Addressing the interaction between food insecurity, depression risk and informal work: findings of a cross-sectional survey among informal women workers with young children in South Africa

BACKGROUND: There is a high burden of depression globally, including in South Africa. Maternal depression is associated with poverty, unstable income, food insecurity, and lack of partner support, and may lead to poor outcomes for mothers and children. In South Africa one-third of working women are in informal work, which is associated with socioeconomic vulnerability. METHODS: A cross sectional survey explored work setting and conditions, food security and risk of depression among informal working women with young children (0-3 years). Depression risk was assessed using the Edinburgh Postnatal Depression Score (EPDS) and Whooley score. Food insecurity was evaluated using Household Food Insecurity Access Scale. Data was analysed using SPSS and Stata. RESULTS: Interviews were conducted with 265 informal women workers. Types of work included domestic work, home-based work, informal employees and own account workers, most of whom were informal traders. Most participants (149/265; 56.2%) earned between US$70-200 per month, but some participants (79/265; 29.8%) earned < US$70 per month, and few earned > US$200 per month (37/265; 14.0%). Many participants experienced mild (38/267; 14.3%), moderate (72/265; 27.2%) or severe (43/265; 16%) food insecurity. Severe food insecurity was significantly higher among participants with the lowest income compared to those with the highest income (p = 0.027). Women who received financial support from the baby’s father were less likely to be food insecure (p = 0.03). Using EPDS scores, 22/265 (8.3%) women were designated as being at risk of depression. This was similar among postnatal women and women with older children. Household food insecurity was significantly associated with depression risk (p < 0.001). CONCLUSIONS: Informal women workers were shown to be vulnerable with low incomes and high rates of food insecurity, thus increasing the risk for poor maternal health. However, levels of depression risk were low compared to previous estimates in South Africa, suggesting that informal workers may have high levels of resilience. Interventions to improve social protection, access to health services, and support for safe childcare in the workplace could improve the health and wellbeing of these mothers and support them to care for their children. A cross sectional survey methodology was employed to explore the work setting and conditions, income, food security and risk of depression among informal working women with children under the age of 3 years. The study was conducted in three primary health care (PHC) clinics in three townships in Durban, South Africa. In these areas unemployment is high at around 20%, with high rates of poverty, and an average annual income of R29 400 (USD2000) [33]. In addition, there is poor access to basic services including water and sanitation and informal housing is common, but most households do have access to electricity [34]. A recent study in Durban found that women in the informal economy work long hours and many only earn < R1000 (USD70) per month [27]. Primary caregivers of children living in a low-income household are entitled to an SA government provided Child Support Grant (CSG) of R420 (USD30) per month for each child [35]. Health care for pregnant women and children is provided free of charge in PHC clinics, and there are high rates of antenatal clinic attendance (94%) and facility-based delivery (96%) [36]. HIV prevalence is high among young women in KZN, with 41.1% of pregnant women attending government antenatal clinics testing HIV positive in 2017 [37]. Women were recruited while waiting in the queue in the three participating clinics. Trained fieldworkers approached women in the clinic waiting area and explained the study. Eligibility was determined using a structured screening tool based on the definition of informal work set out above (Additional file 1). Eligible participants were informal workers, aged 18 years or older, with a child under the age of 3 years. Informal workers were either informally employed or own account (self-employed) workers. Informally employed women were defined by: (1) receiving money from an employer; (2) having no formal work contract; and (3) not contributing to SA’s mandatory Unemployment Insurance Fund (UIF). Own-account workers provided goods and services directly to customers but were not tax registered or paying tax. Women were excluded from the study if they worked fewer than 3 days per week or had been in informal work for < 6 months in order to focus on those women with longer term experience of informal work. A stratified sampling approach was employed to recruit equal numbers of eligible women with children aged < 1 year (defined as the postnatal period) and mothers of children aged 12–35 months. Given the well-documented high risk of postnatal depression in SA, this approach was used to determine depression prevalence in the postnatal population compared to the wider population of informally working mothers. A sample size of 192 women (96 with children aged < 1 year and 96 with children aged 12–35 months) was required to detect a 20% difference in depression risk among postnatal women compared to the general population of informal working mothers. This sample size calculation assumes a 35% rate of depression risk in women with children 1– < 3 years of age with 95% probability and 80% power. Data collection was undertaken among women waiting in the queue in the three participating clinics. Trained fieldworkers approached women in the clinic waiting area and explained the study. Eligibility was determined using a structured screening tool based on the definition of informal work set out above. Data were collected using a structured questionnaire administered in the local language (IsiZulu) (Additional file 1), and data collection continued until the sample size was reached. All eligible women agreed to participate and the survey was conducted in a separate room in the clinic to ensure privacy. We recorded the type of work based on the mother’s reported occupation, workplace (home/street/employer’s premises) and working conditions (own account/self-employed/unpaid or paid in kind). Risk of depression was assessed using two tools validated for use in SA and translated into IsiZulu: the Edinburgh Postnatal Depression Scale (EPDS) [38, 39] and the Whooley score [6]. The EPDS is the most widely used screening tool for depression among postnatal women, and has been widely used and validated in many settings including in South Africa [40], including among populations of non-postnatal women [41] and in high HIV prevalence settings [42]. The EPDS employs a 7 day recall of depressive symptoms including anxiety, tearfulness and poor sleep. A cut-off score of ≥ 13 was used. Although a range of cut-off scores have been used in different settings, this cut-off has been most commonly used in South Africa [3, 6, 9, 10, 39, 43]. The Whooley questions are three questions to assess low mood and lack of interest as well as suicidal ideation. These two tools were used together to allow triangulation of the findings from two screening tools for depression risk. The USAID Household Food Insecurity Access Scale (HFIAS), adapted and translated to IsiZulu, was used to assess food insecurity. The HFIAS tool is a nine-item experience-based measure with questions about experiences of food insecurity of the past 4 weeks. The tool is designed to be easy to use and applicable to diverse sociocultural settings, and examines food security related to three domains: anxiety about food access; insufficient quality of food; and insufficient quantity of food [44]. The HFIAS can been shown to be broadly applicable across different cultural contexts and has been widely used and validated in Africa, including in South Africa [9, 22, 45] A professional psychologist, who was IsiZulu-speaking and had experience with both the EPDS and Whooley tools, trained the fieldworkers to assess depression risk using the two tools. The psychologist also supervised data collection to ensure the correct administration of the tools in the field. Quantitative analysis was undertaken using SPSS version 24 and Stata V15.1. Background characteristics, feeding and childcare practices, and conditions of informal work are presented as frequencies and proportions. To describe the types of work undertaken, participants were retrospectively categorised using a series of steps. Firstly, all women who reported they worked from home were categorised as home-based workers. Women were categorised as domestic workers if they reported they worked in a private household undertaking domestic or childcare work. All other women were categorised either as employees if they reported receiving a wage from an employer, or as own account workers if they were self-employed. Sub-categories of own account workers included informal traders and a variety of other own account jobs. For the EPDS high risk of depression was determined by a score of ≥ 13. Two positive answers were considered a positive score for the Whooley questions. For both tools a positive answer to questions about suicidal ideation was considered high depression risk, regardless of the answers to the other questions. Household food security was scored on a scale of 0–27, where 27 is the most food insecure, using the established and validated scoring for this tool [46]. For the multivariable analysis the food insecurity index was dichotomised into food secure (none or mild food insecurity) and food insecure (moderate or severe food insecurity), and the EPDS depression score was analysed on a continuous scale and means and standard deviations reported. The analysis was a two-step process. Initially a multi variable logistic model was used to identify factors associated with food insecurity. Due to the over dispersion of the EPDS depression score, a multi-variable negative binomial model including food insecurity was used to identify factors associated with PND score. In both cases only factors significant at p < 0.3 in bivariate analysis were included in the multi-variable model. Adjusted and unadjusted odds ratios and 95% confidence limits are reported. Stata V15.1 statistical software was used in the analysis All participating women provided written informed consent. Screening, consent processes and data collection were conducted in a private area in the PHC clinics. Referral procedures were established with staff at each of the PHC clinics to offer additional clinical and social support to participants as required. All women identified as at risk of depression or with any thoughts of self-harm were immediately referred to clinic staff for further assessment. Follow up support from a study counsellor was offered to all women identified as being  at risk of depression. Women identified as having moderate or severe food insecurity were referred to a clinic-based social worker.