Model development for anemia prediction in pregnancy

Abstract

Background: The number of pregnant women who suffer from anemia is still high. Anemia has an impact on the health status of mothers, fetuses, and newborns. This paper aims to develop an anemia risk prediction model for pregnant women in Indonesia. Methods: The data was obtained from the Indonesian basic health research database years 2013 and 2018. A bivariate method the user to select which variables are significantly affecting anemia in pregnant women. The result was then analyzed using the multivariate for model development and the data was then plotted into a receiver operator characteristic (ROC) to illustrate the relationship between clinical sensitivity and specificity. Results: The variables that significantly to the anemia risk status of pregnant women are iron supplement during pregnancy (p = 0.003) and MUAC (p = 0.001) from 2013 data. After analyzing using multivariate, only MUAC and iron supplement during pregnancy could be included in the prediction model since the p-value is less than 0.05. The ROC value is 60.7 % with a sensitivity of 87.8 % and a specificity of 75.7 % for the 2013 data and we got the same ROC score is 60.7 % for the 2018 data with a sensitivity of 94.3 % and a specificity of 90 %. Conclusion: Results suggest that the anemia risk prediction model developed based on data from pregnant women in 2013 and 2018 consisted of 2 variables, namely iron consumption and mid-upper arm circumference.

Keywords

• Anemia
• Pregnancy
• Prediction model
• Undernutrition