Longitudinal food insecurity patterns and cognitive performance among midlife Latina women: Findings from the CHAMACOS Maternal Cognition Study

Abstract

Background

Although food insecurity (FI) has been linked to cognitive aging, few studies have examined associations in midlife or socioeconomically disadvantaged populations or considered longitudinal FI patterns.

Objective

To examine cross-sectional and longitudinal associations between FI and cognitive performance in a prospective cohort of midlife women.

Methods

We analyzed data from 512 women in the CHAMACOS Maternal Cognition Study (2018–2024). FI was measured twice (approximately 3.2 years apart) with the six-item Short Form of the USDA Household Food Security Survey Module (range: 0–6). We categorized FI cross-sectionally (secure, scores of 0–1; insecure, scores 2–6) and longitudinally (persistent security, persistent FI, new FI). Cognitive performance was assessed across multiple domains with the SOL-INCA-AD assessment. We estimated associations using linear regression adjusted for sociodemographic confounders and, among a subset, APOE ɛ4 carrier status.

Results

Cross-sectionally, FI was associated with lower global cognition z-scores (β = −0.11, 95% CI: −0.21, −0.01), with similar trends for executive function (β = −0.10, 95% CI: −0.19, 0.01) and memory domains (β = −0.13, 95% CI: −0.29, 0.03). Longitudinally, new FI (versus persistent security) was associated with lower scores across all domains (Global β = −0.19, 95% CI: −0.32, −0.06; Executive function β = −0.15, 95% CI: −0.27, −0.02; Memory β = −0.22, 95% CI: −0.43, −0.01; Verbal fluency β = −0.17, 95% CI: −0.36, 0.02). Persistent FI showed weaker associations.

Conclusions

In this cohort of midlife women, FI and particularly new onset FI were associated with worse cognitive performance after adjusting for sociodemographic and genetic confounders.

Keywords

aging Alzheimer's disease cognitive function dementia executive function nutrition

Introduction

Food insecurity (FI), defined as limited or uncertain access to adequate food, is a critical public health issue that affects millions of individuals globally.^1,2 In 2023, 18 million U.S. households experienced FI at some point during the year, with 11.2 million facing low food security and 6.8 million experiencing very low food security.³ FI disproportionately impacts low-income and people of color, exacerbating health disparities.⁴ For example, the rate of FI among Latino households was 21.9% in 2023, more than double the 9.9% observed in non-Latino white households in the same year.^5,6 There is growing observational evidence linking FI to cognitive impairment and dementia.^7–17 Some of the potential mechanisms underlying the relationship between FI and dementia include stress, reduced nutrient intake, depression, cardiometabolic health, and sleep, although exact pathways remain uncertain.^7–11^,16,18–23

Several important gaps remain in our understanding of the link between FI and dementia risk. First, although FI disproportionately impacts populations with low socioeconomic status (SES), most studies assessing its effects on cognition have been in populations of relatively high SES.¹⁰ Given that these higher-income groups benefit from better education and healthcare, the estimated effects may not fully capture the impacts faced by lower-SES populations. Second, most of the research on FI and cognition has focused on older adults, with median ages ranging above 60 years,^7,10–13^,17 leaving FI in midlife relatively understudied. Dementia-related neuropathology begins long before symptoms manifest, and early and midlife exposures have been associated with later outcomes.^24,25 Therefore, studying midlife may help identify potential risk and protective factors and inform the timing of future prevention efforts.²⁶

Very few studies have explored associations between midlife FI and cognition in US Latinos, despite their disproportionate experiences of FI. US Latinos are also at a higher risk for cognitive impairment, earlier onset of Alzheimer's disease and related dementias (ADRDs), and fewer dementia-free years compared to non-Latino white adults.^27–30 To our knowledge, only one study assessed mid-life FI and cognition among a US Latino population: in a cohort of Puerto Rican adults in Massachusetts, very low food security was associated with lower cognitive performance and faster cognitive decline.^9,14

Lastly, chronicity of FI has rarely been assessed in studies exploring FI and cognition. Most studies measure FI at a single point in time, which may obscure important differences between individuals who experience persistent FI and those who have encountered it more recently. Limited longitudinal research in US and Mexican cohorts suggests that both persistent and new FI may be linked to worse cognitive performance.^15,31,32 In the Mexican Health and Aging Study, persistent and new FI had similar associations with verbal fluency, executive function, and memory acquisition scores, but persistent FI had stronger effects for memory retention scores.³² In a US-based cohort, persistent and new FI had similar associations with executive function scores while new FI had stronger associations with memory, verbal fluency, and global cognition scores.³¹

In this study, we evaluated the relationship between FI and cognitive performance among a prospective cohort of mid-life, predominantly Mexican-American women with relatively low socioeconomic status. We first evaluated cross-sectional FI in association with both domain-specific and global cognitive performance measures. Because cognitive function may influence an individual's ability to maintain food security, we also aimed to address temporality and reverse causation by evaluating longitudinal measures of FI (persistent and new FI) captured across approximately 3.2 years. We accounted for multiple other components of socioeconomic status to reduce residual confounding by related factors (e.g., poverty) as well as APOE genotype (in sensitivity analyses). We hypothesized that both cross-sectional and longitudinal FI would be associated with lower cognitive performance in midlife. We considered competing hypotheses for associations with persistent versus new FI: first, that persistent FI would be associated with worse cognitive performance compared to new FI, given the potential for cumulative impacts; and second, that new FI would be associated with worse cognitive performance compared to persistent FI, given that it might arise in the context of other acute financial stressors.

Methods

Study design and population

The CHAMACOS cohort

We included women from the long-standing Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) cohort.^33,34 The initial CHAMACOS study was launched in 1998 as a long-term, community-based cohort of low-income, primarily Mexican American families in the Salinas Valley, an agricultural area in California.³⁴ The first cohort (CHAM1) consisted of women recruited during pregnancy from local health clinics between October 1999 and October 2000. Eligible participants were over 18 years old, spoke English or Spanish, qualified for Medi-Cal, were less than 20 weeks pregnant, and intended to deliver at the county hospital. Of the 1130 women initially contacted, 597 enrolled, 525 remained until delivery, and 348 were still engaged in the study as their index child(ren) turned 9. A refresher cohort (CHAM2) of 287 mothers with 9-year-old children was recruited between January 2010 and September 2011 through local health clinics, schools, churches, community service agencies, and by referral from CHAM1 participants, with similar eligibility criteria, for a total of 635 participants. From that point, CHAM1 and CHAM2 families proceeded to complete study visits every ∼1.5–2 years. Study visits always included an in-depth maternal questionnaire administered by bilingual staff in the participant's preferred language (Spanish or English).

The Maternal Cognition Study (MCS)

In 2022–2024, we invited 577 maternal participants who had completed a study visit between 2014 and 2021 to participate in the CHAMACOS Maternal Cognition Study (MCS).³⁵ A total of 519 maternal participants completed the CHAMACOS MCS study visit. The MCS visit included a battery of neurocognitive tests and an in-depth questionnaire, both administered in the participant′s preferrred language (Spanish/English) by a bilingual interviewer; the visit also included blood pressure and body measurements, blood draw, and urine collection. The outcome (i.e., cognitive performance) for both our cross-sectional and longitudinal analyses was assessed at the MCS visit. Our cross-sectional analysis assesses FI exposure at the MCS visit only, while our longitudinal analysis uses exposure information from both the MCS visit and the previous CHAMACOS visit (2018-2021) to assess change in food security status. Study procedures were approved by the University of California, Berkeley Office for the Protection of Human Subjects (Protocol ID 2021-02- 14,055) on 12/08/2021. All participants provided written informed consent prior to participating to conduct and publish the study. Participant information has been anonymized.

Final analytic samples

For the cross-sectional analysis, we included CHAMACOS MCS participants, excluding 4 respondents with missing exposure information and 3 with missing covariate information. Given variation in missingness on outcome measures (n = 16 for executive function tasks), our final analytic sample ranged from n = 496 (for executive function and global cognitive performance scores) to n = 512 (for memory and verbal fluency scores).

For the longitudinal analysis assessing change in food security, our analytic sample includes women who participated in the CHAMACOS study visit (2018–2021) and CHAMACOS MCS. Exposure assessment (i.e., assessment of change in food security status) begins at the mothers’ 2018–2021 CHAMACOS visit, when index children were ∼18 years old. We excluded 25 participants with missing exposure information and 3 with missing covariate information from the 2018–2021 visit. We further excluded 46 participants who experienced new food security in the MCS visit because this group was too small to meaningfully compare. This resulted in an analytic sample of n = 426 (for executive function and global cognitive performance scores) to n = 438 (for memory and verbal fluency scores) for the longitudinal analysis.

In sensitivity analyses including APOE ɛ4 carrier status as a covariate, the cross-sectional analytic sample ranged from n = 477 (for executive function and global cognitive performance scores) to n = 488 (for memory and verbal fluency scores), and the longitudinal analytic sample ranged from n = 411 (for executive function and global cognitive performance scores) to n = 419 (for memory and verbal fluency scores). See Supplemental Figure 1 for a full flowchart of participant recruitment and final analytic samples.

Exposure assessment: food insecurity

Food insecurity measurement

We used the Six-Item Short Form of the USDA Household Food Security Survey Module to assess FI at each visit.³⁶ We asked all participants to report the frequency of the following two experiences for their household in the past 12 months: (1) “The food that we bought just didn't last, and we didn't have money to get more” and (2) “We couldn’t afford to eat balanced and nutritious meals” Response options for these two items were “often true”, “sometimes true”, or “never true”. Respondents were additionally asked (3a): “In the last 12 months, did you or other adults in the household ever cut the size of your meals or skip meals because there wasn't enough money to buy food?” (Response options: Yes, No). If participants answered “Yes”, they were asked a follow-up question: (3b) “How often did this happen in the last 12 months?” (Response options: Almost every month, Some months but not others, Only one or two months). If participants answered “Often true” or “Sometimes true” to either of the first two questions or “Yes” to the third question, we asked two additional follow-up questions: (4) “In the last 12 months, did you ever eat less than you felt you should because there wasn't enough money to buy food?” (Response options: Yes, No); and (5) “In the last 12 months, were you ever hungry but didn’t eat because you couldn’t afford enough food?” (Response options: Yes, No).

In accordance with USDA scoring guidelines, we created a continuous FI score ranging from 0 to 6 based on these responses. We assigned one point for each of the following: responses of “often true” or “sometimes true” for question (1), “often true” or “sometimes true” for question (2), “yes” for question (3a), “almost every month” or “some months but not others” for question (3b), “yes” for question (4) and “yes” for question (5). From the continuous FI score, we created a categorical FI variable, following USDA's categories of “high or marginal,” “low,” and “very low food security.” Low food security is defined as reducing the quality, variety, and desirability of diets without substantially disrupting food intake and eating patterns, while very low food security is defined as disrupting eating patterns and food intake due to financial or other resource constraints.³⁷ We classified scores of 0 or 1 as high or marginal food security, scores of 2–4 as low food security, and scores of 5 or 6 as very low food security. Only 13% of participants experienced very low food security. To reduce the risk of participant re-identification and improve statistical power and model stability, we collapsed “low” and “very low food security” into a binary “FI (yes/no)” variable. This approach is consistent with USDA reporting practices.³⁷

For our longitudinal analysis, we generated a measure of change in food security status using the binary food security status variable from the CHAMACOS MCS visit and the 2018–2021 CHAMACOS visit, completed on average 3.2 years prior to the MCS visit. Those who were food secure at both visits were classified as having “persistent food security,” while those who were food insecure at both visits were classified as having “persistent FI.” Those who were food secure at the prior visit but food insecure at the MCS visit were classified as having “new FI.” Lastly, those who were food insecure at the prior visit, but food secure at the MCS visit were classified as having “new food security,” but this very small group (n = 46) was dropped from the analysis.

Outcome assessment: cognitive performance

Cognitive performance measurement

An adapted version of the Study of Latinos – Investigation of Neurocognitive Aging – Alzheimer's disease (SOL-INCA-AD) neurocognitive battery was administered by trained, certified bilingual and bicultural interviewers.³⁸ The assessment included several cognitive tasks: the Brief Spanish English Verbal Learning Test (B-SEVLT),^39,40 the Digit Symbol Substitution (DSS) test,⁴¹ the Trail Making Test (TMT) Parts A and B,⁴² the Digit Span Test (both Forward and Backward),⁴³ the B-SEVLT Delayed Recall, two phonemic fluency tasks (letter-based), and one semantic or categorical fluency task (see Supplemental Table 1 for additional details). Scores for each task were standardized to z-scores, with a mean of zero and a standard deviation of one, where a higher (positive) z-score represents better performance compared to the analytic sample mean and a lower (negative) z-score represents worse performance compared to the analytic sample mean.

Cognitive performance endpoints

To construct our cognitive performance endpoints, we grouped individual task z-scores into composite scores reflecting domains of memory, verbal fluency, and executive function. The memory composite reflects the mean of the z-scores for the B-SEVLT tasks, the verbal fluency composite reflects the mean of the z-scores of the phonemic and semantic fluency tasks, and the executive function composite represents the mean of the z-scores of the DSS, TMT Parts A and B, and the Digit Span Forward and Backward tests. We additionally created a global cognition composite z-score that represents the mean of the three domain-specific z-scores.

Covariates

We used prior epidemiologic literature on determinants of FI and cognitive performance, as well as on shared risk factors linking socioeconomic disadvantage to cognitive outcomes, to inform our covariate selection. We used a directed acyclic graph (DAG) to evaluate plausible common causes of midlife FI and cognitive performance scores and identify a minimally sufficient adjustment set (Figure 1). Our minimal adjustment set included APOE ɛ4 carrier status (an indicator of genetic risk for neurodegenerative disease; no ɛ4 alleles or one or two ɛ4 alleles), age at neurocognitive exam (continuous), nativity and age-at-migration (US-born; born in Latin America and first arrived in the US in childhood [<18 years old]; born in Latin America and first arrived in the US in adulthood [≥18 years old]), educational attainment (<6th grade, 7–12th grade, > 12th grade), parental educational attainment (no formal education, any formal education, or don’t know/don’t remember for both parents), current marital status (married/living as married or not married/living as married), current employment status (no work, field work or other agricultural work, other work [non-agricultural work])), and current household poverty status (reporting a household income at or below the Federal Poverty Level [FPL] threshold, above the FPL threshold, or missing information to classify FPL).

Figure 1.

Directed acyclic graph identifying minimal and full adjustment set for analysis of associations between food insecurity and cognitive Z-scores.

All socioeconomic variables included in our minimal adjustment set reflect or are associated with aspects of economic (dis)advantage, which have in turn been shown to lead to both food insecurity and cognitive performance. For example, nativity and age-at-migration were included because immigration history is associated with access to social and economic resources, which influence food insecurity risk⁴⁴ and cognitive health. As another example, being partnered or married is associated with economic stability and other factors upstream of socioeconomic disadvantage (e.g., household size, employment status, household income) for women in particular;⁴⁵ marital status has also been linked to cognitive function.⁴⁶

In sensitivity analyses, we also included APOE ɛ4 carrier status (an indicator of genetic risk for neurodegenerative disease; no ɛ4 alleles or one or two ɛ4 alleles). Because APOE ɛ4 genotype was only available for a subset of participants who provided blood samples and consented to genetic analysis, we included APOE ɛ4 carrier status in a sensitivity analysis restricted to this subset.

We additionally considered language of neurocognitive assessment (English or Spanish), current number of children in the home (continuous), and current household crowding (≤2 people per bedroom or >2 people per bedroom) but did not include them in our minimally sufficient adjustment set because, according to our DAG, they were consequences of upstream socioeconomic factors that were already adjusted for. We instead included these additional covariates in our full adjustment set to assess whether conclusions are robust to more extensive covariate adjustment.

For the cross-sectional analysis, time-updated covariates were from the CHAMACOS MCS (2022–2024) visit. For the longitudinal analysis, time-updated covariates were from the 2018–2021 CHAMACOS visit.

Statistical analysis

We first calculated summary statistics for the entire analytic sample and stratified by food security status. We then fit crude, minimally adjusted, and fully adjusted linear regression models of the association between domain-specific and global cognitive performance z-scores and FI at the MCS visit (“food secure” as the reference group). We then estimated crude, minimally adjusted, and fully adjusted linear regression models of the association between cognitive performance z-score and change in food security status across visits (“persistent FI” and “new FI,” with “persistent food security” as the reference group).

Missing data were handled via exclusion from the analytic sample, with the exception of the household poverty status covariate. Because a significant proportion of the analytic sample did not know their household income information (n = 36 or 7%), and because participants who do not know their household income information may be systematically different from those who do, we chose to retain them in the analytic sample using a category of “missing poverty information.”

Analyses were completed using STATA version 15 (StataCorp, College Station, Texas, USA) and the figures were generated using R version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

In our overall analytical sample (n = 512), participants’ mean age was 48.8 years old (SD = 5.4) (Table 1). Less than one-fifth of participants (19%) carried one or two APOE ɛ4 alleles. The majority of participants were born outside the United States (89%); of the entire sample, 26% arrived during childhood and 64% arrived during adulthood. A large proportion of participants had limited education, with 44% having completed less than 6th grade. Additionally, 23% reported that neither of their parents had received formal schooling.

Table 1.

Participant demographic characteristics^a by food security status, CHAMACOS Maternal Cognition Study (n = 512).

	Total (n = 512)	Food secure (n = 309, 60.4%)	Food insecure (n = 203, 39.7%)
Age, years, mean ± SD	48.8 ± 5.4	48.5 ± 5.2	49.2 ± 5.7
APOE ɛ4 carrier status, n (%)
No ɛ4 alleles	396 (81.2)	239 (81.3)	157 (80.9)
One or two ɛ4 alleles	92 (18.9)	55 (18.7)	37 (19.1)
Nativity, n (%)
US-born	55 (10.7)	35 (11.3)	20 (9.9)
Arrived in Childhood (<18 years old)	132 (25.8)	83 (26.9)	49 (24.1)
Arrived in Adulthood (≥18 years old)	325 (63.5)	191 (61.8)	134 (63.5)
Language of exam, n (%)
English	58 (11.3)	40 (12.9)	18 (8.9)
Spanish	454 (88.7)	269 (87.1)	185 (91.1)
Educational attainment, n (%)
≤6th grade	223 (43.6)	125 (40.5)	98 (48.3)
7–12th grade	175 (34.2)	105 (34.0)	70 (34.5)
>12th grade	114 (22.3)	79 (25.6)	35 (17.2)
Highest level of parental educational attainment, n (%)
No formal educational attainment	118 (23.1)	60 (19.4)	58 (28.6)
Any formal educational attainment	337 (65.8)	218 (70.6)	119 (58.6)
Don’t know or don’t remember for both parents	57 (11.1)	31 (10.0)	26 (12.8)
Marital status, n (%)
Married/Living as married	362 (70.70)	234 (75.7)	128 (63.1)
Not married/Living as married	150 (29.3)	75 (24.3)	75 (37.0)
Household crowding
≤2 People per bedroom	408 (79.7)	252 (81.6)	156 (76.9)
>2 People per bedroom	104 (20.3)	57 (18.5)	47 (23.2)
Children in home, mean ± SD	1.3 ± 1.2	1.3 ± 1.2	1.3 ± 1.1
Employment status since last visit^b, n (%)
No work	103 (20.1)	60 (19.4)	43 (21.2)
Fieldwork or other Ag	244 (47.7)	138 (44.7)	106 (52.2)
Other Work (non-Ag)	165 (32.2)	111 (35.9)	54 (26.6)
Household poverty status, n (%)
At or below poverty threshold^c	181 (35.4)	77 (24.9)	104 (51.2)
Above the poverty threshold^c	295 (57.6)	208 (67.3)	87 (42.9)
Missing poverty information	36 (7.0)	24 (7.8)	12 (5.9)

Some categories have been collapsed or suppressed to prevent small cell sizes.

Last Visit completed an average of 1.67 (range: 0.67–8.12) years prior.

Calculated using reported monthly household income and household size according to the 2022 Poverty Thresholds designated by the US Census Bureau.

At the MCS visit, 309 (60%) participants experienced food security and 203 (40%) experienced FI. We observed differences in participant characteristics by current food security status. For example, 51% of those with current FI were classified as having household incomes at or below the FPL, compared to only 25% with current food security. Similarly, 76% of those with current food security were married or living as married, compared to 63% with current FI. Participants experiencing current FI were also more likely to report a sixth-grade education or less, no formal parental educational attainment, more than 2 people per bedroom, and agricultural work compared to participants experiencing current food security, but the differences were lower in magnitude. There were no differences in mean age, APOE ɛ4 carrier status, nativity, language of exam, or children in home by current food security.

Among 438 participants in the analysis with the longitudinal FI measure, 250 (57%) were persistently food secure, 101 (23%) were persistently food insecure, and 87 (20%) became newly food insecure (Supplemental Table 2). Compared to the persistently food secure group, participants who reported persistent FI differed across several sociodemographic factors, including own education, parental education, marital status, agricultural work, and poverty status. In contrast, the newly food insecure group more closely resembled the persistently food secure group in terms of own education and marital status, but showed intermediate levels for parental education, agricultural work, and poverty status. Notably, household crowding reported at the MCS visit was higher among the newly food insecure (31%) than both persistent food secure (18%) and persistent food insecure (25%) participants, suggesting a unique pattern of housing-related vulnerability in this group.

In crude regression models, experiencing current FI was associated with lower scores across all cognitive domains, compared to current food security: memory (β = −0.23, 95% confidence interval [CI]: −0.39, −0.07), verbal fluency (β = −0.16, 95% CI: −0.31, −0.01), executive function (β = −0.20, 95% CI: −0.32, −0.08), and global composite (β = −0.20, 95% CI: −0.31, −0.08). After adjusting for the minimal adjustment set of covariates, experiencing FI was associated with lower global composite z-scores (β = −0.11, 95% CI: −0.21, −0.01) (Figure 2, Supplemental Table 3). Associations between FI and lower memory z-scores (β = −0.13, 95% CI: −0.29, 0.03) and executive function z-scores (β = −0.10; 95% CI: −0.19, 0.01) were also notable, although the upper bounds of the 95% confidence intervals crossed the null. The association with verbal fluency z-scores (β = −0.06; 95% CI: −0.20, 0.09) was in the expected direction but of smaller magnitude. Results from fully adjusted models did not differ meaningfully from those of minimally adjusted models (Supplemental Table 3).

Figure 2.

Minimally adjusted^a associations between food insecurity and cognitive Z-scores (reference group: food secure), CHAMACOS Maternal Cognition Study (n = 512^b).

In analyses using the longitudinal FI measure, crude models showed associations between persistent FI and lower memory (β = −0.24; 95% CI: −0.45, −0.02), executive function (β = −0.24; 95% CI: −0.40, −0.08), and global composite z-scores (β = −0.21; 95% CI: −0.35, −0.06) compared to persistent food security. Persistent FI was also associated with lower verbal fluency z-scores (β = −0.18; 95% CI: −0.37, 0.02), although the 95% confidence interval crossed the null. Similarly, new FI was associated with lower z-scores across all domains compared to persistent food security: memory (β = −0.24; 95% CI: −0.46, −0.02), verbal fluency (β = −0.22; 95% CI: −0.43, −0.01), executive function (β = −0.19; 95% CI: −0.35, −0.03), and global composite (β = −0.22; 95% CI: −0.37, −0.07).

After adjusting for the minimal adjustment set of covariates, associations between persistent FI and cognitive z-scores were attenuated and all 95% confidence intervals crossed the null (Figure 3, Supplemental Table 4). Adjusted models of new FI continued to show associations with executive function (β = −0.15; 95% CI: −0.27, −0.02), memory (β = −0.22; 95% CI: −0.43, −0.01), and global composite z-scores (β = −0.19; 95% CI: −0.32, -0.06). The association with verbal fluency z-scores (β = −0.17; 95% CI: −0.36, 0.02) was of similar magnitude and in the expected direction, but the 95% confidence interval crossed the null. Similar to the cross-sectional analysis, results from fully adjusted models did not differ meaningfully from those of minimally adjusted models (Supplemental Table 4).

Figure 3.

Minimally adjusted^a associations between persistent food insecurity and new food insecurity and cognitive Z-scores (reference group: persistent food security), CHAMACOS Maternal Cognition Study (n = 438^b).

Adjusting for APOE ɛ4 carrier status did not meaningfully change results for the cross-sectional or longitudinal analyses (Supplemental Tables 5 and 6).

Discussion

Within our cohort of midlife, predominantly Mexican-American women (mean age = 48.8), we found that current FI was associated with lower scores in global cognition and domains of executive function, memory, and, to a lesser extent, verbal fluency. When we assessed FI over time, we found limited to no evidence of association between persistent FI (versus persistent food security) and cognitive performance scores. However, new onset of FI was associated with lower scores across all domains. To place our estimates in context, we calculated that in our sample, each year of additional age was associated with 0.026 SD lower global cognition z-score (95% CI: −0.04, −0.02). This means that compared to those who experienced current food security, those who experienced current FI had lower estimated cognitive scores equivalent to a difference of roughly 4 years of age. Compared to those who experienced persistent food security across two timepoints, those who experienced new onset of FI had scores equivalent to roughly 7 years of age. These findings provide insight into the potential relationship between FI and midlife cognition, suggesting that there are associations with all cognitive domains and that the chronicity and timing of FI matters.

Previous research has shown that FI is associated with worse cognitive function.^8,12–14^,17 Research from the National Health and Nutrition Examination Survey (mean age = 69.8) found that older adults experiencing marginal, low, and very low food security, defined using the USDA Household Food Security Survey Module, all had significantly lower executive function scores compared to those who were food secure.¹² The Boston Puerto Rican Health study, which examined midlife adults (mean age = 57.8), found that very low food security, as defined by the USDA, was associated with lower memory, verbal fluency, executive function, and global cognition scores.¹⁴ Our findings are consistent with these studies, further supporting an association between FI and poorer cognitive performance in midlife cohorts. Furthermore, although our study assesses cognitive performance at a single time point, other research on cognitive performance over time has shown an association between FI and cognitive decline.^8–11

Many fewer studies have examined persistent and new onset FI in association with cognitive outcomes. The Mexican Health and Aging Study (mean age = 65.4) linked persistent and new FI to poorer memory, verbal fluency, and executive function scores compared to persistently food secure adults.³² In the US-based Coronary Artery Risk Development in Young Adults Study (mean age at baseline = 40.3), persistent and new FI (versus persistent food security) had similar associations with executive function scores, although associations between new FI and memory, verbal fluency, and global cognition scores 15 years after baseline were of larger magnitude than associations with persistent FI.³¹ In our study, after adjusting for sociodemographic covariates, we found associations with new FI only. One possible explanation is that new experiences of FI may produce acute stress and disruption, which could be linked to cognitive performance.⁴⁷ In contrast, individuals facing persistent FI may develop coping strategies or adapt over time, potentially attenuating its association with cognition.⁴⁸ Our results suggest that sociodemographic characteristics may explain a substantial portion of the observed differences in cognition associated with persistent FI, leaving little independent association after adjustment, whereas new FI may capture other recent instability or shocks, such as acute housing insecurity, that are associated with short-term differences in cognitive performance. It is important to note that the COVID-19 pandemic occurred between our two study visits in which FI was assessed, which could have contributed to new FI and other related stressors.^49,50

Several interacting biological and psychosocial mechanisms may explain the observed association between mid-life FI and cognitive performance. FI-related stress and anxiety can activate the hypothalamic-pituitary-adrenal axis and elevate cortisol, damaging the hippocampus, important for memory function, and the prefrontal cortex, which plays a role in executive function.^{18,19,51–53} Depression, more common among food insecure individuals,^18,54,55 is also consistently linked with lower cognitive performance.^20,21,56,57 Nutritional pathways may also contribute; financial constraints often limit dietary variety and essential nutrients (e.g., omega-3 fatty acids, B vitamins),^23,58 which protect against neurodegeneration,^22,59 and are associated with improved cognition.^2,60 FI is also linked to higher prevalence of obesity, diabetes, and hypertension, established risk factors for cognitive impairment.^61–67 Stress, depression, poor nutrition, and cardiometabolic disease can all disrupt sleep patterns and reduce sleep duration, known risk factors for cognitive impairment across the lifecourse.^68–70 Finally, these processes are all associated with systemic inflammation, including elevated pro-inflammatory cytokines (e.g., c-reactive protein, interleukin-6), which can lead to neurodegeneration and cognitive decline.^71,72

This study has limitations that should be considered. First, the cross-sectional measurement of FI and cognitive performance limits our ability to fully rule out reverse causation, whereby underlying cognitive function could influence an individual's ability to maintain food security. Although our secondary analysis leveraged two timepoints of FI to classify persistent and new food insecurity, cognitive performance was measured only at the later visit and we are therefore unable to determine whether prior cognitive function influenced FI trajectories. Our sensitivity analysis controlled for APOE ɛ4 carrier status which, along with other variables (e.g., educational attainment), would account in part for the bias introduced through pathways linking food insecurity to current cognitive performance scores via pre-exposure cognitive performance (see DAG in Figure 1). Second, although we used two timepoints to assess the chronicity of FI, with only ∼3 years of follow-up, our ability to capture associations with cumulative FI is limited. Extending exposure assessment over additional timepoints could provide a more complete picture, but this was beyond the scope of the current analysis. Third, the self-reported measures of FI may also be subject to measurement error, potentially due to social desirability bias (e.g., participants may underreport FI due to stigma) or recall bias (e.g., participants may not remember FI status across the entire 12-month period). Finally, we are unable to assess the relationship of food insecurity to cognitive decline over time, as we currently have only one wave of cognitive assessments. However, follow-up data collection is currently in progress, which will allow for future longitudinal analyses. Repeated cognitive performance measures would also support efforts to further account for reverse causation beyond the steps taken in our analyses.

Despite these limitations, this study has several important strengths. We administered a comprehensive neurocognitive assessment, allowing for domain-specific cognitive performance scores in memory, executive function, verbal fluency, and global cognition. The richness of our covariate data, collected through a longstanding prospective cohort, allowed us to adjust for a wide range of socioeconomic confounders to isolate the specific association of FI. We assessed two timepoints of FI and were able to examine associations with persistent and new FI separately. Our study focuses on a midlife, predominantly low-income Latina population — an understudied group in research on FI and cognition. By examining FI in midlife rather than later life, we address a critical gap in the literature, as most prior studies have centered on older adults. Moreover, concerns about reverse causation are lessened in a healthy midlife sample; respondents are less likely to be experiencing symptoms of neurodegeneration that might contribute to FI risk as compared to participants in older cohorts. Finally, by focusing on an underrepresented population in cognitive research, our work contributes valuable knowledge on early risk factors that may help inform future efforts to reduce disparities in cognitive health.

Our findings have important implications for shaping future public health initiatives. If the observed associations reflect causal relationships, addressing FI in midlife may improve early dimensions of brain aging, including midlife cognitive performance. Future research should aim to untangle the interrelationship between FI and cognitive health across the lifespan. Additionally, analyzing repeated measures of FI may better capture its chronicity and evolving effects on cognition. Overall, initiatives aimed at improving food security, such as improving benefits and expanding eligibility for nutrition assistance programs like the Supplemental Nutrition Assistance Program (SNAP) in the United States,² could have downstream implications for cognitive health, particularly in vulnerable low-income populations.^60,73

Supplemental Material

sj-docx-1-alz-10.1177_13872877261453615 - Supplemental material for Longitudinal food insecurity patterns and cognitive performance among midlife Latina women: Findings from the CHAMACOS Maternal Cognition Study

Supplemental material, sj-docx-1-alz-10.1177_13872877261453615 for Longitudinal food insecurity patterns and cognitive performance among midlife Latina women: Findings from the CHAMACOS Maternal Cognition Study by Yesli Perez-Rocha, Lucia E. Calderon, Katherine Kogut, Nadia Rojas, Brenda Eskenazi, Marcella Warner, L. Paloma Rojas-Saunero, Robert B. Gunier, Elizabeth Ambriz and Jacqueline M. Torres in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

We would like to thank the continued participation of the CHAMACOS mothers and the hard work of the CHAMACOS field staff.

ORCID iD

Yesli Perez-Rocha

Ethical considerations

Study procedures were approved by the University of California, Berkeley Office for the Protection of Human Subjects (Protocol ID 2021-02-14055) on 12/08/2021.

Consent to participate

All participants provided written informed consent prior to participating to conduct and publish the study. Participant information has been anonymized.

Consent for publication

Not applicable

Author contribution(s)

Yesli Perez-Rocha: Conceptualization; Data curation; Formal analysis; Methodology; Visualization; Writing – original draft.

Lucia E. Calderon: Data curation; Formal analysis; Methodology; Project administration; Supervision; Visualization; Writing – original draft.

Katherine Kogut: Data curation; Methodology; Project administration; Writing – review & editing.

Nadia Rojas: Writing – review & editing.

Brenda Eskenazi: Conceptualization; Funding acquisition; Writing – review & editing.

Marcella Warner: Writing – review & editing.

L. Paloma Rojas-Saunero: Visualization; Writing – review & editing.

Robert B. Gunier: Writing – review & editing.

Elizabeth Ambriz: Writing – review & editing.

Jacqueline M. Torres: Conceptualization; Funding acquisition; Methodology; Supervision; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institutes of Health/National Institute on Aging, R01AG069090 and K99AG093067, and National Institute of Environmental Health Sciences, R01ES026994 and U24ES028529-06.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Supplemental material

Supplemental material for this article is available online.

References

Economic Research Service. Food security in the U.S. – measurement. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/measurement (accessed 19 March 2025).

Center on Budget and Policy Priorities. More adequate SNAP benefits would help millions of participants better afford food. https://www.cbpp.org/research/food-assistance/more-adequate-snap-benefits-would-help-millions-of-participants-better (2019, accessed 14 April 2025).

Economic Research Service. Food security in the U.S. – interactive charts and highlights. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/interactive-charts-and-highlights (accessed 24 March 2025).

National Institute on Minority Health and Health Disparities. Food accessibility, insecurity and health outcomes. https://www.nimhd.nih.gov/resources/understanding-health-disparities/food-accessibility-insecurity-and-health-outcomes.html (accessed 3 March 2025).

Rabbitt

Reed-Jones

Hales

, et al. Household food security in the United States in 2023. Washington, DC: Economic Research Service, US Department of Agriculture, 2024. https://www.ers.usda.gov/publications/pub-details?pubid=109895 .

Bread for the World. The state of hunger in Latino(a) communities. https://www.bread.org/article/the-state-of-hunger-in-latino-communities/ (2024, accessed 3 March 2025).

Koyanagi

Veronese

Stubbs

, et al. Food insecurity is associated with mild cognitive impairment among middle-aged and older adults in South Africa: findings from a nationally representative survey. Nutrients 2019; 11: 749.

Wang

El-Abbadi

. Food insecurity, race and ethnicity, and cognitive function among United States older adults. J Nutr 2024; 154: 233–242.

Wong

Scott

Wilde

, et al. Food insecurity is associated with subsequent cognitive decline in the Boston Puerto Rican health study. J Nutr 2016; 146: 1740–1745.

10.

Kim

Samuel

Thorpe

Jr , et al. Food insecurity and cognitive trajectories in community-dwelling medicare beneficiaries 65 years and older. JAMA Netw Open 2023; 6: e234674.

11.

Qian

Khadka

Martinez

, et al. Food insecurity, memory, and dementia among US adults aged 50 years and older. JAMA Netw Open 2023; 6: e2344186.

12.

Frith

Loprinzi

. Food insecurity and cognitive function in older adults: brief report. Clin Nutr 2018; 37: 1765–1768.

13.

Portela-Parra

Leung

. Food insecurity is associated with lower cognitive functioning in a national sample of older adults. J Nutr 2019; 149: 1812–1817.

14.

Gao

Scott

Falcon

, et al. Food insecurity and cognitive function in Puerto Rican adults. Am J Clin Nutr 2009; 89: 1197–1203.

15.

Tamargo

Meade

Campa

, et al. Food insecurity and cognitive impairment in the Miami adult studies on HIV (MASH) cohort. J Nutr 2021; 151: 979–986.

16.

Srivastava

Muhammad

. Rural–urban differences in food insecurity and associated cognitive impairment among older adults: findings from a nationally representative survey. BMC Geriatr 2022; 22: 287.

17.

Kezios

Jawadekar

, et al. Associations of food insecurity and memory function among middle to older-aged adults in the health and retirement study. JAMA Netw Open 2023; 6: e2321474.

18.

Pourmotabbed

Moradi

Babaei

, et al. Food insecurity and mental health: a systematic review and meta-analysis. Public Health Nutr 2020; 23: 1778–1790.

19.

Marin

Lord

Andrews

, et al. Chronic stress, cognitive functioning and mental health. Neurobiol Learn Mem 2011; 96: 583–595.

20.

Rock

Roiser

Riedel

, et al. Cognitive impairment in depression: a systematic review and meta-analysis. Psychol Med 2014; 44: 2029–2040.

21.

Shimada

Park

Makizako

, et al. Depressive symptoms and cognitive performance in older adults. J Psychiatr Res 2014; 57: 149–156.

22.

Leung

Epel

Ritchie

, et al. Food insecurity is inversely associated with diet quality of lower-income adults. J Acad Nutr Diet 2014; 114: 1943–1953.e2.

23.

French

Tangney

Crane

, et al. Nutrition quality of food purchases varies by household income: the SHOPPER study. BMC Public Health 2019; 19: 231.

24.

Ritchie

Yaffe

, et al.

Is late-onset Alzheimer’s disease really a disease of midlife?

Alzheimers Dement Transl Res Clin Interv 2015; 1: 122–130.

25.

National Institute on Aging. What are the signs of Alzheimer’s disease? https://www.nia.nih.gov/health/alzheimers-symptoms-and-diagnosis/what-are-signs-alzheimers-disease (2022, accessed 25 March 2025).

26.

Novotný

Gonzalez-Rivas

Medina-Inojosa

, et al. Investigating cognition in midlife. Alzheimers Dement (N Y) 2021; 7: e12234.

27.

Weden

Miles

JNV

Friedman

, et al. The hispanic paradox: race/ethnicity and nativity, immigrant enclave residence and cognitive impairment among older US adults. J Am Geriatr Soc 2017; 65: 1085–1091.

28.

Clark

DeCarli

Mungas

, et al. Earlier onset of Alzheimer disease symptoms in latino individuals compared with Anglo individuals. Arch Neurol 2005; 62: 774–778.

29.

Fitten

Ortiz

Fairbanks

, et al. Younger age of dementia diagnosis in a hispanic population in southern California. Int J Geriatr Psychiatry 2014; 29: 586–593.

30.

Garcia

Downer

Chiu

, et al. Racial/ethnic and nativity differences in cognitive life expectancies among older adults in the United States. Gerontologist 2019; 59: 281–289.

31.

Kezios

Yaffe

, et al. Depressive symptoms mediate the relationship between sustained food insecurity and cognition: a causal mediation analysis. Ann Epidemiol 2023; 81: 6–13.e1.

32.

Saenz

Avila

. Late-life food insecurity and cognition: exploring timing, duration, and mechanisms among older Mexican adults. BMC Geriatr 2023; 23: 788.

33.

Eskenazi

Fahey

Kogut

, et al. Association of perceived immigration policy vulnerability with mental and physical health among US-born latino adolescents in California. JAMA Pediatr 2019; 173: 744–753.

34.

Eskenazi

Harley

Bradman

, et al. Association of in utero organophosphate pesticide exposure and fetal growth and length of gestation in an agricultural population. Environ Health Perspect 2004; 112: 1116–1124.

35.

Torres

Calderon

Kogut

, et al. Lifecourse socioeconomic status and cognitive performance among midlife Latina women in a California agricultural region. Am J Epidemiol. Epub ahead of print 24 Feb 2025: kwaf036. doi: 10.1093/aje/kwaf036.

36.

Economic Research Service. Food security in the U.S. – survey tools. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/survey-tools#household (accessed 3 March 2025).

37.

Economic Research Service. Definitions of food security. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/definitions-of-food-security (accessed 19 March 2025).

38.

González

Tarraf

Fornage

, et al. A research framework for cognitive aging and Alzheimer’s disease among diverse US latinos: SOL-INCA. Alzheimers Dement 2019; 15: 1624–1632.

39.

González

Mungas

Reed

, et al. A new verbal learning and memory test for English- and Spanish-speaking older Mexican-American adults. J Int Neuropsychol Soc 2001; 7: 544–555.

40.

González

Mungas

Haan

. A verbal learning and memory test for English- and Spanish-speaking older Mexican-American adults. Clin Neuropsychol 2002; 16: 439–451.

41.

Wechsler

. The measurement of adult intelligence. J Nerv Ment Dis 1940; 91: 548.

42.

Adjutant General’s Office. Army individual test battery. 2014. doi:10.1037/t31500-000.

43.

Fortugno

. Wechsler, David. In: Rieber

(ed.) Encyclopedia of the history of psychological theories. New York, NY, USA: Springer, 2012, pp.1183–1184.

44.

Sharareh

Seligman

Adesoba

, et al. Food insecurity disparities among immigrants in the US. AJPM Focus 2023; 2: 100113.

45.

Light

. Gender differences in the marriage and cohabitation income premium. Demography 2004; 41: 263–284.

46.

Liu

Zhang

Burgard

, et al. Marital status and cognitive impairment in the United States: evidence from the national health and aging trends study. Ann Epidemiol 2019; 38: 28–34.e2.

47.

Bateson

Chevallier

Johnson

, et al. Does food insecurity cause anxiety and depression? Evidence from the changing cost of living study. PLOS Ment Health 2025; 2: e0000320.

48.

Brunet

Machín

Fajardo

, et al. Coping strategies of food insecure households with children and adolescents in Uruguay, a high-income Latin American country: a qualitative study through the lens of Bourdieu’s theories of capitals and practice. Appetite 2024; 198: 107383.

49.

Kim-Mozeleski

. Food insecurity trajectories in the US during the first year of the COVID-19 pandemic. Prev Chronic Dis 2023; 20: 220212.

50.

Bakal

Ambriz

Ortiz-Pivaral

, et al. Impacts of COVID-19 shelter in place across key life domains among immigrant farmworker Latina mothers and young adults. BMC Public Health 2024; 24: 2036.

51.

Nikoonia

Khosravi

Islam

SMS

, et al.

Is household food insecurity related to mothers’ stress, anxiety and depression in Iran?

Prev Med Rep 2023; 35: 102293.

52.

Shansky

Lipps

. Stress-induced cognitive dysfunction: hormone-neurotransmitter interactions in the prefrontal cortex. Front Hum Neurosci 2013; 7: 123.

53.

Lupien

Maheu

, et al. The effects of stress and stress hormones on human cognition: implications for the field of brain and cognition. Brain Cogn 2007; 65: 209–237.

54.

McMichael

McGuinness

Lee

, et al. Food insecurity and brain health: systematic review. Crit Rev Food Sci Nutr 2022; 62: 8728–8743.

55.

Arenas

Thomas

Wang

, et al. A systematic review and meta-analysis of depression, anxiety, and sleep disorders in US adults with food insecurity. J Gen Intern Med 2019; 34: 2874–2882.

56.

Lugtenburg

Oude Voshaar

Van Zelst

, et al. The relationship between depression and executive function and the impact of vascular disease burden in younger and older adults. Age Aging 2017; 46: 697–701.

57.

Zainal

Newman

. Inflammation mediates depression and generalized anxiety symptoms predicting executive function impairment after 18 years. J Affect Disord 2022; 296: 465–475.

58.

Economic Research Service. Can low-income Americans afford a healthy diet? https://www.ers.usda.gov/amber-waves/2008/november/can-low-income-americans-afford-a-healthy-diet (accessed 12 June 2025).

59.

Flanagan

Lamport

Brennan

, et al. Nutrition and aging brain: moving towards clinical applications. Ageing Res Rev 2020; 62: 101079.

60.

Zuo

Heflin

. Cognitive impairment and supplemental nutrition assistance program take-up among the eligible older Americans. J Gerontol B Psychol Sci Soc Sci 2022; 78: 99–110.

61.

Ding

Zhu

Xiong

, et al. The association of age at diagnosis of hypertension with cognitive decline: the China health and retirement longitudinal study (CHARLS). J Gen Intern Med 2023; 38: 1431–1438.

62.

Lespinasse

Chêne

Mangin

, et al. Associations among hypertension, dementia biomarkers, and cognition: the MEMENTO cohort. Alzheimers Dement 2023; 19: 2332–2342.

63.

Black

Kraemer

Shah

, et al. Diabetes, depression, and cognition: a recursive cycle of cognitive dysfunction and glycemic dysregulation. Curr Diab Rep 2018; 18: 118.

64.

Moheet

Mangia

Seaquist

. Impact of diabetes on cognitive function and brain structure. Ann N Y Acad Sci 2015; 1353: 60–71.

65.

Damanik

Yunir

. Type 2 diabetes mellitus and cognitive impairment. Acta Med Indones 2021; 53: 213–220.

66.

Tanaka

Gourley

Dekhtyar

, et al. Cognition, brain structure, and brain function in individuals with obesity and related disorders. Curr Obes Rep 2020; 9: 544–549.

67.

Bischof

Park

. Obesity and aging: consequences for cognition, brain structure, and brain function. Psychosom Med 2015; 77: 697–709.

68.

Killgore

WDS

. Effects of sleep deprivation on cognition. Prog Brain Res 2010; 185: 105–129.

69.

Porter

Buxton

Avidan

. Sleep, cognition, and dementia. Curr Psychiatry Rep 2015; 17: 97.

70.

Scullin

Bliwise

. Sleep, cognition, and normal aging: integrating a half century of multidisciplinary research. Perspect Psychol Sci 2015; 10: 97–137.

71.

Miller

Spencer

. Obesity and neuroinflammation: a pathway to cognitive impairment. Brain Behav Immun 2014; 42: 10–21.

72.

Sartori

Vance

Slater

, et al. The impact of inflammation on cognitive function in older adults: implications for health care practice and research. J Neurosci Nurs 2012; 44: 206–217.

73.

Dou

Brown

, et al. Food insufficiency, supplemental nutrition assistance program (SNAP) status, and 9-year trajectory of cognitive function in older adults: the longitudinal national health and aging trends study, 2012–2020. J Nutr 2023; 153: 312–321.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

4.13 MB