Data analysis

Differences in categorical sociodemographic characteristics between the OC group and the reference groups were assessed using contingency table analyses test, with pairwise post hoc analyses if significant. For continuous characteristics an ANOVA was used with Tukey’s post hoc tests if significant, or Welch’s ANOVA with Games-Howell post hoc tests if Levene’s test indicated unequal variances, which was the case for gestational age (F = 225.49, p < 0.001).

Differences between the OC group and the reference groups in depressive symptoms and bonding were analyzed using ANOVA with planned contrast, or Welch’s ANOVA with planned contrast if Levene’s test indicated unequal variances, which was the case for bonding (F = 3.15, p = 0.044). A contingency table analyses test, with post hoc pairwise comparisons if significant, was used to assess differences between the three groups on the proportions of participants scoring above the cut-off on the EDS.

Pretest versus posttest differences were analyzed using paired sample t-tests, or with Wilcoxon-signed rank test if Shapiro–Wilk’s test the pretest–posttest difference scores were non-normally distributed, which was the case for bonding (D(17) = 0.81, p = 0.002). A McNemar test was used to assess a possible difference in the percentage of participants experiencing depressive symptoms above the before mentioned cut-off scores. For all main effects a Pearson’s correlation coefficient, r, was calculated. Effect sizes were interpreted as small (r ≥ 0.1), moderate (r ≥ 0.3), and large (r ≥ 0.5) [34]. Effects were regarded significant when p < 0.05.

A post hoc power analysis was conducted using the software G*Power [35]. For the calculation, a matched pairs t-test and an alpha error probability of 0.05 was selected. Furthermore, we used the effect sizes that were found. Power ranged between 0.66 and 0.82.