Food Security in the District of iLembe, KwaZulu-Natal: A Comparison of Coping Strategies between Urban and Rural Households

ABSTRACT

Food insecurity varies between urban and rural populations, as do their household characteristics and practices. The aim of the study was to compare the behaviours and practices households in rural and urban areas carry out during times of limited food in the district of iLembe, South Africa. Using a cross-sectional study design, household surveys were carried out to collect information on household characteristics, food, and coping strategies. In total, 376 households were randomly selected from low-income wards, 229 of which were rural, and 147 from urban areas. Water access was significantly better in the urban areas, as was diet diversity. The coping strategies carried out in rural households indicated better access or reliance on natural resources compared to their urban counterparts. Interventions or policies aimed at improving household food insecurity should take into account the location of the population, the natural resources available to them, and the needs of the community.

KEYWORDS:

• Coping strategy
• urban
• rural
• South Africa
• food security

Introduction

Although there is enough food produced globally to feed the world’s population, 815 million people still go hungry. This is an increase from 2015, and thought to be as a result of increasing conflicts and climate-related shocks (Food and Agriculture Organization [FAO] et al. 2017). Hunger is an outcome of food insecurity; when an individual or household has limited or uncertain access and availability to safe and nutritious food to meet their daily needs (FAO 2004; United States Department of Agriculture [USDA] 2018). The majority of people suffering from food deprivation continue to live in developing countries with sub-Saharan Africa (SSA) remaining to have the highest number (FAO et al. 2017). Due to the different elements that contribute to food insecurity, namely food availability, access, utilisation and stability (FAO 2008), a number of different factors can impact food insecurity, such as economics, geographic location and climate-related events.

In terms of location, food insecurity has been found to vary between urban and rural areas. Usfar, Fahmida, and Februhartanty (2007) found food insecurity to be higher in rural households in Indonesia, whereas Walsh and van Rooyen (2015) found the opposite in South Africa. According to Walsh and van Rooyen (2015), there is limited research on the similarities and differences in food insecurity between urban and rural areas. Accurately defining and differentiating between urban and rural can be a challenging task as definitions between countries are not universal. However, rural often implies a less densely populated area with limited infrastructure and an agricultural base for livelihoods and food source (Lerner and Eakin 2011). Urban areas are then often defined as having a high population density and size, and a labour force or land-use in non-agricultural activities such as commercial or financial services (Hugo and Champion 2004; Wu 2014). Rural and urban households may also have different characteristics with regard to their household size, income, and water and electricity access as found by Walsh and van Rooyen (2015). Similarly, Herrador et al. (2014) found that the number of female-headed households and household literacy varied significantly between urban and rural areas in Ethiopia. Such characteristics may influence a household’s vulnerability to food insecurity (Hadley et al. 2011) and their ability to cope with food or economic shocks.

Household food insecurity is economically affected through food affordability. While households will select food based on their personal preferences, they will also choose food based on price, quality and their resources available to purchase food (Institute of Medicine [IOM] and National Research Council [NRC] 2013; Pollard et al. 2014). A potential cause of varying food insecurity levels between rural and urban areas, aside from household characteristics, could be food price. Pollard et al. (2014) found that in Australia, distance from a major city and the cost of food were significantly associated, with food in rural and remote areas costing more. They found the cost of a healthy food basket to be 23.5% more in a remote location, with the quality of fresh food also being less (Pollard et al. 2014). This was also the case in Mozambique where food prices were higher in the rural regions (Garrett and Ruel 1999). Although food prices may be higher in rural areas, it is the urban dwellers who may be more vulnerable to price or income shocks as a result of having less natural resources available to cushion the effects (Garrett and Ruel 1999). Food production has been found to be higher in rural areas and growing vegetables negatively associated with food insecurity (Walsh and van Rooyen 2015). When faced with a sudden shock or situation where there is limited money to buy food, rural and urban households may cope differently based on the resources available to them, as found by Usfar, Fahmida, and Februhartanty (2007). Coping methods range from short-term behaviours such as small dietary changes, to long-term or more extreme behaviours, such as selling assets; and depending on the success or failure of such strategies, a household may then be confronted with a number of direct or indirect impacts on its wellbeing (Heltberg et al. 2012). In turn, these primary impacts affect household resilience to future shocks and level of vulnerability to risk, both of which have the potential for long-term consequences (Miller et al. 2010).

Although considered a middle-income country, South Africa is no exception to the issues around hunger. At national level, South Africa is a food secure country with food supply and availability able to meet the population needs (Drimie and McLachlan 2013). However, the prevalence of food insecurity in South Africa was declared as 22.5% in 2017 (FAO (Food and Agriculture Organization), IFAD (International Fund for Agricultural Development), UNICEF (United Nations Children’s Fund), WFP (World Food Programme) and WHO (World Health Organization) 2017).

Food insecurity is often seen to be associated with poverty (Gonzalez 2015; Naser et al. 2014), which has also historically been linked to rural communities in South Africa (Kepe and Tessaro 2014; Neves and Du Toit 2013). With growing urbanisation during the last 10 years, there has been a shift to urban areas now facing higher levels of poverty (Frayne et al. 2009), possibly impacting upon urban food insecurity levels. In order to sufficiently address food insecurity, it is important to understand community specific conditions to address the key variables, and to provide information for relevant interventions (Garrett and Ruel 1999; Walsh and van Rooyen 2015).

The aim of this study was therefore to compare the behaviours and practices that households in rural and urban areas carry out during times of limited food in the district of iLembe in KwaZulu-Natal (KZN), South Africa. If the households cope or behave differently, the information could be used to inform community specific and acceptable policies or interventions that ensure the highest impact.

Methodology

This paper forms part of a larger study comparing the annual changes in household food insecurity and child nutrition in the district of iLembe, South Africa. The data used in this paper was collected during Round 1 of data collection in November 2016 and followed a cross-sectional study design.

Study setting

This study is located in the district of iLembe in KZN which lies along the eastern coast of South Africa. Throughout much of the year, the area experiences a subtropical climate, with the rainy season being between October and April. At the time of this study, the area was coming to the end of a severe drought. iLembe is further divided into four sub-districts, namely KwaDukuza, Maphumulo, Mandeni and Ndwedwe. The district has a mixture of urban and rural areas, with Stanger, the main business centre, being located in the sub-district of KwaDukuza. The remaining three sub-districts are classified as being rural, with some areas deemed as deeply rural. The district has a heavy reliance on agriculture for income and food source, with 29.4% of the households identifying as agriculture-based (Stats SA 2011a). Unemployment in the area is high at 30.6% (Stats SA 2011b).

Sampling

Two of the main outcomes from the overall study are the prevalence of household food insecurity and child malnutrition. Unlike food insecurity, malnutrition is measured frequently with globally accepted tools; therefore it was decided to use the number of children as basis of the sampling equation. Based on the SMART methodology (Sampling Methods 2012), the following equation was used to determine the sample size:

$n=(z^2)\left(\frac{\left(p\right)\left(q\right)}{d^2}\right)$

$n=({1.96}^2)\left(\frac{\left(0.1\right)\left(0.9\right)}{{0.03}^2}\right)$

$n=384.16$

Based on a global prevalence of acute malnutrition of 10% (p), a confidence interval of 95% (z) and a degree of precision of 3% (d), 384 children needed to be included in the study. Considering a non-response rate of 5%, it was determined that a total of 402 children needed to be included in the survey. Given that a household would only be included in the study if there was at least one child under five years old at the time of the research being undertaken and estimating that a household would have only one child under five years old, 402 households were targeted.

The four sub-districts of iLembe are further divided into 77 local wards. Wards were stratified based on their monthly average income and only those with an income of R1 200 or less were eligible for the sample. This was agreed upon by the researchers based on the assumption that such households would be the most vulnerable to food or income shocks, and due to the link between poverty and food insecurity (Gonzalez 2015; Naser et al. 2014). Once stratified, wards were given a numerical identifier and 10 wards were randomly selected. Within each ward, data collectors were advised to visit every fifth household with a goal of 40 households per ward. If a selected household did not have a child, the next household would be visited until one had a child. From here, they would use the original protocol.

Data collection

A questionnaire which included socio-demographic, food, environment and child information, was administered by a trained data collector in either English or the local language of isiZulu, and completed by the main caregiver of the child. Within these questionnaires, participants were asked to go through a list of 12 coping strategies and to advise if they had to carry them out during times of limited food, or when they had no money to buy food within the last 12 months. The 12 coping strategies were those suggested in the Coping Strategies Index Field Methods Manual (2008), and were also found to be relevant for the population of iLembe through fieldwork to adapt the coping strategies index for implementation in the same study area (Drysdale, Moshabela, and Bob 2019). The need to carry out at least one of the 12 coping strategies would indicate that a household had faced a food shortage and was therefore deemed as having experienced some degree of food insecurity. In addition, households were asked to re-call their weekly consumption of certain food groups. Low income has been cited as a barrier to balanced and diverse diets with poorer households often having to consume the same foods in order to save money (Faber and Drimie 2016; Faber, Schwabe, and Drimie 2009; Labadarios, Steyn, and Nel 2011; Safefood 2011). Based on this and that the researchers were interested in an overview of the types of food eaten rather than specific nutrient intake, it was decided to use a weekly food recall as opposed to 24-hour food recall. In total, data from 376 households was collected, 261 of which were categorised as rural, and 115 as urban. Although this was slightly below the stipulated household target, information from 430 children was collected, therefore reaching the child target for the overall study.

Data analysis

Data from the household survey was captured and analysed using the statistical software, Stata/IC 13. An index of socio-economic status (SES) was created using an asset index approach that uses household characteristics and ownership to categorise each individual household to a socio-economic group. This was completed through the use of the principal component analysis (PCA) in Stata (Vyas and Kumaranayake 2006). The household characteristics included variables such as dwelling type, electricity access and water resources, while ownership included variables such as owning a fridge, washing machine and a car.

Descriptive statistics were used to compute the household characteristics and to show a basic comparison between rural and urban areas. Logistic regression was calculated on the variables highlighting how households access water and store their food to investigate the difference between urban and rural households. Chi-square analysis was calculated to investigate the differences in food groups eaten on a weekly basis and to compare the prevalence of food insecurity between urban and rural households. Logistic regression was conducted to assess whether location had any influence on the likelihood of households being food insecure. In order to investigate whether the location influenced the likelihood of a household carrying out each of the 12 coping strategies, univariate regression was carried out. Household characteristics such as income or education may also influence the types of coping strategies carried out. Based on this, a multivariate logistic regression was also calculated. The confounding variables to control for were decided upon using a stepwise process in Stata. All the possible confounding variables were inputted, and it was concluded that the following should be controlled for: SES, number of people in the household, education, water and electricity access, dwelling type and where the households purchase or secure their meat, fruit and vegetables from. The differences between the two areas were found to be significant with a p-value of less than 0.05.

Reliability and validity

The questionnaire used was designed based on a conceptual framework that guided the overall study. This ensured that all relevant and confounding data was collected to ascertain the completeness of the tool in terms of the study objectives. The coping strategies listed in the questionnaire were based on those presented in the coping strategies index manual (2008), and were also found to be relevant within the study area through focus group discussions conducted and presented by Drysdale, Moshabela, and Bob (2019). Combined, these contributed to the validity and reliability of the coping strategies suggested.

Ethical clearance

Ethical clearance was granted from the Humanities and Social Research Ethics Committee from the University of KwaZulu-Natal (HSS/1500/016D). Permission to complete data collection was obtained from the iLembe District Municipality. Participants were requested to sign an informed consent if they were willing to participate. No names were taken during the data collection process to maintain confidentiality.

Results

Demographic information

Table 1 documents the key socio-economic and demographic information of the participants as a whole and is then separated between the urban and rural households. The overall mean age of the participants was 31 years, with the majority of them being female, likely to be skewed as a result of the data collectors requesting the main caregiver, often the mother, to take part. In terms of education, the proportion of participants having completed secondary school was fairly low at 37.8%, with 37.2% completing up to grade 10. Only 1.2% had completed some form of higher education and 7.5% had no form of education. Overall, the average monthly income was fairly low, with 86.5% earning less than R5 000 a month coupled with a high unemployment rate of 77.4%. This is likely to be heavily weighted towards the lower income households as a result of the sampling strategy only including wards with an average income under R1 200.

Table 1. Household demographics.

		All Households		Rural Households		Urban Households
		n	%	n	%	n	%
Sample Size		376	–	229	61.0	147	39.0
Mean Age		31	–	31	–	30	–
Gender	Male Female	20 356	5.3 94.7	13 216	5.7 94.3	7 140	4.8 95.2
Education	Completed Primary School Completed up to Grade 10 Completed Secondary School Certificate or diploma	28 61 140 143 4	7.5 16.2 37.2 38.0 1.1	16 33 81 96 3	7.0 14.4 35.4 41.9 1.3	12 28 59 47 1	8.2 19.1 40.1 32.0 0.7
Employment	Unemployed Employed	291 71	77.4 18.9	181 41	79.0 17.9	110 30	74.8 20.4
Income	R0-1,000 R1,001-R5,000 R5,001-R10,00 >R10,001	168 157 37 13	44.7 41.8 9.8 3.5	111 95 17 5	48.5 41.5 7.4 2.2	57 62 20 8	38.8 42.2 13.6 5.4
Mean Household Size		5	–	5	–	5	–

There was no statistical significance found between the socio-economic and demographic characteristics between rural and urban households. However, education appeared to be higher in the rural areas, with almost 10% more having completed secondary school than those in the urban areas. Unemployment was slightly lower and the income was higher in the urban areas, with 9.4% more urban households earning more than R5 000 per month.

Food and water

Table 2 shows the information collected on food and water practices in the study area. In terms of drinking water, over 50% of the households had access to a tap managed by the local municipality; however, only 22.1% of these were in the actual dwelling themselves. Of the 44.1% that relied on a public standpipe, violence and fighting over tap use had been reported as a result of the water restrictions put in place during the drought. A further 30.6% relied on a river or dam for their drinking water and only 26.7% of those stated they would purify the water prior to consumption. The most common form of purification was through the use of a local bleach product which with one small drop, made the water safe for drinking. The majority of the participants stated they stored their fresh food in a fridge or freezer, although 10% of the respondents advised they did not have a functional one within their homes.

Table 2. Food and Water Information.

		All Households		Rural Households		Urban Households		P value
		n	%	n	%	n	%
Drinking Water	Piped (in dwelling) Piped (public standpipe) River/Dam	83 166 115	22.1 44.1 30.6	8 95 114	3.5 41.5 49.8	75 71 1	51.0 48.3 0.7	Ref. 0.037 <0.001
Food Storage	Fridge/Freezer Cupboard Pantry	279 81 6	74.2 21.5 1.6	149 67 6	65.1 29.3 2.6	130 14-	88.4 9.5-	Ref. 0.003

Logistic regression on water access and food storage. Dependant variable is location. p = significant at 5% level.

The way in which urban and rural households accessed their drinking water varied significantly. Rural households were more likely to make use of a river or dam (p < 0.001) compared to an urban household, with 49.8% and 0.7% of households using this method, respectively. Only 3.5% of rural households had access to piped water within their dwelling, compared to 51.0% of urban households. As a result of this, they were also more likely to use a public standpipe (p < 0.037) than an urban dweller. The way in which households stored their fresh food also varied significantly, with 23.3% more urban households using a fridge or freezer than rural ones. Those living in rural areas were also more likely to store their food in a cupboard (p < 0.003) than urban households, and no urban dwellers stored their food in a pantry. It is important to consider that there are different methods to store food depending on the type of food and household preferences. The participants in this study were asked to indicate their fresh food storage, rather than perishables, the majority of which may need to be refrigerated. The results could therefore highlight a difference in fridge or freezer ownership.

Households were asked to state if they ate certain food types on a weekly basis, shown in Figure 1. The most commonly eaten food was maize or grain products, followed by beans, peas and lentils. The least common food types consumed was milk and other dairy products. Households in the study were heavily reliant on supermarkets for meat access, with only 5.9% relying on their own farm produce. There was also a high reliance on supermarkets for fresh fruit and vegetables, with only 9.6% of the households growing their own produce.

Figure 1. Weekly food consumption.

The two most commonly eaten food groups in rural and urban households did not differ and were maize and grain products, and beans, peas or lentils. Milk or dairy products remained the least commonly consumed products within both areas, although a significantly higher proportion of urban households consumed these products compared to rural households (X² = 11.799, DF = 1, p < 0.001). Other food types that were significantly more common in urban households, included meat (X² = 13.729, DF = 1, p < 0.001), pumpkin and carrots (X² = 9.905, DF = 1, p < 0.002), green, leafy vegetables (X² = 9.207, DF = 1, p < 0.002) and eggs (X² = 20.218, DF = 1, p < 0.001). From these results, it appears that urban households have better diet diversification compared to rural participants, although both average three meals per day.

Food security

Figure 2 shows the proportion of food secure and food insecure households overall, further differentiated between rural and urban households. From the figure, it is evident that 72.9% of households had to carry out at least one coping strategy and were therefore facing a food shortage and some form of food insecurity. The figure also indicates that there was very little difference in the food insecurity prevalence between urban and rural households, with 71.3% and 74.1% being deemed as food insecure, respectively. Results from the chi-square analysis further supported this (X² = 0.338, DF = 1, p < 0.561), and the results from the logistic regression analysis suggested that the location of a household had no influence on their food security status. (OR = 1.155 p < 0.590).

Figure 2. Household food security.

Coping strategies

Figure 3 shows the proportion of households that had to carry out each of the 12 strategies posed to the participants, both overall and split between the urban and rural areas. The most common coping strategy was to buy less expensive food, with 63.8% of all households doing so, followed by limiting portion sizes at 53.4%. Other strategies that more than a third of the households engaged in were borrowing food (49.2%), reducing meals (36.5%), purchasing food on credit (40.2%), and hunting or harvesting early (33.8%). Overall, the least carried out coping strategy was to go an entire day without eating anything, with only 6.8% of households doing so, followed by sending a household member to beg for food, at 9.4%. When differentiating between rural and urban households, the frequency of a number of coping strategies varied. Sending a household member to beg for food was more common in urban households at 15.6% compared to only 5.2% for rural households. Similarly, consuming reserved seed stock was more common in rural households at 17.9% compared to 7.5% of urban households. Other strategies that varied between areas included limiting portions, sending a household member to hunt or harvest early, restricting others for working members and reducing the number of meals.

Figure 3. Household coping strategies.

Table 3 shows the results of the univariate and multivariate regression analysis to investigate how location may influence coping strategies. From the univariate analysis, three of the coping strategies were significantly associated with location, namely using reserved seed stock (p < 0.011), sending a household member to hunt or gather wild food (p < 0.001) and sending a household member to beg (p < 0.002). Taking into consideration the confounding factors, a multivariate analysis was completed, and an additional strategy was found to be associated with location. From the participants, those who used reserved seeds (adjusted odds ratio [AOR] = 2.62, p < 0.049) or gathered wild food or hunted (AOR = 2.62, p < 0.032) were more likely to live in a rural area. Alternatively, those who sent a household member to beg (AOR = 0.25 p < 0.013) or restricted food by others for working members (AOR = 0.30 p < 0.13), were more likely to live in an urban area.

Table 3. Results of the regression analysis on coping strategies and location.

Coping Strategy	Univariate OR (95% CI)	P Value	Multivariate AOR (95% CI)	P Value
Less expensive food	1.04 (0.65 to 1.64)	0.883	2.13 (0.73 to 6.26)	0.169
Limit portions	1.20 (0.74 to 1.93)	0.463	3.07 (0.86 to 10.90)	0.084
Borrow food	0.94 (0.61 to 1.46)	0.779	0.68 (0.33 to 1.37)	0.270
Buy food on credit	1.10 (0.70 to 1.73)	0.668	0.79 (0.41 to 1.53)	0.479
Use seeds	2.65 (1.26 to 5.61)	0.011*	2.62 (1.01 to 6.82)	0.049*
Hunt	2.31 (1.40 to 3.81)	0.001*	2.24 (1.07 to 4.66)	0.032*
Send children elsewhere	1.23 (0.59 to 2.54)	0.581	0.90 (0.36 to 2.28)	0.828
Beg	0.31 (0.15 to 0.64)	0.002*	0.25 (0.08 to 0.75)	0.013*
Restrict adults	0.86 (0.49 to 1.51)	0.594	0.48 (0.22 to 1.02)	0.057
Restrict others	0.52 (0.27 to 1.01)	0.053	0.30 (0.12 to 0.76)	0.013*
Reduce meals	1.15 (0.73 to 1.81)	0.560	1.18 (0.58 to 2.41)	0.648
Skip eating	1.11 (0.47 to 2.60)	0.810	0.59 (0.20 to 1.76)	0.343

OR (logistic regression) and AOR (multiple logistic regression) on coping strategies controlling for SES, number of people in the household, education, water and electricity access, dwelling type and where meat, and fruit and vegetables are gained from. Dependant variable is location, either rural or urban. *p = significant at 5% level.

Discussion

The aim of this study was to compare the behaviours and practices of households between rural and urban areas in the district of iLembe, South Africa. There was very little difference in the demographic profile of the households between urban and rural areas. Unemployment was only 4.2% less in the urban areas and the average income of the households, although slightly higher, was not statistically significant. South Africa has one of the highest inequality rates in the world (World Bank 2018), with iLembe also experiencing high levels of inequality. For the purpose of this study, only wards with an average income of less than R1 200 were eligible for selection. Although previous studies have found significant differences in urban and rural household characteristics (Herrador et al. 2014; Walsh and van Rooyen 2015), in iLembe, the differences may be more visible when comparing and including wards with higher average incomes. Future research could include a higher income cut-off within the sampling strategy and investigate whether household characteristics do differ between the urban and rural areas.

In terms of access to water, there was a significant difference in how urban and rural households obtain their drinking water. The South African food-based dietary guidelines include clean and safe drinking water as part of a nutritious and healthy lifestyle (Van Graan et al. 2013). In 2014, a South African survey found that 90% of the households had access to piped water; however, in KZN this was slightly lower at 86.5% (Stats SA 2015). From the current study, it appears that there is better infrastructure in the urban areas in iLembe, compared to the rural areas. With almost half of the households in the rural areas obtaining water from a river or dam, it puts them at a higher risk of waterborne diseases should they not filter the water prior to consumption. Inequalities in service delivery in South Africa between urban and rural areas has been documented on a number of occasions (Gnade 2013; Rodina and Harris 2016; Sartorius and Sartorius 2016), but there is still a need for water access to be improved in the rural areas of iLembe to limit the negative health outcomes associated with poor water consumption. Policymakers should consider either building more public standpipes in the area or educating communities on the importance of purifying water before drinking.

Unlike the studies conducted by Usfar, Fahmida, and Februhartanty (2007) and Walsh and van Rooyen (2015), the results of this study found no evidence to suggest that the location of a household influenced their food security status. Usfar, Fahmida, and Februhartanty (2007), although having reported on coping strategies, measured food insecurity through the United States Food Security Survey Module, which may explain the opposing results. Walsh and van Rooyen (2015) on the other hand, recorded food shortages in rural and urban households, yet found significant differences between the two. Food insecurity is made up of four pillars, one of which is food access, related to how people obtain their food, often through financial means (Ike, Jacobs, and Kelly 2015). In South Africa, both rural and urban households rely on food purchases from supermarkets for the majority of their food (Aliber 2009). Food insecurity in the district, for both rural and urban households, could therefore be a result of poor access to food, as opposed to food availability. In order to overcome this, the International Food Security Assessment 2011–2021 states that food production at household level is critical in reducing food insecurity in SSA (USDA 2011). Walsh and van Rooyen (2015) also support this and state that domestic food production in South Africa should be encouraged to improve food access and availability. Households should be capacitated to grow their own food either through seed programmes or through providing them with the skills or tools to do so. This could reduce their reliance on markets and vulnerability to food price fluctuation. An intervention as such may be best suited towards households in rural areas where there is often more available land; however, when considering it, it is important to also introduce agricultural practices and seeds that are resilient to climate impacts, such as a lack of rainfall or higher temperatures.

Despite only a small difference in food insecurity status between households, those in urban areas appeared to have better diet diversification, and consume a wider variety of food types on a weekly basis. While the current study did not investigate the cost of food, previous studies found prices to be higher in rural and remote areas (Garrett and Ruel 1999; Pollard et al. 2014). Income in this study was higher in the urban households. Taking this into account, along with potential higher food costs in rural areas, this may explain the variation in diet. When selecting foods, households will consider certain ones based on their potential to maximise their utility within their limited resources (IOM and NRC 2013). If rural households have limited financial means, they may purchase cheaper food in bulk to ensure it lasts for a longer period of time. In addition, if the rural households rely on local agricultural production, it may limit their food choices. Commercial agriculture in iLembe is dominated by sugar cane (KwaDukuza Municipality 2012), but also has the correct climate for the staple crop of maize and other fruit and vegetables. If households rely only on the products grown locally, they can limit the number of food groups consumed and therefore have a lower diet diversity. This would also be the case with meat and poultry products. As explained by Pollard et al. (2014), policymakers should consider the modifiable factors that may influence food pricing, such as poor road quality or high petrol costs. By reducing the cost of food, rural residents may be able to purchase a better variety of food. In addition, it is imperative to encourage and support agricultural diversity in both subsistence and small commercial farms to ensure nutritional variation and improve household food security.

The most common coping strategy in this study was buying less expensive food, which conforms with the expectation that a household will first make small dietary changes during times of acute food insecurity (Coping Strategies Index 2008). The second most common coping strategy was limiting portion sizes, a similar finding to previous studies in Bangladesh (Farzana et al. 2017) and South Africa (Oldewage-Theron, Dicks, and Napier 2006). Although going an entire day without eating was the least reported coping strategy, it is worrying that over 6% of households in the study reported having to do this at least once. Such a strategy can have severe negative effects on an individual’s health if carried out frequently and indicates a severe problem within the household that needs to be addressed. While a number of coping strategies had similar prevalence between rural and urban households, four were significantly associated with location. Of these, relying on natural resources such as seeds or animals were more common in the rural areas. Begging was a more common practice in the urban households, as was restricting food for working members. Rural populations tend to have better access to natural resources that may help cushion food or income shocks (Garrett and Ruel 1999), whereas urban areas often have a higher population density (Wan et al. 2015), potentially making begging a more beneficial strategy due to the potential to gain more from it. The variations in coping strategies in the current study, as well as previous studies (Usfar, Fahmida, and Februhartanty 2007; Walsh and van Rooyen 2015), not only highlights that households react differently in different locations but also emphasise Garrett and Ruel’s (1999) conclusion that community specific conditions must be understood to best implement interventions, and that such programmes are not easily transferable between rural and urban locations. For example, a rural community with open land may benefit more from home gardening projects, unlike those situated in urban areas where space may be limited. These communities may be more suited to policies that reduce price inflation or improve well-paid job opportunities (Amendah, Buigut, and Mohamed 2014).

Recommendations

Based on the results from this study, some recommendations were integrated into the discussion in the previous section. Additional recommendations are presented in this section. It is recommended that policies or interventions aimed at improving food insecurity take into account the location in which it is being targeted. Although the prevalence of food insecurity did not vary, urban and rural households appear to have different resources available to them and considering this when creating an intervention or policy, would ensure it has the best impact. It is also recommended that further research be carried out in the district of iLembe to better determine the level and severity of food insecurity being faced and to assess to need for action. Finally, infrastructure in the rural areas of iLembe should be improved to ensure that more households have access to drinking water from a municipal tap. Better infrastructure and services will also enable livelihood diversification, including access to markets for surplus agricultural produce, should opportunities arise. This will increase the purchasing power of poor households which will contribute positively to food security.

Limitations

Forty households were selected in each of the 10 wards sampled in the study, not considering that the wards may have varying population sizes. In order to overcome this, the data was weighted in Stata using probability proportional to size (World Health Organization [WHO] 2018) with the available South African census data (Stats SA 2011a). Food insecurity status was determined by a household having carried out a coping strategy at least once. Although this did not indicate the severity of food insecurity being faced, it did highlight that the household was unable to meet their food requirements, and thus still acted as an indication that some degree of food insecurity was present. The researchers acknowledge that only 12 coping strategies were suggested to the participants and that no option for ‘other’ was provided. However, during preliminary fieldwork, the 12 coping strategies were found to be relevant for the study location and the results from this study provide a useful indication of the different types of coping strategies carried out between urban and rural areas.

Conclusion

The prevalence of food insecurity and food shortages in iLembe was high, and although it did not appear to differ between urban and rural households, there were variations in practices and the types of coping strategies carried out during food shortages. Urban households had better access to drinking water from municipal taps and appeared to consume a more diversified diet. Four of the coping strategies carried out by rural and urban households varied and highlighted that rural households appear to have better access or reliance on natural resources. Interventions or policies implemented in iLembe to address food insecurity should take into consideration the resources available and the needs of the communities to ensure they are of high impact.

Disclosure Statement

No potential conflict of interest was reported by the authors.