Predictors of evidence-based knowledge in Italian doctors

Background: In the last decade many countries have legislated the recertification of medical practitioners within their health systems. Evidence-based medical knowledge is increasingly seen as an essential requirement to demonstrate that doctors are clinically competent to perform certain roles. A free-access distant Continuing Medical Education (CME) system, based on Clinical Evidence (CE), a derivative product of systematic reviews, called ECCE (Continuing Education Clinical Evidence), was developed in Italy. The tests consist of fixed- and multiple-choice questions related to selected clinical vignettes to assess CE knowledge. Objectives: Our objective was to assess whether doctor characteristics were predictors of good or bad performance to evidence-based vignettes. Methods: Twenty vignettes were extracted from ECCE. Two types of measures were used to determine how doctors performed on vignettes: total score (number of correct answers) and success/failure. We collected demographic information about gender, age, year of degree, specialty, area of practice and geographical area. For each case we performed logistic and linear regression analysis based on the outcome variable and conducted heterogeneity analysis using RevMan to determine whether the associations detected were generalizable. Results: At data extraction, 26,769 Italian doctors were enrolled, who voluntarily chose the vignettes to solve. Increase in age seems to have the most important influence on success in less then fifty per cent of cases (e.g. headache vignette OR = 1.58, CI 1.12 to 2.22; hypertension OR = 1.33, CI 1.10 to 1.60; etc.). The minor influences of geographical area, residence (i.e. urban vs. rural) and year of degree were statistically significant in one quarter of cases. Gender and specialty had no influence on success/failure. Linear regression analyses confirmed these results. The heterogeneity analyses showed that our clinical vignettes were homogeneous for all detected characteristics (e.g. age: p-value 0.8; specialty: p-value 0.92; etc.). Conclusions: We detected some characteristics which were good predictors of naivety to evidence-based medicine. This study is important to guide educational policies for medical professionals who perform below standards in relevant scenarios and to foster systematic review results.