Maternal Depressive Symptoms and Parenting Stress in Young, Low-Income Mothers: Bidirectional or Unidirectional Effects?

Background/Purpose:  Postpartum depression is common, especially among mothers at social and demographic risk, and can contribute to problems in parenting and maternal and child development.  Parenting stress, specifically the mother’s perceptions of her child and her relationship with her child as being difficult, can lead to similar negative outcomes.  Many studies have shown that parenting stress and maternal depression are strongly associated, but the temporal direction of this relation has seldom been put to rigorous empirical test utilizing longitudinal data analysis.  Understanding whether there are bidirectional/transactional effects (i.e., parenting stress predicts later depression and the other way around) or unidirectional effects (e.g., depression predicts later parenting stress but not the other way around) could help practitioners choose entry points to offer psychosocial interventions for new mothers.  This study examines the associations between parenting stress and depressive symptoms throughout the first two years postpartum in a sample of young, low income mothers.

Methods: 248 young (M=18.3yrs, SD=1.7), low-income, African American women were recruited from the prenatal clinics of an urban teaching hospital to participate in a longitudinal study of pregnancy and parenting.  At 4-, 12- and 24-months postpartum, mothers were interviewed about their parenting stress using the Parenting Stress Index-Short Form (PSI-SF) and depressive symptoms using the Center for Epidemiologic Studies-Depression Scale (CES-D).  Cross-lagged structural equation modeling was conducted via four models: Model 1 (including only stability in CES-D and PSI scores over time), Model 2 (Model 1+cross-lagged effects of PSI predicting future CES-D), Model 3 (Model 1+cross-lagged effects of CES-D predicting future PSI) and Model 4 (Model 1+all cross-lagged coefficients).  Covariates, such as prenatal depression, mother’s cognitive ability, and social support that were significantly associated with PSI and CES-D at each wave were controlled for in all models.  Full information maximum likelihood was used to account for missing data.

Results: All four models demonstrated good model fit (RMSEA<0.05, CFI/TLI>0.90, and chi-square statistics non-significant).  PSI and CES-D both showed stability over time, as shown by significant positive stability coefficients.  The cross-lagged coefficients from PSI to CES-D were significant in Model 2 but the coefficients from CES-D to PSI were not significant in Model 3.  Model 2, but not Model 3, significantly improved model fit compared to Model 1. Model 4 did not improve model fit compared to Model 2.

Conclusions/Implications:  Using longitudinal data and controlling for the stability in parenting stress and depressive symptoms over time, this study identified a unidirectional relation from parenting stress to depressive symptoms, as opposed to a transactional relation or a unidimensional relation from depressive symptoms to parenting stress.  This finding that parenting stress may precede maternal depression, is consistent with that of an existing study of primarily white, middle-class mothers.  Social workers and mental health providers who offer psychosocial support to young, low-income mothers should be aware that the stress of caring for an infant may give impetus to the development of postpartum depression.  Interventions that target the mother-infant relationship and help mothers manage difficult infant behaviors may alleviate both parenting stress and depressive symptoms.