Artificial intelligence tool utilization in deprescribing practices among community pharmacists in Jordan: a cross-sectional study

Introduction Data on artificial intelligence-supported deprescribing in Jordanian community pharmacy are still lacking. This study aimed to assess current deprescribing practices among community pharmacists in Jordan, identify the medication classes most commonly targeted for deprescribing, evaluate perceived outcomes, and determine factors associated with AI tool use in deprescribing. Methods This was a cross-sectional survey-based study conducted among licensed community pharmacists in Jordan from 15 November 2025 to 15 March 2026 using online, in-person, and telephone recruitment. The questionnaire assessed participant and practice characteristics, deprescribing activities, digital infrastructure, and use of AI tools. For the primary analysis, AI use was operationalized specifically as self-reported AI-chatbot use, whereas other computerized decision-support tools were reported separately. The SPSS (version 29) was used to conduct descriptive statistics, chi-square tests, and multivariable logistic regression. Results Of the 1,036 pharmacists who completed the questionnaire, 798 (77.0%) reported practicing deprescribing. The medication classes most frequently reported as targets of deprescribing-related activity were antibiotics (799, 77.1%), proton-pump inhibitors (790, 76.3%), vitamins and mineral supplements (776, 74.9%), non-opioid analgesics/nonsteroidal anti-inflammatory drugs (753, 72.7%), and antidiabetic drugs (638, 61.6%). More than half of the pharmacists (641, 61.9%) reported using a digital tool to identify deprescribing opportunities and (340, 32.8%) reported using AI chatbot. Frequently observed outcomes after deprescribing were better patient-reported health (429, 41.4%), improved adherence (417, 40.3%), and fewer side effects (417, 40.3%), whereas worsening of symptoms was reported less often (89, 8.6%). AI use was independently associated with male gender (adjusted odds ratio [AOR] 1.32, 95% confidence interval [CI] 1.01–1.73), PharmD qualification versus BPharm/BSc (AOR 1.38, 95% CI 1.02–1.87), urban practice versus rural practice (AOR 1.52, 95% CI 1.04–2.22), and reliable internet access (AOR 3.89, 95% CI 1.95–7.76). Conclusion Deprescribing is widely reported in Jordanian community pharmacy practice, and one-third of pharmacists use AI chatbots to support deprescribing decisions. Strengthening internet access, targeted training, and equitable infrastructure may help expand AI-assisted deprescribing across practice settings.