Quantifying Gestational Age Dating Error from Late Ultrasound in a High-Small for Gestational Age Population in Bangladesh
Study Period: April 2026-December 2026
Partners: Johns Hopkins Bloomberg School of Public Health
Project Description:
Background: Gestational age (GA) estimation based on the last menstrual period (LMP) is often inaccurate due to variability in ovulation timing, irregular cycles, and recall errors. These issues are particularly problematic in low-resource settings where literacy and menstrual tracking are limited. Ultrasound dating provides greater accuracy, but only when performed early. Scans in the first trimester are most reliable (±3–5 days), while accuracy declines substantially after 20 weeks (±21–30 days). Pregnancies without an ultrasound before 22 weeks are considered “suboptimally dated”. In South Asia, including Bangladesh, high rates of small-for-gestational-age (SGA) infants (30–40%) exacerbate misclassification. Late ultrasounds underestimate GA in growth-restricted fetuses, leading to term SGA infants being misclassified as preterm, inflating preterm birth rates and distorting trial analyses. Although most women presenting in labor have had at least one ultrasound, the majority occur after 20 weeks, increasing the risk of systematic misdating.
Objective: To quantify both systematic and random errors in gestational age (GA) estimation when the first pregnancy ultrasound is performed at or beyond 14 weeks of gestation, using <14‑week ultrasound dating as the reference standard. Additionally, to assess the extent of misclassification of birth status among infants in a Bangladeshi population with a high incidence of small‑for‑gestational‑age (SGA).
Method: This prospective methodological study will evaluate the accuracy of gestational age (GA) dating when the first ultrasound is performed at or beyond 14 weeks, using <14‑week crown‑rump length (CRL) scans as the reference. Mixed‑effects modeling will be applied to assess systematic and random errors across gestational weeks, with four follow‑up scans scheduled at 20, 25, 30, and 35 weeks (±10 days). Biometric parameters including CRL, biparietal diameter (BPD), femur length (FL), abdominal circumference (AC), and head circumference (HC) will be measured following INTERGROWTH‑21st protocols. Based on simulation‑based calculations, a sample of approximately 200 women with five rounds of ultrasound measurements will provide 80% statistical power to detect GA dating errors, assess growth trajectories, and conduct agreement analysis using the Bland‑Altman method.
Significance: This study will provide the first population‑specific GA dating error functions for South Asia. By improving the precision of GA estimation, the findings will help reduce misclassification between term and preterm births. This will strengthen the validity of studies and interventions targeting preterm outcomes in populations with high SGA prevalence.