The social context and perception of poor health in Colombia : a multilevel analysis
METHODS: A multilevel analysis using data from the National Survey of Health and Quality of Life of 2009 was performed. In relation with survey details, the perceived health was dichotomized and individual level variables related were extracted. Information on HR, ICV and GDP per capita (GDP pc) at the departmental level was taken from DANE. Prior to the completion of the multivariate model, t test were performed and chi-square, to explore the association of each independent variable and the presence of poor perceived health. Logistic mixed effects models with random intercept a municipality and department level were performed. The existence of random parameters were ruled out. Each statistical model evaluation was performed using the Akaike criteria. Subsequently logistic mixed effects models with random intercept were prepared by municipality level and department, including intercepts sequentiallyby town and municipality.
RESULTS: A statistically significant association between the Index of Quality of Life at the departmental level and the presence of poor perceived health was found (OR 0.96; 95% CI :0.94 -0 .97 ), this effect was independent of the different variables and adjusted for departmental GDPpc and the percentage of explained variance was 1% for department level and 4% for the municipality level.
CONCLUSIONS: There is an effect of the departmental and municipal level that clearly outlined the differences in context. It confirms the importance of the social context of these levels on the determination and social construction of perceived health and could enhance the comprenhension on inequities because of the location were people live.