Birth weight is a crucial factor linked to a newborn’s survival and can also affect their future health, growth, and development. Earlier, researchers focused on exploring maternal and fetal factors contributing to low birth weight. However, in recent years, there has been a shift toward effectively predicting low birth weight by utilizing a combination of variables. This study aims to develop and validate a nomogram for predicting low birth weight in Ethiopia.
A retrospective follow-up study was conducted, and a total of 1,120 pregnant women were included. Client charts were selected using a simple random sampling technique. Data were extracted using a structured checklist prepared on the KoboToolbox (Cambridge, Massachusetts in the United States) and exported to STATA version 14 (Computing Resource Center in California) and R version 4.2.2 (University of Auckland, New Zealand) for data management and analysis. A nomogram was developed based on a binary logistic model, and its performance was assessed by discrimination power and calibration. Internal validation was performed using bootstrapping. To evaluate the clinical impact, decision curve analysis was applied.
The nomogram included gestational age, hemoglobin, primigravida, unplanned pregnancy, and preeclampsia. The AUROC of the predicted nomogram was 84.3%, and internal validation was 80.1%. The calibration plot indicated that the nomogram was well calibrated. The model was found to have clinical benefit.
The nomogram demonstrates strong discrimination performance and can predict low birth weight clinically. As a result, it can be used in clinical practice, which will help clinicians in making quick and personalized predictions simply and rapidly, enabling the early identification and medical intervention. For broader applicability, the nomogram must be externally validated.