Original Report: Ethnic Restaurant Nutrition Preventing Cardiovascular and Other Chronic Diseases Environments and Cardiovascular Health: Examining Hispanic Restaurants in New York City

Melissa Fuster, PhD1; Enrique R. Pouget, PhD1; Margaret A. Handley, PhD, MPH2; Krishnendu Ray, PhD3; Brian Elbel, MPH, PhD4; Eddie N. Sakowitz, MS1; Kayla Halvey, MS1; Terry Huang, PhD, MPH, MBA5

Objective: To adapt and apply the Nutri- Introduction tion Environment Measures Survey for Res- from home are an increasingly im- taurants (NEMS-R) to Hispanic Caribbean portant part of consumption (HC) restaurants and examine associations Hispanics have a higher burden and expenditures, accounting for between restaurant characteristics and nutrition environment measures. of cardiovascular disease compared 40% of food spending among US with non-Hispanic Whites,1 with Hispanic households (44% among Methods: We adapted the NEMS-R for HC dietary factors being a leading cause all US homes).5,6 Among Hispan- (Cuban, Puerto Rican, Dominican) 2 and cardiovascular health-promoting fac- of preventable death and disability. ics, the consumption of tors, and applied the instrument (NEMS- The Hispanic Community Health away from home has been associ- HCR) to a random sample of HC restau- Study/Study of Latinos shows that ated with decreased diet quality,7 rants in New York City (NYC) (N=89). Multivariable linear regression was used to most fail to meet evidence-based including increased intakes of satu- assess independent associations between dietary benchmarks, with compli- rated fat and sodium, negatively NEMS-HCR score and restaurant charac- ance being lowest among Hispanic affecting cardiovascular health.8,9 teristics (, size, type [counter-style vs 2 sit-down] and price). Caribbean (HC) communities. Restaurants have the potential While interventions seeking to ad- to positively influence food and Results: None of the menus in the res- dress diet-related conditions focus nutrition environments and popu- taurants studied listed any main dishes 3 as “healthy” or “light.” More than half mostly on changing individual eat- lation health. Most public health (52%) offered mostly (>75%) nonfried ing behaviors, food and nutrition interventions and research target- main dishes, and 76% offered at least one environments can greatly affect food ing restaurants focus on chain res- vegetarian option. The most common 3,4 facilitator to healthy eating was offering choices. Foods consumed away taurants, leaving out community- reduced portion sizes (21%) and the most common barrier was having salt shak- ers on tables (40%). NEMS-HCR scores improve food offerings and environmental 4 Department of Population Health, (100-point scale) ranged from 24.1-55.2 cues to encourage healthful choices. Ethn Grossman School of Medicine, and Wagner (mean=39.7). In multivariable analyses, Dis. 2020;30(4):583-592; doi:10.18865/ Graduate School of Public Service, New scores were significantly related to cuisine ed.30.4.592 York University, New York, NY (with Puerto Rican cuisine scoring lower 5 Center for Systems and Community Design than Cuban and Dominican cuisines), and Keywords: Restaurants; Hispanic/Latino; and NYU-CUNY Prevention Research size (with small [<22 seats] restaurants Cardiovascular Disease; Food Environment Center, Graduate School of Public Health scoring lower than larger restaurants). We and Health Policy, City University of New found a significant quadratic association 1 Department of Health and Nutrition York, New York, NY with midpoint price, suggesting that scores Science, Brooklyn College, City University increased with increasing price in the low- of New York, Brooklyn, NY Address correspondence to Melissa Fuster, est price range, did not vary in the middle 2 Department of Epidemiology and PhD; Department of Health and Nutrition range, and decreased with increasing price Biostatistics, School of Medicine, University Science, City University of New York, in the highest range. of California, San Francisco, CA Brooklyn College, 2900 Bedford Ave, 3 Department of Nutrition and Food Studies, Brooklyn, NY 11210; M.FusterRivera54@ Conclusions: Our application of the Steinhardt School of Culture, Education, brooklyn.cuny.edu NEMS-R to HC restaurants in NYC revealed and Human Development, New York areas for potential future interventions to University, New York, NY

Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 583 Eating Well While Eating Out? - Fuster et al based, non-chain restaurants, such fulness of menu offerings and en- nonfried seafood, and vegetarian op- as those serving ethnic communi- vironmental support for healthy tions. method (fried foods) ties.10 Addressing this gap, the pres- food choices.11 The NEMS-R exam- was determined based on menu item ent study sought to: 1) adapt and ines food availability (main dishes, description, and supplemented with apply the Nutrition Environment sides, and non-alcoholic bever- knowledge of traditional prepara- Measures Survey for Restaurants ages), and environmental promo- tions, where some traditional foods (NEMS-R)11 to HC restaurants; tion of healthy or unhealthy choices (like ) are fried. These ad- and 2) examine associations be- through visual cues (table tents, ditions were based on the HC cui- tween restaurant characteristics and marketing material) and pricing. sines’ over-reliance on fried dishes nutrition environment measures. For this study, the NEMS-R was and offerings,14 and the cardio- adapted for HC cuisines, account- vascular health benefits of seafood consumption.16 We also expanded Methods the NEMS-R facilitators and barri- ers dimensions. The NEMS-HCR Setting assessed whether salt shakers were New York City (NYC) is a large, Foods consumed away present on tables given the role of concentrated urban area where eat- salt intake in cardiovascular health,16 ing out is common practice. The from home are an and expanded pricing comparisons, city has the largest concentration increasingly important by assessing pricing differences be- of Hispanics in the tween comparable nonfried and fried (29%), with and Do- part of food consumption dishes (ie, grilled chicken breast vs minicans being the leading groups fried chicken breast). The resulting (30% and 28% of the NYC His- and expenditures, NEMS-HCR was first piloted using panic population, respectively).12 online menus to refine the scoring accounting for 40% of criteria, and to ensure that the added Study Sample food spending among US item values varied across restaurants. We developed a sampling frame- work using Yelp, a popular busi- Hispanic households (44% Data Collection ness, crowd-sourced review site. 5,6 Data were collected between We searched NYC restaurants among all US homes). June-August 2019. We followed the classified as Puerto Rican, Cu- field procedures recommended by ban, and/or Dominican, follow- the NEMS-R protocol.11 Two re- ing conventional definitions of search assistants (RAs) were trained the Hispanic Caribbean region,13 using the University of Pennsylvania yielding a total of 183 restaurants, ing for dietary recommendations NEMS-R online training module. of which half were randomly se- for cardiovascular health. The ad- Additionally, the RAs completed two lected for the assessment, using the aptation was informed by previous practice assessments with the study MS Excel randomization function. research documenting HC dietary principal investigator, followed patterns7,14 and interviews with by ongoing discussions of sample Measures HC restaurant owners and cooks/ menus and the NEMS-R protocol. We used the Nutrition Envi- chefs.15 The resulting NEMS-HCR RAs were assigned restaurants ronment Measurement Survey for is composed of 25 items. It expand- according to location, facilitating Restaurants (NEMS-R), a widely ed the food availability component multiple assessments a day. Data used, validated tool that scores res- to assess other potentially healthful collection encompassed a site ob- taurants according to the health- items, such as nonfried main dishes, servation, where RAs confirmed

584 Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 Eating Well While Eating Out? - Fuster et al restaurant data (cuisine, location), meetings. A second quality check table). The restaurants were clas- conducted a guided observation was undertaken after data entry. sified as small, medium, or large, (environmental barriers and facilita- based on the tertile distribution of tors for healthy eating, and market- Data Analysis the number of seats. Lastly, we used ing materials), and collected a copy Data were analyzed using SPSS the midpoint price of main dishes as of the restaurant take-out menu (v.25, Armonk, NY). The main out- the measure for restaurant price, cal- for analysis. If a takeout menu come of interest was the NEMS- culated as the difference between the was not available, the raters took a HCR total score. NEMS-HCR highest and the lowest priced main photograph of the onsite menu or components were scored using dish on each menu, divided by 2. used the online menu where onsite the NEMS-R criteria. Most items We analyzed the distributions of and takeout menus were not avail- within the food availability and fa- NEMS-HCR total score, compo- able. Following NEMS-R protocol, cilitator dimension received one nents and sub-scores (healthy food menus were excluded, point where available, except for availability, and facilitators and and if separate and dinner the availability of healthy entrees barriers to healthier eating choic- menus were available, the dinner and healthy main dish salads, where es), and examined the associations menu was used. The assessment fo- points ranged between 0-3, depend- with selected restaurant character- cused on dishes that were available ing on the number of items available istics using Student’s t-tests, chi- every day, excluding special offer- (0 if none, 1 if one choice was avail- square tests, Pearson correlations, ings (ie, Sunday specials). Site visits able, 2 if 2-4 choices were available, and ANOVA with Tukey HSD were unobtrusive, and RAs did not and 3 if five or more were available). post-hoc tests, where appropriate. meet resistance from the restaurants. The presence of barriers received To control for potential confound- Data quality was ensured a -1 point. The resulting NEMS- ing statistically, we assessed the as- throughout the data collection peri- HCR had a range of -7 to 22, where sociation between restaurant char- od. Inter-rater reliability was assessed higher scores denoted a healthier acteristics and NEMS-HCR total at the beginning of the assessment, nutrition environment. The NEMS- scores using multivariable linear when two RAs each independently HCR total score was normally regression. Predictors were restau- assessed 10% of the sample (n=8).11 distributed (skewness=-.26; kur- rant cuisine, type, size and main Analysis showed good to excellent tosis=-.61), and was converted to dish midpoint price. We dichoto- inter-rater reliability, with percent- a 100-point scale to facilitate com- mized the size category, combining age agreements ranging between parisons with previous research. the medium and large categories. 62.5%-100% (mean=86.2%), in- We assessed the NEMS-HCR Based on preliminary analysis that cluding the total calculated score. components and sub-scores (food suggested that there may be a qua- Only one item, whether low-fat availability, facilitators, and barri- dratic relationship between NEMS- dressing was available, had a lower ers dimensions) against selected res- HCR score and midpoint price, we percentage agreement (37.5%), taurant characteristics. Cuisine was included both linear and quadratic which was addressed in subsequent based on the main HC cuisine sold midpoint price terms. To minimize meetings and quality checks. We or advertised (Cuban, Dominican or multicollinearity between the linear had weekly research team meetings, Puerto Rican). Restaurant type was and quadratic terms, midpoint price where RAs debriefed about the as- defined as a dichotomous variable was centered before squaring. Po- sessment, sharing issues (if any) categorizing restaurants as counter- tential multicollinearity among all encountered during the menu as- style (fast casual) or sit-down (waiter the predictors was examined using sessment. Each survey underwent a services). Restaurant size was based standard methods (assessing correla- quality check, where the survey was on official seating capacities or, tions among the predictors, compar- re-checked against the menu, flag- when not available, a visual assess- ing change in coefficients and their ging issues to resolve during team ment (count of tables and chairs per standard errors between full and

Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 585 Eating Well While Eating Out? - Fuster et al reduced models, examining toler- Table 1. Sample description, N=89 ance and variance inflation factors). Variable % (n) or mean (range) Main cuisine served Results Cuban 29% (26) Dominican 55% (49) Puerto Rican 16% (14) Sample Description Restaurant type The study sample consisted of Sit-down/waiter service 53% (47) 89 HC restaurants. Most of the Counter-style 47% (42) restaurants served Dominican cui- Restaurant size (mean number of seats) 47.2 (0-271) sine, with an almost even split by Small (0-21 seats) 32.6% (29) type (counter-style vs sit-down). Medium/large (22-49) 33.7% (30) On average, restaurants had 53.8 Large (52-274) 33.7% (30) Menu size (number of main dishes) 53.8 (5-172) main dishes, ranging from five to Main dish price (mean midpoint, range) $16.25 ($6.88-$30.92) 172. Only two (2%) of the restau- Cash-only 10% (9) rants assessed had a salad bar, and Offering family combos 33.7% (30) 30 (34%) offered family combos. Salad bar 2% (2) Contrary to typical combo , these were family com- bos typically offering a whole rotis- serie chicken, with large sides of can restaurants (64%) were classified and only 2 offered fruit (Table 2). and and a 2-liter soda. A few as small, whereas most Dominican While almost all restaurants offered restaurants (10%) were cash-only. restaurants (67%) were medium- nonfried dishes, only about half of- The mean midpoint price of sized. Cuban restaurants were most- fered a large proportion (>75%) main dishes was $16.25 (Table 1). ly large (50%) or medium (38.5%). of the main dishes as nonfried. The lowest-priced main dishes most- Three-quarters offered at least one ly included hamburgers and sand- Nutrition Environments vegetarian option. In terms of bev- wiches, ranging from $2.00-$25.00 in Hispanic Caribbean erages, only a few (16%) offered (mean=$7.60), while the highest- Restaurants low-fat or non-fat milk (Table 2). priced main dishes were usually The NEMS-HCR assessment Notable differences were found dishes containing seafood and/or provided an overview of the po- by cuisine. Cuban restaurants had steak (eg, lobster, seafood paella, tentially healthful foods available greater healthful food availability surf n’ turf), ranging from $8.00- in the restaurants, as well as the than Puerto Rican or Dominican $49.95 (mean=$24.91). Midpoint barriers and facilitators for health- restaurants. Whole grains were only price did not significantly differ ful eating (Table 2). To provide found in two Cuban sit-down restau- by cuisine, type or size category. context for the results, we com- rants, where one offered brown rice Size category significantly dif- pared NEMS-HCR measures and the other whole wheat . fered by restaurant type (X2 = 41.3, by restaurant cuisine and type. The availability of healthy main dish P<.001) and cuisine (X2 = 16.8, salads differed by cuisine (X2=7.2, P<.01). Counter-style restaurants Food Availability df=2, P<.05), with Cuban restaurants were more likely to be smaller than None of the menus in the res- most likely to offer them (69%), fol- sit-down restaurants. Puerto Rican taurants studied listed any main lowed by Dominican (39%) and restaurants were more likely to be dishes as “healthy” or “light,” based Puerto Rican restaurants (35%). The smaller than Cuban and Dominican on NEMS-R criteria. Only two proportion of nonfried main dish- restaurants. Most of the Puerto Ri- restaurants offered whole grains, es offered also differed by cuisine

586 Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 Eating Well While Eating Out? - Fuster et al

(F=14.0, P<.05), with Cuban restau- Table 2. NEMS-HCR component distribution rants (84%) having a significantly greater proportion than Dominican NEMS-HC Variable n % (69%) or Puerto Rican restaurants Availability of healthful options Healthy main dish(es) available 0 0% (57.9%), and Dominican restaurants Non-fried main dishes (proportion >75%) 46 52% having a significantly greater propor- Non-fried seafood main dishes (proportion >75%) 52 58% tion than Puerto Rican restaurants. Vegetarian main dishes 68 76% Healthy main dish salads 42 47% Fewer differences were found Low-fat or fat free salad dressing 14 16% by restaurant type. A higher per- Fruit 2 2% centage of sit-down restaurants of- Non-fried, non-starch side 76 85% fered vegetarian offerings (89.4% Whole grain bread 1 1% Brown rice 1 1% 2 vs 60.5% in counter-style, X =12.6, Other whole grains 0 0% df=1, P<.01), and a higher mean Beverages: 100% juice 53 60% proportion of nonfried main dishes Beverages: 1%/nonfat milk 4 4% Facilitators of healthy eating (76.1% vs 66.5%, t=-2.6, P<.01). Nutrition information on menu 0 0% Healthy main dishes identified on menu 0 0% Barriers and Facilitators Reduced-sized portions available 19 21% Restaurants presented few visible Healthy requests encouraged 2 2% Healthy less expensive than regular main dishes 0 0% facilitators or barriers to healthful Non-fried less expensive than fried 14 16% eating (Table 2). The most common Nutrition information posted 0 0% facilitators were offering reduced Highlighting healthy options 11 12% Healthy eating encouraged 1 1% portions (21%) and making non- Barriers to healthful eating fried main dishes less expensive than Large portions encouraged 4 4% comparable fried ones (16%). Some Menu discourages special requests 0 0% menus highlighted healthier options “All you can eat” or “unlimited” available 0 0% Combination meal cheaper than sum price of individual items 2 2% (12%), for example showing images Charge for shared main dishes 0 0% of nonfried and in- Salt shaker on table 36 40% stead of fried ones. However, none Unhealthy eating encouraged 8 9% of the assessed restaurants provided Overeating encouraged 1 1% nutrition information or identified healthy options on their menus (ie, having a “healthy” or “light” sec- included, for example, using menu by cuisine, type and size (Table 3). tion). Only one restaurant had en- images to promote fried dishes. Scores of Puerto Rican restaurants vironmental encouragements for Large portion encouragement (n=4) were significantly lower than those healthy eating – a counter-style Cu- was done by, for example, offering of Cuban or Dominican restau- ban restaurant with a tent promoting an extra chop for a few more rants (F=9.1, P<.001). Small restau- a vegetable platter at the entrance. dollars. These barriers were found in rants had significantly lower total The main visible barrier found Dominican and Cuban restaurants. scores than medium and large res- was having salt shakers on tables taurants (F=5.9, P<.05) (Table 3). (40%), which was assessed in emp- NEMS-HCR Scores When examining the NEMS- ty and occupied tables (Table 2). The NEMS-HCR total score HCR sub-scores, Puerto Rican res- Encouragement of overeating was ranged from 0 to 9 (mean=4.5). taurants scored significantly lower only found in one restaurant (a Do- On a 100-point scale, the mean than Cuban and Dominican restau- minican, counter-style restaurant). score was 39.7, ranging from 24.1 rants for healthful food availability Unhealthy eating encouragement to 55.2. We compared total scores (F=11.0, P<.001). Similarly, small

Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 587 Eating Well While Eating Out? - Fuster et al

Table 3. Nutrition environment scores by NEMS-HCR dimension and by restaurant type and cuisine Total Score NEMS-HCR Dimension Sub-scores Healthful food Facilitators for Barriers to Variable Raw 100-point scale availability healthful eating healthful eating Possible range (-7-22) (0-100) (0-15) (0-7) (-7-0) All 4.5±2.1 39.7±7.2 4.5±2.0 .6±.7 -.6±.6 By cuisine Cuban 5.5±2.0 43.1±6.9 5.5±1.2 .6±1.0 -.6±.6 Dominican 4.5±1.8 39.6±6.4 4.5±1.8 .6±.6 -.6±.6 Puerto Rican 2.8±2.0a 33.7±6.8a 2.6±1.8a .5±.5 -.4±.6 By type Counter-style 4.3±2.1 39.0±7.3 4.1±1.9b .7±.8 -.4±.5 b Sit-down 4.7±2.1 40.3±7.1 4.9±2.0 .5±.6 -.7±.7 By size Small 3.5±2.1c 36.7±7.1c 3.3±1.7c .6±.7 -.4±.6 Medium 5.1±1.8 41.6±6.1 4.9±1.8 .7±.8 -.6±.6 Large 5.0±2.1 41.3±7.2 5.3±2.0 .4±.6 -.7±.7

Data shown are unadjusted means±SD. a. ANOVA significantly different by cuisine; Puerto Rican cuisine significantly different from Cuban and Dominican cuisines via Bonferroni post-hoc test, P<.001. b. T-test for equality of means significantly different by type (counter-style vs sit-down), P<.05. c. ANOVA significantly different by size; small size restaurants significantly different from medium and large size restaurants via Bonferroni post-hoc test, P<.05 (total score) and P<.01 (food availability).

restaurants had significantly lower total scores are shown in Table 4. creasing price in the high price food availability scores than me- There was no evidence of multicol- range (approximately $20-$31). dium and large restaurants (F=9.1, linearity in the model. Controlling P<.01). In addition, compared with for restaurant characteristics, to- counter-style restaurants, sit-down tal scores of Cuban and Domini- Discussion restaurants had significantly great- can restaurants were significantly er healthful food availability sub- higher than those of Puerto Rican To our knowledge, this study scores (t=-2.0, P<.05), lower barrier restaurants. Small restaurants had is the first to adapt the NEMS-R sub-scores t=2.1, P<.05) (Table 3). lower scores than those in the for use in HC restaurants, find- There was a small positive associ- combined medium and large size ing significant associations between ation between price and the NEMS- category. The quadratic term for NEMR-HCR scores and restaurant HCR total score (r=.10, P=.35); the midpoint price of main dishes characteristics. The assessment re- however, a scatterplot suggested that was significantly negatively associ- vealed room for potential improve- there may be a quadratic relationship. ated with the NEMS-HCR score ments in food availability, such as Midpoint price was positively corre- (B=-1.6, P=.01), while the raw providing whole grains and fresh lated with the food availability score midpoint price term was positive fruits, but also potential environ- dimension (r=.26, P=.013) and neg- and nonsignificant. This indicates mental changes to encourage and atively correlated with the facilitator a curvilinear relationship, with facilitate healthful choices. On the dimension score (r=-.31, P=.004). scores increasing with increasing other hand, the nutrition environ- price in the low price range (ap- ments in these restaurants did not Factors Associated with proximately $7-$15), scores not include many environmental barri- NEMS-HCR Scores changing across the middle price ers to healthful eating. For example, Multivariable linear regression range (approximately $16-$19), the restaurants in the sample did results predicting NEMS-HCR and scores decreasing with in- not engage in the promotion of

588 Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 Eating Well While Eating Out? - Fuster et al

Table 4. Multilinear regression of NEMS-HCR score (100-point scale) on restaurant characteristics, N=89 Standardized Unstandardized Coefficients Coefficients B Std. Error Beta P (Constant) 34.4 2.7 - <.001 Main cuisine served Cuban 6.9 2.3 .4 <.01 Dominican 4.6 2.0 .3 .02 Puerto Rican (REF) - - - Restaurant type Take-out 1.8 1.6 .1 .26 Sit-down (REF) - - - Restaurant size Small -4.6 1.8 -.3 .01 Medium/large (REF) - - - Midpoint price .2 .1 .1 .18 Centered quadratic midpoint price -1.6 .6 -.3 .01

R2 = .28

overeating, such as supersizing and mation, which increases the score HCR score may be explained by providing “all you can eat” offers. regardless of the healthfulness of subtle differences across these cui- The mean NEMS-R score of the foods provided. The provision of sines when offered in contemporary the HC restaurants included in the nutrition information is only man- restaurants. While HC cuisines have study fell within previously report- datory for chain restaurants with 20 the same staple dishes (white rice, ed ranges (36–51, on a 100-point or more locations nationwide, do- beans, , plantain), they differ scale).17-19 Some of these studies ing business under the same name, on the emphasis given to certain showed lower NEMS-R scores than and offering the same menu items.21 foods, such as the potentially greater found in this study (39.7). For ex- The provision of nutrition informa- emphasis on fried foods in Puerto ample, an assessment of low-income tion is not feasible for community- Rican cuisine found in NYC com- urban settings in Australia revealed a based (non-chain) restaurants, as munities.14 However, the results mean score of 35.718 and a statewide compliance comes with high costs may also indicate differences in how assessment of restaurants in Wiscon- incurred for the nutrition analysis these cuisines are marketed. While sin showed a mean score of 36.1.17 and requires recipe standardization. we did not examine neighborhood- Neckerman et al examined fast food Policies could offset the cost of this level information in this study, field- restaurants in New York City, result- added expense (for example, tax in- work observations indicate that it is ing in a mean score of 48,19 higher centives for non-chain restaurants to possible that Cuban restaurants were than the average NEMS-R score include nutrition information), but located in areas targeting more afflu- found among the HC restaurants this shows an important limitation ent, non-Hispanic markets. Estab- included in this study. Fast food of the NEMS-R, and the need for lishments outside of ethnic enclaves restaurants, while usually associated adaptations, as the one presented may provide menu adaptations to with unhealthy dietary patterns,20 in this study, to capture potentially potentially market to younger and may present factors that lead to a healthful strategies and offerings more health-conscious custom- higher score, compared with small- found in non-chain restaurants. ers, including non-HC customers. er, ethnic eateries. One such factor The significant association -be Moreover, some restaurants offer- is the provision of nutrition infor- tween HC cuisine and the NEMS- ing Cuban and Dominican cuisines

Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 589 Eating Well While Eating Out? - Fuster et al included in the sample marketed ited, as a function of potentially tion of salads may be substituted themselves as “fusion” restaurants, limited and storage space. with promotion of hearty in and included elements of cuisines This may result in the offering of the cooler months. As such, our from Asian or other Latin Ameri- less fresh produce (ie, salads and findings may not be generalizable can countries, often incorporating non-starchy vegetable sides), influ- to the food offerings and marketing healthier, nonfried options. None of encing the score. Therefore, restau- techniques on an annualized basis. the Puerto Rican restaurants exam- rant size should be considered when The assessment of healthy offerings ined were marketed this way. Based developing interventions to address in these establishments was limited on these findings, further research barriers to healthful nutrition envi- by the lack of nutrition informa- should examine the neighborhood ronments in these establishments. tion. While we expanded the food and market-level factors influencing Lastly, our study sought to pro- availability dimension to include potentially different cuisine inter- vide insights into the association be- potentially healthy offerings (such tween price and the relative health- as nonfried and vegetarian main fulness of nutrition environments in dishes), we could not precisely as- restaurants. Pricing matters because sess the nutritional quality of these The assessment revealed of its strong association with diet foods. Lastly, our assessment fo- quality, especially in low-income set- cused on restaurant characteristics, room for potential tings.22,23 Pricing in restaurants has which are inherently connected to been mostly studied in its associa- the neighborhood characteristics in improvements in food tion with consumption, where lower which the restaurants are located. availability, such as prices tend to increase the sale and Some of our significant associations consumption of healthier items.24 (such as the potential influence of providing whole grains We found a curvilinear relationship, cuisine served) may be at least par- where scores rose up to a given price tially explained by neighborhood and fresh fruits, but also point (approximately $18), and socioeconomic and demographic potential environmental plateaued and then decreased. This characteristics. These limitations association is contrary to the gen- point to the need of future re- changes to encourage and eral notions of healthy food being search to examine whether and how more expensive, demonstrating the neighborhood-level factors may in- facilitate healthful choices. complexity of examining the price- fluence the healthfulness of nutri- healthfulness association.25,26 This tion environments in restaurants. curvilinear association needs further research, due to the relatively small pretations. Such understanding will number of restaurants with mid- Conclusion lead to better tailoring of interven- point prices above $20, which makes tions to account for the different predictions in this range less precise. This study provides information food identities that map to health- Our study addresses an impor- that can be used to develop strate- fulness in these diverse settings. tant research gap to improve health gies to promote healthier eating in Our finding showing that small- outcomes in ethnic communities. ethnic restaurants, and an assess- er restaurants had lower NEMS- However, we must also consider its ment tool adapted for use in HC HCR scores than medium and limitations. Seasonality may poten- restaurants. The assessment identi- larger restaurants in adjusted analy- tially influence the availability and fied existing healthful strategies and sis may suggest that smaller restau- promotion of certain foods, such potential healthful options, provid- rants, regardless of type, may be as salads. We conducted the assess- ing a good starting point to work offering menus that are more lim- ment in the summer. The promo- with these establishments to meet

590 Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 Eating Well While Eating Out? - Fuster et al

SOL) results. Am Heart J. 2016;176:134- 13. Duany J. Blurred Borders: Transnational Mi- them where they are. Previous re- 144. https://doi.org/10.1016/j. gration between the Hispanic Caribbean and search demonstrates that the restau- ahj.2016.02.008 PMID:27264232 the United States. Chapel Hill: University 3. Story M, Kaphingst KM, Robinson- of North Carolina Press; 2011. https://doi. rant sector may be open to health- O’Brien R, Glanz K. Creating healthy org/10.5149/9780807869376_duany ful improvements, as long as these food and eating environments: policy 14. Fuster M. “We like fried things”: Negotiat- do not ultimately affect the business and environmental approaches. Annu ing health and taste among Hispanic Carib- Rev Public Health. 2008;29(1):253-272. Communities in New York City. Ecol 15,27 outcomes. Some strategies can https://doi.org/10.1146/annurev.publ- Food Nutr. 2017;56(2):124-138. https:// be low-cost, such as highlighting health.29.020907.090926 PMID:18031223 doi.org/10.1080/03670244.2016.1267007 4. Malambo P, Kengne AP, De Villiers A, PMID:28059558 existing healthier options on menus Lambert EV, Puoane T. Built environment, 15. Fuster M, Guerrero K, Elbel B, Ray K, or training the staff to recommend selected risk factors and major cardiovascu- Huang TTK. Engaging ethnic restaurants those options. More research is lar disease outcomes: a systematic review. to improve community nutrition environ- PLoS One. 2016;11(11):e0166846. https:// ments: a qualitative study with Hispanic Ca- needed to understand how to best doi.org/10.1371/journal.pone.0166846 ribbean restaurants in New York City. Ecol engage small business owners and PMID:27880835 Food Nutr. 2020;59(3):294-310. https:// 5. US Bureau of Labor Statistics. BLS Reports: doi.org/10.1080/03670244.2020.1717481 to find convergence between pub- Consumer Expenditures in 2016; Table 13. PMID:31964188 lic health and business outcomes. Last accessed July 20, 2020 from https:// 16. Folsom AR, Shah AM, Lutsey PL, et al. Given the significant level of food www.bls.gov/opub/reports/consumer-expen- American Heart Association’s life’s simple ditures/2016/home.htm 7: avoiding heart failure and preserving consumption in restaurant settings 6. US Department of Agriculture, Economic cardiac structure and function. Am J Med. and importance of this on health, Research Service. Food-Away-From Home. 2015;128(9):970-976 e972. 2017. Available from https://www.ers.usda. 17. Martinez-Donate AP, Espino JV, Meinen interventions in restaurants are gov/topics/food-choices-health/food-con- A, et al. Neighborhood disparities in critically needed and can contrib- sumption-demand/food-away-from-home. the restaurant food environment. WMJ. ute to the narrowing of health dis- aspx. 2016;115(5):251-258. PMID:29095587 7. McClain AC, Ayala GX, Sotres-Alvarez 18. Storr R, Carins J, Rundle-Thiele S. Assessing parities in minority communities. D, et al. Frequency of intake and type of support for advantaged and disadvantaged away-from- home foods consumed are as- groups: a comparison of urban food envi- Acknowledgment sociated with diet quality in the Hispanic ronments. Int J Environ Res Public Health. This research was supported by the City Community Health Study/Study of Latinos 2019;16(7):E1135. https://doi.org/10.3390/ University of New York (CUNY-PSC Award (HCHS/SOL). J Nutr. 2018;148(3):453- ijerph16071135 PMID:30934887 463. https://doi.org/10.1093/jn/nxx067 19. Neckerman KM, Lovasi L, Yousefzadeh - B Cycle 50) and the National Heart, Lung PMID:29546313 P, et al. Comparing nutrition environ- and Blood Institute (K01 HL147882-01). 8. Jiao J, Moudon AV, Kim SY, Hurvitz PM, ments in bodegas and fast-food restaurants. Drewnowski A. Health implications of J Acad Nutr Diet. 2014;114(4):595-602. Conflict of Interest adults’ eating at and living near fast food https://doi.org/10.1016/j.jand.2013.07.007 No conflicts of interest to report. or quick service restaurants. Nutr Diabetes. PMID:24035459 2015;5(7):e171. https://doi.org/10.1038/ 20. Fryar CD, Hughes JP, Herrick KA, Ah- Author Contributions nutd.2015.18 PMID:26192449 luwalia N. Fast food consumption among Research concept and design: Fuster; 9. Chum A, O’Campo P. Cross-sectional as- adults in the United States, 2013–2016. sociations between residential environmental NCHS Data Brief. 2018;(322):1-8. Acquisition of data: Fuster, Sakowitz, Halvey; exposures and cardiovascular diseases. BMC PMID:30312154 Data analysis and interpretation: Fuster, Public Health. 2015;15(1):438. https:// 21. Food and Drug Administration. Food Pouget, Ray, Elbel, Huang; Manuscript draft: doi.org/10.1186/s12889-015-1788-0 Labeling; Nutrition Labeling of Standard Fuster, Pouget, Handley, Ray, Elbel, Huang; PMID:25924669 Menu Items in Restaurants and Similar Retail Statistical expertise: Pouget; Acquisition of 10. Valdivia Espino JN, Guerrero N, Rhoads N, Food Establishments. Federal Register 79 funding: Fuster; Administrative: Handley, et al. Community-based restaurant interven- FR71155, December 1, 2014. Docket FDA- Ray, Elbel, Sakowitz, Halvey, Huang; Super- tions to promote healthy eating: a systematic 2011-F-0172 vision: Fuster review. Prev Chronic Dis. 2015;12:E78. 22. Darmon N, Drewnowski A. Contribu- https://doi.org/10.5888/pcd12.140455 tion of food prices and diet cost to References PMID:25996986 socioeconomic disparities in diet qual- 11. Saelens BE, Glanz K, Sallis JF, Frank LD. ity and health: a systematic review and 1. NIH National Health, Lung, and Blood Nutrition Environment Measures Study in analysis. Nutr Rev. 2015;73(10):643-660. Institute. Hispanic Community Health restaurants (NEMS-R): development and https://doi.org/10.1093/nutrit/nuv027 Study/Study of Latinos (HCHS/SOL). Last evaluation. Am J Prev Med. 2007;32(4):273- PMID:26307238 accessed February 21, 2018 from https:// 281. https://doi.org/10.1016/j. 23. Rehm CD, Monsivais P, Drewnowski A. www.nhlbi.nih.gov/science/hispanic-com- amepre.2006.12.022 PMID:17383558 Relation between diet cost and Healthy munity-health-studystudy-latinos-hchssol. 12. Greer S, Naidoo M, Hinternalnd K, et al. Eating Index 2010 scores among adults in 2. González HM, Tarraf W, Rodríguez CJ, Health of Latinos in NYC. NYC: New York the United States 2007-2010. Prev Med. et al. Cardiovascular health among diverse City Department of Health and Mental 2015;73:70-75. https://doi.org/10.1016/j. Hispanics/Latinos: Hispanic Community Hygiene; 2017. ypmed.2015.01.019 PMID:25625693 Health Study/Study of Latinos (HCHS/

Ethnicity & Disease, Volume 30, Number 4, Autumn 2020 591 Eating Well While Eating Out? - Fuster et al

24. Powell LM, Chaloupka FJ. Food prices and obesity: evidence and policy implica- tions for taxes and subsidies. Milbank Q. 2009;87(1):229-257. https://doi. org/10.1111/j.1468-0009.2009.00554.x PMID:19298422 25. Carlson A, Frazão E. Are Healthy Foods Really More Expensive? It Depends on How You Measure the Price. May 1, 2012. USDA-ERS Economic Information Bulletin. 2012(96). Last accessed July 20, 2020 from https://www.ers.usda.gov/webdocs/publica- tions/44678/19980_eib96.pdf?v=9357.8 26. Rao M, Afshin A, Singh G, Mozaffarian D. Do healthier foods and diet patterns cost more than less healthy options? A systematic review and meta-analysis. BMJ Open. 2013;3(12):e004277. https://doi. org/10.1136/bmjopen-2013-004277 PMID:24309174 27. Nevarez CR, Lafleur MS, Schwarte LU, Ro- din B, de Silva P, Samuels SE. Salud Tiene Sabor: a model for healthier restaurants in a Latino community. Am J Prev Med. 2013;44(3 Suppl 3): S186-192. https://doi. org/10.1016/j.amepre.2012.11.017

592 Ethnicity & Disease, Volume 30, Number 4, Autumn 2020