This third course, “Multivariable analysis”, was organized by the MediPIET and the Albanian National Institute of Public Health, in Tirana, Albania, from the 2nd to the 6th December 2019.
The module provided participants with basic definitions and concepts for the different types of regression models (linear, logistic, conditional logistic, Poisson, negative binomial) as well as the skills needed to perform and interpret multivariable analysis. In addition, participants had the opportunity to practice their scientific communication skills by engaging in writing and presenting the methods or the results of scientific work involving multivariable analysis. Therefore, participants worked on their competencies in the areas of statistical data analysis, inferential statistics and written scientific communication.
A pre-module exercise on a case study was provided prior to the module to all participants to practice and review data management, descriptive, univariable and stratified analysis in preparation for the content of the module.
Dr Angeliki Lambrou, Dr Adela Paez Jimenez and Dr Iro Evlampidou as MediPIET Scientific Coordinators, supported by Dr Isabel Noguer, ISCIII Leader for MediPIET Consortium, in collaboration with Dr Albana Fico, Executive Director of the Institute of Public Health Institute in Tirana have put together the contents and mobilised the senior experts with a lot of teaching and work experience in multivariable analysis, the specific statistical software, and scientific communication.
A total of 26 participants attended the course, including fifteen fellows of the Third MediPIET Cohort from eleven current training sites: Albania, Armenia, Georgia, Kosovo1, Lebanon, Moldova, North Macedonia, Palestine, Serbia, Tunisia and Ukraine. In addition, there were six external participants, one from each of the following countries: Algeria, Bosnia-Herzegovina, Jordan, Libya, Montenegro and Morocco. Moreover, five local participants from Tirana were nominated by the country and four attended the module. On the third and fourth day of the course, a MediPIET alumnus joined the module for the two days
At the end of the module, participants were able to:
- Describe the principles of statistical modelling and multivariable analysis
- Identify the analysis situations that require the use of multivariable analysis (i.e. presence of third factors, including effect modifiers and confounders)
- Select the type of multivariable analysis (i.e. linear, logistic, conditional logistic, Poisson) that is adapted to a study objective/design
- Identify the relevant variables to build up an optimal regression model
- Control for confounding and assess effect modification at the analytical level
- Interpret the results of a regression model: regression parameters and the corresponding inferences
- Apply univariable and multivariable models using the statistical software STATA
Although for most participants the module content was new and for some too advanced, it was considered useful and interesting and all of the aspects of the module were much appreciated and overall highly rated by the participants.
MediPIET team is very grateful to the Italian National Institute of Health, the UK Public Health Rapid Support Team/Public Health England, and Santé Publique France for their selfless collaboration in this module. Their contribution has been essential and the participation of their professionals throughout the duration of the module is acknowledged and highly appreciated.
Special gratefulness as well goes to the collaborators in the Institute of Public Health in Tirana for their support in the implementation of the activity.