Estimating Gestational Age in Late Presenters to Antenatal Care in a Resource-Limited Setting on the Thai-Myanmar Border

Abstract

Estimating gestational age in resource-limited settings is prone to considerable inaccuracy because crown-rump length measured by ultrasound before 14 weeks gestation, the recommended method for estimating gestational age, is often unavailable. Judgements regarding provision of appropriate obstetric and neonatal care are dependent on accurate estimation of gestational age. We determined the accuracy of the Dubowitz Gestational Age Assessment, a population-specific symphysis-fundal height formula, and ultrasound biometry performed between 16 and 40 weeks gestation in estimating gestational age using pre-existing data from antenatal clinics of the Shoklo Malaria Research Unit on the Thai-Myanmar border, where malaria is endemic. Two cohorts of women who gave birth to live singletons were analysed: 1) 250 women who attended antenatal care between July 2001 and May 2006 and had both ultrasound crown-rump length (reference) and a Dubowitz Gestational Age Assessment; 2) 975 women attending antenatal care between April 2007 and October 2010 who had ultrasound crown-rump length, symphysis-fundal measurements, and an additional study ultrasound (biparietal diameter and head circumference) randomly scheduled between 16 and 40 weeks gestation. Mean difference in estimated newborn gestational age between methods and 95% limits of agreement (LOA) were determined from linear mixed-effects models. The Dubowitz method and the symphysis-fundal height formula performed well in term newborns, but overestimated gestational age of preterms by 2.57 weeks (95% LOA: 0.49, 4.65) and 3.94 weeks (95% LOA: 2.50, 5.38), respectively. Biparietal diameter overestimated gestational age by 0.83 weeks (95% LOA: -0.93, 2.58). Head circumference underestimated gestational age by 0.39 weeks (95% LOA: -2.60, 1.82), especially if measured after 24 weeks gestation. The results of this study can be used to quantify biases associated with alternative methods for estimating gestational age in the absence of ultrasound crown-rump length to inform critical clinical judgements in this population, and as a point of reference elsewhere.

Fig 1. Agreement between CRL biometry and the Dubowitz method.

GA: gestational age. Reference standard: crown-rump length (CRL) biometry. True gestational age determined from CRL biometry. The thick black lines represent the mean bias of the Dubowitz method in reference to CRL biometry; the thin grey lines represent the 95% limits of agreement. Grey dots are observed values for newborns with normal birthweight for GA (left) or term newborns (right); black dots are observed values for SGA newborns (left) or preterm newborns (right).

Fig 2. Agreement between CRL biometry and the SFH formula.

Reference standard: crown-rump length (CRL) biometry. SFH: symphysis-fundal height. True gestational age determined from CRL biometry. Thick black lines represent the mean bias of the SFH formula in reference to CRL biometry; the thin grey lines represent the 95% limits of agreement. Grey dots are observed values for newborns with normal birthweight for GA (left) or term newborns (right); black dots are observed values for SGA newborns (left) or preterm newborns (right).

Fig 3. Agreement between CRL and HC or BPD biometry.

Reference standard: crown-rump length (CRL) biometry. HC: head circumference. BPD: biparietal diameter. Gestation time of HC/BPD measurement determined from CRL biometry. Thick black lines represent the mean bias of HC biometry in reference to CRL biometry; the thin grey lines represent the 95% limits of agreement.

Fig 4. Crude comparison of biases associated with alternative methods of estimating gestational age.

GA: gestational age estimated from CRL biometry. CRL: crown-rump length. SFH: symphysis fundal height. HC: head circumference, measured at 25, 30, 35 or 40 weeks gestation. Solid red vertical lines delineate cut-offs for preterm (<37 weeks) and post-term (>41 weeks) newborns, and small for gestational age (Z-score <-1.28) newborns. Dotted red horizontal lines are mirrors of HC bias to facilitate visual comparison.