Developing an Effective and Simple Digital Screening Tool to Identify Inadequate Calcium Intake in Pregnant Women: a prediction model in the Rotterdam Periconception cohort

Objective: To develop an evidence-based and simple screening tool to estimate calcium intake in pregnant women, suitable for use in daily clinical practice. Design: Cross-sectional analysis within a cohort study Population and setting We extracted all data from the Rotterdam Periconceptional cohort (PREDICT study) conducted at the Erasmus MC, University Medical Centre in Rotterdam, the Netherlands, between November 2014 and December 2020. Methods: Data was extracted from food frequency questionnaires. The estimated average requirement of 750 m/day was defined as the lower limit for an adequate calcium intake. We created a prediction model, using multivariable binary logistic regression with backward stepwise selection. We developed a simple screening tool based on the prediction model. Main outcome measures Probability of adequate calcium intake Results: 694 participants are included, of which 201 (29%) had an adequate calcium intake. Total daily or weekly intakes of cheese, milk, and yogurt or curd were selected as predictors for the prediction model. The model had excellent discrimination (AUC 0.858), a good fit (Brier score 0.136, HL statistic p=0.499) and satisfactory calibration. The test accuracy measures were: sensitivity 80.9%, specificity 77.1%, PPV 89.7%, NPV 62.2%. A color coded digital screening tool was developed for use in clinical practice. Conclusions: This evidence-based and simple screening tool is a reliable and efficient instrument to predict inadequate calcium intakes in pregnancy, which can easily be incorporated in daily clinical practice and existing pregnancy coaching platforms.