2.3. Statistical Analyses

Descriptive statistics for demographic variables, number of life events, and levels of anxiety or depressive symptoms were calculated. To allow for valid comparison of effect sizes, we created z-scores for the antenatal symptoms of anxiety and depression. Subsequently, we performed a series of multivariable linear regression analyses with postpartum scores on STAI and EPDS to quantify the associations under study. The analyses quantified change in anxiety and depression symptom levels by adjusting postpartum levels for antenatal levels. In a separate analysis, potential confounders were added (i.e., socioeconomic position and nulliparity). Confounders were chosen based on their well-known association with the outcomes. To investigate associations with symptoms of anxiety or depression specifically, all analyses were additionally adjusted for depressive symptoms in the analysis of anxiety, and vice versa.

Lastly, we assessed whether associations of specific life events, delivery complications, and unfavorable obstetric outcomes of the newborn with postpartum symptoms of anxiety or depression were modified by antenatal anxiety or depression symptoms.