ObjectivesTo determine the level of knowledge and practice of evidence-based medicine (EBM) and the attitudes towards it and to identify the factors associated with its practice among primary care practitioners in Selangor, Malaysia.SettingThis cross-sectional study was conducted in randomly selected health clinics in Selangor. Data were collected from primary care physicians using self-administered questionnaires on knowledge, practice and attitudes regarding EBM.ParticipantsThe study included 225 respondents working in either government or private clinics. It excluded house officers and those working in public and private universities or who were retired from practice.ResultsA total of 32.9% had a high level of EBM knowledge, 12% had a positive attitude towards EBM and 0.4% had a good level of its practice. The factors significantly associated with EBM practice were ethnicity, attitude, length of work experience as a primary care practitioner and quick access to online reference applications on mobile phones.ConclusionsAlthough many physicians have suboptimal knowledge of EBM and low levels of practising it, majority of them have a neutral attitude towards EBM practice. Extensive experience as a primary care practitioner, quick access to online references on a mobile phone and good attitude towards EBM were associated with its practice.

<b>Aim</b>: This study aimed to examine the construct verification of the knowledge on the Noor evidence-based medicine (EBM) questionnaire using the Rasch measurement model.<br /> <b>Methods</b>: A cross-sectional study was conducted among ninety healthcare professionals working in a government hospital. The Rasch model was used to investigate the distribution of statistics, unidimensionality, polarity, misfit, and Wright map.<br /> <b>Results</b>: The Rasch analysis showed that the 15 items had high reliability of items at 0.96, while reliability for persons were 0.81. Item K15 is a misfit (PtMea Corr=0.22, outfit MnSq=1.95, outfit z-std=5.00, infit MnSq=1.63, infit z-std=3.60); the item does not contribute to the construction of scale but not degrading. Three items (K1, K14, and K9) can easily be answered even by low ability respondents.<br /> <b>Conclusions</b>: The Noor EBM questionnaire knowledge is robust with excellent psychometric properties that can be used for both research and clinical purposes.