In this letter to the editor, we highlight some concerns with a recently published method to estimate gestational age at delivery from DNA methylation data. We conduct novel analyses to highlight the implications of different choices in study design and statistical methods for the prediction of phenotypes from methylation data.