Effect of combining lower- and higher-value monthly cash transfers with nutrition-sensitive agriculture, male engagement and psychosocial intervention on maternal depressive symptoms in rural Malawi: a secondary analysis of a cluster-randomised controlled trial

Abstract

Maternal depression affects one in five women in Malawi. Integrated interventions simultaneously addressing multiple risks are a promising strategy to improve mental health. This study evaluated the impact of a nutrition-sensitive social behaviour change (SBC) intervention (agriculture and livelihoods, male engagement and Caring for the Caregiver) with or without cash transfers on maternal perinatal depression during the lean season in rural Malawi. A midline survey for a cluster-randomised controlled trial was conducted, where 156 clusters were randomly assigned to four arms (39 clusters/arm): (1) standard of care (SoC), (2) SBC, (3) SBC+low cash (US$17 per month) and (4) SBC+high cash (US$43 per month). Pregnant women and mothers of children <2 years of age (n=2677) were enrolled at baseline (May–June 2022). A subsample of 1303 women was followed-up at midline (November–December 2023). Maternal perinatal depression was assessed using the Self-Reporting Questionnaire with a score of ≥8 indicating symptoms consistent with depression. Intervention effects were estimated using linear mixed effects models. At midline, SBC+high cash reduced depression scores relative to SoC (mean difference −1.13 (95% CI −1.96 to –0.31)) but had no impact on the proportion of women with depressive symptoms. Relative to SoC, SBC+low cash and SBC alone had no impact on depression scores or the proportion of women with depressive symptoms. Relative to SBC alone, adding cash to SBC reduced depression scores and the proportion of women with depressive symptoms regardless of the size of the cash transfer. Cash transfers integrated with SBC can benefit maternal perinatal mental health in rural Malawi during the lean season. Trial registration number ISRCTN53055824.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Cash transfers can improve mental health by addressing financial and economic risks, but evidence on perinatal mental health is limited, particularly during periods of peak food insecurity.

• Meta-analyses show no overall impact of parenting interventions on maternal mental health. However, individual interventions, including those engaging male caregivers, have shown positive effects on perinatal mental health in several African countries, although not in Malawi.

• Nutrition-sensitive agricultural interventions can reduce depressive symptoms by improving food security, food access and household income; however, evidence is scarce.

• Evidence on the additive or synergistic effects of cash transfers, nutrition-sensitive agriculture and parenting interventions on perinatal mental health is lacking.

WHAT THIS STUDY ADDS

• Larger case transfers with nutrition-sensitive social behaviour change (SBC) interventions comprising nutrition-sensitive agriculture, male engagement and psychosocial components reduced maternal depressive symptoms during the lean season. Smaller cash transfers with SBC or SBC alone had no impact on perinatal mental health relative to the standard of care.

• Cash transfers had an added benefit for maternal depressive symptoms on top of SBC regardless of the size of the transfer.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Comprehensive intervention packages addressing multiple vulnerabilities and risks for mental health and providing a variety of coping and thriving strategies can help improve perinatal mental health, particularly during peak periods of food insecurity.

• Additional research is needed to understand the precise pathways through which integrated intervention packages work to improve perinatal mental health.

Introduction

Mental disorders are a major public health concern globally and a leading cause of years lived with disability. Depression and anxiety are the two most common mental disorders globally.¹ Maternal mental disorders, particularly perinatal depression which occurs during pregnancy and the first year post-partum, can lead to pregnancy-related morbidity, mortality and adverse birth outcomes.²,⁵ Maternal depression can also adversely affect parenting and caregiving practices throughout infancy and childhood and result in poor child health, growth and development.⁵,⁹

In Malawi, depression is the second most prevalent mental disorder, with the prevalence increasing by 20% over the past decade.¹⁰ Perinatal depression affects one in five women, with prevalence ranging from 10% to 33%.¹¹ Among the primary risk factors for perinatal depression empirically demonstrated in Malawi are low socio-economic status, lack of social support, experience of intimate partner violence and HIV infection.¹²,¹⁶ There is no routine screening for perinatal depression in primary care settings in Malawi.¹¹ There is also a substantial treatment gap defined by a lack of mental health professionals.¹⁷ Therefore, improving prevention by addressing key risk factors can be a viable strategy to reduce the burden of perinatal depression in the country.

Given that poverty is a major risk factor for depression,¹⁸ and that depression and poverty are mutually reinforcing,^{1 19} social protection interventions (ie, interventions aiming to reduce poverty and increase household resilience to shocks) are a promising strategy to improve mental health.²⁰,²³ Consistent evidence shows that cash transfers improve mental health with larger effects for larger or unconditional cash transfers.²⁰,²³ However, much of this evidence is ‘very low quality’, arising from studies with serious risk of bias.²² In Malawi, studies show that the national Social Cash Transfer programme, providing unconditional cash transfers of US$3–7 per month (MWK2600–5200) depending on household size, improved perceived quality of life and subjective well-being.²⁴,²⁶ Further, the Malawi Incentive Programme (cash transfers of MWK500–1000 or MWK2000–4000 (US$0.30–0.6 or US$1.2–2.3, respectively) provided to individuals/couples conditional on maintaining their HIV status for a year) improved mental health, with larger effects among those with worse baseline mental health.²⁷

However, much of the evidence on cash transfers arises from large social protection programmes targeting ultra-poor and labour-constrained households rather than women during the crucial perinatal period.²⁰,²³ Presently, there is limited evidence on whether cash transfers work to improve perinatal depression. The Janani Suraksha Yojana programme in India reduced depressive symptoms in pregnant women who received cash transfers of INR1000–1400 (US$12–16) conditional on them delivering in a government health facility.²⁸ In contrast, in Tanzania, conditional cash transfers for antenatal care and child growth monitoring (TZS10 000 (US$4.30) per antenatal visit and TZS5000 (US$2.20) per growth monitoring visit) delivered through a responsive stimulation, health and nutrition intervention had no impact on perinatal depressive symptoms.²⁹ Further research is needed to better understand the impact of cash transfers on perinatal depression, as well as the effects of combining cash transfers with interventions targeting the non-economic determinants of depression.²²

In addition to poverty, food security (ie, adequate access to ‘“sufficient, safe and nutritious food that meets dietary needs and food preferences for an active and healthy life”)³⁰ is another main determinant of depression.¹⁸ The relationship between depression and food security is bidirectional, operating through multiple biological and psychosocial pathways.³¹,³³ Therefore, interventions effective in improving food security, such as nutrition-sensitive agricultural interventions,³⁴,³⁶ may be a promising strategy to reduce depressive symptoms. Nutrition-sensitive agricultural interventions can also promote mental health by alleviating poverty through increased food access and household income.³⁴ Nevertheless, evidence to date is limited to a single study from Tanzania which demonstrated that a nutrition-sensitive agricultural intervention reduced the odds of probable depression in women with young children by 43%, with improvements in food security explaining approximately one-quarter of the effect.^{37 38} Given the multiple social and cultural risk factors for poor mental health,^{1 18} addressing food insecurity alone is likely insufficient to promote mental health. More research is needed on combining interventions addressing food security with interventions targeting other economic and non-economic determinants of depression.

Among interventions targeting non-economic determinants of depression, parenting interventions (ie, interventions designed to improve parents’ caregiving knowledge, skills and practices) can help promote and protect mental health by improving parenting and caregiving practices, mother-child interactions, maternal mood and psychological well-being.^{1 17 39} Although theoretically parenting interventions can create a stable environment that supports mental health, consistent empirical evidence shows that alone they are insufficient to improve perinatal depressive symptoms,^{40 41} even when explicitly designed to do so.⁴² Some individual interventions show promise for improving perinatal depression, including those in Tanzania,²⁹ Uganda^{43 44} and Zambia,⁴⁵ as do interventions involving male caregivers.⁴⁶ To date, no studies have examined the impact of parenting interventions on perinatal depressive symptoms in Malawi.⁴⁰

Social protection, parenting and nutrition-sensitive agricultural interventions can work in additive or synergistic ways to improve perinatal depression by simultaneously addressing multiple risk factors and providing a variety of coping strategies.⁴⁷ Nevertheless, evidence remains limited to the study in Tanzania mentioned above which combined cash transfers with a parenting intervention and found no effect on perinatal depressive symptoms. However, the cash transfers provided in this study were relatively small.²⁹ Larger cash transfers are likely necessary in resource-constrained settings to adequately reduce poverty and address economic risk factors. Integrating psychosocial and psychological interventions, which are among the most effective mental health interventions,^{48 49} can help generate further synergies and enhance the effects on mental health of combining social protection, parenting and nutrition-sensitive agricultural interventions.

In this paper, we examined the impact of MAZIKO—a programme combining maternal and child cash transfers with nutrition-sensitive social behaviour change (SBC) interventions, comprising agriculture and livelihoods, male engagement and psychosocial components—on perinatal depressive symptoms in rural Malawi during the lean season. The lean season, which typically spans from November until the maize harvest begins in March, is a period of peak food insecurity with recurrent acute food insecurity shocks and large declines in food consumption and dietary diversity and quality.^{50 51}

Methods

Study design

The MAZIKO trial is described in detail elsewhere.⁴⁷ The trial was completed in June 2025. Briefly, the MAZIKO trial is a 3-year cluster-randomised controlled trial implemented in 156 clusters in Balaka and Ntcheu districts in the Southern and Central regions of Malawi. Clusters were defined as village groupings or communities in the catchment areas of community-based childcare centres (CBCCs) targeted by the interventions. Clusters were randomly assigned to one of four intervention arms (39 clusters per arm): (1) standard of care (SoC), (2) SBC, which received agriculture and livelihoods, Male Champions and Caring for the Caregiver interventions, (3) SBC+low cash, which received US$17 per month in addition to the SBC and (4) SBC+high cash, which received US$43 per month in addition to the SBC. The trial aimed to evaluate the effectiveness and cost-effectiveness of the SBC and cash transfer interventions on health and nutrition outcomes in women and young children. Maternal depressive symptoms were a pre-specified secondary outcome.⁴⁷ The trial involved outcome measurements at baseline, midline (1.5 years after baseline) and endline (3 years after baseline). The midline survey was designed to understand the fidelity, uptake and quality of intervention implementation, to measure impacts on selected outcomes likely to be affected by large seasonal changes (eg, diet and expenditures) and to assess short-term impacts. The endline survey was designed to assess the impacts on all pre-specified outcomes at the end of the 3-year programme.⁴⁷ This paper used data from the baseline and midline surveys to estimate intervention effects during the lean season, approximately 18 months post-baseline. All analyses were pre-specified.⁴⁷ The trial was prospectively registered with ISRCTN, number ISRCTN53055824.

Study population

The trial included pregnant women aged 15–49 years and mothers/caregivers (aged 15–49 years of age) of children <2 years of age at baseline living in the study clusters. At the cluster level, all women who met these inclusion criteria were eligible to participate in the MAZIKO interventions. Women were randomly selected for enrolment into the trial. First, a household census was conducted in the CBCC catchment areas prior to the baseline. Using data from the census, we constructed lists of eligible households, ie, those with women aged 15–49 years who self-reported as pregnant and/or were mothers/caregivers of children <2 years of age. Index mother-child dyads were randomly selected for inclusion in the trial. Randomisation was stratified by pregnancy status so that approximately one-half of women were pregnant and one-half were mothers/caregivers of children <2 years of age. A total of 2677 mother-child dyads were enrolled at baseline. At midline, we prioritised surveying women pregnant at baseline. Women with children <2 years of age at baseline served as replacements for the fixed budget survey.

Trained enumerators provided information about the study. Written informed consent was obtained from adult women or parents/guardians of women<18 years of age.

Randomisation and masking

Clusters were randomly assigned to one of four intervention arms using a restricted randomisation procedure, described in detail in the study protocol.⁴⁷ Randomisation was stratified by district to allow for the same approximate number of clusters within each intervention arm in each district. Treatment allocation was modelled using village-level variables to maximise balance across the intervention arms. The algorithm tested 5000 random allocations. The permutation that minimised the R2 was selected.

Given the nature of the interventions, participants and implementers could not be blinded to intervention assignment. The data collection and analysis teams were blinded to intervention allocation. After being finalised, results were unblinded by a non-study researcher.

Interventions description

The SoC intervention comprised Care Groups which delivered the government-approved curriculum to improve maternal, infant and child nutrition and early childhood development through home visits and cooking demonstrations. Home visits were conducted fortnightly. Cooking demonstrations were organised once per agricultural season after harvest time.⁵² This component was the same across all four arms.

The nutrition-sensitive SBC was designed as a package. It comprised agriculture and livelihoods, Male Champions and Caring for the Caregiver interventions. The agriculture and livelihoods intervention provided agriculture inputs (seeds) and training using CBCCs and community gardens and increased access to Village Savings and Loans Associations (VSLAs). Three types of agricultural training were conducted, with households receiving each type once with refresher training provided annually: (1) nutrition-sensitive agriculture with a focus on integrated household farming, dietary diversification, food processing, food safety and hygiene; (2) post-harvest losses with a focus on low-cost technologies; and (3) climate-smart agriculture with a focus on conservation agriculture, including soil fertility conservation and rainwater harvesting.⁵² Agricultural trainings were conducted by lead farmers, trained by members of the area and village nutrition coordination committees. Winter seeds (okra, amaranth and Ethiopian mustard) were provided once per year before planting for the winter season. Orange-fleshed sweet potato plants and rainfed seeds (groundnuts) were distributed once prior to planting for the summer season.⁵² Seed distributions were conducted by Save the Children. Under the VSLA component, Save the Children conducted a one-off training of community-based trainers on the government-approved VSL curriculum. These trainers then supported project VSLAs, including providing training on small enterprise and business skills for VSLA members. The Male Champions approach engaged fathers and husbands through fortnightly group and couple sessions to improve relationships, sharing of decision-making, resources and chores, and to reduce gender-based violence. Save the Children trained male champions, who in turn trained father leads. Father leads conducted fortnightly group sessions with men using role play and games, where men reflected on their roles as fathers and husbands. This component targeted fathers of children <5 years of age and husbands of the women enrolled in the trial.⁵² As a result, households of unmarried women enrolled in the trial (~20% of the sample) would have only been reached by the male champion interventions if a father of a child <5 years of age was a member of the household. The couple sessions were fortnightly home visits with the women enrolled in the trial and their husbands, conducted by lead fathers focusing on equitable distribution of labour, chores and childcare within the home. The Caring for the Caregiver package, originally developed by UNICEF and adapted for Malawi by the Department of Nutrition and HIV/AIDS and UNICEF, aimed to improve the capacity of Care Group promoters/cluster leaders and male champions to support and promote women’s mental health and emotional well-being. The Caring for the Caregiver package was not part of the original trial design. It was added based on consultations with the government, which was planning a wider piloting and training for the package. All nutrition-sensitive SBC interventions were delivered at the same frequency in the three treatment arms that received them. The trial was designed to test the overall effectiveness of the nutrition-sensitive SBC package, not to disentangle the effects of its individual components.

The cash transfer intervention comprised monthly cash transfers to pregnant women and mothers of young children provided through mobile phones linked to bank accounts registered with participants. Cash transfer values were set at the Malawian Kwacha equivalent of US$17 or US$43 per woman per month, at exchange rates set at baseline in June 2022. These values were determined through an analysis of household consumption and cost of diet.⁴⁷ Further details on the MAZIKO interventions are available elsewhere.^{47 53}

Pathways of impact

The MAZIKO interventions addressed multiple social and cultural distal and proximal risk factors for poor mental health.^{1 18} The SoC Care Groups were expected to improve mental health by supporting maternal and child health and nutrition, improving parenting competencies and providing social support. The Caring for the Caregiver package was expected to support mental health both directly through activities specifically designed for this purpose and indirectly by promoting social networks and reinforcing support structures through Care Groups. The agriculture and livelihood interventions were expected to improve food security and access to nutritious food through provision of agricultural inputs and training and to reduce financial constraints through improved access to VSLAs. The Male Champions intervention was expected to improve partner support, fathers’ parenting and sharing of chores and caregiving responsibilities. In addition, the cash transfers can improve mental health by addressing economic and poverty-related risk factors, reducing financial and liquidity constraints, increasing access to healthcare for the mother and child and increasing resources and resilience to negative shocks. Lastly, all intervention components were designed to increase women’s empowerment and gender equity, which are associated with improved mental health.¹⁵⁴,⁵⁶

Study timeline

The baseline survey was conducted in May–June 2022. Cluster randomisation was completed in August 2022. The Care Group and VSLA interventions started in October 2022, and the cash transfers in November 2022. Distribution of agricultural inputs and agricultural training were completed in March 2023 (winter seeds) and November 2023 (rainfed seeds). The Caring for the Caregiver and Male Champions interventions started in September 2023. The midline survey was conducted in November–December 2023. The nutrition-sensitive SBC intervention components were designed separately with their own timelines, training and implementation schedules. As a result, the staggered or layered implementation of these components was by design.

Sample size

The primary outcomes of the MAZIKO trial are women’s mean probability of adequacy of micronutrient intake and child development. Cluster and resource availability suggested ~40 clusters per intervention arm and 20 mother-child dyads per cluster. Power calculations were conducted only for the primary outcomes and suggested a minimum detectable effect size of 0.31 SD for mean probability of adequacy and 0.26 SD for child development.⁴⁷

Data collection

Structured questionnaires were used at baseline and midline to collect data on socio-economic and demographic characteristics and mental health. At midline, information on self-reported participation in the MAZIKO interventions (care groups, agricultural inputs, VSLAs and cash transfers) since the start of the programme was also collected. Spouses were not interviewed on participation in the Male Champions component. Enumerators were trained over a 3-week period prior to the baseline survey. Topics included research ethics, good clinical practice, questionnaire comprehension, interviewing techniques, CAPI software and standard operating procedures for the study. A 2-week refresher training for the same data collection team was conducted prior to the midline survey.

Outcomes

Mental health was assessed using the Self-Reporting Questionnaire (SRQ-20), which consists of 20 yes/no questions on depression, anxiety and stress symptoms.⁵⁷ Questions were asked in private, out of sight/earshot from others. The SRQ-20 has been validated in Malawi against clinical diagnosis with a score ≥8 indicating symptoms consistent with major/minor depression.⁵⁸ Internal consistency was high in our sample at baseline (α=0.84) and midline (α=0.87).

Statistical analysis

We used linear mixed effects models to estimate intent-to-treat intervention effects of the SBC, SBC+low cash and SBC+high cash intervention arms compared with SoC. Clusters (the unit of randomisation) were included as a random effect. Per protocol, primary analyses were unadjusted.⁴⁷ We also estimated adjusted estimates controlling for district and a priori selected baseline covariates, namely woman’s age (<18 vs ≥18 years), education (primary/lower vs secondary/higher), marital status (married vs unmarried), household head status (self vs other), household head gender (female vs male) and household wealth (lowest vs higher quintiles of total household expenditures). All estimates were mean differences (MDs) and 95% CIs. Missing data on SRQ-20, the outcome variable, were not imputed. Missing data on covariates were imputed using mean cluster imputation.

Monitoring data revealed that 50 women assigned to cash transfer arms were not receiving cash transfers. As a sensitivity analysis, we re-estimated the primary analyses excluding these women. As a second sensitivity analysis, we estimated the added benefit of the low cash and high cash intervention arms in addition to SBC by comparing the SBC+low cash and SBC+high cash arms to the SBC arm, as well as the added benefit of extra cash by comparing the SBC+high cash to the SBC+low cash intervention arms. Both unadjusted and adjusted models were estimated. We calculated standardised mean difference (SMD) as a measure of effect size by dividing the unadjusted MD by the pooled SD at baseline.

In addition, we examined effect modification by pre-specified baseline factors: child sex (male vs female), pregnancy status (pregnant vs child <2 years of age) and household wealth (lowest one vs highest four quintiles of household expenditures).⁴⁷ Missing data on effect modifiers were imputed using mean cluster imputation. Interactions were considered statistically significant at p<0.10. Post-hoc contrast estimation was used to generate effect estimates for each subgroup of the effect modifiers. All analyses were conducted in Stata V.18.

Results

Sample characteristics

The trial enrolled 2677 mother-child dyads (figure 1). At midline, 1303 women had mental health data at both baseline and midline. At baseline, women in the analytic sample were on average 25 years old, 78% were married, 22% had some secondary or higher education and 89% were pregnant (table 1). Their households had four members, on average, and 32% were headed by women. Mental health was poor with 30% of women experiencing symptoms consistent with depression. Online supplemental table 1 shows the proportion of women experiencing each of the SRQ-20 symptoms by intervention arm and timepoint. At baseline, characteristics were similar across intervention arms with few exceptions. A higher proportion of households in the SoC arm were in the lowest wealth quintile relative to other intervention arms. In addition, a higher proportion of household heads were female in the SBC arm compared with other intervention arms. Accordingly, a lower proportion of women were married in the SBC arm compared with other intervention arms.

Figure 1. Study flow diagram. SBC, social-behavioural change.

Table 1. Baseline characteristics of the women included in the analytic sample.

	Standard of care	Social behaviour change (SBC)	SBC+low cash	SBC+high cash
	Mean±SD or %	Mean±SD or %	Mean±SD or %	Mean±SD or %
N	348	287	357	311
Household characteristics
Household size	4.2±1.9	4.1±1.8	4.1±1.9	4.0±1.7
Total household expenditures (MWK)	35 835±29 256	37 264±31 406	37 052±29 868	36 284±29 487
Lowest household expenditure quintile	23.3	17.1	18.2	20.9
Household head characteristics
Age (in years)	33.5±11.6	32.0±10.7	33.3±11.3	33.0±10.8
Head is female	30.2	36.6	32.2	29.9
Secondary or higher education	28.5	25.8	29.4	26.4
Woman’s characteristics
Age (in years)	25.3±7.0	25.3±7.2	25.5±7.1	24.8±6.6
Woman is <18 years	11.5	11.2	8.7	10.6
Pregnant at baseline	87.1	90.2	89.3	90.8
Married or co-habitating	81.0	76.0	78.2	77.9
Secondary or higher education	21.6	21.3	22.7	21.2
Woman is head of household	22.7	25.4	23.8	18.3
Self-Reporting Questionnaire (SRQ-20) score	5.6±4.2	5.8±4.2	5.5±4.3	5.6±4.1
Has symptoms consistent with depression (SRQ-20≥8)	31.3	28.6	29.1	30.2
Child characteristics
Child is a boy	50.0	51.1	49.8	44.4
Child age (in months)	15.4±5.7	15.5±5.3	15.8±5.1	15.4±5.1

Programme participation

Self-reported programme participation indicated that Care Group membership was low, ranging from 17% in the SoC arm to 45% in the SBC+high cash arm. In terms of the SBC interventions, 64% of women in the SBC and SBC+low cash arms and 74% in the SBC+high cash arm reported receiving agricultural inputs in the past 12 months. VSLA membership ranged from 36% in the SoC arm to 48% in the SBC+low cash arm. With respect to cash transfers, 58% of women in the SBC+low cash arm and 68% in the SBC+high cash arm reported being registered in a cash transfer programme. Notably, monitoring data showed that only 50 women (4% of the analytic sample) were not receiving cash transfers. A handful of women in the SoC arm reported being registered in any cash transfer programme (3%) or receiving agricultural inputs (3%). Likewise, 5% of women in the SBC arm reported being registered in any cash transfer programme, likely reflecting receipt of the Malawi Social Cash Transfer programme.

Programme impact

The SBC+high cash intervention significantly reduced SRQ-20 scores at midline by MD −1.13 (95% CI −1.96 to –0.31) as compared with SoC, but had no impact on the proportion of women experiencing depressive symptoms (figure 2, online supplemental table 2). We observed a shift in the distribution of SRQ-20 scores such that the SBC+high cash intervention benefited all women (online supplemental figure 1). The SBC and SBC+low cash interventions had no effect on SRQ-20 scores or the proportion of women experiencing depressive symptoms relative to SoC. Intervention effects were similar after adjusting for covariates, although point estimates were smaller (online supplemental table 3). Results from the sensitivity analyses excluding the 50 women who were not registered to receive cash transfers were also consistent (online supplemental table 4).

Figure 2. Effect of the MAZIKO interventions relative to the SoC on depression scores (A) and the proportion of women with depressive symptoms (B). SBC, social behaviour change; SoC, standard of care; SRQ, self-reporting questionnaire.

The sensitivity analyses examining the added benefit of the cash transfer compared with SBC alone showed that both SBC+low cash and SBC+high cash reduced SRQ-20 scores relative to SBC: MD −1.00 (95% CI −1.84 to –0.17) and MD −1.48 (95% CI −2.33 to –0.63), respectively (figure 3A, online supplemental table 5). Likewise, SBC+low cash and SBC+high cash reduced the proportion of women with depressive symptoms by 9 and 11 percentage points relative to SBC, respectively (figure 3B, online supplemental table 5). When comparing SBC+high cash and SBC+low cash, there was no added benefit of the higher cash transfer on SRQ-20 scores or the proportion of women with depressive symptoms (figure 3B, online supplemental table 5). Results were consistent after adjusting for baseline covariates (online supplemental table 6).

Figure 3. Added benefit of the low and high cash transfers in addition to the nutrition-sensitive social behaviour change interventions on depression scores (A) and the proportion of women with depressive symptoms (B). SBC, social behaviour change; SoC, standard of care; SRQ, self-reporting questionnaire.

We found limited evidence of effect modification of the effect of the MAZIKO interventions on SRQ-20 scores or the proportion of women with depressive symptoms (online supplemental tables 7 and 8). Specifically, child sex modified the effect of the SBC+high cash intervention on the proportion of women with depressive symptoms (p-value for interaction <0.10) with larger effects for mothers with boys and not girls (online supplemental table 8).

Discussion

This study assessed the impact of cash transfers integrated with nutrition-sensitive SBC interventions on perinatal mental health during the lean season in Malawi. Eighteen months post-baseline, the SBC+high cash interventions reduced depression scores as compared with SoC, with reductions corresponding to an effect size of −0.27 SMD. The SBC+high cash interventions benefited all women, not just those with more severe depressive symptoms. The SBC+low cash and SBC alone interventions had no impact on maternal mental health relative to SoC. Our study also demonstrated that those women who were provided with cash transfers with SBC in comparison to just SBC had lower depression scores regardless of the size of the cash transfer. Reductions corresponded to an effect size of −0.24 SMD for SBC+low cash versus SBC and −0.35 SMD for SBC+high cash versus SBC. When comparing the SBC+high cash to SBC+low cash groups, there was no evidence of added benefit of the larger cash transfer relative to the smaller cash transfer. In addition, we found that mothers of boys appeared to benefit more from the SBC+high cash intervention than mothers of girls.

Our study demonstrated that the most comprehensive package (combining SoC, SBC and large cash transfers) was the most effective relative to the SoC in improving maternal perinatal depression. We hypothesised that the interventions worked by addressing the highest number of vulnerabilities and stressors and supported the largest variety of positive coping and thriving strategies during the lean season. Post-endline analyses will unpack the specific pathways through which the interventions worked to improve maternal mental health.⁴⁷ Compared with existing literature, the effect size of −0.27 SMD we observed was nearly four times higher than the effect size of −0.07 SMD of parenting interventions alone⁴⁰ and 2.7 times higher than the effect size of −0.10 SMD of cash transfers alone.²³ Taken together, these estimates support the hypothesis that cash transfers, nutrition-sensitive agriculture and parenting interventions can have additive or synergistic effects on maternal mental health during periods of high food insecurity.

Further, the size of the cash transfer appeared to matter in generating additive or synergistic effects. Although depressive symptoms declined in the SBC+low cash arm, we found no impact on depressive symptoms of the SBC+low cash interventions relative to SoC. These findings are consistent with a study in Tanzania which found no impact of combining small cash transfers (TZS5000–10,000 or US$2.20–4.30 per visit) with a parenting intervention on perinatal mental health.²⁹ They are also consistent with the social protection literature indicating larger effects on mental health with larger cash transfers.²⁰,²³ Likewise, we found that SBC alone had no impact on mental health relative to SoC. In fact, depression scores increased in the SBC arm. This may be because of the low self-reported participation in SBC activities. Although anecdotal evidence indicated misreporting of participation for fear of losing programme benefits, it is more likely that women limited their participation in more labour or effort-intensive activities in the last trimester of pregnancy and first few months post-partum. Moreover, most of the benefits from the agriculture and livelihoods activities (the only SBC interventions fully rolled out by midline) are likely not immediate, constant every month or predictable like the cash transfers. Thus, there were likely insufficient activities to promote and protect women’s mental health or benefits that had yet to be generated. Yet another explanation is that without cash, women had insufficient resources to act on the advice received through SBC,⁵⁹ which may have increased their stress. Given the staggered roll-out of SBC interventions and that some components had only been fully running for 2–3 months, the SBC effects should not be interpreted as those of a fully mature SBC intervention. Understanding dose-response effects from staggered roll-out is an important area of future research.

That SBC+low cash or SBC alone was insufficient to help women and their households cope with existing or new vulnerabilities is not surprising given the multiple complex shocks in the country in the months prior to the midline survey. In 2023, Malawi experienced nearly 30% inflation following several rounds of currency devaluation.⁶⁰ In March 2023, tropical cyclone Freddy displaced 659 000 people and caused loss of homes and livelihoods for many others.⁶¹ El Niño induced a late onset of the 2023/2024 rainy season, inadequate rains and reduced the number of rainy days negatively affecting crop production.⁶² Humanitarian emergencies increase risks to mental health not only by disrupting livelihoods and financial/economic stability but also by disrupting health services, increasing food insecurity, poverty, overcrowding and breakdown of social networks.¹ Considering all these events, many respondents in the MAZIKO midline qualitative evaluation, conducted in April–June 2024, reported feeling worried and stressed because of their inability to provide sufficient and diverse food for their families.⁵⁹ Another MAZIKO analysis showed that while SBC+high cash protected household consumption and reduced food insecurity at midline, there were no effects in the SBC+low cash or SBC arms.⁶³ These results support our hypothesis that the SBC and SBC+low cash were insufficient to address economic vulnerabilities and reduce food insecurity during the lean season following major economic and climate shocks. Notably, only the agriculture and livelihoods components of the SBC were fully rolled out by midline. Had the Caring for the Caregiver and Male Champions interventions been rolled out earlier, larger and more diverse impacts on maternal perinatal depression may have been observed. The late roll-out for both components was necessary to align activities with the government’s wider piloting and training plans for these two interventions.

Relative to SBC, both cash amounts improved women’s mental health. Although the effect size was larger for the SBC+high cash than for the SBC+low cash interventions (−0.35 and −0.23 SMD, respectively), the difference was not statistically significant. Notably, these effect sizes are similar to the −0.27 SMD effect size for the effect of SBC+high cash versus SoC. Coupled with the increase in depression scores in the SBC arm, these findings indicate that the positive impacts on maternal perinatal depression are likely driven by the cash transfers. However, given the integrated nature of the interventions, we cannot fully disentangle which intervention components contributed to the improvement in mental health, and we cannot rule out additive or synergistic effects.

Our study is among a handful to evaluate the effect of cash transfers on perinatal depression and the first to our knowledge to evaluate the effect of combining cash transfers of different amounts with nutrition-sensitive SBC (comprising agriculture and livelihoods, male engagement and psychosocial interventions) on perinatal depression. The interventions were assessed through a rigorous cluster-randomised controlled trial, relying on a theory-driven approach. Since the study was implemented in a high number of clusters in two districts, results are likely generalisable to the rest of the country. Nevertheless, some limitations of this work should be noted. First, we only surveyed half the sample at midline, prioritising women who self-reported as pregnant at baseline. As a result, our findings may not generalise to the rest of the sample, consisting of women with children 1.5–3.5 years of age at the time of the midline survey. Whereas 80% of women in the midline sample were still breastfeeding, most of the women in the remainder of the baseline sample would no longer be breastfeeding and may likely have been more worried and stressed about providing food for their children. Thus, it is unclear whether the SBC+high cash interventions would have been sufficient to improve mental health for these women. Second, the trial did not include a pure control group or cash only arms. Since the government SoC was available to all pregnant women and mothers of young children, including a pure control group was not relevant in the Malawian context. We did not include a cash-only arm because we preferred to maximise the number of new comparisons, considering the extensive social protection literature on the impacts of cash transfers alone on variety of household and individual outcomes. Lastly, intervention assignment could not be blinded, which may have introduced participant bias. This was mitigated by blinding the data collection and analysis teams.

Conclusion

Overall, larger cash transfers coupled with nutrition-sensitive SBC interventions comprising nutrition-sensitive agriculture, male engagement and psychosocial components were effective in improving maternal perinatal depression during the lean season, benefiting all women, not just those with more severe depressive symptoms. Cash transfers provided an added benefit on top of SBC. The interventions likely addressed multiple vulnerabilities and risks and supported a range of coping and thriving strategies. Targeted cash transfers and SBC interventions during the key perinatal period can help promote and protect mental health during peak periods of food insecurity. In low-income countries like Malawi, there are <1 mental health worker per 100 000 population,¹ and substantial financial and human resources investments are needed before the health system can implement other cost-effective interventions for mental health.⁶⁴ Nevertheless, the SBC package is comprehensive and both values of the cash transfer are large relative to other cash transfer programmes in Malawi or programmes targeting perinatal mental health. Planned cost-effectiveness analyses at endline⁴⁷ will further build the evidence base for cost-effective community-based interventions for perinatal mental health and guide next steps in optimising the scale-up of the MAZIKO interventions, and in designing new programmes to improve women’s well-being.

Data availability statement

Data described in the manuscript, code book and analytic code will be made available by the corresponding author upon request.