Introduction: The short-and long-term health positive impact that exclusive breastfeeding (EBF) confers to children are well documented. Despite the secular trend for EBF in Brazil has increased, the prevalence of 36% observed in the last national survey conducted in 2013 is still below the 50% considered satisfactory by the World Health Organization. In this scenario, it is important to identify relevant modifiable key risk factors for the premature interruption of EBF. Although pacifier use has been pointed as one of the multiple determinants of exclusive breastfeeding there is no consensus on the effect of this habit in the early interruption of EBF. Objective: To analyze the association between pacifier use and EBF interruption as well as the influence of temporal trends of pacifier use on the EBF in children under 6 months old in Brazilian state capitals and Federal District, 1999-2008. Methods: 4 manuscripts were produced. The first consisted in systematic review and meta-analysis to investigate the association between pacifier use and interruption of exclusive breastfeeding in infants less than 6 months. A wide search in 5 databases (CINAHL, Scopus, Web of Science, LILACS, Medline) from inception through 30 December 2104 without restriction of language yielded 1.866 publications submitted to predetermined inclusion/exclusion criteria peer reviewed (PROSPERO protocol CDR42014014527). In the second and third manuscripts data from 2 waves of infant feeding surveys conducted in 1999 and in 2008 in the Brazilian state capitals and Federal District (I and II PPAM) was used. Methodologically equivalent surveys used representative samples selected based on complex sampling procedures and systematic random selection of children in the queue of each selected immunization center. Questionnaires were composed of closed-ended questions regarding the pacifier use and infant feeding on the day before the survey. In the second manuscript, to find out whether or not pacifier use is an independent risk factor for the EBF interruption were used multivariate regression models in the pooled sample and for each survey wave adjusting for socioeconomic, demographic and biomedical confounders. In the third paper, we calculated the potential impact fraction (IF) of the influence of temporal trends of pacifier use on the decline of EBF interruption. Manuscript 4 resulted from methodological deepening for application of the analysis strategy applied in the previous manuscript, into it we present how estimates of IF that might be used in the analysis of variation in the population outcomes considering shift in its determinants, taking as an example the data about EBF interruption and pacifier in Brazil. Results: The meta-analysis pooled effect was OR 2.48 (CI95%=2.16-2.85) for the association between pacifier use and EBF interruption. The majority heterogeneity was explained by study design (40.2%). Two RCT's with very limited external validity found a null association, but 44 observational studies, including 20 prospective cohort stu...