Father involvement, maternal depression and child nutritional outcomes in Soweto, South Africa

Father involvement in South Africa is low, despite evidence that it can improve maternal and child health and wellbeing. Within a larger randomised controlled trial, we assessed whether father involvement during and after pregnancy increased birth weight and exclusive breastfeeding through improved maternal mental health. At 6-week postnatal, mothers completed questionnaires on birth, feeding practices, social support, father involvement and postnatal depression. Father involvement during pregnancy was measured by their attendance at antenatal care and the study intervention, whereas postnatal involvement was measured by attendance at antenatal care and type of paternal support provided. Structural equation modelling was used to identify associations between father involvement, maternal depression, low birth weight and exclusive breastfeeding. Among the 212 mother–baby pairs, father involvement was very low with only 43%, 33% and 1% of partners attending early ultrasound, antenatal care and the birth of the child, respectively. Twenty-nine percent of the mothers showed signs of depression during pregnancy, compared with 7% after birth. Eighteen percent of the infants were born low birth weight, and 57% of mothers reported exclusively breastfeeding at 6 weeks. Father involvement was directly associated with postnatal depression, but it did not directly or indirectly impact exclusive breastfeeding or low birth weight. We conclude that postnatal father involvement can improve postnatal maternal depression and that men would benefit from specific guidance on how they can support mothers during and after pregnancy. This paper reports on the 6‐week follow up from a randomised controlled trial assessing the impact of augmented ultrasound during pregnancy on healthy child growth and development compared with standard practice of care. Recruited participants were blindly allocated to the intervention or control arms. A routine ultrasound scan was offered to all participants in the study. The ultrasound examination in the control group followed standard practice of care, whereas the intervention group received additional messages to promote early childhood development, an educational baby book and a printed and electronic image of their ultrasound scan. Full details of the trial intervention and methods can be found in Richter et al. (2020). The study is situated in Soweto, South Africa’s largest township located near Johannesburg. All study procedures took place between March 2019 and August 2020 in a research unit at Chris Hani Baragwanath Hospital (CHBH), a tertiary hospital in Soweto. Participants were recruited from the Foetal Medicine Unit (FMU) at CHBH and invited to participate if they had a singleton pregnancy <25 weeks gestation and were over the age of 18 years and lived within Soweto. Women who were found to have severe maternal comorbidities or major foetal abnormalities were excluded from the study but remained in clinical care at CHBH. Participants who attended the ultrasound were requested to attend follow‐up visits with their infant 6 weeks after birth. Mothers in both the intervention and control groups were encouraged to invite the father of the baby to attend all study appointments and were provided with invitation cards to give to their male partner on recruitment. During the ultrasound visit, women completed individual questionnaires on their sociodemographic characteristics and relationship status with the father of the baby. Antenatal depression was measured using the Edinburgh Postnatal Depression Scale (EPDS), a 10‐item self‐reported depression scale that has been validated for use during the antenatal period (Choi et al., 2012) and in South Africa (Lawrie et al., 1998). As per recommendations by Cox et al. (1987), mothers with a score of ≥10 were referred to a specialised nurse for further assessment and to receive counselling if required. At the 6‐week follow‐up visit, the mothers were asked questions related to the birth of their baby, including the type of delivery and complications experienced by the mother or baby, as well as questions on their feeding practices, such as when the baby was put to the breast after birth. Mothers were asked if they had given their baby formula, water or porridge, and also asked to list all the foods and liquids they had given their baby up until the time of the interview. Those who did not report giving formula, water, porridge or other substances were recorded as having exclusively breastfed up to 6 weeks. To limit the overestimation of infants exclusively breastfed at 6 weeks, we chose not to follow the WHO infant and young child feeding guidelines that use a 24‐h recall method (WHO, 2008). Mothers were also asked about partner support and involvement during and after their pregnancy, specifically men's attendance at health checks or the birth, and the types of support partners were providing. Finally, mothers completed the EPDS for an assessment of postnatal depression and were referred for further assessment with a score of ≥10. Details of the infant, including their date of birth and birth weight, were taken from the Road to Health Book that is maintained for all children from birth in South Africa to keep track of their health, growth and development. An index of socio‐economic status (SES) was calculated using a principal component analysis (PCA) based on data collected at baseline. Variables included in the index were crowding (number of rooms used for sleeping), household water and sanitation, internet access, and ownership of a washing machine, satellite television, DVD player, computer and car. The index was divided into tertiles. Antenatal father involvement was based on whether the partner attended the study ultrasound or additional antenatal health visits and was recorded as none, at least one or both. Postnatal father involvement was measured using a cumulative score comprising different forms of father involvement, such as responsibility or provisioning as per recommendations from Greene et al. (2001). This included whether the father had provided information, practical, emotional or financial support, was rated by the mother as providing her with the most help with the baby or her own health and how she rated this help, whether he made things harder for the mother by his behaviour or attitude and his attendance at antenatal health visits. We opted to use maternal reports of father involvement as perceived support is important, and we anticipated that father attendance at 6‐week follow‐up would be low. Low birth weight was classified as <2.5 kg. Binary variables were coded using 1 and 0 for yes and no responses, respectively. Data were captured and analysed using Stata I/C 14. Categorical descriptive statistics are presented as frequencies and proportions. Continuous descriptive statistics are presented as minimum, maximum, mean and standard deviations. Using structural equation modelling (SEM), two individual models were built to test our hypotheses; Model 1 assessed whether antenatal father involvement improved birth weight by improving antenatal maternal mental health, and Model 2 assessed whether father involvement during and after pregnancy increased exclusive breastfeeding through improved postnatal mental health. Antenatal variables were used in Model 1, whereas a combination of antenatal and postnatal variables was used in Model 2. Simple linear regression was undertaken to identify potential confounding variables for antenatal and postnatal maternal mental health. Variables with a p value < 0.10 were included in the SEM. For the main outcome variables (birth weight and exclusive breastfeeding), simple linear and univariate logistic regression, respectively, was conducted with all potential confounding maternal variables included in the SEM if the p value was <0.10. Intervention arm was controlled for both models. Figures 1 and ​and22 show the path diagrams used to guide each of the models. Due to the differing types of outcome data (continuous and binary), the results of the SEM are presented as total, direct and indirect β coefficients. We assess the fit of each model using the root square mean error of approximation (RMSEA), the comparative fit index (CFI) and the Tucker–Lewis index (TLI). A well‐fitting model has RMSEA < 0.05, and CFI and TLI > 0.95. All statistical tests were considered significant with a p value of <0.05, and 95% confidence intervals are presented. Pathway diagram for Model 1: antenatal father involvement, antenatal maternal mental health and birth weight. SES, socio‐economic status Pathway diagram for Model 2: father involvement, postnatal maternal mental health and exclusive breastfeeding. SES, socio‐economic status The required sample size for the trial was 100 participants per arm. In total, 249 pregnant women were enrolled in the trial, with 212 attending the 6‐week follow up. We conducted a reverse power analysis using G*power to determine whether the sample size of 212 was sufficient to conduct an SEM analysis. Based on an effect size of 0.17, alpha of 0.05 and two predictors, the power was 0.99 and the sample size sufficient for analysis. The study was approved by the Human Research Ethics Committee (Medical) of the University of the Witwatersrand, South Africa (M181915). An amendment was approved by the University of the Witwatersrand in order to film some of the ultrasound appointments. Permission was obtained from CHBH to conduct the study and to oversee adherence to the study protocol. The trial is registered with the Pan African Clinical Trials Registry (PACTR201808107241133). Participants provided written consent and were given a unique study identifier to maintain confidentiality.