OBJECTIVES: To compare physicians and patient’s reports of health resource utilisation using the PGRx-3 datasets. METHODS: PGRx (Pharmacoepidemiologic General Research eXtension) is a multiusers, multi-countries, prospective Real World Datasets originally assembled for effectiveness and relative risk of medicines research. In 2015, PGRx was extended to HEOR research in six countries (France, UK, Italy, Germany, Spain and in the US). Diseases studied with the PGRx methodology are: acute coronary syndrome, arrhythmias, stroke, lupus, rheumatoid arthritis, multiple sclerosis, Guillain Barré Syndrome, type I diabetes, autoimmune thyroiditis, depression and suicide attempts, paediatric autoimmune disorders, breast cancer, prostate cancer (on-going) and others. Patients are recruited by board-certified specialists or general practitioners and they can fill self-questionnaires. More than 65% consenting patients also undergo a telephone interview. All PGRx datasets are submitted for ethics and privacy protection committees’ approval in the contemplated countries. Medical records of more than 5,000 patients have been obtained in a series of validation studies including several of the PGRx disorders and results compared to interviews of the same patients on health resources and drug utilisation. RESULTS: An agreement superior to 95% between patients and their physician’s medical records was observed for report of chronic drug use and vaccines, but the agreement was low (<30%) for drugs available OTC or in the patients’ pharmacy (eg: NSAIDs) and for patients’ health resources utilisation (eg: physician visits) (agreement less than 25%). CONCLUSIONS: Physicians have poor knowledge of acute drug use and of health resource utilisation by patients. Patient interviews are more valid sources of information for health economics and outcomes research than chart reviews