Burn Out Among Medical Family Doctors

Overview

The main objective of this study is to assess the burnout prevalence among French general practitioners in private practice. As secondary outcomes, this study aim to measure the impact of sociodemographic variables, organizational practice models and workload. the investigators also intend to characterize the prevalence of depression, anxiety, fatigue and stress, drugs and alcohol consumption, use of psychotropic medication, and preferred strategies to cope with their symptoms.

The BOUM protocol is designed to provide a better understanding of the current burnout prevalence among French general practitioners, as well as its risk and protective factors. In order to fulfil those objectives, the methodologist estimated a 1536 individuals sample size, using an expected prevalence of severe burnout ranged between 10% and 20% (from the the EGPRN study - Soler, J.K., et al., Burnout in European family doctors: the EGPRN study. Family Practice, 2008. 25(4): p. 245-265), thus allowing the investigators to retrieve results with a 95% confidence level and a 2% error margin. All analyses will be performed in a bilateral formulation for a 5% alpha error under the Stata® software (version 13, StataCorp, College Station, US). A difference will be considered statistically significant for p\<0.05. The population will be described as frequency and percentage for the categorical variables, and as mean ± standard deviation or median \[interquartile range\] for the quantitative variables, according to their statistical distribution (normality studied by the Shapiro-Wilk test). Confidence intervals for population prevalence will be characterised with a binomial test. Comparisons between groups of burnout will be performed using Chi2 or Fisher's exact test for the categorical variables (followed by a Marascuillo post-hoc tests if needed), and by an ANOVA or Kruskal-Wallis test if normality and homoscedasticity not respected (Bartlett test) for the quantitative variables (followed by Tukey-Kramer or Dunn tests when necessary). Finally, considering the variables clinically relevant and those with a pertinent univariate analysis results, a multivariate analyse of ordinal polynomial regression type will be proposed; its results will be expressed as relative risk with 95% confidence intervals and presented as a Forest-plot.