Domestic violence, food insecurity and mental health of pregnant women in the COVID-19 lockdown in Cape Town, South Africa

Introduction Common mental disorders (CMD) such as depression and anxiety are associated with low household income, food insecurity and intimate partner violence in perinatal women. The national COVID-19 lockdown in South Africa resulted in increased levels of poverty and food insecurity. We aimed to explore the relationship between CMDs, food insecurity and experiences of violence among pregnant women during the COVID-19 lockdown. Methods Perinatal women, attending 14 healthcare facilities in Cape Town, were enrolled in the study during baseline data collection before the COVID-19 lockdown. During the lockdown period, eldworkers telephonically contacted the perinatal women who were enrolled in the study and had provided contact details. The following data was collected from those who consented to the study: socio-demographic information, mental health assessment, food insecurity status and experiences of abuse. Poisson regression was used to model the associations of a number of risk factors with the occurrence of CMDs. Results Of the 2149 women enrolled in the ASSET study, 885 consented to the telephonic interviews. We found that 12.5% of women had probable CMDs and 43% were severely food insecure. Psychological distress increased signicantly during the lockdown period, compared to before the COVID-19 outbreak. The strength of the association between key risk factors measured during the lockdown and psychological distress increased during the COVID-19 lockdown. Using multivariate Poisson regression modelling, we showed that the risk of CMDs was almost three times more likely in women who were severely food insecure or who experienced psychological or sexual abuse. Conclusions This study provides evidence of the effect of the COVID-19 lockdown on the mental health status of perinatal women living in low resource settings in Cape Town, and highlights how a crisis such as the COVID-19 lockdown amplies the psycho-social risk factors associated with CMDs in perinatal women.


Introduction
Common mental disorders (CMDs) such as depression and anxiety are highly prevalent during the perinatal period, especially in low-and middle-income countries where approximately 19% of perinatal women develop depression, 1 and approximately 34% develop anxiety. 2 In South Africa the prevalence is particularly high, with several studies reporting that one in every three pregnant women develop symptoms of depression, [3][4][5][6][7] and one in every four pregnant women develop symptoms of anxiety. 3 8 However, studies have also shown that the majority of women who experience mild to moderate symptoms of depression during their rst and second trimester of pregnancy, show an improvement in symptoms during the remainder of the perinatal period, even without intervention. 7 9-11 Studies in Cape Town, South Africa have identi ed several psycho-social risk factors for CMDs which include food insecurity, intimate partner violence, and lack of social support. 4 7 12 The Coronavirus (COVID-19) outbreak was declared a global pandemic by the World Health Organisation on 11 March 2020. 13 A few days later the President of South Africa announced that urgent and drastic measures were needed to manage the disease. 14 This included a travel ban, increased screening and testing and the closure of all schools. Two weeks later, a 35-day national lockdown came into effect. 15 16 This initial lockdown period became known as the Alert Level-5 and required South Africans to stay home except for essential purposes. All non-essential activities were suspended until the end of April 2020, and included the sale of alcohol. 17 During May 2020, South Africa moved to Alert Level-4, which resulted in a slight easing of restrictions. However, South Africans remained con ned to their place of residence, except for those performing essential services. 18 On 1 June 2020, Alert Level-3 came into effect, further easing restrictions, which included unbanning the sale of alcohol and the opening of schools and some places of employment. 19 According to the World Health Organisation, 20 experiencing fear, worry and distress is an understandable and normal response in the context of the COVID-19 pandemic. The South Africa Depression and Anxiety Group (SADAG) reported that the COVID-19 lockdown in South Africa resulted in an increase in the number of calls from people feeling overwhelmed, anxious, worried and depressed. In an online survey completed between 2-15 April 2020, SADAG found that 65% of respondents were feeling stressed during the lockdown. 21 South Africa is a country characterised by high levels of inequality and poverty. 22 The majority of South Africans (55%) live below the upper bound poverty line, 23 with many living in overcrowded shacks with intergenerational families. Intimate partner violence rates are high and is thought to be associated with poverty and food insecurity. 24 The lockdown has been especially di cult for the already vulnerable groups living in poverty, as it has resulted in increased levels of unemployment, food insecurity 25 and domestic violence. 26 However, little is known about the speci c effects of the COVID-19 crisis on the mental health of vulnerable groups such as antenatal women in South Africa. As unemployment, food insecurity and domestic abuse is associated with CMDs, this study aims to explore the relationship between CMDs, food insecurity and experiences of violence among pregnant women attending midwife obstetric units (MOUs) in Cape Town during the COVID-19 lockdown.

Setting
This quantitative study was initially conducted in 14 randomly selected midwife obstetric units (MOUs) or basic antenatal care (BANC) clinics in the Cape Metropolitan Health District in Cape Town. MOUs and BANC clinics are public sector antenatal and obstetric care facilities managed by the Western Cape Department of Health. The facilities are free at the point of contact and attended predominately by women living in low socio-economic status communities. This study forms part of the bigger Health Systems Strengthening in sub-Saharan Africa (ASSET) study 27 -which includes a cluster randomised control trial (ISRTCN41483663) to evaluate an intervention to strengthen detection, referral and care for antenatal women with depression, anxiety and experiences of domestic violence in Cape Town.
In February and March 2020 (as part of baseline data collection) 2149 perinatal women were enrolled in the study, which included a le review of their current Maternity Case Records (MCR) -the national stationery used to record all aspects of the pregnancy. 28 Women who were pregnant (irrespective of gestation period) or had given birth in the past three months were included. Information collected included contact details, gestational and medical history, and results of a 3-question mental health screening questionnaire used to assess psychological distress, 29 routinely administered by health professionals during patients' rst antenatal visit. As a result of the COVID-19 pandemic, data collection was suspended after completion of the baseline data collection, but before the intervention could begin. During the period when face-to-face data collection was not permitted, the study was opportunistically pivoted to focus on the impact of the COVID-19 lockdown on women already enrolled in the ASSET study.

Testing procedures
During June and July 2020 (Alert Level-3 lockdown), eldworkers telephonically contacted the perinatal women who were enrolled in the ASSET study and had provided contact details and invited them to participate in a telephonic survey. Questionnaires were telephonically administered to perinatal women who consented to the study. Questionnaires were available in English, Afrikaans or IsiXhosa, and included (1) a socio-demographic questionnaire that was used to collect data on participants' age, obstetric information, relationship status, income status, and effect of the COVID-19 lockdown, (2) a 3-question mental health screening questionnaire, 29 (3) the Edinburgh Postnatal Depression Scale (EPDS), 30 (4) the Household Food Insecurity and Access Scale (HFIAS), 31 and (5) the short form of the Composite Abuse Scale (CAS-SF). 32 The EPDS is commonly used as a screening tool in research settings and has been validated against the Diagnostic and Statistical Manual (DSM-IV) 33 34 for depression and anxiety in a sample of postnatal women in South Africa 30 35 with a cut-off of ≥13 indicating a probable CMD. The questionnaire consists of 10 items with a seven-day recall period. The 3-question mental health screening questionnaire (the same questionnaire routinely administered by health professionals during patients rst antenatal visit to healthcare facilities) has been validated against the EPDS. 29 Using a cut-point of ≥2, this screening tool is able to identify perinatal women with psychological distress (sensitivity=85.7%; speci city=92.9%). The HFIAS was used to assess household food insecurity and hunger. 31 This 9-item scale measures the household's frequency of running out of food or eating inadequate amounts of food during the past 30 days. The CAS-SF 32 is a 15-item instrument that captures women's self-reported experience of physical, sexual and psychological abuse.

Data analysis
Data analysis was carried out using STATA/SE statistical software package version 15.1 (StataCorp., College Station, TX, USA). Variables were described using frequency and percentages, and associations measured using Chi-square tests.
Logistic regression was used to model the associations of socio-demographic risk factors with the occurrence of CMDs. Household food security status was calculated 36 by using the HFIAS to categorise households into four levels of household food insecurity experienced during the past 30 days: food secure, mildly food insecure, moderately food insecure and severely food insecure. The CAS-SF was used to categorise experiences of domestic violence during the past 12 months into physical, sexual and psychological abuse. 32 Purposeful selection 37 was used to build the Poisson regression model. 38 Unadjusted associations between variables and the outcome (probable CMDs) were explored using univariate analysis. During the multivariate model building process, partial likelihood ratio tests were used to compare the models.

Ethical approval
Ethical approval for the study was obtained from the Human Research Ethics Committee at the Faculty of . Consent forms were available in English, Afrikaans and isiXhosa. Participants provided informed consent after the procedure had been verbally explained to them. All participants were informed that they were free to withdraw from the study at any time without consequences. No nancial incentives were provided for participating in the study.

Results
Data from MCRs were available for 2149 perinatal women (Supplementary table). More than half the women (n=1248; 58%) who were initially enrolled in the ASSET study, were not contactable due to missing or incorrect contact details. An additional 16 women declined to participate when contacted. In total, 1264 women were lost-to-follow-up. Gravidity (p=0.672), pregnancy status (p=0.097) and the results of a routinely administered mental health screening questionnaire (p=0.199) were similar for those included in the study and those who were lost to follow-up. Signi cantly more young women aged 15 to 24 years were lost-to-follow-up (p=0.047).
Bivariate associations between CMD and the socio-demographic and psychological characteristics of participants are presented in Table 1. The study sample consisted of 885 participants, of whom 110 (12.4%) were classi ed as having a probable CMD and 775 (87.6%) were not. Almost half the women were severely food insecure (n=378), while more than half were unemployed (n=475). The proportion of women with CMD signi cantly increased with increasing number of pregnancies (p=0.017) and severity of food insecurity (p<0.001). Signi cantly fewer women who had experienced their rst pregnancy (20.9 vs. 30.2%; p=0.045) and had already given birth at follow-up had a probable CMD (63.6 vs. 74.1%; p<0.022).
Signi cantly more women with probable CMD had experienced a previous miscarriage or stillbirth ( A comparison of the results of the routinely administered mental health screening tool is presented in Table 2. Five hundred and fty-ve participants completed the screening questionnaire at their rst clinic visit and during the lockdown. During the lockdown, the proportion of women who felt unable to stop worrying or thinking too much increased more than three-fold (12% to 40%), while the number of women who felt down, depressed or hopeless increased by six times (5% to 30%). Signi cantly more women experienced psychological distress during the lockdown, compared to their rst clinic visit (3.1 vs. 26.2%; p<0.001).
Univariate logistic regression of risk factors against the outcome, psychological distress at the rst clinic visit and during the lockdown is presented in Table 3. Only experiences of psychological abuse were signi cantly associated with psychological distress at the rst clinic visit (Odds ratio (OR)=3.65; 95% Con dence interval ( In the multivariate model, after controlling for experiencing psychological distress at the rst clinic visit, women were at increased risk of a probable CMD if they had a previous miscarriage or stillbirth (RR=4.37; 95%CI: 1.52-12.56), experienced severe food insecurity (RR=2.79; 95%CI: 1.51-5.14), or experienced psychological abuse (RR=2.50; 95%CI: 1.32-4.72) or sexual abuse (RR=2.70; 95%CI: 1.07-6.80).

Discussion
In this study we explored the relationship between the mental health status of perinatal women, and their experiences of food insecurity and domestic violence during the COVID-19 lockdown. We found that 12% of women had probable CMDs and 43% were severely food insecure. Levels of psychological distress increased signi cantly during the lockdown period, compared to before the COVID-19 outbreak in South Africa. While we did not nd an increase in women who felt suicidal, signi cantly more women reported feeling anxious and depressed. Using multivariate Poisson regression modelling, we showed that the risk of CMD was almost three times greater in women who were severely food insecure, or who experienced psychological or sexual abuse. Importantly, the strength of the association between key risk factors (having more than four pregnancies or a previous miscarriage or stillbirth, experiencing increased crime in the community, decreased income or less food in the household, severe food insecurity, or any form of abuse) and psychological distress increased during the COVID-19 lockdown.
Compared to other studies in South Africa, 3 4 7 12 our study found a relatively low prevalence of CMDs (12.5%) in pregnant women during the COVID-19 lockdown, given that more than 80% of those interviewed reported feeling anxious about getting infected, more than 60% were unemployed and fewer than 20% of households were food secure. This is contrary to a recent review assessing the psychological impact of the lockdown, which found that being quarantined resulted in a higher prevalence of psychological problems such as depression and anxiety. 39 We found that the proportion of women experiencing psychological distress during the lockdown was signi cantly more than those who were distressed at their rst clinic visit. Yet, we did not observe a change in the number or proportion of women who felt suicidal. This is not out of keeping with other research; when examining studies reporting on suicidal thoughts and behaviour during pregnancy and the postpartum period, several studies have reported a decreased risk. [40][41][42] In LMIC, poverty is a well-documented risk factor for CMDs, especially during the perinatal period. 43 44 The relationship is considered to be complex and bidirectional with several social issues interacting. 45 46 Similar strong associations between food insecurity and CMDs have been consistently reported. 4 47 48 In a particularly low-resource setting in Cape Town, the odds of depression was ve times greater in perinatal women who were food insecure compared to those who were food secure, while the odds of experiencing food insecurity was four times greater in women who were depressed compared to those who were not. 4 Our study found that being severely food insecure doubled the odds of CMDs during the COVID-19 lockdown, and that 80% of participants reported experiencing various levels of food insecurity. The high prevalence of food insecurity can be attributed to the COVID-19 lockdown. During April and May 2020, when South Africa was at Alert Level-4 and Level-5 of the lockdown, all non-essential services were halted. This resulted in high levels of unemployment, affecting the most vulnerable workers, who were low-skilled and less educated, the most. 49 Six weeks into the lockdown, Statistics South Africa, using an online survey, found that 4.3% of respondents reported experiencing hunger during the month prior to the lockdown, while 7% experienced hunger during the lockdown. 50 This is likely being underreported as those living in poverty and experiencing hunger would be unlikely to have access to a web-based survey.
In our already vulnerable group of perinatal women living in low-resource settings in Cape Town, we found that more than 40% were severely food insecure. This translates to 40% of perinatal women living in households where they were eating fewer meals than needed, lacked the resources to acquire more food, went to sleep hungry or went a whole day and night without eating. 36 We found that the risk of CMDs was almost three times greater in women experiencing psychological or sexual abuse. However, we found quite a low prevalence of domestic abuse in our study compared to ndings in a similar population. 51 While we found that 15% of participants reported experiencing psychological abuse and less than 2% reported experiencing sexual abuse, Malan et al. 51 reported that more than 40% of perinatal women experienced psychological abuse and 25% experienced sexual abuse.
The low prevalence in our study is possibly due to the alcohol ban, implemented during Alert Level-4 and Level-5, 18  Our study has a number of limitations. More than half the women were not contactable. While levels of psychological distress before the lockdown were similar in those who were and were not contactable, it is possible that those who were not contactable could not afford a cellular phone. This study could therefore have underreported the level of unemployment and food insecurity. We used only quantitative questionnaires which did not allow us to fully explore the perceived causes of CMDs, food insecurity and experiences of violence. We were only able to compare psychological distress as measured by a brief screening tool across the two time points, instead of using a validated tool such as the EPDS. In addition, the rst screening was done at healthcare facilities by healthcare workers, while the second screening was telephonically administered by eldworkers. We started data collection on the same day that the alcohol ban was lifted, which did not give us enough time to measure the impact of alcohol sales on domestic violence.
Further research is needed (1) to understand the coping mechanisms used by perinatal women to mitigate the stress of living in households with high levels of unemployment, food insecurity and domestic violence; (2) to investigate the effect that lifting the alcohol ban has had on domestic violence; and (3) to examine longer term trends in mental health, domestic abuse and food insecurity among perinatal women during the COVID-19 lockdown.

Conclusion
This study provides evidence of the effect of the COVID-19 lockdown on the mental health status of perinatal women living in low resource settings in Cape Town. While the COVID-19 lockdown triggered high levels of unemployment and increased the prevalence of food insecurity and CMDs, the accompanying alcohol ban appears to have protected women from experiences of domestic violence during a period when living in close proximity with an abuser was unavoidable. Our ndings highlight how a crisis such as the COVID-19 lockdown ampli ed the psycho-social risk factors associated with CMDs in perinatal women.

Declarations
Acknowledgments We are grateful to the women who participated in the study for generously giving their time and energy to complete interviews.
Contributions ZA and CL proposed the study. All authors provided input in the design of the study. ZA performed the data analysis and prepared the rst draft of the manuscript. All authors reviewed the manuscript and providing input. All authors approved the nal version of the manuscript. Competing interests The authors have no competing interests to declare. Patient and public involvement No patients or members of the public were directly involved in the design, recruitment, conduct, reporting or dissemination plans for this study Ethical standards Ethical approval for the study was obtained from the Human Research Ethics Committee at the Faculty of Health Sciences, University of Cape Town (139/2018) and the Psychiatry, Nursing and Midwifery Research Ethics Subcommittee at Kings College London (17/18-7807). In addition, the Western Cape Department of Health approved the use of the research sites (WC_201807_008). Participants provided informed consent after the procedure had been verbally explained to them.
Availability of data and materials The dataset included for the analyses within this manuscript can be obtained on request from the corresponding author.