**4. Discussion**

In our study on community-dwelling adults aged ≥50 years in South Africa, we found a high prevalence of food insecurity (31.8%). After adjustment for potential confounders, moderate and

severe food insecurity were associated with significant 2.82 and 2.51 times higher odds for MCI, while the OR was particularly elevated for severe food insecurity among those aged ≥65 years (OR = 3.87). Depression was not a major explanatory factor in this association.

To the best of our knowledge, this is the first study to specifically examine the relationship between food insecurity and MCI. Our findings are in line with previous studies that have examined the association between food insecurity and cognitive function in the USA although these studies were on general cognitive function and not MCI. One community-based cross-sectional study conducted among a representative sample of 1358 Puerto Ricans aged 45–75 years living in Massachusetts found that the adjusted difference in the Mini-Mental State Examination (MMSE) score was 0.90 lower among those with very low food security as compared to those who were food secure [8]. A longitudinal follow-up study among 597 participants aged 40–75 years based on the same study population found that food insecurity at baseline was associated with lower global cognitive function after two years of follow-up [7]. Finally, one cross-sectional US study using data from the NHANES found that among 1851 adults between 60 and 85 years, food insecurity was associated with poor cognitive function [6].

The exact mechanisms linking food insecurity and MCI are unknown, but there are several hypotheses. It is possible that stress resulting from food insecurity may increase the risk of MCI. For example, prolonged elevation of cortisol (an HPA axis response to chronic stress) can lead to alterations in brain structure and function (e.g., in the hippocampus) and subsequent cognitive decline [29]. Also, increases in pro-inflammatory cytokines induced by stress [30] may lead to an increased risk of dementia [31]. Furthermore, food insecurity often compromises diet quality, as people tend to switch to more affordable but less nutritious food when food is scarce (e.g., high fat and carbohydrates, low vitamins and micronutrients) [32]. Poor diet has been associated with increased risk of cognitive decline. For example, vitamin B, C, and E have all been shown to have a protective effect against dementia [33]. Furthermore, randomized clinical trials have shown that supplementation of n-3 polyunsaturated fatty acids [34] and folic acid [35] may improve cognitive function in older people with MCI. Finally, high carbohydrate intake has been associated with higher risk of MCI [36].

Our finding that a particularly strong association was observed among older individuals concurs with that of the study by Gao and colleagues [8], which found that the difference in MMSE scores between very low food insecurity and food secure were −1.75 and −0.48 for those aged ≥60 and <60 years, respectively [8]. This age difference may be related to increased vulnerability of the brain among the older population. For example, older adults have smaller hippocampal volumes and this may increase susceptibility to cognitive deficits when exposed to stress [37]. The fact that people aged ≥65 years had 3.87 times higher odds for MCI compared with those without food insecurity is an important finding as the risk of dementia is particularly elevated in this age group [38].

Strengths of the study include the large sample size and the use of nationally representative data. However, some limitations should be taken into consideration when interpreting the findings. First, participants with mild forms of dementia could have been included in our study sample as the study did not include a clinical assessment of dementia. Second, there is no consensus regarding the acceptable level of functional impairment in MCI [39]. The definition of preservation of independence in functional abilities used in our study, which has been used in previous publications [17,18,28,40], was rather conservative. This was done to avoid the omission of MCI cases with disability not related to their cognitive ability. Despite these potential limitations, it is worth noting that the prevalence of MCI in our study was consistent with previously reported figures [41]. Third, data on biomarkers such as cerebrospinal fluid Aβ and tau were not available. These data could have provided a better understanding on how food insecurity affects brain pathology [4]. Fourth, the estimates in the analyses stratified by age should be interpreted with caution as the sample size was small and some estimates had wide confidence intervals. Next, we lacked information on HIV infection, which has been reported to be associated with a higher risk of food insecurity [42] and impaired cognitive function [43]. However, given that the link between food insecurity and HIV infection is likely to be mainly explained by poverty [42], we believe that the adjustment for wealth in our study is likely to have minimized the

potential for residual confounding due to HIV infection. Furthermore, our measure of food insecurity was based on two questions and did not constitute a comprehensive food insecurity measure. Finally, because this was a cross-sectional study, causality cannot be inferred. For example, it is possible that people with cognitive impairments have difficulty in utilizing social safety net services, and this might have led to food insecurity.
