Study design

The U-BIRTH study, an extension of the BASIC study,18 is hosted at the Department of Obstetrics and Gynaecology at Uppsala University Hospital. The BASIC study (from September 2009 to November 2018) involved 5492 women (in 6478 pregnancies, ca. 22% of all Uppsala County births). During the BASIC study recruitment, all Swedish-speaking women aged 18 years and older with a routine ultrasound at Uppsala University Hospital were invited to participate. Exclusion criteria included age under 18 years, inadequate Swedish proficiency, protected identities, bloodborne infections and/or non-viable pregnancy detected through routine ultrasound. These criteria remain the same in the U-BIRTH study. Women received written information by post and provided their written consent for modalities they wished to take part in (eg, questionnaires, blood samples, tissue samples, or microbiota samples). Participating women were followed up from gestational weeks 16–18 through 12 months post partum. In the BASIC study,18 data were collected using various psychometric self-report instruments and biological samples (more information: https://www.basicstudie.se/). For the U-BIRTH study (https://www.ubirth.se/), initiated in 2012, all the BASIC study participants not opting out in any way were (will be) invited when their children reached (or reach) 18 months, 6 and 11 years of age. The youngest participants were born in April 2019. Data acquisition for 6 years old and 11 years old is expected to finish in the summer 2025 and 2030. Prior to each time point for assessments, consent from all participating mother–child dyads is (will be) obtained. Many participating mother–child dyads already have data from multiple time points acquired within the BASIC study, with the sociodemographic information available in table 1. In the U-BIRTH study, all recruited mother–child dyads are invited to a series of assessments when the child is at age of 18 months, 6 years and 11 years (table 2). Figure 1 illustrates the data timeline included in the protocol. For a full list of assessment tools for mothers and children in both BASIC and U-BIRTH studies, see online supplemental table 1. Currently, 3525 families have completed at least 1 assessment and 52 mother–child dyads dropped out due to external factors (eg, death of the mother or the child or the family moved away). Overall, all children have reached 18 months of age and the data acquisition of this phase is completed. There were 6083 invitations to the study for the 18 months post partum, and 3373 mother–infant dyads (55%) consented to participate. Among those consenting to participate in the survey, 85% completed the survey at 18 months post partum. Seventy-three per cent of the children have today reached 6 years of age. There were 4340 invitations sent, and 1409 (32%) mother–child dyads have fully completed the assessments. Moreover, 1547 (25%) children have reached 11 years of age and 624 have completed the assessments. Families not responding to the invitation receive one email reminder. Families who agree to participate but do not answer the survey receive three email reminders. The flowchart for participants is presented in figure 2. For those who decided to drop out without any external cause (eg, death of the child or the mother), some provide reasons such as moving away, while others did not specify. For non-respondents, we send a reminder but cannot ascertain the reasons.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

U-BIRTH study timeline.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Flowchart of the U-BIRTH study participants. BASIC, Biology, Affect, Stress, Imaging and Cognition.

Data collection

In the U-BIRTH, three follow-up timepoints include mother-reported instruments and saliva sampling from children at 18 months and 6 years of age. For detailed measures and timeline, see table 2 and figures 1 and 2 (a full list covers information in both BASIC and U-BIRTH, see online supplemental table 1). Mothers are also invited to complete the Patient Health Questionnaire, at ages of 6 and 11 years of their children. Furthermore, mothers are administered the Diagnostic and Statistical Manual of Mental Disorders, fifth edition-TR Self-Rated Level 1 Cross-Cutting Symptom Measure30 when their children reach 11 years of age. Other repeated maternal measures include lifestyles (eg, body mass index, diet, consumption of alcohol and tobacco), sleep habits, social environment, quality of life (the 36-Item Short Form Health Survey31), health-related quality of life,32 home environment (the Confusion, Hubbub, and Order Scale33), parenting experiences (the Parenting Stress Index short form34; the Swedish Parenthood Stress Questionnaire,35 and general living situation details, and so on). Regarding biological data, as illustrated in figure 1, the BASIC study collected maternal blood samples, microbiota samples, amniotic fluid, umbilical cord biopsy and fetal blood samples at birth. For the description of biological data collection, handling and storage, see the BASIC cohort profile.18

Children’s outcome measures include general physical measures, sleep behaviour, social–emotional development, behaviour, language development, neurodevelopment, school participation and performance, and more (eg, diet and free time activities, and so on; table 2). General physical measures include growth data collected at birth, 18 months, 6 years, and 11 years. Sleep behaviours measured at 18 months of age focus on a child’s sleep habits or problems. Evaluations of the relationship with parents are scheduled at the age of 18 months, 6 and 11 years. Children’s social skills are measured at 6 and 11 years old. Parent-reported behavioural measures are obtained at 18 months (the Early Childhood Behaviour Questionnaire,36 the Child Behaviour Checklist for 1.5–5 years old37); and 6 years (the Child Behaviour Checklist for 1.5–5 years old37) and 11 years of age (the Child Behaviour Checklist for 6–18 years old38). Language development is measured at 18 months, 6 years (the Language Development Survey39), and 11 years of age. The Five to Fifteen questionnaire,40 41 is applied to identify symptoms or problems related to attention deficit hyperactivity disorder at the age of 6 and 11 years. The survey also covers daily life, including school participation, household activities and free time activities, at 18 months, 6 and 11 years of age. Additionally, with parental consent to link data to national register, we collect information on maternal diagnoses, children’s diagnoses, child growth, medical referrals, breastfeeding information during the first years post partum for further analysis. Importantly, with parents’ consent to access the information from the medical journals and register data, we are able to identify children at the age of 11 years with diagnoses (eg, attention deficit/hyperactivity disorder or autistic spectrum disorder) and potentially establish predictive models. Moreover, a subset of the participants was assessed from pregnancy to 30 months post partum in collaboration with the Uppsala Child and Baby Lab at the Department of Psychology at Uppsala University (the BASIC Child Project, for the project descriptions, https://nyu.databrary.org/volume/828,42) between 2014 and 2018. Infants’ cognitive and social–emotional development was assessed using behavioural observations, eye-tracking measures and parent reports. Inhibitory control, executive functions, working memories and parental scaffolding behaviours were evaluated by behavioural observations. Data of this subset will also be included in the analysis.

Patient and public involvement

Patient and public involvement are essential to the current cohort profile. From pilot design to long-term follow-ups, a high degree of engagement from patients and the public continues to shape the development of this study. Our research team has also established open communication through a variety of platforms with the public from the beginning of the study. Study participants have been encouraged to contact the team regarding any comments and questions related to procedures. We have reviewed patients’ and the public’s feedback and we will continue to take this valuable information into account when examining our research approach and acquired data. The research team has close contact with several stakeholders, including clinical staff and patient organisations.

What have we learnt about peripartum depression correlates from the BASIC study?

The BASIC study aimed to identify depression and anxiety symptom patterns and their biophysiological correlates from pregnancy to 12 months post partum (for the full publication list: https://www.basicstudie.se/publicerade-arbeten). In this cohort, we characterised five distinct trajectories of peripartum depression based on symptom onset: (1) healthy (60.6%), (2) chronically depressed (14.6%), (3) depressed with early postpartum onset (10.9%), (4) depressed only in pregnancy (8.5%), and (5) depressed with late postpartum onset (5.4%).16 Interestingly, these groups exhibited significant variations in background and psychosocial characteristics. For instance, all trajectories were associated with smoking prior to pregnancy, migraine, premenstrual mood symptoms, intimate partner violence, interpersonal trauma, negative delivery expectations, pregnancy nausea and symphysiolysis. Early postpartum onset was linked to nulliparity, instrumental delivery, or a negative delivery experience, while early and late postpartum onset, along with chronic depression, were associated with other postpartum factors like low partner support, and bonding difficulties. Furthermore, we demonstrated that, in healthy pregnancy, most inflammatory markers increased in the postpartum period.43 Intriguingly, several markers such as monocyte chemotactic protein 4 tumour necrosis factor ligand superfamily member, hepatocyte growth factor, interleukin, fibroblast growth factor 23, and C-X-C motif chemokine 1, and signal transducing adaptor molecule-binding protein (STAM-BP) were related to different depression trajectories.44–47 Compared with healthy women, women with antenatal depression or with selective serotonin reuptake inhibitor (SSRI) treatment showed lower levels of peripheral inflammatory markers.48 When examining metabolic profiles in our cohort, there were no significant depression-specific differences in untargeted gas chromatography-mass spectrometry between women with depressive symptoms and those without depressive symptoms during the antenatal phase.49 However, five metabolites (glycerol, threonine, 2-hydroxybutanoic acid, erythritol and phenylalanine) were higher among women with postpartum depression than controls.50 Moreover, pregnancy is associated with a dysregulation of the hypothalamic–pituitary–adrenal axis and an increase in basal cortisol levels. Though the endocrine changes are prominent during pregnancy, there is no significant difference between women with depressive symptoms and those without depressive symptoms,51 the difference became significant post partum.52 Intriguingly, among women who did not have depressive symptoms during pregnancy, the levels of corticotropin-releasing hormone in the second trimester were positively correlated with the levels of postpartum depressive symptoms.53 Additionally, while evidence suggests that placental glucocorticoid receptors are affected by maternal depression or anxiety,54 55 results from one of our subsets contrast with some previous reports.56 57 At the gene level, single nucleotide polymorphisms in the hydroxysteroid 11-beta dehydrogenase 2 gene, a key hypothalamic–pituitary–adrenal axis gene, are significantly linked to depressive symptoms in 6 weeks post partum and the stress levels during pregnancy.58 Our data showed that women with SSRIs treatment exhibited higher levels of serotonin receptor 1A and neuropeptide Y2 receptor in the placenta compared with healthy women or women with untreated depression.59 Overall, the BASIC study provides evidence that can support the identification of individual risk profiles. Several psychosocial and emotional risk factors are consistent with previous literature.28 We have also gained a deeper understanding of several biological and psychological factors that are important for the clinical implications. For instance, biological factors, including corticotropin-releasing hormone,53 cortisol levels,52 60 inflammatory/anti-inflammatory markers,43 47 48 stress-related genetic polymorphisms58 and plasma metabolic profiling,50 might be related to different timing of the observed depressive symptoms which can support the diagnosis, treatment and further pathophysiological investigation of peripartum depression. Moreover, our evidence suggests that preventive measures during delivery should be taken to avoid anaemia, negative experience of delivery, and severe obstetric lacerations in order to reduce the risk for peripartum depression.61 62 For psychological factors, our evidence suggests that when screening peripartum depression, combining evaluation of attachment and neuroticism/trait anxiety63 64 might facilitate early identification. Currently, several ongoing analyses focus on the integration of biopsychosocial factors using conventional statistical approaches as well as machine learning techniques. Our results will provide evidence to support the development of accurate prediction as well as personalised preventive intervention and treatment.

What is the impact of peripartum depression on an infant’s outcome?

The interplay between maternal well-being, peripartum exposure and childhood environment contributes to the cognitive, social and emotional development of the child.19 20 The BASIC cohort underscored many important factors and potential underlying mechanisms behind peripartum depression. These findings have further laid the foundation to investigate the associations between peripartum depression and children’s outcomes.

From the BASIC Child longitudinal subset, we observed that among mothers with lower levels of maternal postpartum depression, infants showed more gaze-following behaviour at 10 months.65 We also observed that a higher level of exposure to interpersonal traumatic events moderated the negative impact of anxiety during the second trimester on an infant’s attentional control from 6 to 18 months.66 In the developmental literature, gaze-following is an ability to synchronise visual attention with others67 and this ability is linked to social learning and emotional regulation in the early years.68 69 Attentional control in infancy has been postulated as fundamental to later learning and cognitive development.70 The negative impact of peripartum depression on infants’ development was validated in our recent publication using U-BIRTH data at 18 months. We investigated the associations between different peripartum trajectories and an infant’s behavioural problems at 18 months. Our results demonstrated that compared with infants of healthy mothers, infants of depressed mothers showed a higher risk of behavioural problems. Maternal antenatal and persistent depression were associated with a higher degree of behavioural problems and girls were affected to a greater degree. When we analysed the mediating effect of maternal postpartum bonding, maternal persistent and postpartum depression showed a high level of indirect effect on infants’ behavioural problems, while antenatal depression had a stronger direct effect.71 We are currently investigating how maternal depression and anxiety symptoms during pregnancy are associated with infant temperament trajectories (manuscript submitted). This investigation points to a specific importance for maternal anxiety symptoms during pregnancy that are associated with infant temperament being experienced as increasingly difficult throughout the first 18 months. This is in comparison with depressive/anhedonia symptoms that are associated with infant temperament being stably medium–high in difficulty. Also, sex-specific effects were found with girls being more vulnerable towards maternal anxiety during pregnancy and boys towards maternal postpartum depression. Taken together, results from our subset (BASIC Child) and data from U-BIRTH support the hypothesis that there are negative influences of peripartum depression on an infant’s outcome at the behavioural level.

At the biological level, several studies reported a significant impact of peripartum depressive symptoms on child biological and emotional outcomes. For instance, we looked at DNA methylation in the cord blood of newborns, which is a form of pregenetic modification and most of the changes happen during embryogenesis.72 According to the fetal programming hypothesis,73 it is assumed that DNA methylation might be a possible underlying mechanism that contributes to cross-generational transmission. In our subset with 373 mother–infant dyads, we observed the differential DNA methylation in infants with mothers combined peripartum depressive symptoms and anxiety symptoms or mothers with SSRIs treatment compared with infants with healthy mothers.74 In our recent, we used DNA extracted in cord blood at birth and examined if DNA methylation differences could affect an infant’s behavioural problems at 18 months while taking peripartum depression into account. However, we did nt find evidence to support the fetal programming hypothesis.75 At the physical level, we investigated the relationship between prenatal depression, the mother’s body mass index (BMI) and infant birth weight. Large birth weight is associated with risks for several long-term adverse outcomes, such as obesity, type II diabetes and cardiovascular disease.76 The main results of our study showed that maternal BMI during the first trimester modulates the association between antenatal depressive symptoms and birth weight. Consequently, in mothers who had a higher BMI and depressive symptoms during pregnancy, their infants were born with a greater body weight compared with infants with healthy mothers with higher BMI or depressed mothers with normal BMI.77 Previously, we also demonstrated that in mothers without diabetes, placental gene expression in large gestational birth weight infants influences lipid metabolism, including insulin-like growth factor binding protein 1 and leptin.78 Notably, low leptin levels have been associated with postpartum depressive symptoms.79 Although the exact mechanisms linking peripartum depression, BMI, and child outcomes need further research, our findings offer valuable insights into the early identification of high-risk groups for prevention programmes and facilitate a deeper understanding of the intricate relationship between depression and obesity. Furthermore, previous evidence has shown that skin-to-skin contact might help the infant regulate stress.80 However, mothers with peripartum depressive symptoms during pregnancy showed a delayed first breastfeeding session compared with mothers without depressive symptoms. Mothers who experienced depressive symptoms during pregnancy and had a delayed initiation of breast feeding after birth continued to provide less exclusive breastfeeding at 6 weeks post partum.81 Our results support the notion that both infants and depressed mothers might benefit from targeted breastfeeding support during the first hours after delivery.

Strengths and limitations

This project has one of Scandinavian’s largest longitudinal cohorts, providing comprehensive and rich data across pregnancy, infancy, childhood and preteen phases. To disentangle the complex interaction between maternal well-being, peripartum exposure and childhood environment and how they contribute to the cognitive, social and emotional development of the child, this study includes multidimensional data from biological, psychometrical and behavioural measures. However, several limitations may limit the generalisability of the findings. First, despite including a relatively large sample, the acceptance rate at the baseline of the U-BIRTH cohort from the BASIC cohort was low (ca. 50%). Second, from the participants’ sociodemographic information, most parents enrolled hold a degree in higher education and this has been an indication of a low-risk profile.82 This might be attributed to the study being conducted in a university town, which limits the diversity of the sample. Third, assessments of maternal psychometric data were primarily self-report questionnaires. This approach resulted in a lack of important objective information relating to maternal cognitive functioning and IQ, which could have an impact on, children’s outcomes.83 84 Fourth, mothers experiencing depressive symptoms might have a higher dropout rate compared with mothers without depressive symptoms. Similarly, the cohort’s representability might decrease over time due to the increased rate of missing data and decline in participation. Consequently, our data might not fully capture data on the exact severity level and the possible fluctuation of depressive symptoms over time after the peripartum period.

With regard to outcome measures, though assessments of child development included in the project are well accepted and widely applied in the field, most of them rely heavily on parent-report questionnaires. This could introduce information bias. Moreover, our study only recruited Swedish-speaking mothers. This admittedly excludes many families with an immigrant background who might have other risk factors, such as lack of social support from extended family, language barriers and so on. This could also limit the generalisability of our results. Finally, our project covers a wide range of information, including socioeconomic background, living conditions, maternal health conditions, various lifestyle factors and so on. When analysing data and interpreting our results, we shall very carefully consider the potential heterogeneity present in the dataset. This entails recognising and accounting for the diverse characteristics within our cohort, as they may influence the outcomes and conclusions drawn from our study.