Revista Brasileira de Epidemiologia
Print version ISSN 1415-790X
NEUMANN, Nelson A. et al. Impact of Pastoral da Criança on the nutrition of children under 5 years of age in Maranhão: a multi-level analysis. Rev. bras. epidemiol. [online]. 2002, vol.5, n.1, pp.30-40. ISSN 1415-790X. http://dx.doi.org/10.1590/S1415-790X2002000100005.
BACKGOUND: Pastoral da Criança monitors more than 1.5 million children monthly in the promotion of mother-child health. Independent assessments are necessary in order to evaluate its impact. OBJECTIVE: To compare child nutrition indicators in communities served by Pastoral da Criança and in control communities through traditional and multi-level analysis. OUTLINE: Quasi-experimental study, comparing conglomerates of children monitored by 60 community leaders of Pastoral and control conglomerates. SITE: Urban and rural areas (cities of São Luís and Timbiras) in Maranhão. PARTICIPANTS: 60 conglomerates in each group (Pastoral and Control) and an average of 14 children per conglomerate. The study effectively included 1,629 children, of whom 820 were monitored by Pastoral. RESULTS: Result analysis was conducted by using traditional methods, which do not consider data hierarchy structure, and by considering intra-group correlations by means of multi-level analysis. The latter showed that the regression model variability was distributed along all 4 levels studied. In the traditional analysis of the height-for-age Z-score, Pastoral children were, on average, 0.184 Z-score shorter than control children (P=0.03). After adjustment for the confounding factors score, this difference increased. However, the effect decreased when adjusted for children's ages and was no longer significant. The multi-level analysis did not reveal a significant effect of Pastoral on any of the adjustment models, differently from the traditional analysis. Both traditional and multi-level analyses were similar as to the results of exclusive or predominant breastfeeding up to the third month of life, both showing a positive Pastoral da Criança influence, both in the general and adjusted analyses (odd ratios of 4.37 and 4.63 respectively, both with p<0.001). CONCLUSION: The assessment showed that, by not considering intra-class correlations, traditional analysis may present different results from multi-level analysis. In the present study, Pastoral da Criança was associated to improvements in exclusive or predominant breastfeeding, but not in terms of height-for-age.
Keywords : Children; Primary Health Care; Intervention; Community Agent; Service Evaluation; Epidemiology; Malnutrition; Breastfeeding.