Performance and risks of ChatGPT used in drug information: an exploratory real-world analysis.

OBJECTIVES: To investigate the performance and risk associated with the usage of Chat Generative Pre-trained Transformer (ChatGPT) to answer drug-related questions. METHODS: A sample of 50 drug-related questions were consecutively collected and entered in the artificial intelligence software application ChatGPT. Answers were documented and rated in a standardised consensus process by six senior hospital pharmacists in the domains content (correct, incomplete, false), patient management (possible, insufficient, not possible) and risk (no risk, low risk, high risk). As reference, answers were researched in adherence to the German guideline of drug information and stratified in four categories according to the sources used. In addition, the reproducibility of ChatGPT's answers was analysed by entering three questions at different timepoints repeatedly (day 1, day 2, week 2, week 3). RESULTS: Overall, only 13 of 50 answers provided correct content and had enough information to initiate management with no risk of patient harm. The majority of answers were either false (38%, n=19) or had partly correct content (36%, n=18) and no references were provided. A high risk of patient harm was likely in 26% (n=13) of the cases and risk was judged low for 28% (n=14) of the cases. In all high-risk cases, actions could have been initiated based on the provided information. The answers of ChatGPT varied over time when entered repeatedly and only three out of 12 answers were identical, showing no reproducibility to low reproducibility. CONCLUSION: In a real-world sample of 50 drug-related questions, ChatGPT answered the majority of questions wrong or partly wrong. The use of artificial intelligence applications in drug information is not possible as long as barriers like wrong content, missing references and reproducibility remain.

Blood culture sampling rate in hospitalised children as a quality indicator for diagnostic stewardship.

PURPOSE: Recommendations regarding the optimal number of blood cultures in children are not available. The aim of this article is to describe the correlation between blood culture (BC) rates and laboratory-confirmed bloodstream infection (LCBSI) rates, on different paediatric wards of a tertiary-care centre in Germany. METHODS: We conducted a retrospective cohort study in a paediatric university hospital, from 1st January to 31st December 2018. All blood cultures collected from neonatal (NICU) and paediatric intensive-care units (PICU), haematology/oncology, and general paediatric wards were included. There were no exclusion criteria. BC taken/1000 patients-days (BC rates/BCR) and LCBSI/1000 patient-days at risk (LCBSI rates) were calculated for each unit. RESULTS: A total of 6040 patients were admitted to the hospital with 3114 of them into wards studied. Of the 3072 BCs collected, 200 (6.5%) were positive. Collection of BCs was performed in 51/77 (66.2%) of admitted patients on NICU, in 151/399 (37.8%) of PICU patients, in 163/755 (21.6%) of haematology/oncology patients, and in 281/1883 (14.9%) of children on general paediatric wards. Gram-positive bacteria were the most commonly detected organisms in blood cultures from all wards with exception of NICU. The BCR in NICU, PICU, haematology/oncology wards, and general wards were 61.6, 196.2, 358.4, and 52.3, respectively. Excluding commensal pathogens and possible contaminations, the LCBSI rates in the same units were 2.4, 5.6, 4.4, and 1.0, respectively. CONCLUSION: We found different BCR values according the ward studied, being higher in patients with high risk of bloodstream infection such as haematology/oncology patients.

Patterns of antimicrobial consumption in neonatal and pediatric intensive care units in Germany and Brazil.

Antibiotic consumption (AC) is a key component of antimicrobial stewardship programs to recognize local patterns of antibiotic use. Our aim was to measure AC in neonatal units, including neonatal (NICU)/paediatric (PICU) intensive care units in different countries. We conducted a multicenter, retrospective, cohort study in three NICUs, one neonatal ward, and three PICUs with a total of 84 beds. Global and individual AC in days of therapy (DOT) and DOT per 1000 patient-days were assessed. During the study period, 2567 patients were admitted, corresponding to 4961 patient-days in neonatal units and 9243 patient-days in PICUs. Multidrug-resistant Gram-negative bacteria and methicillin-resistant Staphylococcus aureus were more frequent in Brazil than in Germany. Average AC was 386.5 and 1335.5 DOT/1000PD in German and Brazilian neonatal units, respectively. Aminopenicillins plus 3rd generation cephalosporins were the most commonly prescribed antibiotics in German neonatal units, while aminopenicillins plus aminoglycosides were the class most commonly used in Brazilian NICU. Average AC was 888.1 and 1440.7 DOT/1000PD in German and Brazilian PICUs, respectively. Antipseudomonal penicillins were most commonly used in the German PICU, and glycopeptides were the most frequently prescribed in Brazilian PICUs. Carbapenems represented 2.3-14% of total DOTs in German neonatal units and 4% in the Brazilian NICU and 13.0% in the German PICU and 6-12.2% in Brazilian PICUs. We concluded that different patterns of most commonly prescribed antibiotics were observed in neonatal units and PICUs in these two countries, probably related to different local patterns of antibiotic resistance, with a higher antibiotic consumption in Brazilian study units.