Fast-food for thought: Retail food environments as resources for cognitive health and wellbeing among aging Americans?

Abstract

In this exploratory sequential mixed-methods study, interviews with 125 adults aged 55–92 (mean age 71) living in the Minneapolis (Minnesota) metropolitan area suggested that eateries, including coffee shops and fast-food restaurants, represent popular neighborhood destinations for older adults and sources of wellbeing. Thematic analysis of how older adults perceived and utilized local eateries included sites of familiarity and comfort; physical and economic accessibility; sociability with friends, family, staff, and customers; and entertainment (e. g., destinations for outings and walks, free newspapers to read). To test the hypothesis that these sites, and the benefits they confer, are associated with cognitive welfare, we analyzed data from urban and suburban community-dwelling participants in the Reasons for Geographic And Racial Differences in Stroke (REGARDS) study, a national racially diverse sample of older Americans followed since 2003 (n = 16,404, average age at assessment 72 years). Results from multilevel linear regression models of these data demonstrated a positive association between kernel density of local eateries and cognitive functioning, which corroborated qualitative findings. Taken together, these results complicate our understanding of casual eatery settings as possible sites of wellbeing through social interaction and leisure activities. Results prompt further research investigating whether and how retail food environments can serve as community spaces for older adults that may help buffer against cognitive decline.

For the first time in history, Americans are spending more money dining out than in grocery stores (Elitzak and Okrent, 2018). Fast-food constitutes a large and growing market (Thompson, 2017), including limited-service restaurants where food can be eaten on-site, taken away, or delivered, and establishments with on-premise brewing and baking (e.g., coffee, donuts, ice cream, and bagels). On any given day in the United States (US), an estimated 36.6% or approximately 84.8 million adults consume fast-food (Fryar et al., 2018). While fast-food consumption tends to decline with age (Fryar et al., 2018), coffee drinkers tend to be older: 74% of adults aged 55 and older reported consuming coffee daily (Gallup, 2016). Coffee is consumed both in-home and in retail food establishments such as cafes, restaurants, and coffee bars (Statista, 2019). Fast-food represents a staple in many people’s lives.

The United States Department of Agriculture (Saksena et al., 2018) reports that frequent consumers of fast-food tend to live close to these sites. Both density and distance to nearest fast-food restaurant have been associated with fast-food consumption. Most research to-date compares rising rates of obesity to the ease and abundance of fast-food establishments available in many communities throughout the US. Findings are mixed regarding associations between availability of fast-food, local obesity rates, and consumption of healthy foods (e.g., Li et al., 2009; Pruchno et al., 2014; Oexle et al., 2015; Jiao et al., 2015). There is sufficiently strong evidence that a healthy diet and weight can reduce risk for cognitive decline and dementia. High saturated fat and cholesterol intake (frequently associated with fast-food consumption) are linked to increased risk for dementia and Alzheimer’s disease (Solfrizzi et al., 2017; Shakersain et al., 2016). The management of cardiovascular risk factors – including diabetes, obesity, and hypertension – reduces risk of cognitive decline and may reduce risk for dementia (Baumgart et al., 2015). Aside from short-term effects of heightened alertness and improved cognitive performance, there is some evidence that long-term caffeine consumption may be protective against late-life cognitive impairment, decline, dementia, and Alzheimer’s disease (Liu et al., 2016; Panza et al., 2015).

Given the geographical underpinning of our work focused on the spaces and activities of local eateries (as opposed to nutritional value or lack thereof of items served and consumed), we are interested in the role that these environments play in the everyday lives and wellbeing of aging individuals. Walk into a McDonald’s or Tim Horton’s mid- morning and you will likely find a group of retirees clustered around tables drinking coffee and talking. McDonald’s locations can represent centers of community, particularly among lower-income and older individuals (Arnade, 2016). Qualitative studies report on naturally occurring groups of older adults who regularly frequent fast-food restaurants and coffee shops to socialize and get out of the home (Finlay et al., 2018). Cheang (2002) witnesses how a fast-food restaurant provided structure, meaning, and opportunities for older Hawaiians to engage in personal expression, laugh, play, and be among friends. Broughton et al. (2016) observe the importance of a coffee group hosted regularly at a McDonald’s for aging men to express their feelings in a supportive environment. The group promoted social engagement and connectedness, and generated a sense of purpose, structure, and belonging in addition to social interaction. Coffeehouses regularly frequented by older customers can generate a sense of at-homeness, restorative stimuli, companionship, and emotional support (Rosenbaum et al., 2007). Local eateries can serve as key sites of entertainment, activity, and information sharing (Torres, 2018). These ‘third places’ may offer one of few opportunities for social contact and support outside of more formal and age-graded settings such as a doctor’s office, church, or senior center (Finlay et al., 2019). With widespread desires among older adults to age in place in familiar homes and communities (Kan et al., 2020), local retail food environments may be essential elements of neighborhood context. Fast-food outlets might be reconsidered as age-friendly infrastructure: personally-meaningful, socially engaged, affordable, and accessible locations to age well in place (Scharlach, 2017; Finlay et al., 2019).

In this mixed-methods study, thematic analysis of interviews and ethnographic fieldwork across the Minneapolis (Minnesota) metropolitan area explored how older adults perceived and utilized their local retail food environment. Noted benefits, including sociability and entertainment, prompted new questioning into how eateries might impact the cognitive health of aging Americans. Prior research finds that higher social participation and perceived support is associated with better cognitive function in older adults, while social isolation tends to be associated with poorer cognitive outcomes (Krueger et al., 2009; Evans et al., 2018). Further, eateries may also benefit cognitive health through mentally-stimulating activities centered in these sites, including conversing with staff and other customers, reading the newspaper, and working on crosswords and logic/math puzzles supplied at many venues. The qualitative results also suggested that local eateries were frequently enfolded into everyday walking routines and represented enjoyable destinations outside the home. These qualitative findings led us to an exploratory quantitative examination of whether access to eateries was associated with cognitive functioning in a large, national sample of aging Americans in the REasons for Geographic And Racial Differences in Stroke (REGARDS) study.

This mixed-methods study is the first to examine potential links between access to eateries and cognitive function by integrating evidence from qualitative interviews and ethnographic fieldwork with quantitative survey data. We tested whether residential proximity and density impacts trajectories of cognitive decline in this national sample of older Americans followed since 2003. Results from multilevel linear regression models demonstrated a positive association between kernel density of eateries and cognitive functioning. The study contributes new evidence towards an emerging ecological model of cognitive health (Cerin, 2019).

1. Exploratory sequential mixed-methods design

This study occurred in two distinct phases (Fig. 1). Qualitative analysis inspired the hypothesis that we tested in a large population- based sample (Creswell et al., 2011).

Fig. 1.

Exploratory sequential mixed-methods design.

1.1. Aging in the right place qualitative study

This research aimed to examine how older adults perceived and navigated aging within everyday contexts and pursued wellbeing through aging in place. Situated in the Minneapolis (MN) metropolitan area, the purposive design of three case study areas selected for socio- demographic and geographic variability. Potential participants volunteered in response to project flyers and advertisements placed in senior centers, gyms, coffee shops, sites of worship, residential buildings, civic group newsletters, and health fairs. Eligibility criteria included self-identifying as an older person and at least 55 years old, not institutionalized in a care setting (e.g., nursing home or assisted living), residence in a case study area, and demonstrated cognitive capacity to participate. Semi-structured interviews conducted with 125 older adults from June to October 2015 probed for daily routines, social interactions, housing quality, service provision, and perceived wellbeing (details available in Finlay and Bowman, 2017; Finlay, 2017; Finlay and Kobayashi, 2018; Finlay et al., 2018; Finlay et al., 2019). These interviews were audio-recorded and ranged in duration from 30 to 90 min. Ninety-six participants engaged immediately afterwards in a mobile interview (Finlay and Bowman, 2017) where participants determined the route, speed, and mode of travel. Most participants walked or used a wheelchair or motorized scooter for these go-along sessions (Carpiano, 2009), which were on average 17 min in duration and 0.86 km in length.

In order to better understand participant experiences and perspectives through broader spatial and temporal scales, a subset of six participants engaged in ethnographic fieldwork over 12 months (September 2015 to August 2016). Participants were purposefully selected to represent a variety of socioeconomic backgrounds, health statuses, and locations. The ethnographic approach captured individual lived experiences nested within broader societal and cultural contexts through direct observation and unstructured interviewing (Marshall and Rossmann, 2016). As discussed in-detail in Finlay, 2020, the lead author (JF) typically began with extended tours of participants’ living spaces, sharing food and drink, and looking at pictures of family members. Subsequent interactions ranged to routine places outside the home including grocery stores, senior centers, coffee shops, fast-food restaurants, faith services, medical offices, pools, and parks. JF visited participants at least once a month (some much more frequently), and they conversed often between sessions by phone and email. JF employed unstructured interview techniques within conversations to seek personal accounts and perceptions as determined by participants in their own words (Dunn, 2005). JF wrote notes in a small notebook while talking and took photos during some activities together. Following all interactions (including impromptu calls), JF recorded reflections in a digital journal. This included detailed recalls of events as well as hunches and evolving understandings (Crang and Cook, 1995).

1.2. Analysis

Audio files were transcribed verbatim by a professional service. In order to enhance methodological rigor, JF cross-checked transcripts against original audio files for quality and completeness. All transcripts and field notes were organized using the software package NVivo 11. The guiding research question for the secondary qualitative analysis presented in this paper was: “How did study participants perceive and utilize local eateries?” We define local eateries for this study as quick- serve food and drink outlets such as McDonald’s, Starbucks, Dunkin’ Donuts, Burger King, Wendy’s, Caribou Coffee, Dunn Brothers, and Tim Horton’s. In these sites, customers pay before receiving food and drinks and can eat/drink either on- or off-site. JF analyzed the qualitative data according to six steps of thematic analysis (Braun and Clarke, 2006): (1) familiarization, (2) generation of initial codes, (3) search for themes, (4) review themes, (5) define and name themes, and (6) write up of themes analyzed. During the process, regular debriefing with co-authors and audit trails (clear pathways detailing how data were collected and managed) enhanced transparency and credibility (Marshall and Rossmann, 2016). Through this iterative process, we identified themes, linkages, and explanations.

1.3. Quantitative data source

REGARDS is an ongoing, national prospective cohort study examining regional and racial differences in stroke and cognitive function. Using mail and telephone contact methods, community-dwelling adults aged 45 years or older were recruited from January 2003 to October 2007. The cohort includes 30,239 Black and White individuals with a mean age of 64 years at baseline (details available in Howard et al., 2005).

At baseline, a telephone interview collected information on self- reported socio-demographic, behavioral and lifestyle information, and medical history. A cognitive battery was first implemented in 2006 and conducted during follow-up telephone calls at 2-year intervals. Residential address was documented over the follow-up period and geocoded to latitude and longitude coordinates by study investigators at the University of Alabama at Birmingham. The study procedures are reviewed and approved annually by the University of Alabama at Birmingham and all participants provided written informed consent.

1.4. Measures

Cognitive Function.

Measures of verbal learning, memory, and executive function were administered biannually using the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) Word List Learning (WLL) and Word List Delayed Recall (WLD), as well as the Animal Fluency Test (AFT) and Letter Fluency Test ([LF] Moms et al., 1989; Morris et al., 1989). The WLL measures verbal learning (score range, 0–30) and the WLD measures verbal memory (score range, 0–10). The AFT and LF assess language and executive function: complex cognitive processing used in problem solving or complex action sequences. Scores are based on the number of unique animals (AFT) and unique words beginning with “F” (LF) named in 1 min. These cognitive measures are consistent with the Vascular Cognitive Impairment Harmonization Standards (Hachinski et al., 2006) and have been validated for Black and White individuals (Lucas et al., 2005). In addition, we used recall and orientation items (score range, 0–30) from the Montreal Cognitive Assessment (MoCA), a screening tool for mild cognitive impairment (Nasreddine et al., 2005).

Because we had no a priori hypothesis about the specific domains of cognitive functioning that would be associated with retail food environments, we created a composite index of cognitive function (Zhu et al., 2015; Kobayashi et al., 2015; Weuve et al., 2015) based on the mean of standardized scores from the 5 cognitive tests at each follow-up. The index was calculated for each follow-up assessment regardless of the number of tests that were administered (range 1–5) in order to include the full range of participants who completed a varying number of tests and to minimize bias due to missingness. The composite cognitive index was highly correlated with scores from each separate test (Pearson correlation coefficients ranged from 0.67 for LF to 0.84 for WLD).

Retail Food Environment.

The National Establishment Time-Series (NETS) database (Walls, 2007) provides annual records of the US economy with detailed business microdata for more than 60 million private for-profit and nonprofit establishments, in addition to government agencies, from 1990 onwards. In order to study the retail fast-food environment of REGARDS participants, we selected establishments open from 2006 to 2015 belonging to one of two categories in the North American Industry Classification System (NAICS), the standard used by Federal statistical agencies to classify American business establishments (U.S. Census Bureau, 2019). First, we included limited-service restaurants (NAICS 722513) where patrons select items and pay before eating, such as fast-food restaurants, takeout sandwich shops, and limited-service pizza parlors. Second, we included snack and non-alcoholic beverage establishments (NAICS 722515) with on-premise brewing and baking that serve items such as coffee, bagels, pretzels, cookies, doughnuts, and ice cream.

For all REGARDS participants, we created a measure to describe their annual retail food environment through a kernel density (KD) method (Guagliardo, 2004). Measures were annual to reflect the potentially-shifting food environments of REGARDS participants over the 10-year study period. To calculate the kernel density, we fit a smoothly curved surface over each eatery location. The surface value was highest at the eatery’s exact location, and diminished in value over a 2-mile circular buffer (reaching zero at the edge of the radius). We summed overlapping kernel values at each REGARDS participant’s home location. A high kernel value represented multiple eatery locations in close proximity, while a kernel value of zero reflected the absence of such businesses within the participant’s surrounding area.

Covariates.

Additional covariates included individual demographic characteristics known to be associated with cognitive function, including age at baseline, gender (0 = male, 1 = female), race (0 = White, 1 = Black), highest level of attained education (0 = less than high school or high school diploma, 1 = some college or college degree or higher), marital status (0 = single, divorced, widowed, 1 = married), urbanicity (0 = other urban, 1 = urban core); neighborhood of residence (census tract that respondent resided in at the time of their interview) and years of follow- up since baseline observation (from 1 to 12 years of follow-up). Additional census tract-level covariates derived from the 2008–2012 American Community Survey (U.S. Census Bureau, 2013) included population density (100,000 population per square mile); proportion of population living below the poverty line; proportion of residents who were Non-Hispanic Black; and proportion of housing units that were owner occupied.

1.5. Analytical sample

Our analytic sample was composed of REGARDS participants active in the study between 2006 and 2015. Respondents varied in when they contributed their first cognitive test score within this period, with the majority (86%) contributing their first score between 2006 and 2008. Most respondents were tested 4–5 times over this interval, approximately every two years. Individuals with at least one cognitive test score and a valid kernel density score were included in the analyses. Further, we restricted the sample to individuals living in urban areas (identified as “urban core” or “other urban” by Rural-Urban Commuting Area Codes [United States Department of Agriculture, 2019]) and between the ages of 55 and 92 to match participants in the qualitative study. Our final analytic sample included 16,404 individuals with 54,377 observations.

1.6. Analysis

We used Bayesian multilevel linear regression models to describe between-person differences in cognition as a function of the retail food environment. Specifically, we regressed individual cognitive function scores on our kernel density measure of the retail food environment—with the individual sociodemographic and tract-level predictors described above as covariates. Given that respondents contributed multiple observations to the data, we included years since baseline observation as a fixed term and person-specific random intercepts as additional model parameters. Similarly, given that observations were collected across multiple calendar years and clustered within tracts, we included year-specific and neighborhood-specific random intercepts in the models. Model priors were set to be generic and weakly informative (i.e., priors over slope and intercept parameters were set as normal distributions with mean 0 and standard deviation 1; while priors for random components were half-Cauchys with scale 1 and location 0), as we had little prior substantive information to incorporate into the analysis. Note that models allowing for temporal autocorrelation produced similar substantive results to what is presented below.

In separate models, we specified cognitive function as sharing a: (1) linear association with eatery kernel density; and (2) a non-linear relationship with eatery kernel density via thin plate regression splines (Wood, 2017). Median point estimates and uncertainty intervals, derived from model posterior distributions, were used to summarize results. Models were fit using brms (Bürkner, 2017) in the R statistical programming language (R Core Team, 2019).

2. Results

2.1. Qualitative

Characteristics of the qualitative sample are shown in Table 1. The average age at assessment was 71.3 years (SD = 7.8), and 57% were White. Two thirds of the sample were female, about one-third were married, and 43% had at least some college education.

Table 1.

Descriptive statistics of qualitative sample (n = 125): Aging in the right place study (2015–2016). Note: “Other” self –identified races/ethnicities include (in alphabetical order) African, American Indian, Arabic, Asian, Bohemian, French, German, Hispanic/Latin American, Irish, Jewish, Norwegian, Polish, Swedish.

Measure	Mean or %	Std. deviation
Age (years)	71.3	7.8
Female	67%	–
Race/ethnicity: White	57%	–
Race/ethnicity: Black	25%	–
Race/ethnicity: Other	18%	–
Married	34%	–
Education: high school or less	57%	–

Thematic analysis identified usage and perceived benefits of eatery settings for older adults through four main themes: (1) familiarity and comfort; (2) physical and economic accessibility; (3) sociability with friends, family, staff, and customers; and (4) entertainment (e.g., destination for outings and walks, free newspapers and crosswords).

Comfort.

Participants discussed eateries as low-pressure spaces with familiar layouts and routines. They appreciated the physical comfort of these sites, such as wide booths, sizable tables, and ample seating to sit undisturbed alone or in a group. For these reasons, Thomas (67y)¹ regularly frequented two fast-food restaurants for lunch “with the guys.” When JF and Thomas frequented one such place together for lunch during an ethnographic session, Thomas explained that this place was his group’s “office” where they sat at a corner booth for meetings to plan and construct a local veteran’s memorial. As recorded in JF’s fieldnotes: “Thomas was clearly comfortable in and familiar with this [fast-food restaurant]. He knew exactly where to park, and spoke with ease to the staff when ordering lunch.” Eateries were perceived as generally welcoming and relaxed environments available to diverse ages, racial and ethnic groups, and income levels.

Accessibility.

The second main theme of these spaces was the low- cost food and drinks. For numerous low-income participants, especially those in subsidized housing, these sites were reported as the only places they could afford to frequent. Higher-end restaurants were out of their budgets for social gatherings, as witnessed with Denise (72y) when she declined several lunch invitations during ethnographic sessions to eat out at restaurants. Denise explained that “it was an expensive month” with a wedding and two unexpected funerals, and these events “ate up [her] budget.” Though she could not afford restaurant meals regularly, Denise still enjoyed inexpensive coffee with her girlfriends as a valued opportunity to socialize. As recorded in JF’s fieldnotes: “Denise loves the garden and old building at this [coffee shop]. She usually goes there once a month with her friend as a treat. They both love the garden and walk around.” With most restaurants out of her budget, Denise treasured outings to this affordable place.

Denise could not drive following a stroke and had limited transportation options on publicly subsidized paratransit. She particularly appreciated the coffee shop’s nearby location and accessible layout. Eateries were viewed as disability- and mobility-friendly sites. Participants – especially those in wheelchairs and using walkers – noted the large parking spaces, locations near bus lines and major transit routes, clean and spacious bathrooms, and predictable restaurant layouts. Audibility and quietness were also valued, in contrast to dining locations with loud music where people “practically have to shout at each other” (Ellen, 73y).

Sociability.

Participants discussed planned gatherings with friends and family, such as David (75y) who had a regular group of ten to fifteen fellow retirees who met up several times a week for coffee. In addition to his participation in this group over the past ten years, David had “a coffee group that meets five days a week that’s been going on since the late [19]60s.” He also went out for lunch frequently with longtime friends. Participants mentioned taking their grandchildren out to eat, and meeting siblings, children, other family members for easy meals in convenient locations. These sites represented easy, low-pressure ways to interact with friends and make new social connections – particularly for participants such as Martha (75y) who lived alone and sought companionship. Participants like Dennis (64y) who felt lonely appreciated the ambient contact from staff and other customers. With no close children or family, on a typical week Dennis replied that he only sees and talks to “the people in the shops and the Skyway food shops.” During the mobile interview, Dennis enthusiastically waved and said hello to eatery employees in open stores along the Skyway from his wheelchair. For fellow Downtown resident Ellen, the local coffee shop near her home was a social anchor in her life. During frequent visits during ethnographic sessions, the baristas greeted her warmly and started her order without asking. As a widow and transplant to Minneapolis, Ellen expressed loneliness. She strategically sought opportunities for closeness and contact, including her visible enjoyment of lively conversations with the coffee shop baristas. The manager was one of the first people she came to know in the city, and she referred to the shop with affection as it affixed warmth, care, and connection.

Entertainment.

Eating, as explained by Nancy (77y), was described as a major pastime: “Now it seems all you do in retirement is eat. We go to lunch, we go to breakfast, it all centers around eating.” She continued: “The lady next door and another lady go out walking twice a week, and I thought, that’ll be nice. I found out that what they consider a walk is just over to [a coffee shop], and then they sit there for an hour and drink coffee, and walk back.” Multiple participants enjoyed going for coffee and having lunch with friends to “catch up on everybody’s news and gossip” (Susan, 80y). These local sites were staples to everyday routines by providing structure and purpose. They often represented key neighborhood destinations: local places to visit and convenient things to do. Participants mentioned enjoying the free newspapers, crosswords, and number puzzles available.

Urban dwellers often discussed their enjoyment of having a plethora of convenient and affordable dining and drinking options nearby, while suburban participants enfolded these sites into walkability as some of the only places within walking distance. Michelle and Kurt’s (74y and 82y, respectively) daily walk, for example, was up to a local eatery for a cup of coffee and back. Though it was “getting harder for him” with increasing reliance on a cane, Kurt was determined to keep up his daily routine. When asked if the neighborhood offered services they needed, Michelle replied: “Certainly nothing within walking distance except [the coffee shop].” Several participants expressed sadness at the closure of fast-food outlets and required change in routine, particularly in low- income areas where amenities and services were scarce. Brady (60y), for example, was disappointed that a local McDonald’s closed. Raquel (74y) observed that many of her older friends were isolated: “I think it’s because we don’t have places where you can just go sit down, have a cup of coffee, see who comes in, visit with one another.” Some in the suburbs expressed a desire for more services nearby to visit, such as Harry (75y) who stated that it would be nice to be able to “walk to a coffee shop or a bar or something. I sort of miss that. That’s the disadvantage of living in a suburb, you know.” There was one particular suburban chain coffee location that multiple participants mentioned enjoying frequenting because it had expansive surrounding gardens and blooming seasonal flowers. Overall participants frequented multiple eateries for different purposes, such as a nearby coffee shop while walking and fast-food restaurants to meet friends or family for lunch. They discussed eateries as enjoyable destinations and valued routine sites of daily later life.

2.2. Quantitative

Characteristics of the quantitative sample are shown in Table 2. The mean cognitive functioning score for participants was −0.08 (SD = 0.86). The average age at assessment was 71.9 years (SD = 7.9). About 41% of the sample was Black and more than half female. Two-thirds had at least some college education. Participants lived in diverse retail food environments with varying kernel density measures. Respondents were observed across 10,164 unique census tracts.

Table 2.

Descriptive statistics of quantitative sample (n = 16,404): REasons for Geographic And Racial Differences in Stroke (REGARDS) Study (2006–2015).

Measure	Mean or %	Std. deviation
Cognitive function score	−0.08	0.86
Retail food environment kernel density	0.09	0.15
Age at baseline (years)	71.90	7.90
Female	55%	–
Black	41%	–
Married	59%	–
Education: high school or less	34%	–
Years since baseline	5.60	3.00
Lives in urban core	90%	–
Tract population density (100k pop per square mile)	0.25	0.08
Tract proportion black	0.42	0.36
Tract proportion owner occupied homes	0.63	0.20
Tract proportion of population living below poverty line	0.19	0.14

Table 3 displays parameter estimates for our models of cognitive functioning. In the first model, a positive, linear association among eatery kernel density and cognitive function was observed. Respondents residing in the most sparse retail food environments (i.e., areas with kernel density scores of 0) displayed estimated cognitive scores that were approximately 0.10-points lower than respondents living in the highest density environments (i.e., areas with kernel density scores of 1, which approaches the maximum of observed kernel densities in the sample). To contextualize the size of this association, the estimated difference in cognition between individuals in low versus high kernel density spaces approximates our estimated marital status effect (married compared to not married = a +0.07-point difference), or a two-year difference in age between individuals at baseline. The 95% credible interval around this estimate (95% CI: 0.03, 0.19) bounds away from 0, indicating significance.

Table 3.

Multilevel linear regression models of cognitive functioning score. Note: All values rounded to the nearest hundredth. The parameters SD(person-specific intercepts), SD(tract-specific intercepts) and SD(year-specific intercepts) give the estimated standard deviation of the person-specific, tract-specific, and year- specific “random” intercepts, respectively. The term SD (density spline terms) summarizes the standard deviation of the terms forming the density spline.

Parameter	Model 1 (linear specification)			Model 2 (smooth specification)
	Estimate	Sth. error	95% CI	Estimate	Sth. error	95% CI
Intercept	0.22	0.05	(0.13, 0.31)	0.22	0.05	(0.13, 0.32)
Retail food kernel density	0.10	0.04	(0.03, 0.19)	0.44	0.76	(−1.15, 1.88)
Age at baseline (centered, 65)	− 0.04	0.00	(−0.04, − 0.04)	− 0.04	0.00	(−0.04, − 0.04)
Female	0.23	0.01	(0.21, 0.25)	0.23	0.01	(0.21, 0.25)
Black	− 0.35	0.01	(−0.37, − 0.32)	− 0.35	0.01	(−0.37, −0.32)
High school degree or less	− 0.37	0.01	(−0.39, − 0.35)	− 0.37	0.01	(−0.39, − 0.35)
Married	0.07	0.01	(0.04, 0.08)	0.07	0.01	(0.04, 0.09)
Years since baseline	− 0.03	0.00	(−0.04, − 0.03)	− 0.03	0.00	(−0.04, − 0.03)
Lives in urban core	0.08	0.02	(0.05, 0.11)	0.08	0.02	(0.04, 0.11)
Tract population density	0.05	0.07	(−0.10, 0.18)	0.03	0.08	(−0.12, 0.19)
Tract % below poverty line	− 0.39	0.05	(−0.49, − 0.28)	− 0.37	0.05	(−0.48, − 0.27)
Tract % black	0.00	0.02	(−0.04, 0.04)	0.00	0.02	(−0.04, 0.04)
Tract % owner occupied homes	0.01	0.03	(−0.06, 0.07)	0.03	0.03	(−0.04, 0.09)
SD(person− specific intercepts)	0.49	0.00	(0.48, 0.50)	0.49	0.00	(0.48, 0.50)
SD(tract− specific intercepts)	0.12	0.01	(0.10, 0.14)	0.12	0.01	(0.10, 0.14)
SD(year− specific intercepts)	0.06	0.02	(0.03, 0.11)	0.06	0.02	(0.03, 0.10)
SD (density spline	–	–	–	0.76	0.49	(0.10, 1.95)

In the second model, retail food kernel density was allowed to exert a non-linear influence on cognitive function. Parameter estimates from this model are not directly interpretable—though the uncertainty interval around the variance parameter of the smooth terms bounds away from 0, which provided weak evidence that a specification of kernel density beyond a simple linear form was needed.

For a better understanding of the association implied by Model 2, Fig. 2 displays predicted cognitive scores across a range of densities observed in the sample.

Fig. 2.

Predicted cognitive functioning scores across retail food environment kernel density calculated from Model 2. Note: 50%, 75% and 90% credible intervals marked by shaded regions. The retail kernel density range represented here encodes the 0 to 99th percentile of kernel density scores observed in the data.

Fig. 2 demonstrates a positive, non-linear association among the retail food environment and cognitive functioning. Individuals residing in environments with low, near 0, kernel density scores displayed predicted cognitive values of around 0.07, while individuals situated in high kernel density environments (i.e., individuals at 0.65, or the 99th- percentile of observed kernel densities) displayed predicted cognitive scores of approximately 0.16. For comparison, this 0.09-point difference in scores among respondents residing in sparsely and highly populated retail food environments approximates our estimated marital status effect. Note that the positive association among area kernel density and individual-cognition slows at higher kernel density scores, indicating diminishing returns to residing in more heavily populated retail food environments.

3. Discussion

The results prompt new consideration of local retail food environments and contribute evidence towards an emerging ecological model of cognitive health (Cerin, 2019). The merits of mixed-methods studies are increasingly recognized in health research given the combination of rich, subjective insights on complex realities from qualitative investigation with standardized, generalizable data derived through quantitative inquiry (Regnault et al., 2017, p. 2). Integrating different types of data together can generate novel inquiry and more comprehensive understanding of multifaceted and complicated health research questions (Tariq and Woodman, 2013; Creswell et al., 2011). In this study, we used real-life, contextual understandings and behaviors of older adults to explore a novel and unconventional research question in a national sample of aging Americans.

From qualitative fieldwork, fast-food establishments represented familiar, low-pressure, comfortable community spaces where participants gathered with friends and family, and soaked up ambient contact from staff and other customers. They were essential to aging in place for many participants as sites of community and social connectedness (Au et al., 2020). Only one participant expressed disdain for chain fast-food outlets, and one mentioned health concerns about eating in fast-food restaurants after having heart surgery. Overall, when discussing the retail food environment participants focused on the benefits and comfort of predictable restaurant layouts, clean bathrooms, warm heating, low-cost drinks and food, and ample seating for extended stays. Many participants enfolded eateries into walkability and a reason to “get out the door” as nearby and accessible neighborhood destinations. In low-income areas where safe, warm, and welcoming spaces to gather and socialize were limited, the absence of eateries was criticized. While some participants did not feel comfortable or eager to frequent senior centers given disdain that “they’re for old people,” they expressed appreciation for those same benefits (e.g., socialization) in local eateries. Male participants in particular frequently discussed and used eateries for entertainment and socialization. These observations reflect popular media observations (e.g., Arnade, 2016) and qualitative studies reporting on older adults frequenting fast-food restaurants. These sites can enable daily routine, social connectedness, entertainment, information sharing, emotional support, and a sense of at-homeness (Cheang, 2002; Rosenbaum et al., 2007; Broughton et al., 2016; Torres, 2018; Finlay et al., 2018).

Retail food environments are sites to support aging in place beyond traditional foci on age-friendly housing, health and social services, and public infrastructure (Scharlach, 2017; Finlay et al., 2019). Qualities of these places, including social and civic engagement, social support, and informal service access, can prevent unwanted relocations to higher levels of residential care among older adults (Graham et al., 2018). They are mundane settings that can provide essential opportunities for aging individuals, including those with physical and cognitive impairments (Calkins, 2018), to continue to be engaged with their communities and remain part of the fabric of everyday civic life. Eateries may be novel sites to support wellbeing and independence in later life, and help buffer against widely-feared and increasingly-common cognitive decline (Alzheimer’s Association, 2019).

Lifestyle interventions are an increasing research focus to delay and prevent cognitive impairment, Alzheimer’s disease, and dementia. Behaviors include physical activity, sleep, cognitive stimulation, leisure and social activity, and nutrition (Polidori et al., 2010; Kivipelto et al., 2018; Isaacson et al., 2019). As suggested in the qualitative results, local retail food environments may positively impact cognitive-related behaviors in older adults. For some participants, living close to one particular place enabled regular socialization and leisure. Others frequented a multitude of different places for varied purposes, such as meeting friends at a fast-food restaurant or walking to a nearby coffee shop. The kernel density method (Guagliardo, 2004) enabled us to capture retail food environments by both frequency and proximity. In examining the hypothesis that residing in an area dense with eateries is positively associated with cognitive function, we found evidence to further support our idea. The quantitative results in a national sample of urban-dwelling REGARDS participants affirmed the qualitative: residence in closer proximity to a higher number of local eateries was associated with better cognitive function, net of education, age, race, marital status, and indicators of neighborhood structure. Cognitive function appeared to scale positively with eatery kernel density, though the cognitive returns to living in a well-populated retail food environment appeared to diminish after a point. Individuals at the lowest/highest kernel densities had cognitive scores that differed by 0.09 points, about equivalent to a two-year age difference in age among study participants at baseline. Living near supportive neighborhood resources may be important for participants who are attempting to retain cognitive functioning with age. Above the age of 65, a person’s risk of developing Alzheimer’s disease or vascular dementia doubles roughly every 5 years (Alzheimer’s Society, 2016).

There are a number of study limitations to note. For Phase I: Minneapolis is generally-supportive to older adults through investment in services and supports, parks, care provision, and active transit. The qualitative findings may or may not apply in other settings with distinct built, sociocultural, political, and natural environments. Further, the study did not include rural areas or older adults living in long-term care environments. While many participants chose to discuss the local retail environment, the research protocol did not contain direct questioning about eateries, nor explicit investigation of potential behaviors linked to cognitive health. Given these limitations, the qualitative results should be considered as hypothesis-generating to enable further exploration into associations between retail food environments and wellbeing in later life.

In Phase II, a central limitation is our inability to rule out sources of selection bias. Factors like personal wealth, which may drive a respondent’s neighborhood of residence, access to dense environments, and level of cognitive functioning, are unavailable in the REGARDS study and thus unaccounted for in our models. Relatedly, a form of reverse causation may be at play, with cognitively-healthier individuals selecting into neighborhoods with greater kernel density scores before the observation period began. As such, we stress that our results should be interpreted as descriptive associations, rather than as causal processes. We encourage further data collection on the topic to enable researchers to more precisely identify the causal component of substantive results presented here.

We were also unable to determine REGARDS participant usage of local eateries. The kernel density method accounts for the proximity and density of nearby establishments, but nearby access to and options for eateries does not necessarily mean that participants frequented these establishments. Further, we could not directly test the proposed lifestyle pathways (e.g., higher levels of socialization) that may impact cognitive health. We did not account for the dietary habits and nutrition of participants. We did not differentiate between limited-service restaurants and snack/non-alcoholic beverage establishments, nor between types of eateries (e.g., national chain establishments versus small local businesses). Some businesses may not be included in the NETS data, and some geocodes and sub-classifications may be inaccurate (Kaufman et al., 2015). The analyses did not account for neighborhood and county-level characteristics that may interact with the effects of the local retail food environment, such as level of walkability and proximity to other destinations (e.g., groceries, recreational facilities, healthcare). The quality of local eateries may be clustered geographically, such as varied establishment types and quality of service by neighborhood socioeconomic status. Findings may be confounded by proximity to other retail establishments. Greater proportion of land dedicated to retail space has been associated with better global cognition (Besser et al., 2019). The inclusion of only urban-dwelling REGARDS participants and urbanicity as a covariate in the quantitative analyses minimized confounding. The results may vary in rural and non-American contexts with different availability of institutional resources, public goods, built and social infrastructure, and sociocultural norms.

4. Conclusions and next steps

The findings complicate understanding of fast-food settings and motivate further research on the potential role of retail food environments for cognitive decline. Often overlooked in public health, unassuming third places including local coffee shops and fast-food restaurants may critically support health and wellbeing through mechanisms of stimulation, support, protection, and care (Finlay et al., 2019). Instead of focusing on the foods purchased and consumed, we expand attention to the broader environment of eateries and their role in the everyday lives, service provision, and wellbeing of older adults aging in place (Scharlach, 2017; Finlay et al., 2019). A combination of multilevel modeling and detailed qualitative research is necessary to understand and intervene in the relationship between aging individuals and neighborhood effects on health outcomes. Better understanding of these mechanisms can inform policy decisions and community interventions regarding resource allocation and urban development. Fast-food abundance and availability across communities prompts novel consideration of supportive aging programs and services anchored through retail food environments. Investment in neighborhood eateries may be a novel method to support wellbeing in later life and maintain or improve cognitive health.