Abstract Purpose To evaluate the impact of multimodal interventions, including electronic cross-reactivity alert suppression, clinician education, and implementation of a β-lactam allergy guideline, on β-lactam prescribing in hospitalized adults with reported penicillin or cephalosporin allergy. Methods Single-center, retrospective pre-post study included adults (≥ 18 years) with documented penicillin or cephalosporin allergy. Patients were excluded if they received fewer than two antibiotic doses, died within 48 hours of admission, had allergies to both penicillins and cephalosporins, or received antibiotics solely for surgical prophylaxis. The primary outcome was penicillin or cephalosporin administration. Secondary outcomes included hospital length of stay, allergic reactions, Clostridioides difficile infections, and antimicrobial discontinuation due to adverse reaction. Results Among 234 patients included (pre-intervention n = 123; post-intervention n = 111), baseline demographics, Charlson Comorbidity Index, and allergy characteristics were similar between groups. Penicillin or cephalosporin use increased significantly following interventions (65% vs 81%, P = 0.006), while non-β-lactam use decreased (26.8% vs 14.4%, P = 0.020). There were no differences in duration of therapy, frequency of allergic reactions, hospital length of stay, 30-day C. difficile infection rates, or antimicrobial discontinuation rates due to allergy or intolerance. Conclusion A bundled strategy combining guideline implementation and education with removal of β-lactam cross-reactivity alerts significantly increased penicillin and cephalosporin prescribing among adults with documented penicillin or cephalosporin allergies without increasing allergic reactions or other safety events. These findings support multimodal stewardship approaches to safely optimize β-lactam utilization in hospitalized patients with reported allergies.
Abstract Purpose This primer is designed to provide a broad overview of psilocybin and its use in medically supervised settings as part of psychedelic-assisted psychotherapy. This article will act as a framework to give practicing pharmacists the knowledge they need regarding pharmacology, pharmacokinetics, contraindications, adverse effects, and drug interactions that would impact safe use of psilocybin in medical settings. Summary Psychedelic-assisted psychotherapy has re-emerged in the last twenty years as a potential pathway to address a variety of mental health disorders. Interest in psilocybin has experienced a renaissance, and there have been pushes at federal, state, and local levels to expand access to this agent in research settings and within psychotherapy offices. While generally considered a very safe therapy, psilocybin does have several contraindications and clinically relevant drug-drug interactions that need to be considered when evaluating patient candidacy for psilocybin-assisted psychotherapy. Conclusion Pharmacists play an important role in ensuring safe and effective use of psilocybin as it becomes more readily available and utilized across the globe. Recognition of relevant contraindications and drug-drug interactions is particularly essential to this process.
Abstract Purpose To systematically evaluate ChatGPT Plus (GPT-4o) and Google Gemini Pro as decision-support instruments for the crushability analysis of immediate-release (IR) oral solid dosage forms, benchmarked against a human expert reference using a validated four-question analytical checklist. Methods Sixteen IR oral solid dosage forms approved by the Food and Drug Administration (FDA) in 2024 were independently assessed for crushability (YES/NO) by a human expert team and each AI tool using an identical structured prompt. Evaluator reasoning was coded across four analytical domains. Inter-rater agreement across all three evaluators was quantified using Fleiss kappa (κ) with a 95% confidence interval (CI) and two-tailed z-test; kappa values were interpreted using the Landis and Koch scale. All computations were performed in Microsoft Excel. Results Human experts classified 11 of 16 products (68.8%) as crushable and 5 (31.3%) as not crushable. Eight products achieved unanimous three-way agreement. The Fleiss κ was 0.329 (Fair; 95% CI, 0.046-0.612; z = 2.28; p = 0.023), indicating statistically significant agreement above chance. Both tools substantially over-identified pharmacokinetic concerns relative to human experts (Gemini, 11/16; ChatGPT, 8/16; Human, 3/16), the primary driver of disagreement. Gemini's identification of QTc interval prolongation risk for mavorixafor prompted expert revision of the crushability recommendation to not crushable. ChatGPT's false positive recommendation for lazertinib, failing to detect a reactive acrylamide degradation pathway, represents the highest-risk error in the cohort. Conclusion Generative AI demonstrated statistically significant but fair overall agreement with human expert analysis. Systematic scope deviations, including out-of-scope pharmacokinetic reasoning and hallucination-adjacent misclassification, suppressed concordance and produced clinically consequential errors. These findings support AI as a structured first-pass screening tool, contingent on pharmacist-led verification, institutional governance, and heightened scrutiny for hazardous drugs and novel formulation-technology products.
Abstract Purpose This descriptive report evaluated the accuracy and response quality of ChatGPT 5.1, OpenEvidence, Clair, and a hybrid evidence-assisted ChatGPT 5.1 model in responding to drug information questions, using drug information specialist–developed reference responses as the comparator. Summary Twenty drug information questions were selected from the UNC Health Drug Information Center archive from July 2021 through July 2025, stratified across 10 categories, and classified as simple or complex. Responses were generated in November 2025 using ChatGPT 5.1, OpenEvidence, Clair, and a hybrid evidence-assisted model in which drug information specialist–curated evidence packets were provided to ChatGPT 5.1. Drug information specialist–developed reference responses included a definitive answer, evidence synthesis, a recommendation when applicable, and verifiable references. Deidentified and randomized responses were evaluated through a blinded two-phase review process assessing accuracy and response quality. ChatGPT 5.1 had the numerically highest accuracy at 70%, followed by the hybrid evidence-assisted model at 60%, OpenEvidence at 40%, and Clair at 15%. Drug information specialist–developed reference responses and OpenEvidence had the highest mean overall quality performance at 82.5% each. Conclusion AI platforms demonstrated variable accuracy and quality when responding to drug information questions. In this limited sample, ChatGPT 5.1 had the numerically highest accuracy, while OpenEvidence had the highest mean overall quality performance among AI-generated responses. Despite strengths in organization and readability, limitations in accuracy, reference identifiability, and evidence synthesis support the continued need for drug information specialist oversight.
