2.2. Statistical Analysis

Based on the pilot data, the required sample size in each group to demonstrate a 15% difference in mean B12 with 90% power and at 5% significance was calculated to be 144. Statistical analysis was performed using SPSS version 22.0 [25]. Since BMI, serum B12, and folate were not normally distributed, they were log-transformed for statistical purposes. For comparison of GDM and no-GDM mothers, the Student’s t-test was used for continuous variables (e.g., B12, folate, and BMI) and the Chi-square test for categorical variables. Stepwise multiple linear regression was performed with B12 and folate as the dependent variables with the predictors entered or removed from the model according to the following criteria: Probability-of-F-to-enter ≤ 0.050, Probability-of-F-to-remove ≥ 0.100. Logistic regression was performed to determine the odds of maternal obesity and GDM according to B12/folate insufficiency status and the risk of macrosomia, LGA, LBW, and SGA according to quartiles of B12/folate. The regression models included the following co-variates: age, parity, ethnic origin, smoking, gestation of bloods, BMI, B12, and folate (where appropriate). For macrosomia and LBW, sex and gestational age were additionally added to the models.

Our institution has obtained ethics approval to collect B12 and folate data from pregnant women in an anonymised form (NHS ethics committee reference number 12/LO/0239).