Customized compared to population-based centiles for detecting term small for gestational age infants in Greece.

BACKGROUND Applying customized centiles may improve the accuracy of detecting small for gestational age (SGA) infants; however, the evidence is inconclusive whether adjusted centiles are more sensitive in identifying infants at increased risk of morbidity. We aimed to examine the validity of customized centiles in a Greek cohort and evaluate their performance compared to population-based centiles in predicting infants at risk of increased morbidity. METHODS We prospectively recorded the neonatal and maternal characteristics of singleton, low-risk, term infants over a year. Infants were defined as SGA if their birth weight was under the tenth centile, classified both by population-based centiles and customized centiles, adjusted for maternal and innate factors. We performed a comparative analysis utilizing linear regression analysis and calculating the receiver operating characteristics (ROC) curves. RESULTS Overall 657 infants were identified. Population-based centiles detected 42 (6 %) SGA infants, while customized centiles 80 (12 %). Perinatal morbidity was associated with an odds ratio of 1.02 with customized centiles [95 % confidence interval (CI): 1.01-1.04] and with an odds ratio of 1.02 with population-based centiles (95 % CI: 1.02-1.02). In predicting perinatal morbidity, no significant difference was detected between customized centiles [area under the ROC curve 0.773 (95 % CI: 0.699-0.847)] and population-based centiles [area under the ROC curve 0.737 (95 % CI: 0.662-0.813)] (p =0.272). CONCLUSIONS Customized centiles provided increased accuracy in comparison to the population-based centiles in detecting SGA term infants. However, customized centiles had no better impact on predicting a poor perinatal outcome. HIPPOKRATIA 2020, 24(3): 133-137.