Effect of socioeconomic vulnerability, family structure patterns and non-acceptance of pregnancy on preterm birth: a structural equation model. Londrina, Brazil

Oliveira, Adelaide

INTRODUCTION: Preterm (PT) birth is associated to perinatal morbidity and mortality. Its prevalence is increasing in many countries and the complexity of factors associated has been investigated.

METHODS: To analyze the effects of observed variables and latent variables was used a conceptual framework via exploratory structural equations modeling and the confirmatory approaches in a case-control study in Londrina/PR, Brazil (Jun/06 to Mar/07).

Continuous latent variables generated represent the constructs socioeconomic vulnerability(SEV), family structure pattern(FSP) and non-acceptance of pregnancy(NAP). The classification of prenatal care was defined according to the time of the 1^st appointment and tests performed. PT births were evaluated through gestational age (GA) as a continuous variable. The latent variables involved and the full model generated were built and validated with the weighted least square estimator in the software MPlus.

RESULTS: SEV comprises number of residents per rooms (std.coef.:0.66;p<0.001), income per capita (-0.87;<0.001), maternal education (0.84;<0.001), education of household head (0.73;<0.001), place of residence (0.65;<0.001). FSP was formed by family type (0.91;<0.001), having a partner <2years (0.72;<0.001), presence of elderly (0.76;<0.001), mother relationship with household head (0.82;<0.001). NAP considers negative reactions to the pregnancy of mother (0.81;<0.001), father (0.84;<0.001), family (0.83;<0.001).

The effects on GA included direct effects of prenatal care (-0.32;<0.001), pregnancy complications (-0.50;<0.001); NAP (0.19;<0.03), alcohol use (-0.11;<0.01); multiparity (-0.25;<0.01). There is also an indirect effect of NAP (0.20;<0.02) and multiparity (-0.28;<0.001) was mediated through prenatal care. SEV (0.12;<0.05) and FSP (0.17;<0.05) showed only a indirect effect on GA mediated through prenatal care. It was found a direct effect of NAP (0.11;0.05) and SEV (0.20;0.04) on the prenatal care.

CONCLUSIONS: SEM allowed dealing with latent variables and also observed variables to understand the effects on the outcome. FSP and SEV only express its effect mediated by prenatal care.