This study was a secondary analysis of data from a household-based matched case-control study conducted in the Manicaland and Matabeleland South provinces of Zimbabwe. The parent study enrolled caregiver–child pairs using a 1:2 matched design to investigate determinants of child nutritional outcomes. Detailed procedures of the parent study have been described elsewhere.²¹ For the purposes of the present analysis, we focused specifically on the association between maternal CMD symptoms, caregiving practices and child nutritional outcomes.

The study was conducted in Manicaland and Matabeleland South provinces, predominantly rural regions of Zimbabwe characterised by high levels of food insecurity, limited access to specialised mental health services and a high burden of childhood undernutrition.

The parent dataset comprised 450 caregiver–child pairs. For this secondary analysis, we restricted the sample to biological mother–child dyads in which the mother completed the Shona Symptom Questionnaire (SSQ-14). Caregivers who were not the child’s biological mother or had missing SSQ-14 data (n = 53) were excluded. The final analytic sample, therefore, consisted of 397 mother–child dyads.

Although the original study applied matching, the current analysis did not retain the matched structure. Matching was broken because of the exclusion of non-maternal caregivers and missing mental health data. Accordingly, standard regression modelling was applied, with adjustment for relevant maternal, child and household covariates to account for potential confounding.

Maternal CMD symptoms were assessed using the SSQ-14, a locally developed and validated screening tool widely used in Zimbabwe.²² The SSQ-14 consists of 14 binary (‘yes/no’) items capturing symptoms of depression, anxiety, somatic distress and psychotic experiences over the preceding 7 days.²¹

Outcomes were grouped into child nutritional outcomes and caregiving and health-service utilisation indicators. Anthropometric indicators were defined using the World Health Organization child growth standards. A child was considered stunted if their height-for-age Z-score was below −2 standard deviations (s.d.), wasted if their weight-for-height Z-score was below −2 s.d. and overweight if their weight-for-height was above +2 s.d.

Caregiving and health services indicators included:

Potential confounders were selected a priori based on existing literature and included maternal age, maternal education, maternal occupation, partner education, partner occupation, household asset index, marital status, parity, number of siblings, sibling age spacing, child age, child sex, child human immunodeficiency virus (HIV) status and birthweight history. Unless otherwise stated, all covariates were included in each multivariable model to ensure consistency across analyses.

Descriptive statistics were used to summarise participant characteristics. Means and s.d. were reported for continuous variables, while frequencies and percentages were used for categorical variables.

Associations between maternal CMD symptoms and caregiving and nutritional outcomes were examined using multivariable logistic regression models, with the SSQ-14 total score entered as a continuous exposure. To calculate the prevalence of probable CMD, a score of ≥ 7 was adopted from the validation study in Zimbabwe.²⁰ For all regression analyses, maternal CMD symptoms were treated as a continuous variable, represented by the total SSQ-14 score (range: 0–14). This approach preserves information on symptom severity and avoids misclassification associated with dichotomisation.²³ This approach is consistent with methodological literature demonstrating that depressive and anxiety symptoms exist along a continuum rather than as discrete categories, and that continuous modelling improves statistical power while reducing misclassification bias.^24,25 Results are presented as adjusted odds ratios (AORs) with 95% confidence intervals (95% CI). Statistical significance was set at p < 0.05. All analyses were conducted using R Studio (version 4.4.2).

Permissions were granted by the Zimbabwe Ministry of Health and Child Care. Research participants signed a consent form after explanations of the study procedure in their preferred language. No personal identification details were collected during data collection. The Biomedical Research Ethics Committee at the University of KwaZulu-Natal (BE109/19) and the Medical Research Council of Zimbabwe (MRCZ/A/2479) provided ethical clearance for the study.

Table 1 presents the socio-demographic, household and child characteristics of the 397 biological mother–child dyads included in this secondary analysis. The mean maternal age was 26.86 years (s.d.: 7.14), while the mean child age was 17.82 months (s.d.: 12.80), and the mean partner age was 32.61 years (s.d.: 8.53). Most mothers (65.00%) and their partners (79.10%) had attained at least a high school level of education, and over half of the mothers (58.40%) were unemployed. Among children, 32.50% were stunted, 8.60% were wasted, 21.91% were overweight and 5.00% were living with HIV. Nearly half of households (48.36%) were within the middle household asset index category.

The prevalence of probable maternal CMD was 34.0% (95% CI: 29.4–38.9), with a mean SSQ-14 score of 4.52 (s.d.: 4.49). Anxiety-related symptoms were the most frequently reported. Nearly half of mothers (49%) reported excessive worrying or overthinking, 47% reported difficulty sleeping, 42% experienced nightmares and 41% reported problems with concentration. Depressive symptoms were also common, with 43% reporting that they found life very difficult and 27% reporting persistent unhappiness. Somatic symptoms included persistent tiredness (33%) and frequent stomach pain (31%). Psychotic symptoms, assessed through reports of hearing or seeing things that others do not, were reported by 17% of participants (Figure 1).

Several maternal, child, and household factors were associated with maternal CMD symptom scores (Table 2). Increasing child age was positively associated with higher CMD scores (β = 0.015; 95% CI: 0.002–0.028). Higher maternal education (β = −0.550; 95% CI: −1.018 to −0.082) and higher spousal education (β = −1.267; 95% CI: −1.684 to −0.850) were associated with lower CMD symptom scores. Mothers engaged in informal employment (β = 1.299; 95% CI: 0.826–1.771) and those who were unemployed (β = 0.801; 95% CI: 0.184–1.418) reported higher CMD symptom scores, as did mothers whose partners were unemployed (β = 0.703; 95% CI: 0.249–1.157). Higher parity was associated with increased CMD symptom scores (β = 0.408; 95% CI: 0.101–0.714). Having a child who was HIV-infected, wasted or stunted, as well as a history of low birth weight, was also associated with higher maternal CMD symptom scores.

Figure 2 summarises the associations between maternal CMD symptom scores and caregiving practices, health-service utilisation and child nutritional outcomes. Higher maternal CMD symptom scores were associated with lower odds of ANC attendance (AOR = 0.55; 95% CI: 0.21–0.88), lower likelihood of children being up to date with vaccinations (AOR = 0.88; 95% CI: 0.71–0.97) and reduced uptake of vitamin A supplementation (AOR = 0.46; 95% CI: 0.11–0.70). No statistically significant associations were observed between maternal CMD symptoms and iron-folate supplementation during pregnancy or delivery at a health facility.

Regarding child-feeding practices, higher CMD symptom scores were associated with increased odds of early weaning (AOR = 1.67; 95% CI: 1.19–2.38). Positive associations with early introduction of complementary feeding and low meal frequency did not reach statistical significance.

Higher maternal CMD symptom scores were strongly associated with childhood stunting (AOR = 2.60; 95% CI: 1.78–3.80) and childhood wasting (AOR = 3.61; 95% CI: 1.65–5.57). No significant association was observed between maternal CMD symptoms and child overweight or obesity. Maternal CMD symptom scores were not significantly associated with reported recent childhood diarrhoeal episodes (AOR = 0.83; 95% CI: 0.61–1.05).

By modelling a comprehensive set of household-, maternal- and child-level variables simultaneously, this study provides a more nuanced understanding of the interrelated drivers of maternal CMDs and child nutritional outcomes than many earlier single-factor analyses. The inclusion of both maternal and spousal socioeconomic indicators, child health characteristics and household composition variables allowed a multidimensional exploration of risk and protective factors in the Zimbabwean rural context.

However, several limitations should be acknowledged. The study utilised secondary data derived from a case-control design that was not originally intended to examine maternal CMDs as primary exposures or outcomes. Consequently, some relevant psychosocial or contextual factors that could influence maternal mental health and caregiving behaviours may not have been captured, potentially introducing residual confounding. Additionally, maternal CMDs were assessed at a single time point, limiting the ability to infer temporal or causal relationships. Given that chronic malnutrition, such as stunting, develops cumulatively over time, a cross-sectional measure of maternal mental health may not fully reflect the dynamic and potentially fluctuating nature of CMDs in relation to child growth trajectories.

Moreover, data collection for key historical measures, such as birth weight, height, HIV status and maternal ANC attendance, relied on information extracted from child health cards. This approach, while practical, led to the exclusion of participants without these records, potentially introducing selection bias by disproportionately excluding more socioeconomically disadvantaged households or those with weaker engagement with health services.

While the original dataset was derived from a matched case-control design, matching could not be retained in this secondary analysis, so standard regression methods were applied. Residual confounding remains possible, particularly from unmeasured psychosocial stressors such as intimate partner violence or social support.

Finally, the study assessed CMDs only among mothers of children aged 6 months and older, potentially overlooking the critical perinatal and early infancy periods when maternal mental health exerts the strongest influence on caregiving and nutritional outcomes. Social desirability effect from self-reported sensitive outcomes, such as mental health symptoms and household socioeconomic status, could also result in reporting bias.

The findings underscore the importance of integrating mental health screening and support into routine maternal and child health services. Identifying maternal CMD symptoms at the primary healthcare level could help flag households at increased risk of suboptimal caregiving and child malnutrition. Integrating mental health support with nutrition counselling and social protection interventions may offer a pragmatic approach to addressing the interconnected challenges faced by mothers and children in similar settings. Strengthening intersectoral collaboration between the health, nutrition, social welfare and community development sectors may further enhance the coordination and sustainability of such interventions. Community-based partnerships could also support early identification, referral and continuity of care for vulnerable mothers and children.

To address the limitations of a cross-sectional dataset, conducting longitudinal research will more accurately capture how fluctuations in maternal mental health over time influence child growth trajectories and care practices. In addition, future studies should employ inclusive, diverse data collection methods that capture the experiences of mothers outside the formal health system. Intervention studies that integrate maternal mental health support with nutrition and social protection programming are needed to provide causal evidence of impact.

Randomised and quasi-experimental designs could help determine whether improving maternal psychological well-being translates into sustained improvements in child nutrition and health outcomes. Qualitative research will also be crucial in unpacking cultural conceptions of maternal distress, caregiving expectations and barriers to mental health care, ensuring that interventions are both contextually appropriate and socially acceptable.