[Elaboration and validation of longitudinal reference intervals of fetal weight with a sample of the Brazilian population].

PURPOSES: To elaborate models for the estimation of fetal weight and longitudinal reference intervals of estimated fetal weight (EFW) using a sample of the Brazilian population.

METHODS: Prospective observational study. Two groups of patients were evaluated: Group EFW (estimation of fetal weight): to elaborate (EFW-El) and validate (EFW-Val) a model for the prediction of fetal weight; Group LRI (longitudinal reference intervals): To elaborate (LRI-El) and validate (LRF-Val) conditional (longitudinal) percentiles of EFW. Polynomial regression analysis was applied to the data from subgroup EFW-El to elaborate a model for the estimation of fetal weight. The performance of this model was compared to those of previously published formulas. Linear mixed models were used for the elaboration of longitudinal reference intervals of EFW using data from subgroup LRI-El. Data obtained from subgroup LRI-Val were used to validate these intervals.

RESULTS: Group EFW consisted of 458 patients (EFW-El: 367; EFW-Val: 91) and Group LRI consisted of 315 patients (LRI-El: 265; LRI-Val: 50). The model obtained for EFW was: EFW=-8.277+2.146xBPDxACxFL-2.449xFLxBPD². The performances of other models were significantly worse than those obtained with our formula. Equations for the prediction of conditional percentiles of EFW were derived from the longitudinal observation of patients of subgroup LRI-El and validated with data from subgroup LRI-Val.

CONCLUSIONS: We described a method for customization of longitudinal reference intervals of EFW obtained using formulas generated from a sample of the Brazilian population.