British journal of clinical pharmacology最新文献

N-of-1 trials offer a unique and rigorous methodology for evaluating individualized treatment responses, particularly within the context of personalized medicine. This article provides a comprehensive explanation of the conceptual and methodological underpinnings of N-of-1 trials, with particular emphasis on statistical techniques and considerations critical for their design and analysis. Existing guidelines for planning and conducting these studies are summarized, along with a discussion of the practical and theoretical challenges to their implementation in clinical practice. We provide an overview for clinicians and researchers who may be unfamiliar with the design. As most of the existing guidance has focused on design and implementation considerations, we expand on the statistical analysis. We aim to support researchers and methodologists in understanding and advancing the methodological toolkit necessary for high-quality N-of-1 research.

Aims: Prescribing medication is a complex process, influenced by many factors that can be weighed differently. Four prescriber profiles have been identified: pragmatic-contextual, guideline-oriented, experience-driven and vulnerability-focused. However, the extent to which European prescribers identify with these profiles and the role of these profiles in clinical pharmacology and therapeutics (CPT) education is unclear. Knowledge of this might improve prescribing education and thereby prescribing practice.

Methods: A cross-sectional survey was conducted among physicians, physician assistants, advanced nurse practitioners and prescribing pharmacists across Europe. Participants ranked how strongly they identified with the four prescriber profiles, whether they recognized them in fellow healthcare professionals and whether these were covered in CPT education.

Results: A total of 170 prescribers from 23 European countries and over 40 different medical disciplines participated in this study. Most participants identified themselves as being pragmatic-contextual (36.5%) or guideline-oriented (42.9%) prescribers, whereas few considered themselves experience-driven (11.2%) or vulnerability-focused (9.4%) prescribers. Pragmatic-contextual and guideline-oriented prescribing were covered in current CPT education, whereas vulnerability-focused prescribing was poorly covered in current education. Participants thought that future CPT education should include these prescriber profiles.

Conclusions: These findings highlight the influence of professional background, clinical setting and experience on prescribing approaches across Europe. The strong presence of pragmatic-contextual and guideline-oriented profiles in both practice and education is consistent with current CPT curricula. However, experience-based and vulnerability-focused profiles, though regularly seen in practice, are under-represented in education despite their acknowledged importance. To support the development of prescribing practice, educational strategies should include all profiles.

Current policies regarding artificial intelligence (AI) technology are facing difficulty to keep pace with its rapid development in healthcare due to the complexities involved in regulating this dynamic field. Similar to any pioneering technology, the application of AI in healthcare presents new ethical and legal dilemmas that extend beyond traditional bioethics, legislation, and governance. The process of establishing definitive professional guidelines and standardized regulations for AI in healthcare has been slow. This article examines the ethical challenges that arise at the intersection of AI and healthcare research. Key ethical principles that are essential for the responsible use of AI in healthcare include justice, data stewardship, explainability of AI, accountability and sustainability. We will discuss the ethical considerations that may arise at each stage of the AI development lifecycle, from early problem identification to post-deployment evaluation.

Aims: Tranexamic acid (TXA) stabilizes clot formation by inhibiting fibrin degradation and improves postoperative outcomes. However, rare adverse events (e.g., thrombosis, seizures) warrant dose-risk evaluation. This study examines how perioperative blood loss and transfusion practices affect TXA concentrations during paediatric scoliosis surgery.

Methods: Forty-three patients undergoing scoliosis surgery with TXA were retrospectively analysed. The study assessed the impact of packed red blood cell (PRBC) transfusion on plasma TXA levels and whether maintaining concentrations ≥10 μg/mL correlated with intraoperative blood loss. TXA levels were measured using UHPLC-MS/MS.

Results: Median TXA concentration 30 min after the loading dose was 37.8 μg/mL (IQR: 31.4-39.6 μg/mL), decreasing to 10.6 μg/mL (IQR: 9.7-13.5 μg/mL) after transfusion. At surgery end, the mean concentration was 12.9 ± 2.5 μg/mL. Thirty-one patients maintained TXA levels ≥10 μg/mL, associated with ~80% inhibition of tissue plasminogen activator. Of six patients below this threshold, five had received transfusions. A significant correlation was found between higher intraoperative blood loss and lower TXA levels, consistent with a dilutional effect. In contrast, among patients with TXA ≥ 10 μg/mL, correlation with blood loss was weak (Spearman's ρ = -0.11, p = 0.54). Findings suggest homologous PRBC transfusion reduces plasma TXA through volume expansion.

Conclusions: Sustaining TXA concentrations >10 μg/mL is essential for antifibrinolytic efficacy and haemostatic outcomes. The dilutional impact of PRBC transfusion underscores the need for intraoperative dose adjustment. Optimizing TXA dosing requires understanding pharmacokinetics and patient variability.

Aims: Post-mortem detection of a medicine following suicide can be due to two main reasons: the decedent was taking that medicine therapeutically before death, and/or the medicine was involved in the suicidal act (poisoning-related suicide). We aimed to investigate how antidepressant concentrations differed between poisoning and non-poisoning suicides. We hypothesized that the predictive value of these concentrations and the separation between poisoning and non-poisoning concentrations would improve by adjusting for dose dispensed to the decedent.

Methods: We analysed post-mortem toxicology results from suicides in Australia, July 2013 to October 2019, linked to the individual's dispensing history. Suicides were classified as poisoning- or non-poisoning-related by coroners. We analysed the distribution of concentrations through descriptive statistics, precision-recall curves and quantile regression to compare poisoning and non-poisoning concentrations. We adjusted concentrations by estimated daily dose and total drug quantity dispensed in 90 days and re-assessed model performance.

Results: We had sufficient sample size to analyse nine antidepressants: amitriptyline (n = 149), mirtazapine (n = 399), citalopram (n = 116), escitalopram (n = 297), fluoxetine (n = 183), sertraline (n = 253), duloxetine (n = 122), venlafaxine (n = 261), desvenlafaxine (n = 194). Selective Serotonin Reuptake Inhibitor non-poisoning and poisoning concentrations were similar, with no high certainty threshold for poisoning for citalopram and sertraline. Amitriptyline had the best separation between poisoning and non-poisoning concentrations. Adjustment by estimated daily dose improved the separation of lower quantiles through quantile regression but did not help identify thresholds that separated poisonings and non-poisonings.

Conclusions: Dose adjustment generally did not improve the separation of poisoning vs non-poisoning suicides, indicating that post-mortem concentrations may not have clear dose-concentration relationships.

Aims: Incorrect antibiotic dosing remains common, particularly in patients with renal impairment or obesity, contributing to treatment failure and antimicrobial resistance. Although clinical decision support systems (CDSS) are increasingly used, few offer real-time, patient-specific dosing recommendations based on a curated antibiotic dosing database integrated directly into prescribing workflows, limiting the optimization of antimicrobial therapy.

Methods: We implemented a mandatory CDSS for restricted antibiotics at a 1100-bed tertiary hospital. The system automatically integrated renal function and body weight from the electronic medical record, along with prescriber-selected dialysis modality, to generate predefined, guideline-based dosing options tailored to individual clinical scenarios. Prescribers were required to either accept a recommended dose or provide justification for overriding it. We retrospectively analysed 81 164 prescriptions after implementation to assess CDSS-adherence and compared 41 550 pre- and 33 994 post-implementation prescriptions to evaluate dosing accuracy.

Results: CDSS-adherence increased from 62.2% to 92.6% during the post-implementation period, while the incorrect antibiotic dosing rate declined from 3.7% to 2.1% (P < 0.001). Among non-adherent prescriptions, 72.4% of manually entered doses were concordant with CDSS recommendations and were associated with a lower incorrect dosing rate than discordant overrides (0.5% vs. 9.7%; P < 0.001). Key system refinements, including support for inhaled colistin and agent-specific loading dose guidance, further improved usability and CDSS-adherence.

Conclusions: By integrating real-time, patient-specific parameters into the prescribing process and offering guideline-aligned recommendations, this CDSS significantly improved prescriber adherence and reduced incorrect antibiotic dosing. The system shows strong potential to improve medication safety and support antimicrobial stewardship.

Predicting adverse drug events (ADEs) in outpatient settings is crucial for improving medication safety, identifying high-risk patients and reducing health-care costs. While traditional methods struggle with the complexity of health-care data, machine learning (ML) models offer improved prediction capabilities; however, their effectiveness in ADE prediction remains unclear. This systematic review evaluated ML algorithms used for this purpose, analysing studies that focussed on outpatient care or utilized large-scale data sources (e.g. electronic health records, administrative claims and spontaneous reporting systems) that primarily represent the outpatient continuum. We systematically searched MEDLINE and Embase up to December 2024 to identify studies developing or validating ML models for ADE prediction. Study characteristics, ML methods, ADE types, model performance and risk of bias were assessed using the PROBAST tool. From 59 included studies comprising 191 ML implementations, Logistic regression, Random forest and XGBoost emerged as the most commonly used algorithms. The majority of studies (67.8%) reported area under the curve (AUC), with 85% demonstrating moderate to high performance (AUC > 0.70) for internal validation. However, only 33.9% of studies addressed class imbalance, and merely 18.6% conducted external validation, raising concerns about methodological rigour, particularly in missing data handling and validation procedures. Our findings indicate that ML models, especially ensemble methods, show promise in predicting ADEs, although challenges with class imbalance and limited external validation currently hinder their clinical applicability. Future research should focus on adopting more rigorous methodologies and developing specialized frameworks for ML-based ADE prediction that build upon established pharmacovigilance practices to ensure models are accurate, generalizable, and seamlessly integrated into clinical workflows for ongoing monitoring and improved medication safety.

Background: This study aimed to explore the 90% effective dose (ED90) of remimazolam in paediatric same-day bidirectional endoscopy (BDE) and provide a more scientific dosage selection for the anaesthetic induction application of remimazolam in children.

Methods: The children were divided into a preschool (PS) group and a school-age(S) group, the ED90 of remimazolam was determined using the biased-coin up-and-down sequential test, the initial dose of remimazolam was 0.4 mg/kg in the PS group, 0.3 mg/kg in the S group, with a dose interval of 0.02 mg/kg. And remifentanil 5 μg/kg was combined as the analgesic drug.

Results: A total of 110 children were included in the study, 55 in each of the two groups. In the PS group, the dose range of remimazolam was 0.38-0.68 mg/kg, and the ED90 and ED50 were 0.67 mg/kg and 0.60 mg/kg using Centered Isotonic Regression (CIR); in the S group, the dose range of remimazolam was 0.30-0.48 mg/kg, the ED90 and ED50 was 0.45 mg/kg and 0.37 mg/kg using CIR. No serious adverse effects of remimazolam were seen in both groups.

Conclusions: In the induction of anaesthesia for same-day BDE, combined with 0.5 μg/kg of remifentanil, the younger the age, the higher the dose of sedative medication required. The ED90 of remimazolam was 0.67 mg/kg [90% confidence interval (CI), 0.66-0.76] in preschool children and 0.45 mg/kg (90% CI, 0.42-0.59) in school children.

Clinical trials: NCT06121609, date of registration: Nov. 07, 2023.

Aim: To evaluate quality use of medicines (QUM) in patients admitted to the Psychiatric Alcohol and Non-Prescription Drug Assessment (PANDA) Unit through analysis of polypharmacy risk, prescribing safety indicators and national QUM indicators.

Methods: Retrospective review of electronic medical records for 787 patients (1245 episodes) admitted to PANDA Unit, St Vincent's Hospital Sydney, November 2020-December 2021. We assessed polypharmacy risk using NSW Therapeutic Advisory Group's Inappropriate Polypharmacy Risk Assessment Tool, evaluated nine prescribing safety indicators (PSIs) and assessed four national quality use of medicines indicators (NQUMIs) in a randomly selected subset of 103 patients.

Results: Mean age was 40.2 years; 67.2% were male. Alcohol-related diagnoses comprised 32.1% of presentations. Medium-risk (43.5%) and high-risk (37.2%) polypharmacy were prevalent, largely reflecting protocol-driven prescribing for acute presentations. Co-prescription of QT-prolonging drugs with antipsychotics affected 17.2% of patients, with 64.7% associated with rapid sedation protocols; only one case of potential QT prolongation was documented. Major gaps were identified in medication reconciliation (40.8% had documented medication history) and discharge documentation (13.1% of discharge summaries contained complete medication information). Patients experiencing homelessness demonstrated poorer outcomes across all QUM indicators, with no discharge summaries meeting quality standards.

Conclusion: While PANDA demonstrated appropriate protocol-driven prescribing for acute psychiatric presentations, substantial gaps in medication reconciliation and discharge communication highlight challenges in care transitions for priority populations. Standard polypharmacy and prescribing safety indicators may require adaptation for acute psychiatric settings to distinguish appropriate acute prescribing from potentially inappropriate practices.