Mental health related determinants of parenting stress among urban mothers of young children – results from a birth-cohort study in Ghana and Côte d’Ivoire

Becoming a mother and caring for an infant is both a joyful experience and a challenging and stressful one, with stress levels usually lessening as mothers gain confidence with their parenting abilities [1]. Parenting stress (PS) refers to specific difficulties in adjusting to the parenting role and arises when demands associated with parenting cannot be met by perceived resources [2]. PS has been associated with a range of negative outcomes, including less sensitive caregiver-child interaction, aversive and coercive disciplinary strategies, and increased risk of family dysfunction and child maltreatment. High PS may have long-term effects on child cognitive and behavioral outcomes [3‐5].

PS is a function of both parent and child characteristics as well as of caregiver-child interaction [2]. Previous studies have identified several psychosocial and medical factors in mothers and children associated with PS. These include lower education, high work load, unemployment, both younger and older maternal age, higher number of children, poor social/partner support, negative life events, preterm birth, child physical or mental disorder, child temperament and child caretaking hassles [6‐10]. Chronic or severe impairment of children’s health has consistently been shown to impact on parenting and interacting behaviors. Children’s ill health may also modify other risk factors and increase PS [9, 11‐13].

Most research on PS has been done in high-income countries, targeting at-risk populations, such as parents of disabled or very premature children. In sub-Saharan Africa, PS has been rarely examined. In the few published studies, the focus was on specific high-risk populations like caregivers of HIV-infected children [14] or grandparents of children orphaned by AIDS [15, 16]. In these studies, caregivers indicated higher levels of PS compared with populations in high-income countries. Community-based studies examining PS in a more general population of mothers, infants, and toddlers from sub-Saharan Africa have not been published.

Previous research indicates that the transition to parenthood is associated with increased psychological vulnerability because of major physical, family and social role changes and requirements to adjust to the multiple needs of the infant [17, 18]. This period may be particularly stressful for many women in African settings, who are exposed to increased self and offspring directed health risks around pregnancy, delivery, and the early years of motherhood [19, 20], and often tasked with handling the challenging aspects of early childcare without substantial spousal support [21]. Parenting difficulties and maternal mental health problems overlap to some extent, indicating complex interactions [22]. Common mental disorders like ante- and postpartum depression are frequent in African women with prevalences of about 10-15% [23‐25], comparable to western countries [23]. Postpartum distress and depression in the mother are linked to negative affect and attributions and have been found to influence the degree to which the infant’s behavioral characteristics are experienced as demanding and stressful [7, 26, 27]. Also, antepartum maternal mental health problems are associated with infant traits, e.g. the degree of affective reactivity at 4 and 5 months [28, 29], infant crying and fussing up to six months after delivery [30], and other determinants of child behavior difficulties, contributing to the subjective parenting experience [31]. Maternal depression has recently been associated with an increased hazard of severe febrile illness in the offspring [32], which may contribute to or modify PS. While several studies have shown associations between postpartum depression and dysfunctional parenting, little work has been done to explore the relationship between antepartum mental health problems and postpartum PS [33].

We aim to explore both the maternal PS levels and the influence of ante- and postnatal depression and anxiety on PS among a cohort of sub-Saharan African women in two cities with low-risk pregnancies. A longitudinal birth cohort was followed from the 3^rd trimester in pregnancy to 2 years postpartum, which allowed assessment of trends in maternal anxiety, depression, and PS over time [34]. We hypothesized that mean PS levels would be high, especially during the first year postpartum, where potential health and developmental risks (e.g., maternal and child anemia, infections) are common and infants are almost exclusively cared for by their mothers, restricting the women’s functioning in other life roles and day-to-day activities. Perinatal maternal depression and anxiety are hypothesized to be associated with PS because of typical disorder-related difficulties to cope with burdensome parenting role requirements and cognitive perceptions of the child as being difficult by behavior and temperament [35‐37].

Women who did not return for any of the follow up visits (3, 12, or 24 month) were excluded from this analysis (n = 60). Table 1 shows the baseline demographic characteristics of the included and excluded women. The included women had a mean age of 29 years, and 55% of them had basic or no education. The included and excluded women were not statistically different in age and education level, but the women who were excluded had higher proportions of depression and anxiety. Prevalence of depression and anxiety antepartum were 26.5% and 11.5% in Ghana and 27.8% and 14.1% in Côte d’Ivoire, respectively, and decreased after delivery (Table 2). Among women who were depressed at baseline, 18.2%, 17.0% and 14.0% were still depressed at 3, 12, or 24 month. Among women who were anxious at baseline, 22.5%, 20.3% and 9.1% were still anxious at 3, 12, or 24 month. A large proportion of women who were depressed/anxious postpartum were also depressed/anxious antepartum. For example, among 67 women who were depressed at 3 month, 30 (44.8%) were also depressed antepartum. However, only one woman was constantly depressed and anxious at all time points.

A total of 577, 531 and 264 women completed the PS assessment at 3 month, 12 month and 24 month postpartum across the two sites and the mean total parenting stress score at each time point was 85.1, 78.3 and 72.3 in Ghana and 85.9, 86.2 and 84.0 in Côte d’Ivoire, respectively (Tables 3 and 4). At all three time points, depressed mothers had higher PS scores than non-depressed mothers. Likewise, women who were classified as anxious had higher PS scores than those who were not. The difference in the total PS score between depressed and non-depressed mothers and between anxious and non-anxious mothers was about 10-20. The average PS score among mothers classified as depressed or anxious was almost always higher than 90, indicating clinically significant PS. The proportion of clinically significant PS was higher in mothers who met criteria for depression or anxiety compared with those who did not at all three time points (Tables 5 and 6).

Results from the PS subscale analyses (Tables 3 and 4) revealed that from early infancy to toddlerhood, the parental distress and parental-child dysfunctional interaction scale scores decreased in Ghana, while the difficult child subscale score, which captures perception of child temperament and affect regulation capacities, stayed constant. In contrast, values of the three PS subscales in Côte d’Ivoire did not change significantly over time.

Table 7 shows the multivariate regression results from Ghana and Côte d’Ivoire. In all univariate analyses, antepartum depression and anxiety, postpartum depression and anxiety, and antepartum anemia were significantly associated with PS. These variables were included in the multivariate models. Women’s education level, occupation, their spouses’ education level and occupation, HIV status, and child sex were not associated with PS in all unadjusted models and were not included in subsequent multivariate analyses. Women’s age was included in the multivariate regression in Ghana only, while the number of sick child visits was included in the multivariate regression in Côte d’Ivoire only.

In the multivariate regression analyses, antepartum and postpartum depression were consistently associated with PS. The difference in total PS score between women who met criteria for postpartum depression and those who did not was 8.67 and 8.17 respectively in Ghana and Côte d’Ivoire, after adjustment for other variables. Women’s anemia and younger age were associated with more PS in Ghana (B_anemia = 3.23, 95% CI: 0.47, 6.00; B_age = -0.35, 95% CI: -0.66, -0.04), but these associations were not seen in Côte d’Ivoire. PS decreased with time in Ghana (B_time = -6.63, 95% CI: -7.73, -5.53), but not in Côte d’Ivoire. Antepartum and postpartum anxiety were not associated with PS in either country.

We assessed the overall levels of parenting stress (PS), as well as different components of PS according to the three-factor model of the PSI-SF, at 3, 12 and 24 months postpartum in a sample of sub-Saharan African women after low-risk pregnancies and the birth of an infant without major disability or disease. Women who exceeded cut-off scores indicating clinically significant depression and anxiety had significantly higher PS scores than those who did not in Ghana and Côte d’Ivoire. The total PS level was similar in the two countries at 3 month postpartum, and decreased with time in Ghana, but remained consistent in Côte d’Ivoire. Antepartum and postpartum depression were consistently associated with total PS scores in both countries.

The observed mean PS score, even in mothers not classified as depressed or anxious, was relatively high compared with previous studies from high-income countries [11, 50, 51]. This was somewhat unexpected, since previous research on PS predominantly addressed parents with ill or disabled children, whose condition per se is assumed to cause distress and negatively impact family functioning [49, 50, 52]. Only one study in sub-Saharan African caregivers whose children are HIV positive reported PS scores higher than ours, which was ascribed to a “double burden” from the transgenerational disease hazard [14].

Given the lack of comparable studies on PS in sub-Saharan Africa, we can only speculate about the reasons for this finding. As most studies assessed PS when the children were older, a comparison of mean values with our sample may not be reasonable. According to a recent meta-analysis, PS in mothers of infants and toddlers with and without medical risks was found to be highest at 3 months and to decline rapidly in the first year of the children’s lives, reflecting normal psychological adaptation during early parenthood [8]. The high initial PS scores that we observed in the mothers may be partially explained by phase-specific insecure feelings about parenting, perceived lack of caregiving skills and confidence while confronting the infant’s needs and her difficult or ambiguous signals [53].

This psychobiological condition, first described by D.W. Winnicott as “primary maternal preoccupation” [54], comprises increased levels of distress, worries and anxiety and may be even functional because it can result in frequent checking of the fragile infant’s safety and health [55]. Although we did not investigate attitudes towards childbearing, our observed high levels of perceived maternal PS may be linked to the high proportion of unintended pregnancies occurring in sub-Saharan Africa. Safe birth control can be difficult to access or may be avoided by the women for a variety of reasons [56, 57]. Unintended pregnancies may put strain on maternal role assumption and mother-infant relationships [58, 59]. Other causes for high PS may include poverty [60], exhaustion following childbirth due to iron deficiency anemia [61, 62], pregnancy and childbirth not providing relief from daily duties, and the continued maternal responsibility for family income in addition to early childcare, that is imposed mainly on the mothers [63].

According to our longitudinal data, the trajectory of PS from 3 months to 24 months postpartum revealed a clear decline over time in Ghana. Mean scores two years post birth were consistent with those in a sample of disadvantaged single African American mothers of preschool children [49]. However, scores remained at a high mean level throughout the investigation in Côte d’Ivoire. Since the Ivorian women experienced a severe political crisis and acute threat of harm due to armed conflict and displacement during the assessment period, these mothers were likely to perceive higher distress, which may have also affected their parenting capacities.

In our sample, Parental Distress made the largest contribution to mean general PS scores in both countries, while Parent-Child Dysfunctional Interaction scores contributed less. This result is consistent with findings from a South-African population of HIV-positive children [14] and implies that in our sample, parental role requirements and contextual conditions were perceived as more burdensome than the relationship with the child as such. Parental Distress and Parent-Child Dysfunctional Interaction resolved with time in Ghana, but not in Côte d’Ivoire. Interestingly, scores on the Difficult Child subscale, which represents parental interpretation of basic behavioral characteristics of the children, kept constant and at a similar level in the two countries. Means were high, even in relation to studies on infants and toddlers with biological risks or from disadvantaged backgrounds in high-income countries [49, 51]. This may imply that the mothers’ perception of their children as being difficult and demanding do not only reflect stable traits of temperament in their offspring, but also normative cultural expectations of obedience and adaptability to parental authority [64]. In Western surveys, small children reported as being more difficult have been found to receive less responsive mothering, putting them at increased developmental risk [5, 65], while this link has not yet been established for African populations.

As hypothesized, we found that antepartum and postpartum depression were associated with PS, while antepartum and postpartum anxiety were not after adjusting for confounders. The total PS score may range from 30 to 180. The difference in total PS score between women who met criteria for postpartum depression and those who did not was 8.67 and 8.17 respectively in Ghana and Côte d’Ivoire, after adjustment for other variables. The magnitude of the difference is comparable to the differences detected in studies using PSI-SF to compare PS between parents of children with impairment such as cerebral palsy and parents of healthy children [52]. The association between postpartum depression and PS has been replicated in a number of studies. The direction of the association, however, may be complex and reciprocal [66‐69].

Contrary to our expectation, depressed and non-depressed mothers did not differ significantly on the Difficult Child scale. Our study did not provide support for the assumption that mothers who are depressed feel emotionally overburdened, are biased to negative cognitions and attributions and may judge their child as less adaptable and more difficult to manage than non-depressed mothers [35]. Given that the children in our study were generally perceived as more difficult than in other settings, maternal mental health does not seem decisive for this finding.

The association between antepartum depression and PS is still inconclusive. The woman’s depressed mood may influence the prenatal expectations and representations concerning herself as a parent and her relationship with the infant, and this may impact on the way she feels and acts postnatally [70]. Some studies suggested that antepartum depression was the strongest predictor of postpartum depression, which in turn was the predictor of PS [22]. However direct evidence of association between antepartum depression and postpartum PS controlling for postpartum depression is rare. Misri et al. found that treatment of antepartum depression did not impact PS [71]. Yet, we found that antepartum depression was independently associated with PS.

We did not replicate some previously described predictors for PS in our study. For example, the number of children in the household was not associated with PS in our study. It is possible that help with the older offspring from other family members mitigated the effects of having many children in the household. Lower education and decreased income have been associated with PS in some previous studies [72], but this association was not seen in our sample. Mothers in both urban settings were comparably well off and childhood malnutrition as an indicator of poverty was rare [32].

This study has limitations. The GAD-7 has rarely been used in African women, and its content is not specifically related to the perinatal mental condition. Almost all of the studies on psychometric properties of the GAD-7 focused on high-income western countries. The reliability of the PHQ-9 and the GAD-7 was moderate, which impairs the precision of the measurements. We examined reasons for this moderate reliability elsewhere (Barthel et al.; manuscripts submitted). In short, a different understanding of item wordings may have contributed to this. Moreover, 60 women who did not return for any follow up visits after birth were removed from the analysis. These women had higher PHQ-9 and GAD-7 scores when compared with the women who returned for follow up visits. The drop out of the 60 high risk mothers likely resulted in an underestimation of the PS. Lastly, there may be unmeasured confounders, such as social support from partners or relatives, marital conflict and domestic, intimate partner violence, and children’s mental health and attachment.