Global and South African hunger context

Hunger is an enormous global challenge. The United Nations’ (UN) sustainable development goals (SDGs) have included the eradication of hunger by the year 2030 as Goal 2 [1]. Eliminating hunger and efforts towards attaining food security are crucial, as both are linked to lost productivity due to difficulties with concentration and links to malnutrition, which impede healthy development as well as increased susceptibility to illnesses [2]. South Africa, like other low- and middle-income countries (LMIC), faces persistent hunger challenges, primarily due to structural unemployment, inflation in basic food and fuel prices, and frequent power outages that disrupt food production and distribution systems [3].

Hunger and food insecurity are often conflated, but they refer to related yet distinct experiences [2]. According to the Food and Agriculture Organization (FAO) hunger is the physical sensation caused by insufficient food intake, while food insecurity refers to unsafe or limited nutritious food necessary for growth, development, and general healthy [2,4]. The United States Department of Agriculture (USDA) further distinguishes between food insecurity without hunger, which involves reduced diet quality or variety and food insecurity with hunger, which includes disrupted eating patterns, such as skipping meals, or going without food [5].

Student hunger and food insecurity

Much previous work has examined university student food insecurity, while literature on university student hunger, is limited, especially in Africa [6–9]. Prior to the COVID-19 pandemic, studies reported that 66–84% of South African university students experienced food insecurity, while 5–60% reported going hungry due to insufficient access to food [6,9–11]. Financial mismanagement has also been identified as an important cause of food insecurity among university students, with some students prioritizing clothing and entertainment over food [12,13]. In addition, students often use financial aid to send remittances home, a practice prevalent in the African context [10]. South African literature has also raised concern, about the possible impact of the shift from catered to non-catered university residences, noting that while this approach reduces accommodation costs, it may unintentionally increase students’ risk of food insecurity and hunger [14].

Study aim and rationale

Empirical research on the impact of the COVID-19 pandemic on student hunger has predominantly taken place in Western countries, with limited insights from LMICs, such as South Africa. To our knowledge, this is one of the first quantitative studies in South Africa, and among the first in Africa, to examine student hunger during the pandemic. By situating the study within the context of national lockdown measures, which included the mandatory return of students to their homes, we explore how these policy-driven changes in living arrangements may have influenced hunger outcomes. The current research therefore aims to assess the prevalence and determinants of hunger among students at a South African university during the COVID-19 lockdown, with particular focus on the impact of returning home.

Impact of the COVID-19 lockdown

The COVID-19 pandemic had serious implications for both hunger and food insecurity in South Africa [15]. In response to the COVID-19 pandemic, the South African government declared a national state of disaster in March 2020 and implemented a five-level lockdown system, with Level 5 being the most restrictive. The pandemic and strict lockdown led to job loss which resulted in many households not having money to buy food [16–18]. Statistics suggest that the prevalence of hunger fluctuated between 16–23% in 2020, compared to around 10% in 2017 [19].

In 2020, the government implemented various targeted social protection measures to mitigate the impact of the pandemic and the associated economic disruptions on its citizens [20]. These measures were generally regarded as successful in reaching intended populations [21]. Temporary social protection included: increases in social protection measures such as child support grants and old age grants, a Temporary Employee/Employer Relief Scheme (TERS) which provided financial relief to employers and employees unable to work due to the operational downturn caused by the lockdown and other public health requirements [20,21]. Further to this, the government issued a special COVID-19 Social Relief of Distress (SRD) grant, whose stringent eligibility criteria included: not receiving any other social grant from the government, for students – not receiving financial aid, and not receiving any benefits from the unemployment insurance fund [20,22]. A means test validating applicants’ income through commercial banks formed part of the verification check [21]. Although the above may not have benefitted students directly, it may have contributed to improved household food security when pooled.

Specific challenges for university students

Universities responded to the pandemic by shifting to remote learning and requiring students to vacate residences, encouraging them to return home. The challenges that students subsequently faced whilst learning remotely have been well documented [17,23]. These included poor access to devices and connectivity (data, bandwidth, and network reception) especially among students in more remote areas [24,25]. Some reported that their home environments were not conducive to studying, with many lacking dedicated study spaces, facing emotional distress, and facing financial stress as household situations worsened [17,23,25].

In terms of wellbeing, student hunger has been highlighted as an important psychosocial factor impacted by the pandemic [17,25,26]. Most research on this topic has emerged from the United States of America and has focused on food insecurity, with mixed findings showing both increases and decreases in its prevalence [13,27,28]. However, the most consistent finding is that returning home often improved students’ food security status [13,27,28].

Despite these findings, there is a dearth of literature on how the COVID-19 pandemic and lockdown impacted the hunger status of university students in South Africa. Understanding this impact is important, as food insecurity and hunger have been linked with poor academic performance, negative health outcomes as well as an increased likelihood of risky sexual behaviour [9,29].

Sample

The study was carried out at an urban university in Johannesburg, South Africa with ~41,000 students enrolled in 2020. Although the university is located in an urban area, its student body is drawn from a mixed catchment, including both rural and urban areas across South Africa. Sixty percent of these students were pursuing undergraduate studies, with 55% of them female and the majority (61%) being Black Africans. The university has a diverse student body representing all races and all South African official languages.

The current article analyses a sub-sample of a larger study, whose overall aim was to understand the impact of COVID-19 on the university community [23]. For the present analysis, the sampling frame comprised all first-time, first-year students registered full-time for an undergraduate degree in 2020 (approximately 5,700 students). And the inclusion criteria required that students be 18 years or older at the time of the survey. Only students meeting these inclusion criteria were included in the analysis.

Whilst the entire population was invited to participate in the study, the minimum sample required to power the study sufficiently was 434. This was determined using Cochran’s equation, corrected for finite populations [30].

Expressed as an equation:

Where:

n: Required sample size.

Z: z-score corresponding to the confidence level.

p: Estimated proportion of the population.

e: Margin of error.

n: Corrected sample size for a finite population.

n₀: Initial sample size calculated using Cochran’s formula.

N: Total population size.

The assumptions included a population size 5 700, 99% level of confidence, 5% of marginal error, and a hunger prevalence of 23%. The final sample size also met adequacy criteria for logistic regression, exceeding the conventional threshold of at least 10 events per variable.

Questionnaire development and validation

The questionnaire used in this study formed part of a broader institutional survey on the impact of COVID-19 on the university community and was developed by a multidisciplinary team including epidemiologists, psychologists, and higher education researchers [23]. The instrument combined standardised validated measures with study-specific items informed by prior institutional surveys, evidence on South African students’ and households’ experiences during the lockdown [17,23–25], and national and international studies documenting the socioeconomic impacts of COVID-19 [2,15,16,18,19].

Hunger was measured using the three-item Household Hunger Scale (HHS), a validated tool widely employed in South Africa and other LMICs [31,32]. Given the established psychometric properties of the HHS, including strong cross-cultural validity, no additional test–retest reliability or measurement invariance analyses were undertaken in this study.

Draft items were reviewed by content experts for face and content validity, and the full survey was piloted with a small group of first-year students to assess clarity, comprehensibility, and completion time, which resulted in minor wording adjustments. The questionnaire was administered in English, the primary language of instruction at the university; while formal translation was not required, items were phrased in simple, accessible language to support students for whom English is an additional language.

Data collection

Using a cross-sectional research design, data collection took place between September and October 2020, during South Africa’s COVID-19 lockdown levels Two and One, which represented the post-peak phase of the national lockdown. Following ethical clearance, a list of student email addresses meeting the inclusion criteria was extracted from the university’s data warehouse. As part of the larger institutional study, all eligible first-time, first-year students were invited to participate in an online survey. The recruitment process for the study was initiated by sending emails to all students at the University who fulfilled the inclusion criteria. Following acceptance via an online consent process, participants completed a self-administered online survey, using the Research Electronic Data Capture (REDCap) platform [33]. It is important to acknowledge that students without stable internet access may be underrepresented in the sample. An automated follow-up reminder email was sent to students who had not participated within the first week. No multistage cluster sampling was applied; all data were collected from a single institution using the invitation procedure described above.

Household hunger

The study employed the three-item Household Hunger Scale (HHS) to assess household hunger [31]. The HHS is a tool commonly used in South Africa to assess household hunger and has been validated for cross-cultural use [32]. The three-item HHS assesses whether, in the past four weeks, there was no food in the household, anyone went to bed hungry, or anyone went a whole day without eating due to lack of food. The scale uses a four-week recall and is adaptable to various settings allowing for applicability to different contexts. Participants respond to a four-point Likert scale for each item – ranging from 0 (never) to 3 (often – more than 10 times in the last 4 weeks). Following the recoding of data in accordance with the HHS guidelines [31], the summed responses are then used to determine the level of hunger (a score of 0–1 indicating little to no hunger; 2–3 indicating moderate hunger; and 4–6 indicating severe hunger). These parameters have been used previously in similar studies [9,34]. The current study defined ‘hunger’, the outcome of interest in the analyses, as those scoring 2–6, and defined ‘no hunger’ as those with a score of 0–1. This tool exhibited good internal consistency in the current study, with Cronbach α value of 0.92.

Socio-demographic characteristics

Socio-demographic characteristics examined in the study included sex (‘male’ or ‘female’), race (‘Black African’, ‘White’, ‘Coloured/mixed race’, ‘Indian and Chinese’), first-generation status (‘yes’ or ‘no’) denoting students who were first in their family to attend university, discipline of study (‘Commerce, Law and Management’, ‘Engineering’, ‘Health Sciences’, ‘Humanities’, and ‘Sciences’), financial aid status (coded as ‘yes’, for those who received funding, or ‘no’ for those who did not receive funding). Age was treated as a continuous variable in descriptive statistics, where the mean and standard deviation were used to summarise central tendency and variability across the sample. For the logistic regression, however, age was categorized into four groups (18, 19, 20, and >20 years). The categories were selected to reflect typical first-year entry ages (18–20), with the > 20 year age group capturing older or non-traditional entrants. Linearity with the log odds of hunger was assessed before creating these categories. Indian and Chinese students were grouped due to the small number of Chinese students in the sample which made separate statistical analysis unfeasible. Finally, high schools were classified by quintile (1–5 and ‘other’ denoting those who attended private or international high schools or those whose high schools were unknown). The school quintile is used in South Africa to classify public schools based on the socio-economic conditions of the communities they serve. Quintile 1 schools are found in low resource areas, while quintile 5 schools are found in the most affluent communities [35]. The high school quintile classifications were grouped into three in the multivariate analysis: ‘Quintile 1-3’ and ‘Quintile 4 & 5’ and ‘Other’. This categorization aligns with the South African Department of Basic Education’s classification system and distinguishes schools serving low-income communities from those serving more affluent ones.

COVID-19 lockdown factors

Factors arising from the pandemic were considered. The variable ‘Place of residence during lockdown’ reflected where the students stayed during lockdown and was coded as ‘on- or off-campus residence’, ‘with relatives’, ‘at home’ and ‘no stable accommodation’. COVID-19 exposure was captured through the variable ‘Self-reported COVID-19 infection (of participant and/or close friends and family)’, based on a single yes/no item asking whether the respondent or someone close to them had tested positive by the time of the survey. The number of people currently living in the household was also captured and coded as ‘Living alone’, ‘2’, ‘3-4’ and ‘5 or more’. Other covariates included where the household was situated during lockdown (coded as ‘City/ Suburb’, ‘Township’, ‘Town’ and ‘Village/farm’); whether there was financial stress/ worry during lockdown (coded as ‘Yes’ or ‘No’); and finally the variable ‘Limited workspace at home’ to capture whether the student had a quiet and dedicated area conducive to studying and attending remote classes. This was self-reported and coded as ‘Yes’ if the student indicated workspace constraints and ‘No’ if adequate workspace was available.

South Africa’s lockdown context

In response to the COVID-19 pandemic, the South African government instituted a national state of disaster on 15 March 2020 [36]. The government then introduced a lockdown, guided by 5 alert levels [37]. Lockdown restrictions were applied at a national level. Alert level 5 [38] was the most stringent level, confining people to places of residence, with alert level 1 having more relaxed restrictions including a curfew between 00h00 and 04h00 daily [39]. Universities adopted a new learning mode, allowing the academic programme to continue remotely [25,40,41]. Groups of students were reintroduced into campus as the lockdown levels eased [40–42]. During the lockdown, ‘essential services’ including the operation of grocery stores and corner/convenience shops were not restricted [38]. Similarly, the transportation and logistics of essential goods, such as food, faced no limitations [38]. Commuter transport services were allowed to operate, albeit with capacity restrictions, to facilitate the movement of individuals involved in rendering or obtaining essential goods or services [38]. Moreover, the acquisition of essential goods, like food, was explicitly permitted [38]. These guidelines were relevant country wide, including in rural and urban settings.

Analytic approach

Data were cleaned and analysed using STATA (version 17; College Station, Texas, USA). Descriptive analyses were performed for all variables, with proportions and 95% confidence intervals (95% CI) reported when appropriate. The Pearson’s chi-square test was used to compare categorical variables with hunger, and the Mann-Whitney U test was used to compare the continuous variable (age).

Data were weighted using post-stratification weighting based on joint distribution (sex and race) from institutional administrative records to adjust for the differences between the students who participated in the study and the underlying student population. These weights were applied to all descriptive and regression analyses. Missing data were handled using listwise deletion, resulting in the analytic sample of 596 students described above. Variables included in the logistic regression model were selected using a forward stepwise regression, with a cut-off inclusion of p-values 0.20 for the model and adjusted Odds Ratios (aOR) with confidence intervals were reported. Significance was defined at p-values <0.05. To assess model diagnostics, we tested for multicollinearity using the Variance Inflation Factor (VIF). Model fit was assessed using the Hosmer–Lemeshow goodness-of-fit test, and explanatory power was evaluated using McFadden’s pseudo R².

Ethical statement

Ethical approval was received from the University of the Witwatersrand Human Research Ethics Committee (Non-medical) (H20/06/22). Written permission was received from the Office of the university Registrar to conduct the study. Each participant provided digital consent before participation.