Child and adolescent food insecurity in South Africa: A household-level analysis of hunger

Food insecurity impacts childhood nutritional status, physical and cognitive development, and increases lifetime risk for chronic disease. Previous South African studies have examined hunger at the sub-national level without a specific focus on children and adolescents. This study determines the national prevalence of childhood food insecurity, from birth to adolescence, and identifies factors associated with hunger within the household. Individual and household-level data were extracted from the South African National Health and Nutrition Examination Survey (SANHANES-1). Prevalence of food insecurity was assessed using the Community Childhood Hunger Identification Project (CCHIP) index. Multinomial logistic regression analyses were conducted on all households (with and without children) to determine the predictors of food insecurity, with additional analyses adjusting for child dependency and sociodemographic characteristics of household heads in households with children. Of 5 098 households surveyed, 68.6% had children and adolescents present (0–19 years). Of these households, 32.5% (95% Confidence Interval [CI]: 29.5–35.7) were experiencing hunger and 26.3% (95% CI: 23.9–28.8) were at risk of hunger. Among all the households, significant associations for experiencing hunger were the presence of children and adolescents: Adjusted Odds Ratio (AOR) = 1.68 (95% CI: 1.12–2.53); being female-headed: AOR = 1.53 (95% CI: 1.21–1.94) and informally-located; AOR = 1.6 (95% CI: 1.07–2.43). Of the racial groups, having a non-African household head, Coloured: AOR = 0.29 (95% CI: 0.19–0.44) and White/Indian/Asian: AOR = 0.12 (95% CI: 0.04–0.33) conferred lower odds of experiencing hunger; and, the household head having secondary/tertiary education conferred lower odds of experiencing hunger; AOR = 0.40 (95% CI: 0.28–0.56) as well as being at risk of hunger; AOR = 0.69 (95% CI: 0.52–0.92). Receiving social grants, pensions, or remittances more than doubled the odds of experiencing hunger; AOR = 2.15 (95% CI: 1.49–3.09). After adjusting for child dependency in households with children, having at least one older child (age 15–19 years old) did not change the odds of food insecurity. In summary, only 41% of South African households with children and adolescents were food secure. The associations between household head sociodemographics, household location and size on household food insecurity indicate a need for multi-sectoral interventions to bolster sustainable food systems for households with children and adolescents and to improve public protections for female-headed, African-headed and informally-located households dependent on social grants.

Line 221: HSRC is now spelled out with first time use in the manuscript (Human Sciences Research Council) Line 239: The CCHIP Index is spelled out in full with first time use in Line 132 of the manuscript

Reviewer #4
Thank you for the opportunity I have to provide a review for this important article on Childhood food insecurity in South Africa. The authors have done a good job of identifying the extent of childhood hunger and factors which contribute to it. Title • The title clearly depicts the research that was carried out with focus on the outcome, however the period or time of investigation is missing -We do not specify the dates of the original SANHANES study (2011)(2012) or when we undertook the secondary analysis (2019-2020) in the title of the manuscript. We have, however, now indicated these timelines in the Methods section to ensure greater clarity. -We have updated the references as requested; however, many of the "older" references provide an historical context to this work. -In Methods, we clarified when the SANHANES-1 data were collected originally, and when we undertook this data extraction and conducted our analyses.

Introduction
-We clarified when the original SANHANES-1 study was conducted, as well as when the secondary data analysis was undertaken. -The Ethical Considerations sub-section has been moved to the end of Methods. they want.

Results
• Well written results and clearly stated • In lines 354-6 -larger household as a predictor of experiencing hunger seems not to be related. LCL of 1.00 shows that if repeated a researcher can get AOR of 1.00. I think 4 digits LCL should be used if not it should not be reported because it is an unlikely association.

Discussion
• The discussion is well written and provides the necessary reasoning behind the results of the study • It is was nice that they provided sub-headings which made an easy understanding of the discussion.
• The limitation of the study was clearly stated which eventually clarifies some of my earlier questions on when this study was conducted. But I think the authors should clearly state the year of study in the title.

Conclusion
• Provides answers to the objectives of the study • The recommendations made by the authors may not be relevant since the authors agreed that this is a very old data, but the only data available for this kind of secondary data analysis.
-As regards the association between household size and experiencing hunger, we re-ran the analysis -The LCL was 1.009218, which we've now rounded up to 1.01, and updated both the manuscript (Line 485) and the Fig2B to reflect these changes.
-We have clarified earlier in the manuscript when the original SANHANES-1 data collection happened, as well as when we undertook the secondary analysis.
-Until the SANHANES-2 is conducted, these data provide the only baseline for food insecurity in South African households with children and adolescents.

Reviewer #5
It -We have changed the title of the manuscript to: "Child and adolescent food insecurity in South Africa: A household-level analysis of hunger" -Please see lines 271-275 where we have adapted the language to address this point and clarify the research gap: "There is a paucity of nationally representative data to assess the granular detail about food insecurity across the country. The critical need to address hunger in childhood necessitates particular attention to the factors affecting food insecurity in households with children and adolescents-such as the household size, household head sociodemographics and dependency status of children-so that nutrition-sensitive interventions can be developed and/or strengthened" -We have reconfigured the presentation of the AORs and 95% CIs to make it clearer throughout the text. -This is how we had previously expressed our results; however, earlier reviewers indicated that using the language of "preventive/protective risk for hunger" was not appropriate in this context. Therefore we changed the language according to their preferences.
-We removed three key words: "household", "hunger" and "CCHIP Index" -The time period throughout the study is the same. We have stated this in the Methods and in all other references to the SANHANES-1 as the original data source -We have included the implications of our study in the Conclusion section of the manuscript -The main outcome indicator for our study was the degree of household food insecurity as measured by the Community Childhood Hunger Identification Project (CCHIP) Index. The CCHIP has been internationally validated according to a particular scale with three different outcome measures: 'food secure', 'at risk of hunger' and 'experiencing hunger'. These three categories are standard, so it would not possible to re-categorise into 'low', 'medium' or 'high' as suggested.

Journal Requirements
Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article's retracted status -We went through each reference to determine its retraction status. We do not cite any publication that has been retracted.