Background: Preterm birth is the leading cause of death among children <5 years of age. Accurate determination of prematurity is necessary to provide appropriate neonatal care and guide preventive measures. To estimate the most accurate method to identify infants at risk for adverse outcomes, we assessed the validity of two widely available methods-last menstrual period (LMP) and the New Ballard (NB) neonatal assessment-against ultrasound in determining gestational age and preterm birth in highland Guatemala.
Methods: Pregnant women (n = 188) were recruited with a gestational age <20 weeks and followed until delivery. Ultrasound was performed by trained physicians and LMP was collected during recruitment. NB was performed on infants within 96 hours of birth by trained study nurses. LMP and NB accuracy at determining gestational age and identifying prematurity was assessed by comparing them to ultrasound.
Results: By ultrasound, infant mean gestational age at birth was 38.3 weeks (SD = 1.6) with 16% born at less than 37 gestation. LMP was more accurate than NB (mean difference of +0.13 weeks for LMP and +0.61 weeks for NB). However, LMP and NB estimates had low agreement with ultrasound-determined gestational age (Lin's concordance<0.48 for both methods) and preterm birth (κ<0.29 for both methods). By LMP, 18% were judged premature compared with 6% by NB. LMP underestimated gestational age among women presenting later to prenatal care (0.18 weeks for each additional week). Gestational age for preterm infants was overestimated by nearly one week using LMP and nearly two weeks using NB. New Ballard neuromuscular measurements were more predictive of preterm birth than those measuring physical criteria.
Conclusion: In an indigenous population in highland Guatemala, LMP overestimated prematurity by 2% and NB underestimated prematurity by 10% compared with ultrasound estimates. New, simple and accurate methods are needed to identify preterm birth in resource-limited settings worldwide.