(Original Article) Development and Validation of Thai Ethnic-Specific Reference Ranges for Down Syndrome Screening : A Case Study in Buriram Provincial Screening Network

Abstract

Down syndrome is the most common chromosomal abnormality in newborns, affecting child development and family quality of life. Second-trimester screening using the Quadruple Test remains widely utilized due to its acceptable performance and cost-effectiveness. However, the median values used to calculate the multiple of median (MoM) are typically derived from European populations, which may not represent the genetic characteristics of Thai women. This study aimed to develop and evaluate a Down syndrome screening model using Thai ethnic-specific medians and compare its performance with the conventional Western-European model. This retrospective developmental study included 15,851 singleton pregnant women at 14–20 weeks of gestation who underwent screening in the Buriram provincial network between 2021 and 2025. Maternal biochemical marker levels and weights were used to calculate MoM values based on two models: the Western-European and Thai-specific models. A risk cut-off of >1:250 was considered screen positive, and results were confirmed by karyotyping or postpartum data. The Thai-specific model demonstrated a sensitivity of 85.0%, specificity of 93.1%, false-positive rate (FPR) of 7.0%, and an AUC of 0.94 (95% CI: 0.89–0.98), higher than the Western model (AUC 0.92; 95% CI: 0.82–0.99). Although the difference in AUC was not statistically significant (p = 0.68; z-test for independent ROC curves), the higher AUC and lower FPR indicated that ethnic-specific medians could improve accuracy and reduce bias caused by genetic and anthropometric differences. When analyzed by individual markers, uE3 showed a significant AUC difference (p = 0.009), while AFP decreased slightly (p=0.24), and Inhibin-A and b- hCG remained comparable. Although the study included a large cohort, only 27 Down syndrome cases limited the statistical power for detecting small AUC differences (D 0.02). Nevertheless, the significant uE3 difference (D=0.19) suggests a strong effect size, Therefore, a large-scale validation study is warranted to confirm these findings to enhance statistical power.