"Continuous variables were summarized using means and SD. Categorical variables were summarized using counts and percentages. Multivariable linear regression was used to assess associations between baseline characteristics (fasting OGTT result, 120-minute OGTT result, maternal age, and pre-pregnancy weight or pre-pregnancy BMI), and insulin requirements. Separate models were constructed for initial insulin requirement and peak insulin requirement as defined above. For each outcome, separate models were built with either pre-pregnancy weight or BMI given the collinearity between the two measures. Regression coefficients (β), 95% CIs, and p values were reported. Model assumptions were assessed visually using residual plots. R-squared was used to evaluate the proportion of variation in the outcome variable that the models explain, thereby assessing the model’s explanatory power. To compare relative variable importance, models were re-run with standardization of all continuous variables."