Skip to main content
NCBI home page
Neurology logoLink to Neurology
. 2023 Feb 7;100(6):e595–e602. doi: 10.1212/WNL.0000000000201499

Association Between Supplemental Nutrition Assistance Program Use and Memory Decline

Findings From the Health and Retirement Study

Peiyi Lu 1,✉, Katrina Kezios 1, Jongseong Lee 1, Sebastian Calonico 1, Christopher Wimer 1, Adina Zeki Al Hazzouri 1
PMCID: PMC9946186  PMID: 36351816

Abstract

Background and Objectives

Studies on the effect of the Supplemental Nutrition Assistance Program (SNAP) on the cognitive health of older adults are scarce. We sought to examine the associations between SNAP use and memory decline among SNAP-eligible US older adults.

Methods

Participants aged 50+ years and SNAP-eligible in 1996 from the Health and Retirement Study were included. Participants' SNAP eligibility was constructed using federal criteria. Participants also self-reported whether they used SNAP. Memory function was assessed biennially from 1996 through 2016 using a composite score. To account for preexisting differences in characteristics between SNAP users and nonusers, we modeled the probability of SNAP use using demographic and health covariates. Using linear mixed-effects models, we then modeled trajectories of memory function for SNAP users and nonusers using inverse probability (IP) weighting and propensity score (PS) matching techniques. In all models, we accounted for study attrition.

Results

Of the 3,555 SNAP-eligible participants, a total of 15.7% were SNAP users. At baseline, SNAP users had lower socioeconomic status and a greater number of chronic conditions than nonusers and were more likely to be lost to follow-up. Our multivariable IP-weighted models suggested that SNAP users had worse memory scores at baseline but slower rates of memory decline compared with nonusers (the annual decline rate is −0.038 standardized units [95% CI = −0.044 to −0.032] for users and −0.046 [95% CI = −0.049 to −0.043] for nonusers). Results were slightly stronger from the PS-matched sample (N = 1,014) (the annual decline rate was −0.046 units [95% CI = −0.050 to −0.042] for users and −0.060 units [95% CI = −0.064 to −0.056] for nonusers). Put in other words, our findings suggested that SNAP users had approximately 2 fewer years of cognitive aging over a 10-year period compared with nonusers.

Discussion

After accounting for preexisting differences between eligible SNAP users and nonusers as well as differential attrition, we find SNAP use to be associated with slower memory function decline.

America's Supplemental Nutrition Assistance Program (SNAP, formerly known as the Food Stamp Program) is one of the largest government programs to help low-income households achieve food security.1 Currently, more than 4.8 million older adult Americans participate in SNAP.2 SNAP households receive monthly financial benefits and can use the electronic benefit transfer card to purchase qualified food that meets their needs.3 By improving nutritional intake, food security has been linked to better brain function among older adults.4,5 Food security also reduces overall financial strain and instability, which in turn have been associated with premature cognitive aging and compromised brain integrity.4,5 Despite the plausible pathways of SNAP use affecting cognition, no prior studies have examined the effect of SNAP use on cognitive decline in older age.

Prior studies on the health effects of SNAP have mostly focused on health outcomes of children and adults. Among children, overall, studies consistently indicate that SNAP use is associated with increased food security and better physical and mental health.6-8 Among adults, the health effects of SNAP have been mixed.9-13 On one hand, some studies have found that SNAP is associated with better self-assessed health, fewer office-based visits and outpatient visits,10 and lower all-cause mortality.11 On the other hand, other studies have found that SNAP is associated with overweight and obesity12 as well as with psychological distress, especially during the initial application and receipt period.13 Meanwhile, investigations of SNAP use on health in older age remain limited, and none explored cognitive function. For example, one study using data from the Health and Retirement Study (HRS) has shown that SNAP participation mitigated the negative effects of food insecurity on physical health.14 In another study, the authors found no significant differences in self-reported health and number of doctor-diagnosed conditions between eligible older adult SNAP users and nonusers.15 Some other studies have found SNAP associated with reduced hospitalization16 and reduced food insecurity-related chronic health problems17 among older adults.

Our study extends prior literature by evaluating the effect of SNAP use on long-term trajectories of memory function, using longitudinal data from the HRS, which consists of a nationally representative sample of older adult Americans. Furthermore, because of the potential differences—for example, socioeconomic and health-related—between SNAP users and nonusers even when restricted to those who are SNAP-eligible,18-20 selection bias21 is often a concern. In this study, in an effort to untangle the impact of SNAP—which is the effect of interest—from the impact of circumstances that made an individual more likely to participate in SNAP, we estimated the probability of someone “participating in SNAP” in the form of a propensity score (PS), which we then used to account for potential bias using inverse probability (IP) treatment weighting22 and propensity score-matching23 techniques.

Methods

Study Population and Analytical Sample

The HRS is a nationally representative population-based cohort of Americans 50 years and older. HRS used area-level probability sampling of US households with oversampling of Black and Hispanic individuals and residents of the State of Florida. Extensive demographic, economic, and health information has been collected through face-to-face and telephone interviews every 2 years since 1992 by the investigators at the University of Michigan. We constructed our data set by merging the raw core HRS data and longitudinal HRS data harmonized by the RAND Corporation.24 Our analytical sample consisted of HRS participants who were eligible for SNAP and enrolled in HRS in 1996 (our study baseline, the first survey year when SNAP eligibility could be calculated). Participants were followed until dropout (which included death) or the 2016 study examination.

Standard Protocol Approvals, Registrations, and Patient Consents

All HRS participants gave verbal informed consent for their participation in the study; HRS data collection was approved by the Health Sciences and Behavioral Sciences institutional review board at the University of Michigan.

Assessment of SNAP Eligibility in 1996

Using prior established work and federal criteria,3,15 we constructed SNAP eligibility for all HRS participants based on participants' household income, assets, and expenditures. Because SNAP eligibility differs depending on household type, we first divided HRS households into 2 types: (1) general households and (2) households with an elderly (aged 60+ years) and/or disabled member (receiving disability benefits from the federal or state government). For general households, SNAP eligibility was based on 3 requirements: (1) gross monthly income ≤130% of the federal poverty level (FPL), (2) net income (i.e., income after deductions) ≤ 100% of FPL, and (3) asset ≤ yearly thresholds ($2,000 in 1996). For households with an elderly or disabled member, SNAP eligibility criteria are less restrictive, with only 2 requirements: (1) net income ≤100% of FPL and (2) assets ≤ yearly thresholds ($2,000 in 1996). It is important to note that there were no state variations in main SNAP eligibility criteria in 1996 because all states adopted federal rules at that time.25 Please see eMethods in the Supplement (links.lww.com/WNL/C455) for further details on how we constructed SNAP eligibility.

Assessment of SNAP Use in 1996

In 1996, HRS respondents were asked whether they or other family members received government food stamps at any time since the previous interview (yes/no). A confirmative answer (“yes”) indicated that the household participated in the SNAP program in the past 2 years (i.e., a SNAP user). Otherwise, the household was considered a SNAP nonuser.

Assessment of Trajectories of Memory Function in 1996–2016

Memory function was assessed biennially from 1996 to 2016 using a previously developed composite score based on the modified HRS items from direct and proxy memory assessments.26 In brief, at each assessment wave, HRS respondents were first asked to complete the tasks of immediate and delayed recall of a 10-word list. For those too impaired to answer for themselves, the proxy (usually the spouse and other family member) rated the respondent's memory on a 5-item Likert scale and assessed the 16-item Informant Questionnaire for Cognitive Decline. By using the full neuropsychological tests and dementia diagnoses from Aging, Demographics, and Memory Study, a subsample of HRS respondents, Wu et al.26 used immediate and delayed word memory assessments and proxy tests, to create a continuous memory function score. This memory score has been shown to reduce the potential attrition bias of HRS data and, thus, improve the estimates of cognitive change rate and risk factor of cognitive aging.26 The memory function score data have been used in many longitudinal studies of cognitive decline.27-31 Memory scores ranged from −2.30 to 2.65 standardized units. A higher score indicated better memory performance.

Assessment of Covariates in 1994

We used covariate information from 1994, which is the survey year before SNAP assessment in 1996. By doing so, we ensured that these covariates/potential confounders were measured before the exposure of interest. Covariates included individuals' age (continuous), sex (female vs male), race(White, Black/African American vs other), educational attainment (≤ high school [HS] vs > HS), marital status (married vs other), highest parental education (≤HS, >HS vs missing), whether born in South (yes vs no), log-transformed household income (continuous), and labor force status (employed, retired/others, vs missing). We also constructed the number of doctor-diagnosed chronic conditions (range 0–8) by including high blood pressure, diabetes, cancer, chronic lung diseases, heart problems, stroke, psychiatric problems, and arthritis.

Statistical Analysis

We first compared participants' characteristics measured in 1994 (i.e., pre-SNAP exposure) according to categories of SNAP use in 1996. Then, to estimate the effect of SNAP use on memory function decline, we used linear mixed-effects models with random intercepts. Years since the beginning of cognitive follow-up (1996) was used as the time scale. The primary coefficients of interest were (1) the main effect for time (i.e., the slope of the memory trajectory for SNAP “nonusers”) and (2) the interaction between time and SNAP use (i.e., the difference in the slopes of the memory trajectories for SNAP “users” vs “nonusers”).

Analyses Using IP-Weighting Approach

To account for preexisting differences between SNAP users and nonusers, we used a predefined set of covariates from 1994 to model the probability (i.e., propensity) of SNAP use in 1996. This process of creating a propensity score (PS) involved running a logistic regression model where the dependent variable was SNAP use in 1996 (yes/no) and predictors included the following 1994 covariates: age, sex, race, education, marital status, highest parental education, birth place, household income, labor force status, and number of chronic conditions. The HRS person-level weights from 1996 were applied to the PS model. We then used the resulting PSs in an IP of treatment weight (IPTW) approach and a PS matching (PS-matched) approach. The IPTW was calculated as the inverse of PS of SNAP use and was later stabilized using established methods.32 Furthermore, to address survey attrition over the study period, we calculated IP attrition weights (IPAWs) using the same methodology listed above for the IPTWs. Attrition was defined as dropping out (including because of death) from the study anytime between 1998 and 2016. In particular, we modeled the probability of survey attrition using a logistic regression model and included the same 1994 covariates. For each participant, the final IP weight was the multiplication product of their IPTW and IPAW. The final weights were applied to the linear mixed-effect model when examining the relationship between SNAP use and trajectories of memory function.

Analyses Using Propensity Score Matching Approach

We also repeated our main analyses for the associations of SNAP use and memory decline using linear mixed-effects models in a PS-matched sample (N = 1,014). SNAP users and nonusers were matched on the PS, and matching was performed using the nearest neighbor method with replacement, setting 1:1 matching and caliper to be 0.005. To account for differential attrition, we generated IPAWs for the PS-matched sample, using the same procedure described above, and applied the IPAW to the linear mixed-effects models.

Translation of Our Estimates

For ease of interpretation, we translated time × SNAP use estimates into terms of “excess/fewer years of cognitive aging over a 10-year period,” following the procedures in prior research by Weuve et al.33 To get a sense of how much years of cognitive aging per 10-year period occurred among the SNAP users compared with nonusers, we computed fewer years of cognitive aging in the following way: We first assumed that the annual rate of change in memory scores among SNAP nonusers (captured by the main effect term for time) represented “cognitive aging among SNAP nonusers”. Then, we divided the differences in the annual memory rate between SNAP users and nonusers (which is the coefficient of the SNAP use × time interaction) by the annual memory decline among SNAP nonusers and then multiplied it by 10. A negative value indicates that SNAP users had fewer years of cognitive aging than SNAP nonusers over a 10-year period. Or put in other words, it means that SNAP nonusers had excess years of cognitive compared with users.

Sensitivity Analyses

Although our predictor of interest was SNAP eligibility in 1996, our outcome period spanned from 1996 to 2016. As such, we conducted sensitivity analyses to check whether SNAP eligibility and SNAP use were stable across the period after 1996. We also examined rates of attrition over time across categories of SNAP use. All analyses were performed in R using the packages “lme4”, “ggplot2,” “MatchIt,” and “cobalt.”

Data Availability

This study used a completely deidentified data set from HRS. HRS data are publicly available. For more information, please refer to hrs.isr.umich.edu/about.

Results

Among the 16,649 respondents who participated in the 1996 HRS survey, 4,085 participants lived in households that were eligible for the SNAP program. Of those, a total of 3,555 participants had at least one memory function score between 1996 and 2016 (our outcome period) as well as information on their demographic and health characteristics at the survey year before their SNAP exposure (i.e., in 1994) and, thus, constituted our final analytical sample.

Our analytic sample in 1996 included a total of 559 eligible SNAP “users” (15.72%) and 2,996 eligible SNAP “nonusers” (84.28%) (Table 1). There were significant differences in the preexisting characteristics between eligible SNAP users and nonusers. Specifically, compared with SNAP nonusers, SNAP users were more likely to be female, Black, and born in the South. SNAP users were less likely to have completed greater than HS education, less likely to be employed, and more likely to have lower household income. Finally, compared with SNAP nonusers, SNAP users were less likely to be married and had a higher median number of chronic conditions.

Table 1.

Distribution of Baseline Sample Characteristicsa Across Categories of SNAP Use in 1996, Health and Retirement Study (N = 3,555)

graphic file with name WNL-2022-201361t1.jpg

Open in a new tab

Figure 1 shows that there was sufficient overlap in the distribution of the PS for SNAP use across SNAP users and nonusers. Furthermore, the covariates/potential confounders were well-balanced in the PS-matched sample (N = 1,014). The absolute standardized mean differences were less than 10% for all covariates in the matched sample (eFigure 1, links.lww.com/WNL/C455). There was also sufficient overlap in the distribution of the PS for attrition across SNAP users and nonusers (eFigure 2).

Figure 1. Distribution of the Propensity Score for SNAP Use Across Categories of SNAP Use in 1996, Health and Retirement Study.

Figure 1

Open in a new tab

The probability of SNAP use is computed based on the logistic regression model adjusting for pre-exposure (i.e., 1994) characteristics (age, sex, race, education, marital status, highest parental education, whether born in the South, log-transformed household income, labor force status, and number of chronic diseases). SNAP = Supplemental Nutrition Assistance Program.

In Table 2, we present the findings from the linear mixed-effects models for the relationship between SNAP use and memory function decline using IP weights and PS-matched approaches. The model results using IP weights suggested that SNAP users had worse memory scores at baseline (standard units, β = −0.134, 95% CI = −0.213 to −0.043), but a slower rate of memory decline than nonusers (slope of time for “nonusers” = −0.046, 95% CI = −0.049 to −0.043; slope of time for “users” = −0.038, 95% CI = −0.044 to −0.032). The model results using the PS-matched sample showed that SNAP users and nonusers had no significant differences in baseline memory function (β = −0.046, 95% CI = −0.164 to 0.061), but SNAP users again had slower rates of memory decline than nonusers (slope of time for “nonusers” = −0.060, 95% CI = −0.064 to −0.056; slope of time for “users” = −0.046, 95% CI = −0.050 to −0.042). Furthermore, our findings suggested that SNAP users experience 1.74 to 2.33 fewer years of cognitive aging over a 10-year period than SNAP nonusers, from the IP weights model and PS-matched model, respectively. Put in other words, SNAP nonusers experience 1.74–2.33 excess years of cognitive aging over a 10-year period than SNAP users, that is, the amount of cognitive aging a SNAP nonuser experiences in 10 years is what a SNAP user experiences in 11.74–12.33 years.

Table 2.

Linear Mixed-Effect Models for the Association Between SNAP Use (1996) and Memory Function Trajectories (1996–2016), Health and Retirement Study

graphic file with name WNL-2022-201361t2.jpg

Open in a new tab

Figure 2 further illustrates the predicted memory function trajectory for SNAP users and nonusers, from the IP weights and PS-matched models. In a sensitivity analysis, we added a quadratic term for time and its interaction with SNAP use in the mixed-effects model. The findings were generally similar to the model with only the linear time term.

Figure 2. Comparison of the Predicted Memory Function Trajectory in 1996–2016 Between SNAP Users and Nonusers in 1996 Based on Linear Mixed-Effect Models, Health and Retirement Study.

Figure 2

Open in a new tab

Associations are based on Table 2 model results. Abbreviation: SNAP = Supplemental Nutrition Assistance Program.

Because we only analyzed SNAP use in 1996, we conducted several sensitivity analyses to check whether SNAP eligibility and SNAP use were stable across the years (i.e., the outcome period). Our findings suggested that respondents included in our sample who were SNAP-eligible in 1996 were still very likely (69%–93%) to be SNAP-eligible in the following years (eTable 1, links.lww.com/WNL/C455). Furthermore, those who were SNAP users in 1996 were likely to continue using SNAP in the following years while those who were SNAP nonusers in 1996 remained unlikely to participate in/use SNAP in the following years (eTable 2). For example, approximately 36%–66% of SNAP users in 1996 continued to use SNAP later on, and 85%–96% of nonusers remained as such in later years.

We further checked whether there was differential attrition across categories of SNAP use. In the Supplements, eTable 3 (links.lww.com/WNL/C455) summarizes that SNAP users had higher attrition rates than SNAP nonusers over time, highlighting the importance of accounting for attrition, as we did, in the form of IPAWs.

Discussion

To the best of our knowledge, this is the first study to examine the effects of SNAP use on long-term trajectories of memory function. Overall, our findings showed that compared with nonusers, SNAP users had slower memory decline. Put in other words, our findings suggested that among SNAP-eligible adults, nonusers experienced 1.74 to 2.33 more (excess) years of cognitive aging over a 10-year period compared with users.

In our study, which only included those eligible for SNAP, we found users had lower socioeconomic status (e.g., lower education and income), had more chronic conditions at baseline, and were more likely to dropout, compared with nonusers. These findings are consistent with previous literature on the differing characteristics between SNAP users and nonusers.18,19 They also validate the concerns about selection bias when evaluating the health effects of SNAP use.15,21 To account for the differences in preexisting characteristics and survey attrition between the users and the nonusers, we used IP weights22,23,33,34 for SNAP use and for attrition in all of our regression models. We also repeated the analyses matching on the propensity of SNAP use (PS-matched technique). The consistency of our findings across the IP weights and PS-matched models is reassuring.

To the best of our knowledge, this analysis presents the first empirical evidence on the long-term influence of SNAP use on older adults memory function. Compared with the substantial evidence on the effects of SNAP on children and adults' health,7,8,11 the evidence on the effects of SNAP on health in older age remains scarce, and none have examined the cognitive function. One prior HRS study indicated SNAP participation alleviated the negative effects of food insecurity on physical health14 while another study did not find significant differences in self-reported health and number of doctor-diagnosed conditions between older adults SNAP users and nonusers.15 More studies are warranted to examine the effect of SNAP on various health dimensions in older age, including other cognitive domains and brain integrity.

There are several hypothesized mechanisms through which SNAP may influence cognition and brain health. SNAP's primary goal is to reduce food insecurity among low-income households and to increase access to higher quantity and quality foods.35 By consuming better-quality diets and improving nutritional intake, SNAP may benefit brain health by reducing oxidative stress and neuroinflammation and subsequently promoting neuronal integrity.4,5 SNAP may also reduce stress36 and overall financial hardship, which have in turn been linked to premature cognitive aging and compromised brain integrity.37,38 SNAP may also increase the purchasing power and investment in other health preserving behaviors, not only resulting in less healthcare expenditures39,40 but also resulting in better access to care,10,16 which may in turn result in better disease management and management of risk factors of cognitive function. For example, prior research has shown that individuals experiencing financial strains may be less likely to visit a doctor or take their medication, consequently resulting in increased risk of brain-related diseases such as stroke.41,42

Examining SNAP use and cognitive health has clear policy implications, especially considering the large number of older adults that are eligible for SNAP. Approximately 4.8 million older Americans (aged 60+ years) participate in SNAP. Yet, this represents less than half of older adults who are actually eligible for SNAP.1 SNAP usage is especially low among older adults—according to the United States Department of Agriculture, in 2017, only 48% of eligible older adults participated in SNAP compared with 84% participation from all-age eligible individuals.2 In parallel, more than 5 million US older adults are currently living with Alzheimer disease and other dementias, and this number is expected to increase to 13 million by 2050.43 As such, the low take-up rate among older adults—whether because of program misinformation,44,45 application difficulty,45 and/or welfare stigma18—is a huge missed opportunity for dementia prevention. We hope our findings open up discussions around ways to increase SNAP program participation in older age.

Our study has some limitations that are worth noting. First, while we restricted our sample to those who are “eligible” for SNAP, this allowed us to make a neater comparison of SNAP users and nonusers who are both eligible. Furthermore, despite this restriction, our study findings are relevant to more than 10 million US older adult population deemed to be eligible for SNAP according to the National Council on Aging.1 In addition, we examined memory decline in relationship to SNAP use at only one point in time. However, our sensitivity analyses suggested that both SNAP eligibility and SNAP use were relatively stable over time. Second, while we addressed differences between users and nonusers in IP weights and PS-matched models, there may be other methodological challenges in SNAP studies such as endogeneity and misreporting issues.12,46 These challenges were beyond the scope of this analysis. Although PS allow us to match SNAP users and nonusers on their probability of using SNAP (treatment of interest), there might be still residual confounding. Third, although in this study we examined whether SNAP use relates to memory decline, future research should explore underlying mechanisms, for example, whether the benefit may be through income benefits and stress reduction, among others. Finally, the process of cognitive decline and dementia pathogenesis may start earlier in the life course than we were able to examine in this study. That is, because HRS is a study focusing on Americans aged 50+ years, it is not possible for us to examine the relationship between SNAP use and cognition earlier in the life course, which might reveal important information about the relevant timing of SNAP use and memory function and help mitigate concerns about potential reverse causation. Thus, future work on SNAP and cognition should identify cohorts in which the relationship can be examined earlier in the life course.

Despite these limitations, this study has some strengths that are worth noting. This study provides novel evidence about the long-term effects of SNAP use on trajectories of memory function in older age. To address methodological challenges associated with bias because of SNAP participation and attrition, we used IP weights and PS-matched methods, and our findings were consistent across the 2 methods. Our findings of a beneficial effect of SNAP use on memory decline represents a huge opportunity for dementia prevention at a population level. Given the low take-up rate of SNAP among eligible older adults, education and outreach programs are needed to disseminate SNAP program benefits while reducing welfare stigma.45 Government agencies can also take steps to simplify administrative procedures that facilitate the application process.47

In sum, our study provides new evidence that SNAP use is associated with slower memory decline in older age. Future studies should replicate our findings and examine the specific mechanisms underlying the relationship between SNAP use and brain function.

Acknowledgment

The HRS (Health and Retirement Study) is conducted by the University of Michigan. HRS data are publicly available. For more information, please refer to HRS webpage: hrs.isr.umich.edu/about.

Glossary

FPL

federal poverty level

HS

high school

HRS

Health and Retirement Study

IP

inverse probability

IPAW

IP attrition weight

IPTW

IP of treatment weight

PS

propensity score

SNAP

Supplemental Nutrition Assistance Program

Appendix. Authors

Appendix.

Open in a new tab

Footnotes

Editorial, page 269

CME Course: NPub.org/cmelist

Study Funding

Authors reported no funding.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.

References

  • 1.McGovern EK. 7 Facts About Older Adults and SNAP; 2021. ncoa.org/article/7-facts-about-older-adults-and-snap. Accessed November 3, 2021. [Google Scholar]
  • 2.Vigil A. Trends in Supplemental Nutrition Assistance Program Participation Rates: Fiscal Year 2010 to Fiscal Year 2017: USDA; 2019. [Google Scholar]
  • 3.Centeron Budgetand Policy Priorities. A Quick Guide to SNAP Eligibility and Benefits; 2021. cbpp.org/research/food-assistance/a-quick-guide-to-snap-eligibility-and-benefits. Accessed August 11, 2021. [Google Scholar]
  • 4.Na M, Dou N, Ji N, et al. Food insecurity and cognitive function in middle to older adulthood: a systematic review. Adv Nutr. 2020;11(3):667-676. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.McMichael AJ, McGuinness B, Lee J, Minh HV, Woodside JV, McEvoy CT. Food insecurity and brain health in adults: a systematic review. Crit Rev Food Sci Nutr. 2021:1-16. [DOI] [PubMed] [Google Scholar]
  • 6.Ettinger de Cuba S, Chilton M, Bovell-Ammon A, et al. Loss of SNAP is associated with food insecurity and poor health in working families with young children. Health Aff. 2019;38(5):765-773. [DOI] [PubMed] [Google Scholar]
  • 7.Gundersen C, Ziliak JP. Food insecurity and health outcomes. Health Aff. 2015;34(11):1830-1839. [DOI] [PubMed] [Google Scholar]
  • 8.Myers CA. Food insecurity and psychological distress: a review of the recent literature. Curr Nutr Rep. 2020;9(2):107-118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Oddo VM, Mabli J. Association of participation in the supplemental nutrition assistance program and psychological distress. Am J Public Health. 2015;105(6):e30–e35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Gregory CA, Deb P. Does SNAP improve your health? Food Policy 2015;50:11-19. [Google Scholar]
  • 11.Heflin CM, Ingram SJ, Ziliak JP. The effect of the supplemental nutrition assistance program on mortality. Health Aff. 2019;38(11):1807-1815. [DOI] [PubMed] [Google Scholar]
  • 12.Almada L, McCarthy I, Tchernis R. What can we learn about the effects of food stamps on obesity in the presence of misreporting? Am J Agric Econ. 2016;98(4):997-1017. [Google Scholar]
  • 13.Heflin CM, Ziliak JP. Food insufficiency, food stamp participation, and mental health. Soc Sci Q. 2008;89(3):706-727. [Google Scholar]
  • 14.Pak TY, Kim G. Food stamps, food insecurity, and health outcomes among elderly Americans. Prev Med. 2020;130:105871. [DOI] [PubMed] [Google Scholar]
  • 15.Kim J. Are older adults who participate in the supplemental nutrition assistance program healthier than eligible nonparticipants: evidence from the health and retirement study. J Soc Welfare 2017;44(3):95. [Google Scholar]
  • 16.Samuel LJ, Szanton SL, Cahill R, et al. Does the Supplemental Nutrition Assistance Program affect hospital utilization among older adults? The case of Maryland. Popul Health Manage. 2017;21(2):88-95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Lee JS, Fischer JG, Johnson MA. Food insecurity, food and nutrition programs, and aging: experiences from Georgia. J Nutr Elder. 2010;29(2):116-149. [DOI] [PubMed] [Google Scholar]
  • 18.Fuller-Thomson E, Redmond M. Falling through the social safety net: food stamp use and nonuse among older impoverished Americans. Gerontologist 2008;48(2):235-244. [DOI] [PubMed] [Google Scholar]
  • 19.Geiger JR, Wilks SE, Livermore MM. Predicting SNAP participation in older adults: do age categorizations matter? Educ Gerontol. 2014;40(12):932-946. [Google Scholar]
  • 20.Haider SJ, Jacknowitz A, Schoeni RF. Food stamps and the elderly: why is participation so low? J Hum Resour. 2003:1080-1111. [Google Scholar]
  • 21.Bitler M. The health and nutrition effects of SNAP: selection into the program and a review of the literature on its effects. In: Bartfeld J, Gundersen C, Smeeding T, Ziliak JP, editors. SNAP Matters: How Food Stamps Affect Health and Well-Being: Stanford University Press; 2015:134-160. [Google Scholar]
  • 22.Weuve J, Proust-Lima C, Power MC, et al. Guidelines for reporting methodological challenges and evaluating potential bias in dementia research. Alzheimes Demen. 2015;11(9):1098-1109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Zeki Al Hazzouri A, Jawadekar N, Grasset L, et al. Statins and cognitive decline in the Cardiovascular Health Study: a comparison of different analytical approaches. J Gerontol Ser A. 2021:1-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Bugliari D, Carroll J, Hayden O, et al. RAND HRS Longitudinal File 2018 (V1) Documentation: RAND Center for the Study of Aging; 2021. [Google Scholar]
  • 25.ERS. SNAP Policy Database, SNAP Policy Data Set; 2019. ers.usda.gov/data-products/snap-policy-data-sets/. Accessed August 11, 2021. [Google Scholar]
  • 26.Wu Q, Tchetgen EJ, Osypuk TL, White K, Mujahid M, Maria Glymour M. Combining direct and proxy assessments to reduce attrition bias in a longitudinal study. Alzheimer Dis Assoc Disord. 2013;27(3):207-212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Mayeda ER, Mobley TM, Weiss RE, Murchland AR, Berkman LF, Sabbath EL. Association of work-family experience with mid- and late-life memory decline in US women. Neurology. 2020;95(23):e3072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Donovan NJ, Wu Q, Rentz DM, Sperling RA, Marshall GA, Glymour MM. Loneliness, depression and cognitive function in older U.S. adults. Int J Geriatr Psychiatry. 2017;32(5):564-573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Whitlock EL, Diaz-Ramirez LG, Glymour MM, Boscardin WJ, Covinsky KE, Smith AK. Association between persistent pain and memory decline and dementia in a longitudinal cohort of elders. JAMA Intern Med. 2017;177(8):1146-1153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Marden JR, Mayeda ER, Walter S, et al. Using an Alzheimer Disease polygenic risk score to predict memory decline in Black and White Americans over 14 years of follow-up. Alzheimer Dis Assoc Disord. 2016;30(3):195-202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kezios K, Zhang A, Kim S, et al. Low hourly wages in middle-age are associated with faster memory decline in older age: evidence from the health and retirement study. Am J Epidemiol. 2022;191(12):2051-2062. doi: 10.1093/aje/kwac166. [DOI] [PubMed] [Google Scholar]
  • 32.Hernán MA, Robins JM. Causal Inference, what if. Boca Raton: Chapman & Hall/CRC; 2020. [Google Scholar]
  • 33.Weuve J, Tchetgen EJ, Glymour MM, et al. Accounting for bias due to selective attrition: the example of smoking and cognitive decline. Epidemiology. 2012;23(1):119-128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zeki Al Hazzouri A, Zhang L, Murchland AR, et al. Quantifying lifecourse drivers of international migration: a cross-national analysis of Mexico and the United States. Epidemiology. 2020;32(1):50-60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Congress. Food and Nutrition Act of 2008 in Washington; 2014:1-157. [Google Scholar]
  • 36.Oddo VM, Mabli J. Association of participation in the supplemental nutrition assistance program and psychological distress. Am J Public Health. 2015;105(6):e30-e35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Grasset L, Glymour MM, Elfassy T, et al. Relation between 20-year income volatility and brain health in midlife: the CARDIA study. Neurology. 2019;93(20):e1890–e1899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Zeki Al Hazzouri A, Elfassy T, Sidney S, Jacobs D, Pérez Stable EJ, Yaffe K. Sustained economic hardship and cognitive function: the coronary artery risk development in young adults study. Am J Prev Med. 2017;52(1):1-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Nicholas LH. Can Food Stamps help to reduce Medicare spending on diabetes?. Econ Hum Biol. 2011;9(1):1-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Berkowitz SA, Seligman HK, Rigdon J, Meigs JB, Basu S. Supplemental Nutrition Assistance Program (SNAP) participation and health care expenditures among low-income adults. JAMA Intern Med. 2017;177(11):1642-1649. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Bartfeld J, Gundersen C, Smeeding T, Ziliak J. SNAP Matters: How Food Stamps Affect Health andWell-Being, 1st ed: Stanford University Press; 2015. [Google Scholar]
  • 42.Basu S. Income volatility: a preventable public health threat. Am J Public Health. 2017;107(12):1898-1899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Association As. Alzheimer's Disease Facts and Figure: Race, Ethnicity and Alzheimer's in America: Alzheimer’s Association; 20212021. [Google Scholar]
  • 44.Wu AY. Why Do So Few Elderly Use Food Stamps? Chicago: The Harris School of Public Policy Studies. The University of Chicago; 2009. [Google Scholar]
  • 45.Pinard CA, Bertmann FMW, Byker Shanks C, et al. What factors influence SNAP participation? Literature reflecting enrollment in food assistance programs from a social and behavioral science perspective. J Hunger Environ Nutr. 2017;12(2):151-168. [Google Scholar]
  • 46.Kreider B, Pepper JV, Gundersen C, Jolliffe D. Identifying the effects of SNAP (food stamps) on child health outcomes when participation is endogenous and misreported. J Am Stat Assoc. 2012;107(499):958-975. [Google Scholar]
  • 47.Keith-Jennings B, Llobrera J, Dean S. Links of the Supplemental Nutrition Assistance Program with food insecurity, poverty, and health: evidence and potential. Am J Public Health 2019;109(12):1636-1640. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

This study used a completely deidentified data set from HRS. HRS data are publicly available. For more information, please refer to hrs.isr.umich.edu/about.

Articles from Neurology are provided here courtesy of American Academy of Neurology

RESOURCES