Abstract
Using data from the National Health and Nutrition Examination Survey (2001–2018; N = 19,602), this study examined whether ultra-processed food (UPF) consumption is associated with cardiometabolic health (obesity, hypertension, high cholesterol, and diabetes), among White, Black, Hispanic, and Asian Americans (AA) US adults 50 or older. Diet was assessed using 24 hour dietary recall. NOVA dietary classification system was used to calculate the percentage of caloric intake derived from UPFs. Cardiometabolic information was assessed through physical examination, blood tests, and self-reported medication information. A median of 54% (IQR: 40%, 68%) of caloric intake was attributed to UPFs and was lowest for AAs (34%, IQR: 20%, 49%) and highest for White adults (56%; IQR: 42, 69%). In multivariable adjusted models, UPF consumption was associated with greater odds of obesity, high cholesterol, and diabetes. UPF consumption is associated with poor cardiometabolic health among all US older adults. For AAs, UPFs may be particularly obesogenic.
Introduction
The diets of Asian Americans – including older adults – are relatively understudied compared with other groups and analyses of national data are revealing previously undocumented dietary disparities for the Asian American population. Compared to peers of other racial/ethnic groups, Asian American adults have greater consumption of refined grains, higher sodium intake, and therefore, a lower likelihood of consuming the recommended amount of sodium per day (≤2300 mg/day) (Awata et al., 2017; Bailey et al., 2016; Firestone et al., 2017). While healthy dietary patterns are important at any age, older Asian Americans are at high risk of cancer, cardiovascular disease (CVD), diabetes, and Alzheimer’s disease and related dementias (Mayeda et al., 2016; Mehta & Yeo, 2017; Rajpathak & Wylie-Rosett, 2011; Shih et al., 2014; Thompson et al., 2016; Tsai et al., 2013; Wong et al., 2014; Yi et al., 2015) – conditions which may be mitigated by improved dietary behaviors (Boeing et al., 2013; Dauchet et al., 2006; He et al., 2006; Morris, 2016). However broad stereotypes both societally and among researchers that this community suffers from few health disparities (Yi et al., 2016) and a lack of reliable and systematically collected data in Asian Americans (Jose et al., 2014) have contributed to limited knowledge of diet (Palaniappan et al., 2010) in Asian Americans.
In addition to the limited availability of data pertaining to Asian American diet writ large, there is also a lack of cultural adaptation with respect to preventative dietary approaches for Asian Americans (Wang et al., 2021). For example, the Mediterranean diet (Salas-Salvado et al., 2018) emphasizes fruit, vegetable, whole grain, and fish consumption (Van Horn et al., 2016), but as its name suggests, it is based off of typical dietary patterns common in the Mediterranean region. The Healthy US-Style or Healthy Vegetarian Dietary patterns, also endorsed by the Dietary Guidelines for Americans (Dietary Guidelines for Americans, 2020), may have limited appeal to non-Vegetarians or to Asian American immigrants unaccustomed to an American diet. Thus, there is a need for evidence-based diets or dietary patterns complemented with preservation of healthy aspects of non-Western Asian diets to prevent diet-related death and disability. In fact, in 2021 the American Heart Association updated its dietary guidance to emphasize the selection of minimally processed foods (Lichtenstein et al., 2021). A dietary pattern that minimizes processed food consumption can be applied cross-culturally (Marron-Ponce et al., 2018). This is particularly salient for Asian Americans, given the great diversity of diets in different Asian populations. The NOVA framework is one such classification system that can be used to assess the level of food processing within a diet as it emphasizes the consumption of minimally processed foods over highly processed foods rather than focusing on specific nutrients (Monteiro et al., 2016). Ultra-processed food (UPF) consumption, as assessed by the NOVA framework, has been linked to less chronic conditions (Monteiro et al., 2018), is easy to measure, and is flexible—it can easily be adapted to any cuisine (Monteiro et al., 2016). Yet the relationships between UPF intake (measured through the NOVA framework) with health endpoints has not yet been assessed for racial/ethnic minorities, or specifically older adults, for whom adverse health outcomes are proximal.
Within this context, the overall goal of this study is to fill the critical gap in our understanding of the associations between UPF consumption, classified according to the NOVA framework, with key cardiometabolic risk factors (obesity, hypertension, high cholesterol, and diabetes) that can result from poor diet (Rakhra et al., 2020; Rumawas et al., 2009; Sacks et al., 2001) and are associated with cardiovascular health (Lloyd-Jones et al., 2022), within older and racial/ethnic minority populations. To do so, we will utilize data from multiple waves of the National Health and Nutrition Examination Survey (NHANES) from 2001 through 2018– which includes an oversampling of Asian Americans (from 2011–2016).
Methods
Study Design and Population
The current study used data from the NHANES 2001–2018 cycles. NHANES is a cross-sectional nationally representative survey, that is designed to assess the health and nutritional status of the United States (U.S.) population (CDC). The survey includes a sample of 5000 non-institutionalized civilian U.S. residents annually that are sampled using a complex, stratified, multistage probability cluster sampling design. We restricted the analytical sample to adults aged 50 years and older with at least one valid dietary recall per United States Department of Agriculture assessment (Steele et al., 2023). We chose to focus on older adults given that cardiometabolic disease is more prevalent in older persons (Sinclair & Abdelhafiz, 2020). Of 24,768 participants who were in the appropriate age-span, we excluded 5166 individuals who lacked valid dietary data, resulting in a final analytical dataset of 19,602 participants. All NHANES data have been approved by the National Center for Health Statistics Research Ethics Review Board.
Data Collection
Data on demographics, socio-economic characteristics and health behaviors were self-reported during an interview by trained personnel in the respondents’ homes. During a health examination performed at a mobile examination center, staff measured participants' height and weight, assessed participant’s blood pressure and collected blood samples (Zipf et al., 2013). Data on dietary intake was collected by 24 h dietary recall interviews performed by trained interviewers (Zipf et al., 2013). One single in-person recall was carried out in 2001–2002, while later survey cycles included an in-person recall followed by a telephone-based recall 3–10 days later (Zipf et al., 2013). From 2003 and onwards, recalls were performed using the validated US Department of Agriculture Automated Multiple-Pass Method (Blanton et al., 2006; Moshfegh et al., 2008).
Food Consumption according to Processing Level
Based on the NOVA framework, we classified all food items recorded in NHANES 2001–2018 into four mutually exclusive processing levels: (1) “minimally processed foods” which include fresh, dry or frozen fruits or vegetables, grains, legumes, meat, fish and milk made by removing unwanted parts of foods, and are minimally processed though grinding, drying, roasting, boiling, non-alcoholic fermenting, etc.; (2) “processed culinary ingredients” which include table sugar, oils, fats, salt, acids (lemon, vinegars used extensively in Asian cooking) and other constituents extracted from foods or from nature and used in kitchens to make culinary preparations with Group 1 foods; (3) “processed foods” which include foods such as canned fish and vegetables, breads and artisanal cheeses which are manufactured by adding salt, sugar, oil, or other processed culinary ingredients to unprocessed or minimally processed foods; and (4) “ultra-processed foods”, which are defined as industrial formulations containing little or no whole foods that are produced using ingredients (e.g. dyes, emulsifiers, flavorings, and preservatives) and/or processing techniques (e.g. extrusion, molding) of exclusive industrial use (Moubarac et al., 2014). Other examples of Group 4 (UPFs) include instant and canned soups; reconstituted meat and fish products; meatless patties; ready-made sauces, gravies, and dressings; french fries and other pre-made potato products such as chips; ready-to-eat and dry-mix desserts such as pudding; confectionary; sweet and savory snack foods including granola bars and protein bars and sugar- or artificially-sweetened beverages including soda, fruit drinks, pre-sweetened tea and coffee, energy drinks and dairy-based drinks; flavored and/or sweetened yogurt; industrially manufactured cakes, cookies and pies; dry cake- and pancake mixes; industrially manufactured breads; sweet breakfast cereals; frozen and shelf-stable plate meals; ice cream, frozen yogurt and ice pops (Moubarac et al., 2014).
A detailed description of the approach used to classify foods and beverages according to the NOVA framework has been published (Steele et al., 2023). Briefly, items were classified by considering the NHANES variables “Main Food Description”, “Additional Food Description”, which describes foods (food codes), and “SR Code Description”, which describes the underlying ingredients of foods (SR codes), as well as “Combination Food Type” and “Source of Food”. When foods were judged to be a hand-made recipe, we applied the classification to the underlying ingredients in order to ensure a more accurate classification.
When needed, we used lists of ingredients from supermarket, Amazon and Fooducate websites, as well as the USDA Branded Food Product Database (applicable to the 2017-2018 NHANES cycle only) to establish the most appropriate NOVA classification. In case of uncertainty, we generally used a conservative approach and classified ambiguous items into a lower degree of processing. However, industrially produced breads, salty snacks and ready-to-eat cereals with ambiguous processing levels were classified as UPF, as most such items meet the criteria for UPF. We obtained SR codes for each survey cycle from the corresponding versions of the United States Department of Agriculture (USDA) Food and Nutrient Database for Dietary Studies (FNDDS) and used Food Code energy values provided by NHANES to calculate energy intakes from each NOVA food group. For hand-made recipes, we used data from both FNDDS and the USDA National Nutrient Database for Standard Reference, Release 18–28 and SR Legacy, to calculate energy values of the underlying ingredients (SR Codes). These calculations are described in detail elsewhere (Martínez Steele et al., 2016). Using dietary data for day 1, we calculated each participants’ intake of UPFs, as the relative contribution to daily energy intake (% of total energy). The whole sample was thereafter divided into quartiles according to UPF consumption.
Definition of Outcome Variables
Body mass index (kg/m2; BMI) values of ≥30 kg/m2 were classified as obesity per World Health Organization criteria (WHO, 2020). We also considered a BMI threshold of ≥27.5 kg/m2 to define obesity for Asian Americans (WHO Expert Consultation, 2004). We defined hypertension as blood pressure ≥140/90 mmHg or current blood pressure medication use; diabetes as fasting serum glucose levels ≥126 mg/dl, hemoglobin A1c ≥ 6.5 or current diabetic medication use (insulin or oral hypoglycemic agents); and hypercholesterolemia as low-density lipoprotein cholesterol (LDL-C) ≥160 mg/dl or current lipid-lowering medication use.
Definition of Covariates
Along with sex, age group was self-reported and categorized as follows: 50–59 years, 60–69 years, 70–79 years and ≥80 years. Race/ethnicity was also self reported. Regardless of race, Hispanic ethnicity was assessed via self-report of being Hispanic or Latino. For individuals who did not self-report being Hispanic or Latino, race/ethnicity was self reported as non-Hispanic white, non-Hispanic Black, Non-Hispanic Asian, or other race including anyone multi-racial. Asian Americans were defined by self-identification as “having origins in the original peoples of the Far East, Southeast Asian, or the Indian subcontinent” (Paulose-Ram et al., 2017). Socio-economic information was also self-reported including: education level (less than high school, high school degree/general equivalency diploma, some college, college graduate or above), family poverty income ratio, defined as the ratio of family income to the year-specific federal poverty threshold (categorized as <130%, 130%–349% and ≥350% of the federal poverty threshold), marital status (married/not married) and health insurance (yes/no). Lifestyle factors of interest included regular alcohol consumption (yes/no) and smoking status (categorized as never smoked, former smoker and current smoker). Physical activity level was classified as low (<150 min of moderate intensity equivalent activity per week), medium (150–300 min of moderate intensity equivalent activity per week), and high (>300 min of moderate intensity equivalent activity per week), based on the fulfilment of the 2008 Physical Activity Guidelines for Americans (U.S. Department of Health and Human Services, 2008).
Statistical Analyses
First, we examined the distribution of participant demographic, socioeconomic, clinical, and behavioral characteristics by quartile of UPF consumption. Differences by quartile of UPF consumption were assessed by Pearson’s chi-squared test of independence. We also estimated the age adjusted prevalence of each cardiometabolic indicator overall, according to race/ethnicity, and by quartile of UPF consumption.
Multivariable adjusted logistic regression analysis was used to evaluate associations between quartile of UPF and cardiometabolic outcomes (separately for obesity, hypertension, high cholesterol, and diabetes). We fit four models for all outcomes: (1) adjusted for demographic characteristics (age group, sex, and race/ethnicity); (2) additionally adjusted for socio-economic characteristics (education, income, marital status, and insurance status); (3) additionally adjusted for behavioral characteristics (smoking, physical activity, and alcohol consumption); (4) further adjusted for clinical characteristics (BMI for all non-obesity outcomes, and systolic blood pressure for diabetes). We also tested the interaction between UPF consumption with race/ethnicity for all outcomes. Significant interactions led to race/ethnicity stratified models. Finally, as a sensitivity analysis, for Asian Americans, we repeated our analysis for obesity using the Asian American specific BMI threshold of ≥27.5 kg/m2 to define obesity.
All analyses were conducted using the NHANES sample weights in order to account for oversampling of certain populations, nonresponse and population coverage (Rothwell et al., 2013). The Taylor Series Linearization variance approximation procedure was used to account for the complex sample design of NHANES in the variance estimation. All analyses were performed using Stata/SE 15.0 and SUDAAN V11.4. Statistical significance was set to alpha<.05.
Results
Characteristics of U.S. Adults Aged 50 or Older, Stratified by Quartile of Ultra-processed Food Intake, NHANES 2001–2018.
NHANES: National Health and Nutrition Examination Survey.
A “missing data” category was created for missing/refused/don’t know responses. The following proportions of above variables were coded as missing in the overall sample: education: .1%, income: 7.5%, marital status: .1%, insured: .1%, smoking: .1%, alcohol consumption: 18.0%, health status: 3.6%, Cardiovascular disease: .7%, emotional support: 67.0%, needed more support: 69.0%.
aIndicates characteristic differs significantly (p < .05) by Ultra processed food quartile, chi-square tests.
Ultra-Processed Foods
Overall UPFs accounted for 53.8% (95% CI: 53.2, 54.3) of daily caloric intake, with substantial differences by race/ethnicity. For example, UPF consumption accounted for 55% of daily caloric intake for non-Hispanic White adults (95% CI: 54.5, 55.8) which was comparable to that of non-Hispanic Black adults (54.5%, 95% CI: 53.6, 55.8), but significantly higher than that of Hispanic adults (47.7%, 95% CI: 46.7, 48.7) and non-Hispanic Asian adults (35.8%, 95% CI: 34.1, 37.6), p-values both <.05.
Cardiometabolic Conditions
Overall, cardiometabolic conditions were common with a high prevalence of obesity (39.0%, 95% CI: 37.8, 40.2), hypertension (55.9%, 95% CI: 54.7, 57.2), high cholesterol (60.7%, 95% CI: 59.4, 62.0), and diabetes (35.7%, 95% CI: 34.5, 37.0), Figure 1 [Insert Figure 1]. Compared with non-Hispanic White adults, the prevalence of obesity (38.5%, 95% CI: 37.0, 40.1) was higher in non-Hispanic Black adults (48.8%, 95% CI: 47.1, 50.4), and Hispanic adults (42.1%, 95% CI: 40.0, 44.1), but lower in non-Hispanic Asian adults using both obesity definitions (BMI ≥30: 10.5%, 95% CI: 8.7, 12.8, BMI ≥27.5: 23.8%, 95% CI: 20.4, 27.6), p values <.05. The prevalence of hypertension was higher in non-Hispanic Black adults (72.8%, 95% CI: 71.1, 74.3) compared with non-Hispanic White adults (53.7%, 95% CI: 52.2, 55.2), p < .05. The prevalence of high cholesterol was higher in non-Hispanic Asian adultss (69.8%, 95% CI: 65.0, 74.3) compared with non-Hispanic White adults (60.4%, 95% CI: 58.8, 61.9), p < .05. The prevalence of diabetes was higher in non-Hispanic Black adults (51.5%, 95% CI: 49.5, 53.5), Hispanic adults (48.8%, 95% CI: 46.2, 51.5), and non-Hispanic Asian adults (48.6, 95% CI: 44.5, 52.7) compared with non-Hispanic White adults (31.2%, 95% CI: 29.8, 32.7), p values <.05. Age adjusted prevalence of cardiometabolic indicators among U.S. adults aged 50 or older by race/ethnicity, NHANES 2001–2018. *Indicates prevalence differs significantly (p < .05) from the reference category (non-Hispanic Whites).
Ultra-processed Foods and Cardiometabolic Conditions
Overall, all four cardiometabolic conditions were more common in the highest compared with lowest quartile of UPF consumption (Figure 2) [Insert Figure 2]. Multivariable adjusted associations between quartiles of UPF intake with cardiometabolic conditions are presented in Table 2. Obesity: From logistic regression models adjusted for age, sex, and race (model 1), compared to the first quartile of UPF, higher quartiles were associated with a 21% greater odds of obesity in Q2 (OR: 1.21, 95% CI: 1.08, 1.36), 46% greater odds in Q3 (OR: 1.46, 95% CI: 1.30, 1.63), and 60% greater odds in Q4 (OR: 1.60, 95% CI: 1.41, 1.83). Results were attenuated but still significant in our fully adjusted model 3, with higher quartiles compared with quartile one associated with an 18% greater odds of obesity in Q2 (OR: 1.18, 95% CI: 1.05, 1.34), 41% greater odds in Q3 (OR: 1.41, 95% CI: 1.25, 1.58), and 50% greater odds in Q4 (OR: 1.50, 95% CI: 1.31, 1.72). Hypertension: UPF consumption was only significantly associated with a greater odds of hypertension in our minimally adjusted model 1 comparing Q4 to Q1 (OR: 1.16, 95% CI: 1.02, 1.32). High cholesterol: From model 1, compared to the first quartile of UPF consumption, higher quartiles were associated with greater odds of high cholesterol by 26% in Q2 (OR: 1.26, 95% CI: 1.08, 1.47) and 24% in Q4 (OR: 1.24, 95% CI: 1.05, 1.47). Results were attenuated in our fully adjusted model 4, with quartile 2 compared to quartile 1 associated with a 23% greater odds of high cholesterol (OR: 1.23, 95% CI: 1.05, 1.45). Diabetes: In model 1, compared to the first quartile of UPF consumption, higher quartiles were associated with greater odds of diabetes by 32% in Q2 (OR: 1.32, 95% CI: 1.15, 1.51), 43% in Q3 (OR: 1.43, 95% CI: 1.25, 1.64), and 39% in Q4 (OR: 1.39, 95% CI: 1.19, 1.62). Results were attenuated but still significant in our fully adjusted model 4, with higher quartiles compared with quartile one associated with a 25% greater odds of diabetes in Q2 (OR: 1.25, 95% CI: 1.07, 1.46), and 30% in Q3 (OR: 1.30, 95% CI: 1.12, 1.52). Age adjusted prevalence of cardiometabolic indicators among U.S. adults aged 50 or older by quartile of ultra-processed food intake, NHANES 2001–2018. *Indicates prevalence differs significantly (p < .05) from the reference category (quartile 1). Associations Between Ultra-processed Food Intake With Cardiometabolic Indicators Among U.S. Adults Age 50 or Older, NHANES 2001–2018. Model 1 is adjusted for age group, sex, and race; model 2 is additionally adjusted for: education, income, marital status, and insurance status; model 3 is additionally adjusted for: smoking, physical activity, and alcohol consumption; model 4 is additionally adjusted for: BMI and SBP (for the diabetes model only).
We found a significant interaction between UPF consumption and race/ethnicity for obesity only (p = .02), leading to race/ethnicity stratified models (Figure 3) [Insert Figure 3]. Results were strongest for non-Hispanic Asian adults. For example, in our fully adjusted model 3, compared to the first quartile of UPF consumption, higher quartiles were associated with a 180% greater odds of obesity in Q3 (OR: 2.80, 95% CI: 1.51, 5.20) and 115% greater odds in Q4 (OR: 2.15, 95% CI: 1.04, 4.45). Results were consistent (and even stronger) in a sensitivity analysis using the BMI threshold of 27.5 to define obesity for non-Hispanic Asian adults (not shown). For example, in our fully adjusted model 3, compared to the first quartile of UPF consumption, higher quartiles were associated with a 226% greater odds of obesity (BMI ≥27.5) in Q3 (OR: 3.26, 95% CI: 1.90, 5.58) and 207% greater odds in Q4 (OR: 3.07, 95% CI: 1.85, 5.09). Associations between ultra-processed food intake with obesity stratified by race/ethnicity among U.S. adults aged 50 or older, NHANES 2001–2018.
Discussion
In a multi-ethnic population-based sample comprised of U.S. adults aged 50 or older, over half of caloric intake was attributed to UPFs. We found strong associations between UPF consumption and a higher prevalence of cardiometabolic conditions (obesity, high cholesterol, and diabetes) among all race/ethnic groups. These results highlight the importance of diet as a major contributor to cardiometabolic health—especially in older adults. Notably, in Asian Americans, the fastest growing ethnic minority population in the U.S. (Gordon et al., 2019; Pew Research Center, 2021), we found that the prevalence of high cholesterol was high (70%), while the prevalence of obesity (11%) and UPF consumption (36%) was lower compared with other race/ethnic groups. Despite this, the association between UPF consumption and obesity was strongest for Asian Americans using standard or Asian-American specific BMI cut-points. Specifically, in older Asian Americans, high (quartile 4) compared with low (quartile 1) UPF consumption was associated with a 115% greater odds of obesity. Asian Americans, who have been historically under-represented by the medical literature, may be particularly vulnerable to obesogenic effects of diets high in UPFs. Given the high degree of ethnic heterogeneity in this fast growing group (Gordon et al., 2019; Pew Research Center, 2021), future studies among disaggregated Asian American sub-groups are critical to corroborate our findings and to better understanding health and its determinants in this important population.
UPFs are defined as a combination of artificial ingredients, additives, and other substances, such as sweeteners, colorings, flavorings, and emulsifiers (Ares et al., 2016). These highly processed products contain a high content of sodium, sugars, oils, and unhealthy fats (Ares et al., 2016) that typically characterize Western diets (Pagliai et al., 2021). In fact, in our sample of older U.S. adults, a striking 54% of caloric intake was derived from UPFs. These data are consistent with other studies of Western populations showing that in the U.S., Canada, and Britain, UPF consumption accounts for 50%–60% of the energy content (Juul et al., 2018; Martínez Steele et al., 2016, 2017, 2019; Pagliai et al., 2021). This high degree of UPF consumption in the U.S. and in particular older adults is concerning given that poor diet has been implicated as the leading cause of poor health (Mozaffarian, 2016). In fact, suboptimal diet has consistently been shown to be associated with adverse health outcomes such as CVD, hypertension, overweight and obesity, metabolic syndrome, and cancer (Chen et al., 2020; Pagliai et al., 2021; Santos et al., 2020). In our study, we found strong associations between UPF consumption with obesity, high cholesterol, and diabetes among older adults. This is particularly troubling given that for older adults, CVD outcomes may be more proximal and that dietary modification, and in particular sugar restriction, is indicated for individuals with cardiometabolic conditions such as obesity or diabetes, which were common in our sample (Johnson et al., 2009; Mozaffarian, 2016; Raynor & Champagne, 2016; Yu et al., 2018).
While we found associations between UPF consumption and cardiometabolic outcomes among all older adults, there were notable differences, particularly for Asian Americans. Older Asian Americans, on average, consumed less UPFs than other groups. This may be explained by non-Western Asian diets, which have been characterized as being high in fruits, vegetables, and legumes, carbohydrates (whole grains, rice, and noodles), and low in fat and meat (Woo et al., 2001). Another potential explanation for this finding may be related to socio-economic status, and in particular educational attainment, which is inversely correlated with UPF consumption (Baraldi et al., 2018). Given that Asian Americans, in aggregate, have the highest level of educational attainment compared with all other race/ethnic groups, (Everett et al., 2011) the lower relative UPF consumption is not surprising. It remains under-explored whether UPF consumption in younger Asian Americans follows similar patterns. Further, we also found Asian American adults to have a substantially lower prevalence of obesity (even at a lower BMI threshold) compared with non-Hispanic White adults, non-Hispanic Black adults, and Hispanic adults. Yet, despite this lower obesity rate, the prevalence of hypertension and diabetes, which are driven by obesity (Poirier et al., 2006) were comparable or worse to that of non-Hispanics White adults, perhaps due to sarcopenia in older Asian persons (Wu et al., 2016). These data are consistent with prior reports showing that obesity driven conditions, such as diabetes (Poirier et al., 2006), may manifest at lower BMI thresholds for Asian Americans (Hsu et al., 2015; Jih et al., 2014; Mui et al., 2018). Finally, in our sample of older adults, we found that there was a significant interaction between UPF intake with race/ethnicity for obesity, with Asian Americans having substantially stronger associations compared with other racial/ethnic groups. As the fastest growing ethnic minority population in the U.S. (Gordon et al., 2019), these data are alarming. In fact, as the proportion of the US born Asian population grows, it is anticipated that UPF intake for Asian Americans will increase (Pachipala et al., 2022). Therefore, preservation of healthy dietary behaviors—including minimizing intakes of UPFs will be critical for the prevention of subsequent cardiometabolic disease.
Asian Americans are a heterogeneous group that includes people who hail from different ethnic backgrounds and cultures, speak different languages, and possess different sociodemographic characteristics (Gordon et al., 2019). Despite this heterogeneity, population-based data tend to present Asian Americans as a larger aggregated group (Holland & Palaniappan, 2012; Islam et al., 2010). Further, though Asian Americans are the fastest growing U.S. group (Pew Research Center, 2021), they often comprise a small percentage of study samples from pooled national survey data further limiting the ability to study disaggregated Asian American ethnic groups (Gordon et al., 2019; Holland & Palaniappan, 2012; Islam et al., 2010). Underscoring the urgent need for disaggregated data, several studies have reported differences in health behaviors and health outcomes according to Asian American ethnic group (Choi et al., 2013; Gordon et al., 2019; Hsu et al., 2015; Huang et al., 2013; Jih et al., 2014; Maxwell et al., 2012; Mui et al., 2017, 2018). Thus, the aggregation of Asian Americans in data collection and analysis fails to reflect differences in dietary and cultural practices, socioeconomic characteristics, and prevalence of disease, comorbidities, and mortality across different Asian American groups (Hastings et al., 2015; Holland & Palaniappan, 2012; Islam et al., 2010). Therefore, differences in social and behavioral patterns, including UPF consumption across Asian Americans groups are masked within the current study. This presents a major limitation to most studies focused on Asian Americans, including the current analysis for which Asian Americans constitute the racial/ethnic group with the smallest sample size. Yet, despite the lack of disaggregated data, it remains important to highlight national trends among diverse groups with as much specificity as is possible. Additionally, our study was limited by its cross-sectional design, and therefore we could not establish temporality. Further, our dietary indicator was assessed via a single 24 h dietary recall, which may introduce imprecision in our measure. However, we note that the use of the Automated Multiple-Pass Method during dietary interviews has been shown to reduce recall bias (Moshfegh et al., 2008), which would likely be non-differential and lead to more conservative estimates. Nonetheless, we found the association between UPF consumption and cardiometabolic indicators to be strong, even with the possible attenuation of this relationship.
The study also has notable strengths. To the best of our knowledge, this is the first study examining the association between UPF and several cardiometabolic indicators using the NOVA food classification in a nationally representative and non-institutionalized older adult U.S. population. The NOVA framework provides a reliable and valid assessment of diet (Chen et al., 2020; Santos et al., 2020) that is not culturally specific. For example, by focusing on level of food processing, rather than individual dietary components, we are better able to compare diet across different race or ethnic groups. Finally, the pooling of multiple survey years from NHANES allowed for a robust study sample of older U.S. adults and which included large enough sample sizes for Asian Americans (due to oversampling) to produce reliable results.
Conclusions
In summary, we found UPF consumption to be very common in U.S. adults age 50 or older and was associated with the presence of adverse cardiometabolic conditions. Our results underscore the importance of diet as an indicator of health, especially in older populations. Further, though we found associations between UPF consumption and adverse cardiometabolic health across all race/ethnic groups, our findings were particularly strong for older Asian Americans with respect to obesity. Given the lack of research among Asian Americans in the U.S. coupled with the lack of disaggregated data in this heterogeneous group, future studies in this population are critical to understanding Asian American health and its determinants.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Elfassy was supported by a grant from the Rutgers University Asian Resource Centers for Minority Aging Research Center under National Institutes of Health and the National Institute on Aging Grant P30-AG0059304 and is currently supported by the National Institutes of Health; and the National Institute on Minority Health and Health Disparities (K01MD014158). Dr. Yi is supported by grant numbers U54MD000538 from the National Institutes of Health (NIH) National Institute on Minority Health and Health Disparities, and R01HL141427 from the National Heart, Lung and Blood Institute. Manuscript contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.
