Study design and participants
We conducted a prospective study involving women delivering a single live birth in one of the three maternities of the Sorbonne University during the first strict lockdown in France from 27th March to 5th May 2020. The study was conducted in accordance with the Helsinki declaration and was approved by the local ethics committee (Institutional Review Board Ile de France II, approval 27042020 - ClinicalTrial.gov ID: NCT04366817). All participants gave written informed consent. The inclusion criteria were: maternal age ≥ 18 years, French social security registration, patients speaking and understanding French, and maternal post-delivery hospitalization in the conventional postnatal ward. The exclusion criteria were: women who did not speak French, mothers hospitalized in other units, and mothers of babies hospitalized in the Neonatal Intensive Care Unit.
Eligible participants were recruited by the midwives in the postnatal units and were offered the possibility of one telephone interview with a psychologist at Day 10 (±1 day) postpartum, and then another one 6 to 8 weeks later. Three attempts were made to call patients who failed to answer. The standardized interviews lasted approximately 45 minutes and were structured in two parts: (i) a structured interview with validated questionnaires, and (ii) a semi-structured interview about the mother's experience of childbirth, the maternity ward, conditions of discharge, and the first days/weeks at home with the baby and their partner.
We extracted sociodemographic and obstetrical data from the obstetrical records. These included: (i) socioeconomic data (Marital/couple status, Educational Level, Profession, Current financial situation, Geographic origin), (ii) Obstetrical background: history of perinatal death, fetal malformation, intra-uterine growth retardation (IUGR), maternal-fetal infection, (iii) Maternal complications during the current pregnancy: hypertension, diabetes, hospitalization for threatened preterm labor, (iv) Maternal psychiatric background: history of PPD, history of mood or anxiety disorders, history of Post-Traumatic Stress Disorder (PTSD), and (v) Specific psychosocial contexts: precarity, conflict within the couple, single parent, transgenerational familial history of obstetrical pathologies.
The following validated self-questionnaires were used during the interviews to assess postpartum depressive and post-traumatic symptoms, couple satisfaction and mother-child interactions:
- 1) The Edinburgh Postnatal Depression Scale (EPDS) , which is a 10-item questionnaire specific to the postpartum period and results in a depression score. For this study, we used the score as a continuous variable. An EPDS score > 12 was used to define a major risk of PPD, and an EPDS between 10 and 12 a mild to major risk.
- 2) The Perinatal Posttraumatic Stress Disorder Questionnaire (PPQ)  (validated French version ), which evaluates post-traumatic stress reactions of parents undergoing a childbirth with a high perinatal risk. It consists of 14 items, which refer to the DSM-V criteria of PTSD. A threshold of 6 was used to define a high risk of perinatal post-traumatic syndrome .
- 3) The couple’s relationship (Dyadic Adjustment Scale; DAS-16)  is a self-questionnaire derived from the original 32-item DAS scale, which evaluates dyadic adjustment in the marriage . The revised scale consists of 16 items relative to the quality of the couple’s relationship and includes two dimensions: the degree of accordance (DA), and the quality of interaction (QI). The participants are asked to indicate their responses in relation to their experiences in the preceding 2 weeks, on a 6-point Likert scale. The score is calculated as a summation of all the responses, and ranges from 16 to 96. Scores above 54 are considered to represent satisfactory adjustment. Some questions concern the frequency in which the couple laughs together, or the frequency with which the couple considers divorce. The last question asks, “In general, to what degree do you experience happiness in your relationship” on a scale from “extremely unhappy” to “extremely happy”.
- 4) The Mother-Infant Bonding Scale (MIBS)  was initially developed to detect disturbances of maternal feelings towards their newborns. It is a short, simple questionnaire, used after birth in the maternity ward by different medical professionals: midwives, obstetricians, pediatricians, nurses. The scores range from 0 to 24, with a higher score indicating an impaired mother-infant bond.
As part of this study, psychological or psychiatric support was offered to each woman with an EPDS score > 10 and/or a PPQ score > 6, or if the interview revealed clinical vulnerability indicators (e.g., expression of emotional distress, history of perinatal loss, separation from the partner) whatever the results of EPDS score.
The study was designed to achieve 80 % of power to detect at least a 5-point change in the proportion of major PPD compared to the expected proportion of 15% (15 vs. 20%), considering two-sided alpha risk of 5% and 5% of dropout. Therefore, 452 women needed to be included.
Continuous variables were described as mean and standard deviation or median and interquartile range, depending on the distribution of the variable. Qualitative variables were expressed as numbers and proportions.
The proportion of patients with major PPD during follow-up, defined by an EPDS score > 12, was calculated with its 95% confidence interval (exact method) and the exact binomial test was used to test the superiority of the proportion compared with the expected value of 15%.
The association between depression and baseline characteristics was studied using logistic regression. Unadjusted analysis was performed to select variables at the 20% threshold taking into account missing values and relationships between variables or redundancy.
Post hoc analysis also assessed the association between baseline characteristics and PPD and/or post-traumatic symptoms during follow-up (EPDS >12 or PPQ > 6).
Spearman rank correlation was used to determine post hoc correlations between scores. Missing values were not replaced. All tests were two-sided and a P-value < 0.05 indicated statistical significance. Analyses were performed using SAS version 9.3, SAS institute Inc., Cary, USA.