The prevalence ratio
(PR) was adjusted for the cluster effect of the design. This correction
was used because each participating center was considered as a cluster,
and the correspondent heterogeneity in each variable among clusters
was adequate [20,21]. Sociodemographic and obstetric factors poten-
tially associated with worse outcomes among women with PPH were
evaluated by comparing women with PLTC with those with SMO. Final-
ly, multiple Poisson regression analysis was used to identify factors
independently associated with SMO secondary to PPH. P b 0.05 was
considered statistically significant.