-limb amputations. Vitamin D regulates immunity and wound healing by binding the vitamin D receptor (VDR) and inducing the cathelicidin antimicrobial peptide (CAMP) gene to produce cathelicidin (LL-37). However, its role in DFIs is not fully understood.
Aim: To find the association of vitamin D supplementation with serum vitamin D levels, VDR and CAMP gene expression, and ulcer healing in DFI patients infected with biofilm-forming bacteria.
Method: This exploratory prospective cohort study (September 2022-March 2026) enrolled 16 DFI patients each in the study group (SG) and reference group (RG), alongside 16 healthy controls (HCs). SG participants received 60,000 IU vitamin D with standard care, while RG participants received standard care alone. Both groups were followed for 4 months. Vitamin D levels were assessed using electrochemiluminescence immunoassay, while VDR protein and LL-37 were quantified through ELISA. Biofilm formation was evaluated using the tissue culture plate method. Additionally, VDR and CAMP gene expression were analysed via mRNA quantitation. The relative gene expression was calculated using the 2^−ΔΔCt method, with GAPDH as the endogenous control.
Results: Vitamin D supplementation significantly increased serum vitamin D levels (p<0.001), VDR protein concentrations (p=0.001), and LL-37 levels (p=0.001) in SG participants. However, differences in VDR (p=0.067) and CAMP (p=0.346) gene expression among HCs, SG, and RG were statistically insignificant. Although ulcer healing did not significantly differ between SG and RG groups (p=0.391), favourable healing trends were observed in SG. Furthermore, Vitamin D-sufficient SG patients had lower biofilm-producing bacteria than deficient patients, albeit without statistical significance (p=0.163).
Conclusion: Vitamin D supplementation may positively influence the vitamin D–VDR–LL-37 pathway and enhance wound healing in DFI patients. More extensive studies are required to validate these preliminary molecular and clinical findings.

Keywords: Diabetic foot infection, vitamin D-VDR-LL-37 pathway, wound healing

Introduction: Hepatocellular carcinoma (HCC) is a major complication of chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infection and contributes to liver-related morbidity and mortality. In high viral hepatitis burden settings, real-world evidence on HCC frequency and persistent viral activity in advanced liver disease remains limited. Tissue-level viral persistence may inform surveillance and disease management.

Aims: This study aimed to estimate HCC burden among patients with advanced liver disease and evaluate persistent HBV and HCV viral protein expression in liver cancer tissues.

Methods: A ten-year retrospective real-world study was conducted using clinical data from patients with liver-related complications. HCC frequency was assessed among eligible patients. Viral persistence was evaluated by immunohistochemistry on explanted liver tissues and liver resection samples from patients with HBV- or HCV-associated HCC. HCV non-structural protein 5B and HBV X protein expression were assessed using qualitative immunohistochemistry scoring. Data were analyzed using R software.

Results: Among 1,384 patients with liver-related complications, 256 were diagnosed with HCC, representing 18.5% of the study population. Nearly one in five patients with advanced liver disease had liver cancer. HCC occurrence differed significantly by age and gender, with higher burden among male patients. Among HCV-associated HCC samples, 13 of 16 patients showed detectable low-level HCV NS5B protein expression. Three HCV-associated samples demonstrated moderate-to-high expression. Among HBV-associated HCC samples, HBV X protein expression was detected with variable staining intensity. These findings suggest persistent viral activity within hepatocytes at advanced disease stages.

Conclusions: This study demonstrates substantial hepatitis-associated HCC burden and tissue-level evidence of persistent HBV and HCV protein expression. Findings support earlier hepatitis detection, timely antiviral management, and risk-based HCC surveillance to reduce preventable liver cancer burden.

Keywords: hepatocellular carcinoma; viral persistence; real-world evidence

Introduction: Pediatric long-term valproic acid (VPA) therapy causes weight gain and hyperammonemia, reducing compliance. Understanding concentration impacts and independent risk factors is crucial to optimizing clinical outcomes.

Aims: This study evaluated VPA concentrations, pharmacokinetic ratios, and BMI z-score changes, while analyzing age-specific dose-response relationships in drug-induced hyperammonemia.

Methods: We conducted a retrospective longitudinal study utilizing clinical records collected between March 1, 2024, and May 31, 2026 at a tertiary hospital in Southern Taiwan. Eligible participants were pediatric patients aged 2 to 12 years receiving VPA therapy. Multiple linear regression and age-stratified analyses were performed to evaluate the impacts of trough concentration, daily dosage, serum albumin, and concurrent ketogenic diet.

Results: The 2–6 years group exhibited significantly lower concentration-to-dose ratios compared to the 6–12 years group due to enhanced hepatic clearance（2.12 ± 0.94 vs 4.18 ± 1.05, 𝑃 <0.005）. Conversely, the 6–12 years group demonstrated greater susceptibility to weight gain, showing a higher mean increase in Δ BMI z-score（0.98 ± 1.02 vs 0.15 ± 0.84, 𝑃 <0.03）. After adjusting for age, sex, and baseline weight, high VPA concentration（≥ 80 𝜇g/mL）was a strong independent predictor of increased Δ body mass index z-score（𝑃 <0.001）. Concurrently, concurrent ketogenic diet significantly mitigated this weight-gain trend (ꞵ = -0.214, 𝑃=0.032）. VPA trough concentration strongly correlated with serum ammonia levels (r=0.69, 𝑃 <0.0001）. Furthermore, younger patients showed significantly higher vulnerability to hyperammonemia when serum albumin dropped below (<3.5g/dL).

Conclusions: High VPA exposure independently drives age-specific pediatric obesity and hyperammonemia. While ketogenic diets protect against weight gain, intensive drug monitoring and prospective threshold validation remain vital for safety.

Keywords: Valproic acid, BMI z-score, Hyperammonemia

Introduction: Evidence linking prenatal antiseizure medication (ASM) exposure to childhood neuropsychiatric disorders (NPDs) remains inconsistent, although risks associated with valproate are relatively well established.
Aims: To investigate the association between prenatal ASM exposure and the subsequent risk of NPDs in children.
Methods: This longitudinal, binational birth cohort study included mother-child pairs from nationwide claims databases in South Korea and Japan, with follow-up from 2009 to 2023. Prenatal ASM exposure was defined as maternal ASM prescriptions during pregnancy, and polytherapy was defined as exposure to two or more distinct ASMs during pregnancy. Incident NPDs in children were identified based on at least two diagnostic records in both countries, and adjusted hazard ratios (aHRs) with 95% confidence intervals (CIs) were estimated using Cox proportional hazards models in the full, PS–matched, and sibling comparison cohorts, adjusting for maternal epilepsy history, maternal neuropsychiatric history, and other prespecified covariates.
Results: In the Korean cohort of 3,013,262 mother-child pairs, prenatal ASM exposure was associated with an increased risk of childhood NPDs in the PS-matched cohort (1.10 [1.02-1.18]), whereas the association was not significant in sibling-comparison analyses (0.86 [0.71-1.04]). Comparable findings were observed in the Japanese replication cohort. Increased risks were identified for first-trimester exposure, exposure across multiple trimesters, longer cumulative exposure duration, and higher prescription frequency. Among individual ASMs, clonazepam (1.27 [1.11-1.44]) and sodium valproate (1.50 [1.17-1.92]) were associated with increased NPD risk. Subgroup analyses identified that prenatal ASM exposure significantly elevated risks for intellectual disability, ADHD, autism spectrum disorder, and severe NPDs.
Conclusions: The modest association observed in PS-matched analyses was not replicated in sibling comparisons, suggesting that shared familial or maternal factors may partly explain the association, although limited power in sibling analyses should be considered. These findings support cautious, individualized ASM prescribing during pregnancy.
Keywords: antiseizure medications, neuropsychiatric disorders, pregnancy

Introduction: Escherichia coli poses an escalating pediatric health threat. Globally and locally, the surging prevalence of antimicrobial resistance (AMR) severely restricts therapeutic options and heightens risks for persistent clinical illness. However, the specific behavioral and environmental determinants driving pediatric colonization remain insufficiently characterized.

Aim. To evaluate the epidemiological determinants and antimicrobial resistance status of E.coli carriage among children in Hong Kong.

Methods: This case-control study at Hong Kong Children’s Hospital, recruited pediatric participants (aged 0–18years) and their caregivers. E.coli carriages was screened via child rectal/stool sample and caregiver nasal swab, alongside demographic and antibiotic-use surveys. Isolates were confirmed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOFMS). Antimicrobial susceptibility testing (AST) was performed on 11 antibiotics based on local epidemiology and CLSI guidelines using disc-diffusion and minimum inhibitory concentration (MIC). Children with and without E.coli carriage were designated as cases and controls, respectively. Risk factors for carriage were analyzed using logistic regression, with variable selection optimized via a bidirectional stepwise Akaike Information Criterion (AIC) process.

Results: Among 328 enrolled participants (mean age: 6.82 years; 57.6% male), 69 (21.0%) carried E.coli (cases) and 259 (79.0%) served as controls. Resistance profiles were characterized by high prevalence against cefazolin (91.3%), ampicillin (81.2%) followed by tetracycline (62.3%) and co-trimoxazole (53.6%). Notably, within the colonized subjects, 95.7% presented with resistance to ≥1 antibiotic, and 82.6% exhibited multidrug-resistance (MDR). Multivariable logistic regression identified caregiver colonization (OR: 29.45, 95% CI: 1.75-1284.13) and caregiver hospital exposure (OR: 0.18, 95% CI: 0.04-0.58) as the significant predictors for children carriage.

Conclusions: Hong Kong pediatric E.coli carriage exhibits high prevalence of MDR; first-line antibiotics are predominantly compromised, whereas some last-resort therapeutics remain effective. Multivariable analysis showed that significant children-caregiver carriage associations. These determinants offer critical insights for identifying high-risk pediatric populations and tailoring interventions.

Keywords: Multidrug-resistant Escherichia coli, Pediatric colonization, Caregiver-child dyads

Introduction: Pharmacovigilance relies not only on robust methods and harmonised data, but also on trust among study teams, healthcare providers, health systems, and communities. In multi-country collaborations, cultural assumptions around communication, consent, authority, illness, and death can shape vaccine safety surveillance. The ‘Background Rates of Adverse events for Vaccine Evaluation’ (BRAVE) project recognises cultural awareness as a technical requirement for ethical, high-quality pharmacovigilance, and as such aimed to incorporate it into routine staff training.

Aim: To conduct structured, country-led cultural awareness sessions within BRAVE, strengthening collaboration, site readiness, and data quality.

Methods: Sessions were co-developed and delivered by the Global and African Coordinating Centres alongside the data collection teams from the 5 African countries. Guided by the Purnell model for Cultural Competence, they addressed heritage, communication norms, family and consent structures, workforce dynamics, health-seeking behaviours, spirituality, and death rituals. These domains were examined in relation to local working environments and health-system realities.

Results: The sessions improved familiarity with site-specific contexts and constraints, strengthening empathy and collaboration across teams. Key issues highlighted included the role of community leaders and extended family in consent processes, the impact of religious practices and rapid burial customs on outcome ascertainment and differing professional hierarchies affecting communication and decision-making. Addressing these factors has reduced misunderstandings between teams, informed more context-sensitive engagement strategies, and improved confidence in navigating cultural complexity.

Conclusion: Cultural awareness is a critical enabler of effective pharmacovigilance. BRAVE shows that embedding structured, locally led cultural competence early strengthens trust, alignment, and scientific rigour.

Key words: Cultural competence, Pharmacovigilance, Multi-country collaboration

Introduction: Timely vaccine safety surveillance relies on high-quality clinical data to estimate background rates of adverse events of special interest (AESIs) and detect safety signals. Electronic health records (EHRs) offer potential for both retrospective background rate estimation and prospective vaccine safety surveillance. However, their readiness in low-and middle-income countries (LMICs) is variable due to heterogeneous information systems, variable data quality, limited interoperability, and hybrid paper-electronic workflows.
Aims: To assess the capacity and readiness of hospital EHR systems within the BRAVE (Background Rates of Adverse Events for Vaccine Evaluation) network to support vaccine safety surveillance.
Methods: A standardised EHR readiness assessment was conducted across five BRAVE network sites in Africa. Assessment domains included governance and data security, data completeness, diagnostic coding, timeliness, record linkage, interoperability (including HL7/FHIR compatibility), and data export functionality. Site-reported information was evaluated against a predefined minimum dataset for AESI identification and background rate estimation.
Results: EHR platforms were heterogeneous, ranging from open-source to locally developed systems. Most sites captured core inpatient data sufficient for retrospective AESI identification. Six of ten hospitals reported high data completeness, timely capture, functional record linkage, and automated exports. ICD coding was commonly used to support standardisation; however, dual paper and electronic systems increased workload. Other sites were better suited to hybrid approaches combining EHR-based pre-screening or denominator estimation with manual chart review. Overall, readiness varied across domains, with strengths in data capture and governance, and limitations in interoperability and workflow integration.
Conclusion: Hospital EHR systems within the BRAVE network across multiple LMICs show variable but promising capacity to support pharmacoepidemiologic vaccine safety surveillance. A tiered, pragmatic approach leveraging both electronic data and complementary manual processes may enable retrospective AESI estimation and prospective surveillance. Strengthening interoperability and workflow integration will be critical to support sustainable scale-up.

Keywords: vaccine safety; electronic health records; pharmacovigilance

Introduction: Indonesia has one of the largest social media user bases globally. Following the recent implementation of the national Human Papillomavirus (HPV) vaccination program, public monitoring of vaccine safety has emerged as a critical imperative. However, the utilisation of social media data to monitor activities related to vaccine safety is still limited in the country.
Aims: This study aims to identify post-HPV vaccination safety mentions from social media in Indonesia using a Natural Language Processing (NLP) approach.
Methods: This retrospective observational study extracted data from the Indonesian-language Twitter/X platform during the 2016–2025 period using the search keywords "vaccine" and "hpv". All post texts obtained from the crawling process underwent pre-processing and were then classified using a rule-based NLP model to detect and map vaccine safety mentions. The categorisation was based on the standard clinical terminology categories of the Medical Dictionary for Regulatory Activities (MedDRA), followed by a quantification of the vaccine safety mentions.
Results: Out of a total of 31,870 tweets analysed, only 1,664 (5.2%) were identified as vaccine safety mentions. The trend of HPV vaccine safety mentions was found to increase sharply after 2020. Local symptom complaints were the most frequently discussed, including injection site pain (49.4%), swelling (3.5%), and erythema (1.86%). Meanwhile, the main systemic symptoms reported included pyrexia (8.6%), fatigue (6.0%), and headache (3.6%). Uniquely, this study found a high proportion of mentions related to reproductive function complaints, which were infertility (11.7%) and menstrual irregularities (8.8%).
Conclusion: Vaccine safety mentions following HPV vaccination in Indonesia are relatively infrequent, predominantly highlighting local symptoms. However, signals concerning reproductive issues warrant further detailed verification. While social media and NLP show strong potential for active surveillance, clinical validation remains essential to confirm true signals and mitigate misinformation bias.
keywords: HPV, social media, vaccine safety

Introduction: Romosozumab is an anabolic agent indicated for osteoporosis in patients at high fracture risk. Concerns regarding major adverse cardiovascular events (MACE) have prompted regulatory warnings; however, real-world evidence on treatment patterns and cardiovascular outcomes remains limited.

Aims: To evaluate treatment patterns and cardiovascular outcomes following romosozumab initiation.

Methods: We conducted a retrospective single-center cohort study of patients who received at least one dose of romosozumab between January 2022 and October 2024. Patients with a prior history of myocardial infarction or ischemic stroke were excluded. Participants were followed for up to 545 days after treatment initiation, consisting of a 12-month treatment period and a 180-day post-treatment follow-up. MACE, defined as acute myocardial infarction or ischemic stroke, was identified using diagnostic codes and confirmed through medical record review. Treatment completion, cardiovascular event rates, and time-to-event analyses including Kaplan–Meier estimates were evaluated.

Results: A total of 477 patients were included, with a mean age of 72.9 years; 95.2% were women. Hypertension (38.2%), hyperlipidemia (28.3%), and diabetes mellitus (17.8%) were common baseline comorbidities. Overall, 41.7% completed the planned 12-dose romosozumab course. During 663.8 person-years of follow-up, 7 patients experienced MACE (1.47%), corresponding to an incidence rate of 10.55 per 1,000 person-years. The median time to MACE was 164 days (IQR, 142.5–195.5 days). All cardiovascular events occurred during active treatment and before the tenth dose. Kaplan–Meier analysis showed cumulative incidences of 0.85% at 180 days and 1.49% at the end of follow-up.

Conclusions: Cardiovascular events were uncommon among romosozumab users without prior myocardial infarction or ischemic stroke, whereas fewer than half completed the planned 12-dose treatment course. These findings provide real-world evidence on both treatment utilization and cardiovascular outcomes following romosozumab initiation.

Keywords: Romosozumab; cardiovascular safety; Treatment Patterns

Introduction: Preventable adverse drug reactions (pADRs) pose a critical threat to patient safety and contribute significantly to avoidable patient harm. Conventional voluntary reporting systems consistently suffer from substantial underreporting. Implementing active surveillance methods is therefore essential to overcome these limitations and optimize medication safety.

Aims: This study evaluated the effectiveness of using protamine and naloxone as active trigger tools to detect and subsequently reduce pADRs within an inpatient setting.

Methods: A cross-sectional experimental study was conducted at Nakhon Pathom Hospital , comparing inpatient medical records before and after implementing the trigger tools and clinical interventions (Smart Labeling, Rapid Digital Reporting). Severity was classified using the NCC MERP index, and preventability was assessed via the Schumock and Thornton criteria. Statistical analyses were performed using descriptive statistics and Chi-square tests.

Results: Baseline conventional reporting yielded zero pADRs. In 2024, active trigger tools identified 10 pADRs out of 303 cases: (5 from 295 protamine; 5 from 8 naloxone). These cases (9 category E, 1 category H) were primarily driven by excessive dosing, inappropriate drug switching, and treatment delays, with enoxaparin accounting for 30%. Following targeted interventions in 2025, pADRs significantly decreased to 5 cases out of 462 cases (4 from 457 protamine; 1 from 5 naloxone), all of which were category E with zero instances of severe harm (category F or higher). Heparin became the primary causative agent (80%). Overall, the intervention achieved a statistically significant reduction in pADRs (p = 0.038) with a Risk Ratio of 0.47

Conclusions: Protamine and naloxone trigger tools with clinical interventions demonstrated high clinical efficacy, successfully reducing the risk of pADRs by 53%. The naloxone cohort exhibited a higher pADR rate than the protamine group, indicating an urgent need for more stringent clinical guidelines for opioid management.

Keywords: Trigger tools, Preventable adverse drug reactions, Patient safety

Introduction: Mucolytics like N-acetylcysteine, erdosteine, and bromhexine are widely used for respiratory conditions and generally regarded as safe. However, hypersensitivity reactions occasionally occur, and the clinical spectrum of mucolytic-induced adverse drug reactions (ADRs) remains poorly characterized.

Aims: To identify and describe the clinical features of ADRs associated with various mucolytic agents.

Methods We retrospectively analyzed cases of mucolytic-induced ADRs at a Korean tertiary hospital between January 2016 and December 2025. Inclusion criteria were strictly limited to: 1) cases classified as "certain" by causality assessment; 2) "probable" cases involving a single drug; or 3) cases definitively confirmed via clinical history, skin tests, and provocation tests. Ambiguous cases lacking objective evidence were excluded.

Results: Over the 10-year study period, a total of 37 ADR cases were identified: 15 attributed to erdosteine, 19 to N-acetylcysteine, and 3 to bromhexine. ADRs due to erdosteine included fixed drug eruption (FDE, n=9), maculopapular eruption (MPE, n=4), angioedema (n=1), and urticaria (n=1). For N-acetylcysteine, 15 cases were associated with the intravenous formulation - nausea/vomiting (n=10), bronchoconstriction/dyspnea (n=3), and angioedema (n=2) - while 4 cases involved the oral formulation, presenting as dyspnea (n=1) and nausea/vomiting (n=3), yielding 19 total N-acetylcysteine-related ADRs. Bromhexine was responsible for MPE (n=2) and urticaria (n=1). Cutaneous symptoms were predominant in erdosteine (n = 15) and bromhexine (n = 3) cases, but rare in N-acetylcysteine (n = 2). All skin tests yielded negative results, indicating that controlled provocation tests are essential for definitive diagnosis unless the clinical history is indisputably clear.

Conclusions: Although relatively safe, mucolytics can induce a diverse spectrum of ADRs. Erdosteine-induced FDE was the most frequent cutaneous manifestation, and N-acetylcysteine frequently caused gastrointestinal symptoms. Healthcare professionals should be aware of these potential ADRs to ensure prompt diagnosis and prevent inadvertent re-exposure.

Introduction: Eravacycline is a tetracycline-class antibiotic. Whether hypofibrinogenemia represents a class effect remains unclear.
Aims: To investigate eravacycline's effect on coagulation, particularly fibrinogen (FIB).
Methods: Retrospective study of 98 severe infection patients receiving eravacycline. Coagulation markers were monitored before, during, and after treatment.
Results: FIB levels significantly decreased after eravacycline use (P<0.001). Consequently, 54 cases were classified into the hypofibrinogenemia group (FIB < 2.0 g/L), and the remaining 44 into the normal fibrinogen group.In the hypofibrinogenemia group, hypofibrinogenemia occurred on average 8.57 ± 4.31 days after eravacycline treatment. In 33 patients with post-cessation data, FIB returned to normal after treatment cessation (P<0.001). PT, APTT, and TT prolonged after use and shortened after cessation (PT: non-significant). Multivariate analysis identified decreased eGFR, low baseline FIB, and female gender as independent risk factors. Bleeding events occurred in 5 patients (9.26%), including 3 cases of gastrointestinal bleeding and 2 cases of bloody alveolar lavage fluid.
Conclusion: Eravacycline-related hypofibrinogenemia occurs at 8.6 days, associated with baseline FIB, eGFR, and gender. Patients with lower FIB levels may experience bleeding events. Routine monitoring of FIB and coagulation is recommended.
Keywords：eravacycline, hypofibrinogenemia, risk factors

Introduction: Both Schumock–Thornton (S&T) scale and Liverpool Adverse Drug Reaction Avoidability Tool (LAAT) are designed to assess the preventability of Adverse Drug Reactions (ADRs). Assessing the consistency and reliability of these two tools helps in better clinical decision making.

Aim: To assess the inter-rater reliability of LAAT and S&T scales and evaluate the agreement between the two scales in determining ADR preventability.

Methods: A retrospective observational study was conducted on 80 randomly documented ADRs selected from pharmacovigilance database. Preventability assessment was independently carried out by three trained raters using the Schumock–Thornton scale and the Liverpool ADR Avoidability Tool. Inter-rater agreement for both scales and concordance between the two scales were evaluated using Fleiss’ kappa statistics with 95% confidence intervals (CI). Statistical significance was set at p < 0.05.

Results: A total of 80 ADRs were independently evaluated for preventability by three raters using S&T scale and LAAT. Inter-rater reliability analysis demonstrated moderate agreement among raters using the S&T scale, with a Fleiss’ kappa coefficient of 0.58 (95% CI: 0.46–0.70; p < 0.001). The LAAT scale demonstrated slightly higher agreement, with a Fleiss’ kappa coefficient of 0.64 (95% CI: 0.58–0.79; p < 0.001), indicating substantial agreement among raters. Agreement between S&T scale and LAAT in the classification of ADR preventability was also assessed and demonstrated moderate concordance, with a kappa coefficient of 0.54 (95% CI: 0.42–0.66; p < 0.001). LAAT took less time for preventability assessment. Overall, the LAAT scale showed a greater inter-rater consistency than S&T scale. However, both tools had acceptable reliability for ADR preventability assessment.

Conclusion: Both the LAAT and S&T scales demonstrated moderate to substantial reliability for ADR preventability assessment. The LAAT scale showed slightly better inter-rater agreement compared to the Schumock–Thornton scale.

Introduction: Signal detection using spontaneous reporting systems is a key component of post-marketing pharmacovigilance. While quantitative data-mining methods are widely used, rare but clinically important adverse events may be overlooked. Designated Medical Event (DME) terminology can support qualitative review by prioritizing serious adverse events. This study assessed whether a DME-based review system can contribute to signal detection and regulatory action.

Aims: This study aimed to describe the operation of a DME-based serious adverse event review system. It also evaluated its contribution to identifying drug–event combinations requiring further assessment and regulatory action.

Methods: Monthly serious adverse event reports in the Korea Adverse Event Reporting System (KAERS) were reviewed between January 2021 and October 2023. Signal detection used three complementary approaches: individual case review of deaths and congenital anomalies, DME-based qualitative assessment, and quantitative data-mining analyses. Drug–event combinations identified through DME review were assessed for causality and further evaluated according to the signal management algorithm. Regulatory actions, including labeling changes, were proposed to the regulatory authority when warranted.

Results: A total of 2,533 drug–DME combinations were evaluated. Among these, 781 combinations included at least one case for which a causal relationship could not be excluded. Following further evaluation, 101 combinations were selected for detailed signal assessment. Six combinations were not reflected in Korean product labeling. Safety update proposals were submitted to the regulatory authority, and four regulatory label changes were implemented: pembrolizumab-associated rhabdomyolysis, atezolizumab-associated acute kidney injury, famotidine-associated anaphylactic reaction, and lidocaine-associated anaphylactic reaction.

Conclusions: The DME-based review process complemented conventional data-mining methods by identifying clinically important but infrequently reported drug–event combinations that might otherwise have been overlooked. The system contributed directly to regulatory decision-making, including product labeling changes, demonstrating the value of DME-based monitoring as an effective component of routine pharmacovigilance activities.

Keywords: designated medical event; pharmacovigilance; signal detection

Introduction: Polycystic Ovary Syndrome (PCOS) is one of the most common endocrine disorders affecting women of reproductive age and is frequently associated with psychological distress. College students may be particularly vulnerable due to academic stress, sedentary lifestyles, and poor dietary habits. However, awareness of PCOS and its mental health implications remains limited among young women.

Aims: To determine the prevalence and predisposing factors of PCOS among female students and assess its impact on mental health. To evaluate the effectiveness of a structured educational intervention on knowledge, attitude, and practice (KAP) regarding PCOS.

Methods: A dual cross-sectional and prospective interventional study was conducted among 459 female students. Confirmed PCOS cases (n=53) and undiagnosed participants (n=406) were assessed using the Depression, Anxiety and Stress Scale-21 (DASS-21), a PCOS diagnostic questionnaire, and a validated KAP questionnaire. Participants received a 90-minute educational intervention, followed by KAP reassessment after 3–4 weeks. Data were analysed using appropriate non-parametric statistical tests.

Results: The prevalence of PCOS was 11.5%. Among diagnosed participants, obesity (32.0%), anxiety (22.6%), and hypertension (13.2%) were the most common comorbidities. Among undiagnosed participants, 54.8% reported minimal physical activity, 73.2% experienced active stressors, and 10.8% were classified as having mild PCOS risk. Clinically significant anxiety and depression were reported in 73.4% and 62.3% of diagnosed participants, respectively, compared with 46.5% and 31.7% of undiagnosed participants. Significant differences were observed between groups across all DASS-21 domains (p<0.0001). The intervention significantly improved knowledge and practice scores in both groups (p<0.0001), whereas attitude scores among diagnosed participants showed no significant change (p=0.7593).

Conclusions: PCOS is prevalent among female students and is associated with substantial psychological burden. Educational interventions can improve PCOS awareness and health-related practices, supporting the integration of PCOS education and mental health screening into university health programs.

Keywords: Polycystic ovary syndrome, mental health, educational intervention

Introduction: Guideline-directed medical therapy (GDMT) is central to heart failure (HF) management, but real-world patients often have comorbidities requiring additional drugs, leading to polypharmacy. Whether polypharmacy and drug-related problems (DRPs) are independently associated with survival remains unclear.

Aim: To describe pharmacotherapy patterns and polypharmacy in an ambispective HF cohort, and to evaluate whether polypharmacy and clinical factors were associated with mortality; DRPs were examined descriptively.

Methods: Analysis included 585 adults with HF from a single-centre ambispective cohort (five-year retrospective review plus nine months of prospective data). Polypharmacy was 5–9 medications, hyperpolypharmacy ≥10, and extensive polypharmacy ≥15. DRPs used standard criteria. Cox regression examined mortality predictors; logistic regression explored DRP predictors.

Results: Of 585 patients (72.13% male, largest age group 61–80 years), polypharmacy occurred in 413 (70.59%), hyperpolypharmacy in 94 (16.06%), extensive polypharmacy in 77 (13.16%). Most received core GDMT. During follow-up, 83 (14.14%) died. In Cox models, age, HF subtype, and length of stay significantly predicted mortality; diuretics showed a trend toward lower risk (HR 0.63, 95% CI 0.39–1.02). No drug class independently predicted mortality. In logistic regression, neither medication count nor polypharmacy category was independently associated with higher DRP risk, though comorbidity number showed a borderline effect.

Conclusions: In this real-world HF cohort, polypharmacy was highly prevalent, and pharmacotherapy broadly followed guidelines. Survival was determined primarily by age, HF subtype, and hospitalisation patterns rather than by drug classes or polypharmacy burden, indicating clinical profile drives mortality.

Keywords: Heart failure; Survival; Polypharmacy; Drug-related problems; Pharmacotherapy

Introduction: Pain is highly prevalent in people living with dementia, yet difficult to assess. Opioid use has been increasing in this population despite limited evidence on efficacy and safety. However, evidence on opioid use patterns and outcomes remains limited and heterogeneous. Mapping this evidence is critical to inform safe opioid prescribing in this vulnerable population.

Aims: To systematically map and synthesise evidence on incidence of opioid use, incidence or prevalence of persistent opioid use (POU), predictors of POU, and associated all-cause mortality outcomes in people living with dementia.

Methods: Databases searched included MEDLINE, Embase, International Pharmaceutical Abstracts, APA PsycInfo, Scopus, PubMed and Web of Science, supplemented by hand-searching reference lists. Studies were eligible if they investigated opioid use outcomes in patients with a confirmed dementia diagnosis, with no restrictions on age, dementia type, severity, setting, or geographic location.

Results: Fifteen studies published between March 2013 and June 2025 were identified, across countries in North America, Europe and Australasia, in community and residential care settings. Incidence of opioid use ranged from 13.4% to 42.0%, while incidence or prevalence of POU ranged from 4.7% to 25.1%, with substantial heterogeneity in POU definitions. Key predictors of POU included using ≥3 psychotropic drugs (aOR=3.02; 95%CI [1.04,8.76]; p=0.041), poor physical health (aOR=2.38; 95%CI [1.27,4.35]; p=0.006) and having a partner (aOR=2.06; 95%CI [1.03,4.13]; p=0.041). Opioid use was associated with increased 180-day mortality risk (aHR=5.06; 95%CI [4.86,5.29]), particularly with strong opioids (aHR=7.23; 95%CI [6.82,7.66]). Mortality risk was highest in the first 1-14 days post-initiation (aHR=10.81; 95%CI [9.74,11.99]) and with fentanyl use (aHR=8.26; 95%CI [7.18,9.51]).

Conclusions: Opioid use is common in dementia, but variable, with emerging safety concerns regarding mortality. These findings highlight the need for caution in opioid prescribing, enhanced monitoring, and evidence-based guidelines, alongside future research targeting outcomes such as hospitalisation risk.

Keywords: opioids, dementia, pharmacoepidemiology

Introduction: Medication adherence is a key determinant of therapeutic outcomes in cancer patients, influencing disease progression and overall well being. Poor adherence may result from adverse drug reactions, psychological distress, and financial burden. Quality of life is an important patient centered outcome in oncology care. Understanding this relationship is essential to improve patient outcomes.

Aims: To assess medication adherence and quality of life among cancer patients. To identify factors influencing adherence and their impact on quality of life.

Methods: A prospective observational study was conducted over six months among 60 adult cancer patients aged 18 years and above. Medication adherence was assessed using the Medication Adherence Rating Scale and quality of life using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire EORTC QLQ-C30 version 3. Data on adverse drug reactions and adherence barriers were collected through patient interviews. Descriptive statistics and ordinal logistic regression were applied.

Results: Seventy seven percent of patients showed moderate to high adherence with a mean score of 7.66. Approximately 81.7 percent acknowledged medication benefits, while 21.7 percent discontinued when feeling better and 15 percent when feeling worse. Financial constraints and psychological distress were the major barriers identified. Alopecia, fatigue, and nausea were the most commonly reported adverse drug reactions. Most patients had moderate quality of life. Lower adherence was significantly associated with poorer quality of life (p < 0.05), indicating a higher likelihood of reduced quality of life among non adherent patients.

Conclusions: Medication non adherence is associated with poorer quality of life and is influenced by modifiable factors.

Keywords: Medication adherence; quality of life; cancer patients

Introduction: Tinea is a common superficial fungal infection associated with discomfort, social embarrassment, and reduced quality of life. Poor adherence to antifungal therapy, misuse of topical corticosteroids, and inadequate hygiene practices contribute to recurrent infections. Clinical pharmacists may improve treatment outcomes through counselling and follow-up. However, evidence regarding their impact on dermatology-related quality of life in tinea patients remains limited.

Aims: To evaluate the impact of clinical pharmacist interventions on dermatology-related quality of life among tinea patients using the Dermatology Life Quality Index (DLQI).

Methods: A prospective interventional study was conducted for six months. A total of 210 patients were screened, of whom 151 tinea patients were enrolled. Patients received structured clinical pharmacist-led interventions, including counselling on medication adherence, lifestyle modifications, avoidance of topical steroid misuse, and regular follow-up monitoring. Dermatology-related quality of life was assessed using the validated DLQI questionnaire at baseline and during three follow-up visits at defined intervals. Statistical analysis was performed using SPSS version 29.0.

Results: Among the 151 participants, 58.3% were males, and 34.4% belonged to the 41-50 years age group. Tinea corporis was the most common clinical presentation (31.7%), followed by tinea cruris (22.4%). The mean DLQI score showed a significant reduction from 25.46 ± 3.07 at baseline to 22.10 ± 1.74 at follow-up 1, 7.35 ± 2.17 at follow-up 2, and 0.38 ± 0.55 at follow-up 3 (χ² = 442.85, p < 0.001). Significant improvement in DLQI scores was observed across all sociodemographic and clinical categories. Patients also demonstrated improved medication adherence, better hygiene practices, and reduced misuse of steroid-containing topical preparations.

Conclusions: Clinical pharmacist-led interventions significantly improved dermatology-related quality of life among tinea patients. Structured counselling and regular follow-up enhanced treatment adherence and therapeutic outcomes, supporting the integration of clinical pharmacists into dermatology care.

Keywords: Tinea; Clinical pharmacist; Quality of life

Introduction: Over-the-counter (OTC) self-medication is widely practised and has important implications for Self-medication with over-the-counter (OTC) medicines is widely practiced and has important implications for patient safety and healthcare utilization. Pharmacists play a critical role in ensuring appropriate OTC use; therefore, competency development in pharmacy students is essential. Existing assessment approaches mainly focus on knowledge, with limited validated simulation-based tools to evaluate applied competencies in real-world self-medication scenarios.
Aim: To develop and validate simulation-based training and assessment instruments for evaluating pharmacy students’ competency in OTC self-medication management across multiple OTC drug classes.
Methods: An instrument development and content validation study was conducted. Eight subject matter experts were purposively selected based on predefined criteria (at least 3 years’ experience in pharmacy practice or education and clinical pharmacy; multi-institutional panel). The training and assessment package comprised 57 MCQs, 49 simulation-based case scenarios across nine OTC drug classes, a 54-item pharmacy student competency assessment scale (PSCAS), and an 11-item standardized patient satisfaction form. Content validity was quantified using I-CVI, S-CVI/Ave, and modified kappa (κ*). Face validation was conducted with 15 Pharm D interns.
Results: All four instruments achieved excellent content validity: MCQ bank (56 items, one duplicate was present) S-CVI/Ave = 0.946; case scenarios S-CVI/Ave range 0.961–0.996 (overall 0.983); PSCAS S-CVI/Ave = 0.986 (κ* = 0.867); SP satisfaction form S-CVI/Ave = 1.000 (κ* = 1.000). Face validation confirmed high learner acceptability (mean scores 4.61–4.69/5.00; ≥80% item-level agreement). Psychometric evaluation, including Cronbach’s Alpha, is planned for the pilot phase.
Conclusion: All instruments demonstrated excellent content validity and high learner acceptability, supporting pilot implementation. This simulation-based learning framework offers a structured pharmacy competency assessment model for OTC self-medication counselling, advancing experiential learning and patient safety in pharmacy education.
Keywords: OTC self-medication, Simulation-based learning, Pharmacy competency assessment

Introduction: Medication literacy is defined as the ability to obtain, comprehend, communicate, calculate, and process medication-related information. Inadequate medication literacy leads to medication errors, poor adherence and increased hospitalisation. Limited published data exist on medication literacy among Indian ethnic minority populations. The ethnic minorities are already at a disadvantage due to low literacy rates, geographic isolation and overreliance on informal health advice and alternative medicine.

Aims: To assess medication literacy and to understand which specific medication tasks the participants fail most often. To identify the factors that are associated with medication literacy.

Methods: Cross-sectional survey of 140 ethnic minority adults at a single healthcare facility in a rural district of Southern India (convenience sample). A 14-item adapted MedLitRxSE was scored dichotomously (range 0-14). Internal consistency was assessed with Cronbach's alpha. Pearson correlations and multivariable linear regression tested associations between sociodemographic variables and total scores.

Results: Mean score was 7.54 (SD=3.65). Cronbach's alpha was 0.845. The three worst-performing items were injection angle identification (83.6% failed), warning label interpretation (77.9% failed), and paediatric dose calculation (76.4% failed). Liquid dose measurement was answered correctly by 84.3%. Education correlated most strongly with total score (r=0.842, p<0.001) and was the only variable that remained significant after adjustment (standardised β=0.806, p<0.001; R²=0.714). Age (r=−0.589) and income (r=0.329) were significant bivariate correlates but dropped out of the adjusted model, likely because both track closely with educational attainment in this population.

Conclusions: Participants struggled with applied tasks like calculating doses, reading warnings, and identifying injection angles. These specific failures lead to predictable medication errors. Education drove most of the variance, so different approaches, like hands-on demonstration and the teach-back method, should be implemented rather than generic counselling.

Keywords: medication literacy, ethnic minority, MedLitRxSE

Introduction: Candida auris(C.auris) is an emerging multidrug-resistant fungal pathogen causing nosocomial infections, posing serious challenges to clinical treatment and infection control. Systematic data on its clinical distribution, specimen sources, and drug sensitivity in large tertiary hospitals remain limited.

Aims: Analyzes C.auris cases from a Chinese provincial tertiary hospital with 4,500 beds to clarify the characteristics, informing targeted prevention and treatment strategies.

Method: All clinical C.auris isolates identified at the hospital from January 2024 to April 2026 were enrolled to analyze clinical distribution, specimen sources, and antifungal susceptibility. Screening was targeted on high-touch surfaces of equipment, bed unit, healthcare workers' hands, and specific infected patients.

Results: Between January 2024 and April 2026, 194 clinical C.auris isolates were identified from 176 patients, mainly from ICU(67.05%), Geriatrics(25.00%), Pancreatic center(6.25%), Neurosurgery wards(4.55%), etc. Urine was the most common specimen source (54.64%), followed by sputum (18.04%), bronchoalveolar lavage fluid (6.70%), etc. Another 15.47% of specimens were sterile body fluids. All isolates were resistant to fluconazole; voriconazole yielded MICs > 0.03 μg/mL, and posaconazole MICs ranged from 0.008 μg/mL to 0.5 μg/mL for all isolates. Amphotericin B resistance was observed in 41.91% of isolates, which was higher than the rate reported in previous literature. Flucytosine MICs ≥32 μg/mL were observed in 2.94% of isolates. The isolates were predominantly susceptible to echinocandins, exhibiting resistance rates of 3.68% for caspofungin and 2.94% for micafungin. Screening revealed that C.auris was restricted to patient body sites (axilla, groin, wounds), with no isolates recovered from healthcare workers' hands or environmental surfaces, suggesting that the pathogen was primarily associated with patient colonization.

Conclusions: The isolation rate of C.auris has significantly increased in hospitals, and its resistance to amphotericin B has also increased significantly. Hospitals should promptly adopt a multi-level comprehensive response strategy.

Keywords: Candida auris, Multidrug-resistant, Amphotericin B

Introduction: Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder requiring long-term pharmacotherapy, making medication appropriateness a critical concern in outpatient management. Ensuring prescribing appropriateness is essential to optimize glycemic control while minimizing adverse outcomes.

Aims: To assess the antidiabetic medication use practices, glycemic control over the preceding three months. To evaluate prescribing appropriateness using the Medication Appropriateness Index (MAI), Screening Tool of Older Person's Prescriptions (STOPP), and Screening Tool to Alert doctors to Right Treatment (START) criteria.

Methods: A descriptive cross-sectional study was conducted among 365 outpatients diagnosed with T2DM between November 2025 and May 2026. Data on medication regimens, HbA1c levels were collected, and prescribing appropriateness were evaluated using MAI and STOPP/START.

Results: The population was predominantly male (65.48%), with a mean age of 63.39 ± 12.53 years. Combination therapy was prescribed in 86.58% of patients, most commonly a triple regimen of Metformin, Gliclazide, and DPP-4 inhibitor (17.53%). The proportion of patients achieving target HbA1c was higher in the ≥ 60 age group (66.93%) compared to those under 60 years (54.95%). However, younger patients (<60 years) showed a significantly greater reduction in HbA1c over the past three months (1.10 ± 2.27). Only 6.30% of prescriptions recorded a MAI score ≥1, with dosing errors identified in 5.48%, primarily attributable to failure to adjust Sitagliptin dosage according to renal function. STOPP/START analysis revealed one case of non-selective beta-blocker use (0.27%) and three patients with heart failure lacking SGLT2 inhibitor therapy (0.82%).

Conclusions: Prescribing practices were largely appropriate and consistent with current clinical guidelines. However, attention should be directed toward renal function-based dose adjustment; particularly for Sitagliptin, to optimize therapeutic safety and efficacy in this population.

Keywords: Type 2 diabetes mellitus, medication appropriateness, STOPP/START.

Introduction: Patient representatives, individually or as part of Consumer Advisory Groups, play a crucial role in pharmacoepidemiology and other research on medicines and devices. These representatives are frequently drawn from groups that provide support and advocacy for specific health conditions. Pharmaceutical and device company financing of such groups is widespread however and creates a conflict of interest between a group’s mandate and sponsors’ priorities. At stake is health advocacy policy that reflects both over- and under-representation of selected patient interests. Clear guidance on how consumer groups can best address conflicts of interest that arise from partnerships with industry is lacking.

Aims: To a) analyse interactions between patient groups and pharmaceutical and device companies, b) understand the tools and practices patient groups use to manage these interactions and c) inform the development of a ‘toolkit for independence’.

Methods: From June to July 2026, we are conducting a national online survey of Australian patient groups. The survey covers three key areas of patient group activity: education, advocacy, and research. The current analysis focuses on research.

Results: Survey results will map companies' involvement in patient group research activities and participation of groups' representatives in industry-led research; groups’ management strategies and their perceived impact in controlling potential influence; and perceptions of influence and pressure to align with company interests. These findings will inform a practical toolkit of resources such as policy and contract templates, and disclosure standards that groups can use to better navigate research partnerships with companies.

Conclusions: This first national survey of patient groups' management of conflicts of interest arising from industry funding and subsequent development of a practical toolkit may assist researchers who are recruiting and involving patient representatives from industry-funded patient groups, to mitigate risks of undue influence and enable full representation of patient interests.

Keywords: medicine safety, COIs, governance

Background
Non-Transplant patients with newly diagnosed multiple myeloma represent a highly heterogeneous population with variability in age, and treatment tolerance. Although quadruplet regimens have demonstrated improved outcomes in clinical trials among transplant‑ineligible patients, whether treatment intensity in real‑world practice is aligned with patient fitness—and how this impacts outcomes—remains unclear.
Methods
This retrospective real-world study included adult NDMM patients identified from National Longitudinal Cohort of Hematological Diseases database (January 2013–June 2023). A frailty proxy was constructed using age ≥65 years, renal dysfunction, moderate/severe anemia and thrombocytopenia. Patients were categorized as fit (0 factors), intermediate (1 factor), or frail (≥2 factors). First-line treatment and outcome were analyzed, including triplet vs quadruplet regimens, with subgroup analyses by frailty proxy.
Results
A total of 746 NT NDMM patients were included. Overall, 97.9% were transplant-ineligible and 2.1% were transplant deferred. Mean age was 59.5 years (SD 9.1); 46.1% had ISS stage III disease and 22.0% had high‑risk cytogenetics. Frailty classification was evenly distributed (fit 33.0%, intermediate 32.2%, frail 34.8%).
Triplets predominated (81%); quadruplet use increased over time (2.5% to 19.9%) but showed limited differentiation across frailty groups (fit 27.1%, intermediate 12.6%, frail 20.1%), indicating minimal treatment stratification.
Median PFS declined significantly with increasing frailty (37.5, 34.5, and 25.1 months for fit, intermediate, and frail; p<0.01). Among fit patients, quadruplet regimens were associated with markedly longer PFS (69.6 vs 37.5 months), whereas only modest differences were observed in frail patients (28.9 vs 24.8 months), suggesting a differential treatment effect by fitness.
Conclusions
NT NDMM represents a highly heterogeneous population, with frailty proxy strongly associated with clinical outcomes. Notably, quadruplet regimens provided substantial PFS benefit in fit patients but remained underutilized, indicating a missed opportunity for optimization. These findings underscore significant unmet needs and support more fitness‑guided, personalized treatment strategies.

Introduction: Poorly controlled asthma during pregnancy has been reported to be associated with increased risk of adverse maternal and fetal outcomes, and guideline-concordant maintenance therapy is recommended throughout pregnancy. For moderate/severe patients with asthma, several biologics have been recently available in Japan, but real-world evidence on the association between exposure to biologics during pregnancy and adverse pregnancy outcomes remains limited.

Aims: This study aimed to describe the proportion of adverse fetal outcomes by biologics use during pregnancy among pregnant women with asthma.

Methods: This retrospective descriptive study was conducted using data from the 2010-2024 JMDC administrative claims database. Pregnancies ending in live births and linked to infants in women aged 15–49 years were identified using validated algorithms. Pregnant women who received at least one dose of biologics (omalizumab, mepolizumab, benralizumab, dupilumab, or tezeperumab) during pregnancy were identified as the exposure population. The study outcome was the proportion of adverse fetal outcomes, including major congenital malformations (MCMs), preterm birth, and caesarean section. For contextual reference, we also identified pregnant women who received at least one prescription for inhaled corticosteroid (ICS)-containing inhalation therapy during pregnancy and calculated the corresponding proportions of these outcomes.

Results: We identified 3,583 pregnant women with asthma (mean maternal age 33.7 years). Six women received a biologic during pregnancy, all during the first trimester. Among them, one infant had an MCM (16.7%), one was born preterm (16.7%), and two were delivered by caesarean section (33.3%); the corresponding proportions among women receiving ICS-containing inhalation therapy were 8.1%, 11.0%, and 26.3%, respectively.

Conclusions: We found a few adverse fetal outcomes among pregnant women with asthma exposed to any biologics in the first trimester as a real-world setting in Japan. However, further investigation with a larger sample size is warranted to explore those associations.

Keywords: Maternal asthma, Biologics, Adverse fetal outcomes

Background: Several studies have reported an increased miscarriage risk with antidepressant use during pregnancy, but findings may largely be attributable to key source of biases (time-related, indication, and unmeasured confounding).
Aims: To explore whether antidepressant use during pregnancy is associated with an increased risk of miscarriage using a stepwise approach to systematically address potential biases.
Methods: This cohort study included all pregnancy episodes identified from a nationwide claims database (2016–2024). Antidepressant use was defined as ≥1 prescription between the last menstrual period (LMP)-30 days and the earliest of LMP+140 days or 1 day before pregnancy end or outcome date. The outcome was miscarriage, and outcome date was shifted 7 days earlier to account for antidepressant prescribed for prodromal symptoms. To mitigate time-related and residual biases, a potential non-user set was constructed by matching on index date (±7 days) and number of obstetric-gynecology visits with the user set. Matching variables were then added progressively to further address confounding by indication: (1) hospital level, (2) main indication, and (3) number of psychiatric visits. Relative risk (RR) and 95% confidence intervals (CI) were estimated using modified Poisson regression after 1:1 propensity score matching. Additionally, a within-individual matched analysis was conducted to account for unmeasured invariant factors.
Results: Of 2.6 million pregnancies, 16,045 were identified as antidepressant users. We observed a slightly increased risk of miscarriage with antidepressant use (RR 1.17, 95% CI 1.07 to 1.28), but this attenuated after sequential matching for indication-related factors (variable 1 only: 1.11 [1.03 to 1.20]; variable 1 and 2: 1.10 [1.01 to 1.20]; all variables: 1.09 [1.00 to 1.19]). The estimate further approached the null in within-individual analysis (1.05, 0.83 to 1.32).
Conclusions: Careful attention to biases led to the attenuation of the association between antidepressant treatment and the risk of miscarriage.
Keywords: Antidepressants; Miscarriage; Bias

Introduction: Kampo medicines containing ephedra are widely used in Japan to treat common cold symptoms during pregnancy. However, ephedrine and pseudoephedrine, active components of ephedra, have been suspected to have teratogenic potential. Although two Japanese observational studies have evaluated the association between first-trimester exposure to ephedra-containing Kampo medicines and major congenital malformations (MCMs), the evidence has not been quantitatively synthesized.

Aims: To evaluate the association between first-trimester exposure to Kampo medicines containing ephedra and the risk of MCMs using a meta-analysis of Japanese observational studies.

Methods: We conducted a fixed-effect inverse-variance meta-analysis using adjusted odds ratios (ORs) and 95% confidence intervals (CIs) from previous studies examining association between Kampo medicines containing ephedra and MCMs. One study used the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study, and the other used a nationwide Japanese administrative claims database. In both studies, women using acetaminophen during the first trimester served as the comparator group to reduce confounding by indication related to common cold symptoms. Adjusted ORs were log-transformed, and standard errors were derived from reported CIs. Sensitivity analyses included random-effects models and pooled crude OR analyses.

Results: The meta-analysis included more than 96,000 pregnancies. In the cohort study, first-trimester exposure to Kampo medicines containing ephedra was not significantly associated with MCMs (adjusted OR 1.45, 95% CI 0.57–3.71). Similarly, the claims database study showed no significant association (overlap-weighted adjusted OR 1.077, 95% CI 0.932–1.245). The pooled fixed-effect adjusted OR was 1.08 (95% CI 0.94–1.25). Sensitivity analyses yielded similar findings.

Conclusions: This meta-analysis integrating two independent Japanese data sources did not demonstrate a statistically significant increase in MCM risk associated with first-trimester exposure to ephedra-containing Kampo medicines. Further studies are warranted to evaluate specific malformations and long-duration exposure.

Keywords: Kampo medicine, pregnancy, congenital malformations

Introduction: Bromhexine and its active metabolite, ambroxol, are widely used mucolytic agents in Japan, yet clinical evidence during pregnancy remains limited. Although a Japanese prospective cohort study showed no clear risk of major congenital malformations (MCMs) for ambroxol, a case-control study for bromhexine suggested a potential safety signal of specific malformations. Further evaluation of these medications in pregnancy is essential.

Aims: This study evaluated the association between first-trimester bromhexine/ambroxol exposure and MCMs in infants. We aimed to provide robust safety evidence using a large claims database.

Methods: We conducted a cohort study using a JMDC claims database from 2012 to 2023. The study population consisted of mothers diagnosed with a common cold during the first trimester and their singleton live-birth infants. MCMs were defined based on the European Surveillance of Congenital Anomalies (EUROCAT) definition, mapping to International Classification of Diseases, 10th revision codes. With propensity score–based overlap weighting and log-binomial regression, we estimated the weighted risk ratio (wRR) comparing exposed and unexposed groups.

Results: We identified 32,665 pregnant women with a common cold during the first trimester. Of these, 3,300 were exposed to bromhexine/ambroxol during the first trimester. The prevalence of overall MCMs in infants was 4.8% (157/3,300) among exposed women and 5.1% (1,483/29,365) among unexposed women. Exposure was not significantly associated with overall MCMs (wRR 0.942, 95% confidence interval: 0.802–1.106). Results for overall MCMs were consistent across multiple sensitivity analyses.

Conclusions: First-trimester exposure to bromhexine/ambroxol was not associated with an increased risk of overall MCMs in infants. While these findings did not suggest an increased risk broadly, potential organ-specific associations warrant cautious interpretation and further investigation.

Keywords: bromhexine/ambroxol, pregnancy, congenital malformations

Introduction: Medication safety assessment during pregnancy and lactation requires consideration of maternal treatment needs and fetal, neonatal, and child outcomes. Because interventional studies in pregnant women are ethically constrained, real-world evidence plays a central role in this field. However, the landscape of perinatal pharmacoepidemiology in Japan remains undercharacterized.

Aims: We aimed to map Japanese perinatal pharmacoepidemiology and pharmacovigilance studies evaluating medication-related perinatal outcomes and identify their data sources, exposures, outcomes, and methodological challenges.

Methods: We conducted a scoping review of comparative pharmacoepidemiological studies on medication exposure and perinatal outcomes in Japan. PubMed identified 2,134 records; after excluding 416 non-human studies, 1,717 records were independently screened by two researchers. Reviews, interventional studies, case reports, non-medication exposure studies, and single-site studies were excluded. We included 52 PubMed-derived studies and 11 additional studies identified through hand-searching, yielding 63 studies.

Results: Included studies used diverse data sources, including claims databases, teratology information service records, birth cohorts, spontaneous reporting databases, registries, medical records, and questionnaires. The most common exposure category was neuropsychiatric medications, followed by obstetric medications, antibiotics, immunomodulatory medications, and Kampo medicines. First-trimester exposure was the most frequently evaluated exposure window. Outcomes included congenital anomalies, miscarriages, preterm births, low birth weight, child developmental problems, and maternal adverse events. Across studies, we observed heterogeneity in how methodological issues were assessed and reported, including estimation of pregnancy timing and exposure windows, approaches to mother-child linkage, consideration of live-birth bias, handling of confounding by indication, and the validity of pregnancy identification, exposure assessment, and outcome ascertainment.

Conclusions: Studies on perinatal medication safety in Japan have developed across multiple data sources, but methodological approaches and reporting remained heterogeneous. Strengthening exposure timing assessment, linkage methods, bias considerations, confounding control, and validation practices will improve evidence generation.

Keywords: Pregnancy and lactation; perinatal pharmacoepidemiology; medication safety

Introduction: Medication error (ME) can lead to harm in patients and medication safety is part of Malaysian Patient Safety Goal. The Medication Error Reporting System (MERS), a voluntary reporting system, was established for healthcare professionals in both public and private health sectors to report information on the occurrence of ME in their respective settings.

Aims: Our study aims to describe the frequency and characteristics of medication error reports, focusing on the types, outcomes, and trends.

Methods: This is a retrospective cross-sectional study using data from the MERS database from 2013 to 2023.

Results: Out of 416354 reports, over 90% were from the public health facilities. The volume of ME reports rose from a baseline of 8422 in 2013 to 65767 by 2023, indicating a consistent annual growth of 100-200% over the decade. While the proportion of the total ME reports between hospital and clinic were comparable, trend analysis showed an upward trend in the annual number of reports from the clinic between 2013 and 2023. Most MEs were reported by pharmacists (95%) and occurred in clinics (47%), followed by pharmacy (26%), and ward (15%). More than two-thirds of MEs occurred at the prescribing process, 20% during medication preparation, while 4% took place during data entry. Approximately 4% of the errors reached the patients, with incorrect medication(s) being consumed in over 50% of those cases. Most events were harmless (category A-D, 99%) and 0.8% caused harm or death. Potential causes of MEs were attributed to the work environment and staff factors.

Conclusion: The MERS database provides extensive insights into the occurrence and nature of MEs which is essential for improving service delivery and clinical practices. Enhancing multi-sector reporting and implementing proactive preventive strategies are vital for reducing medication errors and improving patient safety.

Keywords: medication error, medication error reporting system

Introduction: Individuals with type 2 diabetes are at increased risk of infection-related hospitalisations and death. Evidence suggests sodium-glucose cotransporter-2 (SGLT2) inhibitors may influence infection-related outcomes, although findings remain inconsistent.

Aims: To assess the association between initiation of SGLT2 inhibitors and infection-related hospitalisations among individuals with type 2 diabetes following hospital discharge.

Methods: We conducted an active comparator new-user cohort study using linked administrative healthcare data from Victoria, Australia. Adults aged ≥30 years with type 2 diabetes discharged from hospital between July 2013 and June 2018 who initiated SGLT2 inhibitors or DPP4 inhibitors were included. Individuals were followed from treatment initiation until death or 30 June 2018. Weighted incidence rate ratios (wIRRs) and 95% confidence intervals (CIs) were estimated using negative binomial regression with inverse probability of treatment weighting.

Results: The study included 15,717 new users of SGLT2 inhibitors and 27,006 new users of DPP4 inhibitors. New users of SGLT2 inhibitors had lower overall rates of infection-related hospitalisations compared with DPP4 inhibitor users (wIRR 0.74, 95% CI 0.71–0.77). Lower rates were also observed for sepsis (wIRR 0.60, 95% CI 0.54–0.66), pneumonia (wIRR 0.61, 95% CI 0.55–0.66), gastrointestinal infections (wIRR 0.77, 95% CI 0.70–0.85), urinary tract infections (wIRR 0.38, 95% CI 0.34–0.43), respiratory tract infections (wIRR 0.68, 95% CI 0.63–0.74), influenza (wIRR 0.63, 95% CI 0.48–0.83), and kidney infections (wIRR 0.49, 95% CI 0.34–0.71). Higher hospitalisation rates were observed for osteomyelitis (wIRR 1.14, 95% CI 1.02–1.27), foot infections (wIRR 1.25, 95% CI 1.19–1.31), and genital mycotic infections (wIRR 1.48, 95% CI 1.24–1.76). No statistically significant differences were observed for cellulitis-related hospitalisations (wIRR 1.04, 95% CI 0.98–1.10).

Conclusions: Initiation of SGLT2 inhibitors after hospital discharge was associated with lower overall infection-related hospitalisations among individuals with type 2 diabetes. However, associations varied according to infection type.

Keywords: SGLT2 inhibitors; infection-related hospitalisations; type 2 diabetes.

Introduction: Nephrotoxicity has been described with prolonged aminoglycoside use, but is considered low-risk with single-dose empirical therapy. However, limited data exists in patients with chronic kidney disease (CKD).

Aims: To determine the risk of acute kidney injury (AKI) from single-dose aminoglycoside (SDA) use in emergency departments (ED) in patients with and without CKD.

Methods: A retrospective cohort study was performed in four tertiary hospitals within an Australian metropolitan network. Electronic medical records of all adult patients who presented to ED and received SDA between August 2019 and July 2024 were screened for inclusion. The primary outcome was AKI (creatinine increase by 26.5micromol/L from baseline) incidence within one week of SDA. CKD was defined as glomerular filtration rate <60mL/minute.

Results: Overall, 6636 aminoglycoside administrations were identified and 1835 screened. In total, 660 patients were included, receiving gentamicin (74.1%), tobramycin (25.6%), or amikacin (0.3%). Patients with CKD were older (81.3 vs 64.1 years old, p<0.001) and had more comorbidities. AKI incidence within one week of hospitalisation was higher in patients with vs without CKD (30.9% vs 13.4%, p<0.001). CKD was associated with higher likelihood of AKI incidence (OR 2.88 95% CI 1.76-4.71, p<0.001). Patients with CKD had more severe AKI (stage 2: 10.1% vs 2.6%, p<0.001) and a trend towards higher mortality (7.6% vs 4.4%, p=0.082) and intensive care unit admissions (12.9% vs 9.9%, p=0.222) compared to patients without CKD.

Conclusions: Patients with CKD who received SDAs were more likely to have AKI compared to patients without CKD. Clinicians should consider reserving aminoglycoside use for patients with CKD only when the benefit outweighs the risk, such as in cases of septic shock or for patients with high risk of multidrug-resistant pathogens.

Key Words: Aminoglycosides, safety, acute kidney injury

Introduction: Opioids are associated with increased risks of falls immediately following initiation. Clinical trials suggest sustained-release (SR) tapentadol has fewer adverse nervous system effects vs controlled-release (CR) oxycodone, but real-world safety is still unclear.

Aims: To compare risks of falls at 7, 14, and 28 days following initiation of tapentadol (SR) versus oxycodone (CR) in real-world clinical practice.

Methods: We conducted an active comparator, new user cohort study using routinely collected data from New South Wales, Australia (2014-2020). We included adults initiating tapentadol (SR) or oxycodone (CR). Our outcome was a composite measure of fall-related emergency department presentations, hospitalisations or deaths. We used propensity score matching to adjust for confounding and calculated relative risks (RR), absolute risk differences (RD), and numbers-needed-to-harm (NNH). We conducted subgroup analyses among people aged ≥65 and ≥80 years and by recent opioid exposure.

Results: We identified 103,924 tapentadol and 419,732 oxycodone initiators (103,758 matched). 74% were aged between 45-84 years; slightly more than half were female. Within 28 days of initiation, 652 (0.6%) oxycodone and 457 (0.4%) tapentadol initiators experienced a fall. Across all time points, tapentadol was associated with lower fall risks vs oxycodone (7 days: RR 0.57 [95% CI 0.48 to 0.69]; 14 days: 0.62 [0.54 to 0.71]; 28 days: 0.70 [0.62 to 0.79]). Absolute differences were small (RD range: -0.001 to -0.002; NNH range: 739 to 529). These patterns persisted regardless of recent opioid exposure or age. The greatest differences were among opioid-naive people aged ≥80 years (RD range: -0.006 to -0.01; NNH range: 173 to 97).

Conclusions: Tapentadol (SR) was associated with a lower risk of falls than oxycodone (CR) within 28 days of initiation, although absolute differences were small. This may be an important consideration for older patients where the consequences of falls are most severe.

Keywords: tapentadol, oxycodone, falls

Introduction: Opioids are commonly used for pain management after abdominal hysterectomy. Opioid-associated postoperative nausea and vomiting (O-PONV) can worsen patient experience and prolong hospital stay. Despite the well-established opioid-sparing effects of multimodal analgesic strategies, dedicated safety studies evaluating the incidence of opioid-induced nausea and vomiting according to the type of opioid therapy are scarce.

Aims: This systematic review and meta-analysis aimed to evaluate the incidence of O-PONV by analgesic strategy. The risk of O-PONV was compared between opioid monotherapy and multimodal regimens for postoperative pain management after abdominal hysterectomy.

Methods: A systematic search of PubMed, EMBASE, the Cochrane Library, and RISS was conducted in March 2026 according to PICO-based criteria. Eligible studies were randomized controlled trials (RCTs) of patients undergoing abdominal hysterectomy that compared postoperative opioid-based analgesic regimens and reported PONV within 24 hours. Risk of bias was evaluated using the Cochrane RoB 2 tool. Odds ratios and 95% confidence intervals were pooled using a random-effects model in Comprehensive Meta-Analysis software, and heterogeneity was assessed by I2 statistic.

Results: A total of 10 RCTs involving 711 patients were included, comparing morphine monotherapy with morphine-based multimodal regimens. In the overall risk of bias assessment, seven studies were rated as low risk and three as having some concerns. One study with zero events in both groups was excluded from the pooled OR calculation. Morphine monotherapy showed a significantly higher risk of nausea and vomiting (OR, 1.545; 95% CI, 1.114–2.143; p = 0.009). No statistical heterogeneity was observed among the included studies (I² = 0%).

Conclusions: Morphine monotherapy had a significantly higher risk of O-PONV than multimodal regimens after abdominal hysterectomy. These findings suggest that multimodal analgesia may help optimize postoperative pain management by reducing opioid-associated adverse effects.

Keywords: Abdominal hysterectomy; Opioid analgesics; Postoperative nausea and vomiting

Objective:To investigate the correlation between linezolid trough concentration (Cmin) and elevated lactate levels, analyze the risk factors and provide evidence for rational clinical use of linezolid and monitoring of adverse reactions.Methods: Clinical data of 120 critically ill patients treated with linezolid were retrospectively collected. Patients were divided into three groups based on Cmin: <2.0, 2.0-8.0, and >8.0μg/mL. The distribution and influencing factors among the three groups were compared. Patients with lactate >2.0mmol/L were assigned to the elevated lactate group, and those with lactate ≤2.0mmol/L to the non-elevated lactate group. Univariate and multivariate logistic regression analyses were performed, and a nomogram of influencing factors was constructed.Results:In the high-concentration group (>8.0μg/mL), the comorbidity index, age, APACHE II score, SOFA score, serum creatinine, blood urea nitrogen, total bilirubin (TBIL), and direct bilirubin (DBIL) were significantly increased, while estimated glomerular filtration rate (eGFR), albumin, and hemoglobin were decreased. The incidence of hypocalcemia was 59%, and the proportions of urinary tract infections and enterococcal infections were higher (P<0.05). Multivariate regression analysis showed that concentration groups of 2.0-8.0μg/mL (OR=5.5, 95%CI: 1.1–28) and >8.0μg/mL (OR=7.3, 95%CI: 1.4-38), as well as AST (OR=1.0, 95%CI: 1.001-1.022), were independent risk factors, while staphylococcal infection (OR=0.38, 95%CI: 0.15-0.92) was a protective factor. The area under the curve (AUC) of the nomogram was 0.79 (95%CI: 0.71–0.87).Conclusion:Elevated linezolid trough concentration (>8.0μg/mL) is closely associated with televated lactate levels. Monitoring of trough concentration and lactate levels, along with timely dose adjustment in high-risk patients, is recommended in clinical practice.

Introduction: Direct oral anticoagulants (DOACs) have been progressively adopted as an alternative to low-molecular-weight heparin (LMWH) for cancer-associated thrombosis (CAT). However, evidence regarding bleeding outcomes remains inconclusive, with inconsistent definitions of major bleeding across real-world studies. Moreover, head-to-head comparisons of individual DOACs against LMWH remain sparse.

Aims: To conduct a cohort study comparing the safety and effectiveness of DOACs versus LMWH in patients with CAT using territory-wide electronic health database in Hong Kong.

Methods: New users of LMWH or DOACs following a diagnosis of CAT were identified from Jan 1, 2010 to Dec 31, 2023. Primary outcomes included major bleeding, severe major bleeding, recurrent venous thromboembolism (VTE), and all-cause mortality. Outcome risks were compared using inverse-probability-of-treatment-weighting (IPTW) Cox proportional hazards model and Fine-Gray subdistribution hazard model.

Results: Among 7,700 patients, 6,532 (84.8%) were treated with LMWH and 1,168 (15.2%) with DOACs. After IPTW adjustment, DOAC was associated with significantly lower risk of all-cause mortality at both 6 months (HR 0.56, 95% CI 0.50-0.63) and 12 months (HR 0.57, 95% CI 0.51-0.63) compared with LMWH. Risks of recurrent VTE (6-month: HR 0.89, 95% CI 0.63-1.24; 12-month: HR 0.85, 95% CI 0.62-1.16), major bleeding (6-month: HR 1.01, 95% CI 0.78-1.30; 12-month: HR: 0.92, 95% CI 0.73-1.17) and severe major bleeding (6-month: HR 1.05, 95% CI 0.77-1.43; 12-month: HR 0.93, 95% CI 0.69-1.26) were comparable between groups. In agent-specific analyses, apixaban was associated with lower risk of VTE (HR: 0.55, 95% CI 0.35-0.87) and DVT recurrence (HR: 0.57, 95% CI 0.32-1.00) compared with LMWH.

Conclusions: DOACs demonstrated a safety profile comparable to that of LMWH for both major and severe major bleeding, while conferring a significant mortality benefit. Our findings support DOACs, particularly apixaban, as a safe and effective alternative to LMWH for managing CAT.

Keywords: cancer-associated thrombosis; direct oral anticoagulants; low-molecular-weight heparin

Introduction: Intra-articular (IA) injections are widely used to manage knee osteoarthritis (KOA), but septic arthritis is a rare, serious complication with limited population-based evidence on its incidence.

Aims: To estimate septic arthritis incidence after IA injections for KOA by injection type and cumulative injection exposure.

Methods: We conducted a retrospective cohort study using a nationwide administrative healthcare claims database covering 2011–2024. Patients diagnosed with KOA at secondary or tertiary healthcare institutions who received their first IA injection between 2018 and 2023 were included. Patients with septic arthritis before the initial KOA diagnosis were excluded. The first IA injection during 2018–2023 was defined as the index date. To ensure adequate outcome ascertainment, patients were required to have at least one year of potential follow-up after the index date. Follow-up continued until septic arthritis occurrence or end of follow-up. Incidence rates (IRs) were calculated per 100,000 person-years (PY).

Results: Overall, 70,033 patients contributed 418,727 person-years of follow-up. Median follow-up duration was 2,434 days for patients with index injections in 2018, decreasing to 579 days among those with index injections in 2023 (2019: 2,038; 2020: 1,657; 2021: 1,303; 2022: 930 days). Septic arthritis occurred in 1,065 patients, corresponding to an overall IR of 254.3 per 100,000 PY. IRs were 228.8, 292.1, and 283.2 per 100,000 PY for hyaluronic acid (HA)-only, corticosteroid (CS)-only, and both HA/CS injection users, respectively. By cumulative IA injection exposure, IRs were 323.1, 350.9, 273.9, and 184.5 per 100,000 PY among patients with 1, 2–3, 4–6, and ≥7 IA injections.

Conclusions: Septic arthritis occurred at 254.3 per 100,000 PY after IA injections for KOA. Higher rates among corticosteroid-only users warrant further investigation of risk factors after injection therapy.

Keywords: Knee osteoarthritis; Intra-articular injection; Septic arthritis

Background: Metabolic dysfunction-associated steatotic liver disease (MASLD) affects over 39% of the global population and drives a worldwide rise in advanced chronic liver disease (ACLD). Preclinical studies have indicated that finasteride, a first-line therapy for benign prostatic hyperplasia (BPH), may promote hepatic steatosis, hepatitis, and fibrosis. However, population-based evidence remains limited.

Aims: To investigate the association between finasteride and risks of: (1) incident MASLD among non-MASLD patients, (2) incident ACLD among non-MASLD patients, and (3) incident ACLD among MASLD patients.

Methods: Employing the clone-censor-weight (CCW) approach, this target trial emulation study analyzed data from the Exchange Xiamen PLatform Of Resident E-health Records (EXPLORER; 2010-2025) in China. Men aged 40-80 years newly diagnosed with benign prostatic hyperplasia (BPH) were included and classified by baseline MASLD status. The intervention was finasteride initiation within a 30-day grace period after BPH diagnosis. The CCW method was implemented via R package “survivalCCW”. Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated using Cox models. Multi-state Markov (MSM) models were applied to evaluate the impact of finasteride on transitions between normal liver, MASLD, and ACLD states among non-MASLD patients.

Results: 57,942 non-MASLD patients (12,173 on finasteride) and 9,851 MASLD patients (1,546 on finasteride) were included. Among non-MASLD patients, compared with non-users, finasteride use was associated with an increased risk of incident MASLD (HR = 1.18 [95% CI 1.10-1.27]) and ACLD (1.47 [1.17-1.85]). However, among MASLD patients, compared with non-users, finasteride use was not significantly associated with ACLD progression. MSM models further showed that finasteride use increased the risk of transition from normal liver to MASLD (1.14 [1.02-1.27]) and ACLD (1.70 [1.10-2.63]), but not from MASLD to ACLD.

Conclusions: Finasteride use was associated with an increased risk of MASLD and ACLD development among non-MASLD patients, but may not appear to affect ACLD progression among MASLD patients.

Introduction: Sorafenib is used for advanced hepatocellular carcinoma, but treatment-related adverse events may reduce tolerability and impair health-related quality of life. Real-world evidence from resource-limited settings remains limited. Understanding early toxicity and quality-of-life changes may support better treatment monitoring and supportive care.

Aims: To assess sorafenib-related adverse events and changes in health-related quality of life during the first two months of treatment.

Methods: A prospective observational study was conducted from March to November 2024 at a tertiary-care hospital in Rawalpindi, Pakistan. Adults with advanced hepatocellular carcinoma receiving sorafenib were followed at baseline, one month, and two months. Adverse events were assessed using the Common Terminology Criteria for Adverse Events, and quality of life was evaluated using the EORTC QLQ-C30 and QLQ-HCC18 instruments. Paired changes in adverse events were analysed using Wilcoxon signed-rank tests.

Results: Of 105 enrolled patients, 101 completed the first follow-up and 81 completed two months. At month two, common adverse events were weight loss (34.6%), diarrhoea (33.3%), hand-foot skin reaction (27.2%), abdominal pain (27.2%), fatigue (25.9%), and rash/desquamation (21.0%). Weight loss increased from 14.8% at month one to 34.6% at month two (p=0.003), while nausea decreased from 54.3% to 33.3% (p=0.002). Hand-foot skin reaction increased from 13.6% to 27.2% (p=0.054). The mean QLQ-C30 index score worsened from 37.58 at baseline to 53.02 at month one, then improved to 27.47 at month two. Corresponding QLQ-HCC18 scores were 30.44, 40.15, and 28.38. Financial-difficulty scores remained elevated at month two, indicating continued treatment-related economic burden.

Conclusions: Sorafenib was associated with substantial early toxicity and temporary deterioration in quality of life. Early monitoring, nutritional support, skin-toxicity management, and financial counselling may improve treatment tolerability and continuity. Findings should be interpreted cautiously because of attrition.

Keywords: Sorafenib; pharmacovigilance; quality of life

Introduction: Diabetic neuropathy (DN) is a common complication of diabetes associated with oxidative stress, neuroinflammation, cognitive impairment, and neuronal degeneration. Current therapies mainly provide symptomatic relief and fail to target the underlying pathology. Ocimum sanctum (Tulsi), rich in flavonoids and phenolic compounds, possesses antioxidant and neuroprotective properties that may be beneficial in DN management.

Aims: To evaluate the protective effects of aqueous extract of Ocimum sanctum against HFD/STZ-induced diabetic neuropathy and associated cognitive deficits in mice.

Methods: LC–MS/MS characterization of Tulsi extract was performed to identify phytoconstituents. Diabetic neuropathy was induced in C57BL/6 mice using high-fat diet for 10 weeks and streptozotocin (35 mg/kg, i.p.). Tulsi extract (250 mg/kg, p.o.) was administered for the final two weeks. Behavioural assessments included Morris Water Maze, hot plate, and acetone drop tests. Biochemical analyses included blood glucose, lipid profile, antioxidant enzymes, acetylcholinesterase activity, and inflammatory cytokines. Histopathological analyses of sciatic nerve and intestine were also performed.

Results: LC–MS/MS analysis identified flavonoids and phenolic acids including luteolin, kaempferol, and dihydroferulic acid derivatives. Tulsi significantly reduced blood glucose levels (170.08±3.34 mg/dL) compared to disease control (330.1±6.71 mg/dL). Escape latency time decreased from 34.71±2.51 s to 15.58±0.84 s, while time spent in target quadrant increased from 5.91±0.44 s to 14.45±0.72 s. Tulsi elevated SOD (8.48±0.54 vs. 5.51±0.30), GSH (15.82±1.01 vs. 3.54±0.10), and CAT levels (27.83±1.00 vs. 15.02±1.39), while reducing TNF-α levels (1240.92±36.36 vs. 2346.02±61.66 pg/mL). Histopathology showed reduced neuronal degeneration and intestinal inflammation.

Conclusion: Tulsi demonstrated significant neuroprotective, antioxidant, anti-inflammatory, and hypoglycemic effects against diabetic neuropathy and cognitive impairment, suggesting its potential as a multitargeted therapeutic strategy.

Keywords: Diabetic neuropathy; Ocimum sanctum; Cognitive impairment

Introduction:
Parkinson’s disease (PD) is a progressive neurodegenerative disorder marked by dopaminergic neuron loss, oxidative stress, and mitochondrial dysfunction, leading to debilitating motor & non-motor symptoms. Standard interventions provide symptomatic relief but do not arrest disease progression. Natural compounds have garnered attention as multi-target therapeutics due to their pharmacological properties and low toxicity. Piperine, alkaloid from Piper nigrum (black pepper), exhibits antioxidant, anti-inflammatory, & anti-apoptotic effects. Despite growing evidence of its therapeutic benefits, exact mechanism of Piper nigrum in PD remains unclear therefore, targeting molecular pathways might be promising therapeutic potential for disease management & progression.

Aim
This study aimed to investigate the molecular targets underlying the therapeutic effects of Piper nigrum against PD through network pharmacology, molecular docking, and in vitro studies using the SH-SY5Y cell line.

Methods:
Piperine was evaluated using integrated computational & experimental approach. Its physicochemical & ADMET properties were assessed using PubChem, SwissADME, and Schrödinger QikProp. Potential 158 piperine targets and PD-associated genes were identified from SwissTargetPrediction, STITCH, GeneCards, and OMIM databases, & 142 overlapping targets were subjected to protein–protein interaction (PPI) network construction, hub gene identification, and GO/KEGG enrichment analyses. Molecular docking was performed to investigate the interaction of piperine with the potential PD targets, effects of piperine were validated in vitro using SH-SY5Y neuronal cells.

Results:
Protein-protein interaction and pathway network analysis demonstrated Piperine modulation of key signalling cascades, including PI3K-Akt, MAPK, NF-κB, and p53, alongside dopamine synapse regulation & mitochondrial quality control. GO and KEGG enrichment further supported Piperine neuroprotective roles. MTT assays in SH-SY5Y dopaminergic neuron findings validated Piperine dose-dependent cytotoxicity & demonstrated its cytoprotective effects against 6-OHDA neurotoxin-induced toxicity.

Conclusion:
Integration of computational predictions and in vitro validation highlights Piperine multi-target mechanisms, reinforcing its potential as an effective, mechanism-based neuroprotective agent for PD therapy.

Keywords: Parkinson's disease, piperine, network pharmacology.

Introduction: Gridhrasi in Ayurveda (Indian traditional system of medicine) a condition characterised by radiating pain from the lumbar region to the lower extremities, closely resembles Sciatica in modern medicine. Ayurveda, advocates herbal preparations to restore and maintain natural physiological balance. Seeds of Caesalpinia bonducella (SCB) are indicated for the Gridhrasi were evaluated for its use in Sciatica.
Objectives: This study aimed to conduct a therapeutic evaluation of the kwatha (decoction) prepared from SCB (KCB) in the management of Sciatica.
Methods: A small, single-centred, single-blind, exploratory comparative clinical trial was conducted. Thirty patients aged between 20-50 years were randomly assigned to two groups of 15 each. Group A received KCB p.o. alone, and Group B received KCB combined with lepa (an external paste formulation containing SCB) SLL for 21 days. Outcomes were assessed using subjective symptom scores for Ruk (pain), Stambha (stiffness) and Toda (pricking sensation) graded on four-point Likert scale, alongside objective measures comprising the Straight Leg Raising (SLR) test and Braggard’s test. Within-group changes were analyzed using the Wilcoxon signed-rank test, and between-group differences were compared using the Mann–Whitney U test.
Results: In Group A (KCB alone), percentage relief was 93%, 77% and 75% for stiffness, pain and pricking sensation respectively. SLR test range improved by 71% (all P<0.001). In Group B mean percentage relief was 100%, stiffness, 81% for pain, and 87% for pricking sensation; SLR test range improved by 72% (P<0.001). Group B showed numerically better outcomes for most parameters.
Conclusion: This trial provides preliminary evidence suggesting that KCB may offer benefit in Sciatica, and that adding SLL may further enhance outcomes. These findings are exploratory and should be interpreted with caution given the limited sample size, short treatment duration, absence of a placebo control, and single-centre setting.
Key words: Caesalpinia bonducella, Sciatica, Gridhrasi, SLR test.

Introduction: Sex differences in immune response and adverse drug reactions are well described, yet vaccine safety evaluations frequently aggregate outcomes, potentially obscuring important disparities in vaccine benefit–risk profiles.

Aim: To quantify sex-specific differences in adverse event following immunisation (AEFI) reporting.

Methods: Sex-specific reporting rates per 100,000 doses administered were calculated by age group and reaction type for spontaneous AEFI reports submitted to Victoria's population-based vaccine safety surveillance system (Australia) during 2020–2025 and for HPV vaccine safety surveillance data spanning two decades (2007–2025). Sex comparisons were expressed as reporting rate ratios (RR) with 95% confidence intervals.

Results: Among 52,262 AEFI reports across all vaccines, females had higher reporting rates than males (female RR 1.80, 95%CI 1.77–1.84), with the greatest disparity in mid-adulthood and diminishing after 60 years. Reproductive-health-related events accounted for only 2.2% of adult reports. Of reactions with >10 reports, 81 demonstrated ≥3-times higher rates in females compared with 8 in males; only a subset had plausible mechanistic explanations. Adult males were less likely to self-report (10.9% vs 15.4%, p<0.0001) but 25% more likely to have serious AEFI recorded (20.2%, p<0.0001), suggesting under-ascertainment of milder male experiences.
In HPV surveillance, overall reporting rates were similar between sexes (male RR 1.13, 95%CI 0.99–1.29) including serious AEFI (p=0.781). Fever (RR 1.82, p<0.001) and syncope (1.50, p<0.001) were more frequently reported in males, whereas allergic rash was more common in females (RR 0.57, p=0.003). Male preponderance of Guillain Barré syndrome (RR 2.17, p=0.596) was consistent with background incidence expectations. Reporting patterns remained stable over substantial programmatic changes.

Conclusions: Vaccine safety sex differences reflect biological, behavioural, and surveillance-related factors. Failure to disaggregate by sex may misrepresent benefit–risk in specific populations. Routine sex-stratified pharmacovigilance strengthens signal interpretation and supports evidence-based vaccination policy.

Keywords: Vaccine safety, sex differences, pharmacovigilance

Introduction: Long-term vaccine safety studies rely predominantly on observational data and are susceptible to methodological challenges like time-varying confounding. Rigorous methodological approaches are therefore essential to ensure the validity of long-term safety evidence.

Aims: To review the study designs and analytical approaches used in observational studies to assess the long-term safety of vaccines.

Methods: A systematic search was conducted across PubMed/Medline, Web of Science, and Scopus to identify observational studies evaluating vaccine-associated adverse events with long-term follow-up published between 2015 and 2025. Data extracted from the studies included key methodological characteristics, such as follow-up duration, additional vaccination during follow-up, and approaches used to address potential time-varying confounding.

Results: A total of 468 records were identified through database searches, of which 63 studies met the eligibility criteria. COVID-19 vaccines were the most frequently studied vaccine type (n=35, 55.6%), followed by human papillomavirus (HPV) vaccines (n=5, 8%). Cardiovascular outcomes were the most frequently investigated adverse events, followed by autoimmune, neurological, and thrombotic conditions. Prospective or retrospective cohort designs comprised the largest proportion of studies (n=40, 63.5%), whereas three studies (4.8%) utilized a hybrid cohort-SCCS design and one study (1.6%) utilized an SCCS design. The follow-up duration for long-term safety evaluation ranged from a minimum of 30 days to a maximum of 34 years. To control time-varying confounding, 6 studies (9.5%) identified and statistically adjusted for dynamic covariates such as vaccination status (e.g., transitions across vaccine doses), breakthrough infections, and longitudinal medication adjustments. These factors were primarily managed using time-dependent Cox proportional hazards models (n=4, 66.7%)

Conclusions: Despite the long-term periods used in these studies, explicit consideration of time-varying confounding was identified in only 6 of 63 studies. Greater transparency regarding the handling of dynamic factors during extended follow-up is needed to strengthen the long-term safety evidence.

Keywords: vaccine; long-term safety; study design

Introduction: Post-licensure vaccine effectiveness (VE) surveillance can inform immunization policy and individual decision-making. The Zhejiang Provincial Vaccine Real-World Research Platform (ZVRWP) has integrated vaccination records, electronic medical records (EMRs) and free cervical cancer screening data. However, the feasibility of using ZVRWP for VE studies has not been systematically evaluated.

Aims: To assess the feasibility of using ZVRWP to evaluate the real-world effectiveness of human papillomavirus (HPV) vaccines in China. We also aimed to inform future post-licensure VE studies using the platform.

Methods: We first evaluated accessibility, governance, and secure use processes. Then assessed the relevance of key data for effectiveness evaluation of HPV vaccines, including exposure variables from vaccination records and outcome variables (cervical intraepithelial neoplasia (CIN) 2/3, adenocarcinoma in situ (AIS) and cervical cancer) from EMRs and screening data. Data completeness and linkage performance across data sources using a unique identifier were analyzed.

Results: The ZVRWP is accessible for research use under established ethics and data security governance as a comprehensive healthcare information system. As of 2025, the platform had captured 11.5 million HPV vaccine doses among 4.2 million female recipients aged 9–45 years. In the same age group, screening data identified 3,739 cases of CIN2/3, while EMRs documented 6,326 CIN2/3 cases and 64 cases of AIS cases. Key exposure fields showed near‑complete capture (100% for vaccination date and vaccine name; 99.98% for manufacturer). Cross-source linkage demonstrated substantial overlap, including individuals with records present in vaccination, EMRs, and screening data.

Conclusion: Key exposure and outcome data required for post-licensure HPV VE surveillance are obtainable with high completeness and demonstrable linkage across data sources in the ZVRWP. The platform appears fit for conducting HPV VE surveillance studies, with further work recommended to refine cohort definitions and evaluate outcome validation procedures.

Key words: feasibility assessment; human papillomavirus; vaccine effectiveness

Background: Large linked administrative health datasets are increasingly central to population-level research and public health surveillance. Victoria’s Vaccine Safety Health Link (VSHL) links near real-time records from Australian Immunisation Register with statewide hospital, emergency, notifiable disease, perinatal, birth and death datasets. Monthly linkage by the Centre for Victorian Data Linkage provides de-identified extracts for secure analysis, enabling population-wide real-world vaccine safety monitoring. Despite strengths of this comprehensive linkage, pitfalls can trip up novice & experienced analysts alike.

Aims: To share practical challenges working with large, linked health data.

Methods: Vaccine safety analyses within VSHL (with a population >8 million), defining denominator and numerator populations, linking between parents and children, and assessing exposure–outcome associations using methods such as self-controlled case series, logistic regressions and rapid cycle analyses have been performed for four years. These works have highlighted challenges that can impact linked-spine administrative data systems.

Results: Case ascertainment shifts can be subtly impacted by differences in coding completeness, update cycles and temporal alignment across sources. At population scale, even small linkage error rates may create meaningful analytic distortion, including missed links, false matches or implausible records (e.g. multiple maternal links or improbable clusters of vaccinations [>50 in one individual] on a single day). Population continuity presents another critical fault line; when deaths or migration occur outside jurisdictional capture, individuals may persist analytically (like “immortals”) within the cohort, skewing denominators and risk estimates, particularly in older populations. Although near real-time linkage enables rapid surveillance, delays in source data availability (eg. complete admission record coding) and limited clinical granularity can temper immediate signal interpretation.

Conclusions: Linked administrative data brings great richness and representativeness but necessitates considerations when being operationalised. Sharing experience and expertise can lift capability in linked data use and strengthen Australia’s public health intelligence.

Keywords: Linked data, Vaccines

Introduction: Urinary incontinence is up to three times more prevalent among those with cognitive impairment than those without cognitive impairment. The absence of guidance regarding the pharmacological management of urinary incontinence for people living with cognitive impairment or dementia, and evidence to guide decision making, may lead to unwanted clinical practice variation and potentially avoidable medication-related harm.
Aims: To systematically review the safety and efficacy of medications for the management of urinary incontinence in people living with dementia and/or in residential aged care.
Methods: Five bibliometric databases were searched for randomized control trials (RCTs) from inception to March 2026. Randomised controlled trials that evaluated the safety and/or efficacy of a pharmacological treatment for the management of urinary incontinence versus either a placebo, another pharmacological interventions or non-pharmacological interventions among people living with dementia or cognitive impairment or those living in residential aged care were included. Risk of bias was assessed using the Joanna Briggs Institute Critical Appraisal Tool. Efficacy, adverse events and cognitive effects were synthesized narratively.
Results: Ten RCTs with a total of 325 participants were included. Effect on UI symptoms was reported in 8 studies, with oxybutynin (n=5 studies), propiverine (n=1 study), propantheline, (n=1 study) and oral oestrogen (n=1 study) reported as having minimal or no effect on UI symptoms. The most common non-cognitive adverse event reported for oral anticholinergics was dry mouth (42/142 participants). Cognitive effects on attention, alertness and memory were mostly reported within one to two hours after dosing of oral anticholinergics. Risk of bias was common across included studies due to incomplete or unclear reporting of baseline participant characteristics, trial designs and outcome measures.
Conclusions: Medications for urinary incontinence for people with cognitive impairment or dementia should be prescribed cautiously, with consideration of possible benefits and harms.
Keywords: Dementia, Incontinence, Medicine Safety

Introduction: Decentralized clinical trials (DCTs) can overcome the limitations of conventional site-based trials. As DCT utilization increases, international regulatory harmonization is crucial to address barriers in multiregional clinical trials. However, cross-national differences in DCT design and conduction guidelines remain underexamined.

Aims: This study compares regulatory approaches to DCTs between the United States(US) and Australia. Also, we aim to identify how regulatory perspectives on decentralized elements differ between these two leading clinical trial nations.

Methods: This cross-sectional study analyzed DCT guidelines issued by national regulatory authorities and clinical trial organizations, including official guidance from the US Food and Drug Administration, the Cancer Institute NSW, and the Clinical Oncology Society of Australia. We evaluated terminology, general design considerations, local healthcare provider roles, and decentralized element applications. All documents were re-searched as of May 31, 2026, to ensure regulatory data validity.

Results: Regulatory terminology differs: the US defines "DCTs," whereas Australia utilizes both "DCTs" and "Tele-trials." Regarding design, US guidelines focus on investigational product safety and procedural complexity, while Australian guidelines emphasize suitability for rural populations and community barriers. Furthermore, the US restricts local healthcare providers to routine care without informed consent authority. Conversely, Australia permits satellite physicians and local providers to act as investigators and obtain informed consent. Despite structural variations in specific decentralized elements, both countries permit direct-to-patient delivery and remote consent.

Conclusions: Although US and Australian regulatory frameworks share a patient-centric objective, they exhibit distinct operational and structural approaches. These disparities may pose substantive barriers to executing multiregional clinical trials, highlighting the essential need for globally harmonized standards to facilitate international DCT activation.

Keywords: Decentralized clinical trials, Teletrial, Regulatory frameworks

Introduction: Antimicrobial resistance (AMR) is a critical public health threat, largely driven by inappropriate antibiotic use. In Thailand, self-medication with antibiotics for upper respiratory tract infections (URIs) is prevalent, contributing to rising resistance rates. Nakhon Pathom Province consistently ranks among areas with high antibiotic use for these conditions, yet post-COVID-19 evidence on community-level knowledge, attitude, and behaviour (KAB) remains limited.

Aims: This study aimed to assess antibiotic use KAB among residents of Mueang Nakhon Pathom District and to analyse associations between these variables and relevant demographic factors.

Methods: A cross-sectional analytical study was conducted between December 2023 and January 2024. A total of 439 adults aged 18 years and older who had self-administered oral antibiotics for URIs or acute bronchitis within the preceding three months were recruited using multi-stage sampling. Data were collected via a validated self-administered questionnaire assessing demographics, antibiotic knowledge (12 items), attitude (7 items), and behaviour (8 items). Chi-square tests and Spearman's rank correlation were used for inferential analyses.

Results: Most participants demonstrated low antibiotic knowledge (61.50%), negative attitudes towards correct antibiotic use (73.12%), and low accurate antibiotic use behaviour (48.97%). Education level (p<0.05) and occupation (p<0.001) were significantly associated with behaviour. Age showed a statistically significant very weak negative correlation with behaviour (rs = −0.20, p<0.001). Knowledge demonstrated a strong positive correlation with behaviour (rs = +0.63, p<0.001), while attitude showed a weak positive correlation (rs = +0.27, p<0.001). Gender, monthly income, and antibiotic acquisition channel were not significantly associated with behaviour.

Conclusions: Knowledge was the strongest predictor of correct antibiotic use behaviour in this population. Targeted public health interventions are warranted for individuals aged 40 years and older, those without formal education, the unemployed, and those obtaining antibiotics through online channels, to promote rational antibiotic use.

Keywords: Antibiotic use behaviour; antimicrobial resistance; self-medication

Introduction: Drug lag represents a critical challenge that limits pharmaceutical competitiveness and restricts timely patient access to innovative therapies in South Korea. This issue is intrinsically linked to delays within the investigational new drug (IND) review process, which hinders the domestic introduction of advanced treatments. There is an urgent need to identify specific systemic barriers and optimize the regulatory pathway to align with global standards.

Aims: This study aimed to identify key barriers affecting IND review efficiency by analyzing real-world industry experience. Furthermore, it sought to propose concrete strategic actions to mitigate drug lag in the domestic pharmaceutical landscape.

Methods: We employed an exploratory sequential mixed-methods design. Initially, grounded theory was utilized to identify problematic factors within the IND process through semi-structured interviews. Subsequently, the Analytic Hierarchy Process was applied to evaluate these factors, followed by a quadrant analysis of importance and urgency to pinpoint top-priority tasks.

Results: Three core categories—rationality and patient-centricity, professional review, and operational efficiency—were identified as key domains for enhancing IND efficiency. Rationality and patient-centricity emerged as the most important and urgent criterion, followed by professional review. Within rationality and patient-centricity, four sub-criteria were identified in the first quadrant (highly important and urgent): integration of patient needs, global regulatory harmonization, acceptance of scientific evidence for new modalities, and adoption of overseas standards. Two sub-criteria within professional review, specifically expert team review and standardized review guidelines, were also pinpointed as top priorities. Although operational efficiency ranked lowest, simplified translation requirements were identified as the most critical factor within this category.

Conclusions: A transition from focusing solely on procedural speed toward embracing regulatory flexibility to enhance scientific rationality and patient access is essential. These findings suggest a viable pathway for effectively mitigating drug lag in South Korea.

Keywords: Drug lag, Investigational new drug, Analytic hierarchy process

Introduction: The FDA accelerated approval pathway enables earlier access to treatments for serious or life-threatening diseases based on surrogate endpoints. Although widely studied in oncology, regulatory outcomes for non-oncology therapeutics remain less well characterized. This gap is important because non-oncology indications often involve diverse diseases, heterogeneous surrogate endpoints, and varying post-marketing evidence requirements.

Aims: This study evaluated regulatory outcomes and approval timelines for non-oncology therapeutics granted FDA accelerated approval from 1992 to 2024. It also compared approval trajectories by drug type, approval type, and oncology versus non-oncology classification.

Methods: A cross-sectional analysis was conducted using the FDA accelerated approval database. Non-oncology indications approved between June 19, 1992 and December 31, 2024 were included. Indications were categorized by therapeutic area, drug type, approval type, and regulatory outcome. Median approval timelines were analyzed and compared according to drug type and oncology versus non-oncology classification.

Results: Among 328 accelerated approval indications, 98 (30%) were non-oncology indications. Overall, 63 (64%) converted to full approval, 14 (14%) were withdrawn, and 21 (21%) remained not yet converted. Systemic anti-infectives were the largest category, accounting for 49 (50%) indications. Original indications converted more frequently than supplemental indications, with 55/77 (71%) versus 8/21 (38%) converted. The median time to accelerated approval was 7.9 months, and the median time from accelerated approval to full approval was 38.8 months. Non-oncology biologics required significantly longer time to accelerated approval than oncology biologics (8.6 vs 6.0 months; p=0.003).

Conclusions: These findings suggest that regulatory trajectories for non-oncology accelerated approvals may differ by product type, therapeutic area, and post-marketing evidence requirements. Continued post-marketing evaluation is needed to confirm clinical benefit and support evidence-based regulatory decision-making.

Keywords: Accelerated approval; Non-oncology therapeutics; Regulatory outcomes