Cardiometabolic Health Status, Ethnicity and Health-Related Quality of Life (HRQoL) Disparities in an Adult Population: NutrIMDEA Observational Web-Based Study

Precision public health supported on online tools is increasingly emerging as a potential strategy to achieve health promotion and disease prevention. Our aim was to assess the relationships of sociodemographic variables, anthropometric data, dietary habits and lifestyle factors with health-related quality of life (HRQoL), cardiometabolic health status and ethnicity in an online recruited adult population (NutrIMDEA Study). NutrIMDEA Study is a web-based cross-sectional survey that included 17,333 adults. Self-reported sociodemographic characteristics, anthropometric data, clinical and family history of cardiometabolic illnesses, dietary habits, lifestyle factors and HRQoL features were collected. Diseased individuals showed significative poorer MedDiet and worse HRQoL than those in the healthy cardiometabolic status group (p < 0.05). In comparison, European/Caucasian individuals reported a significantly better HRQoL, higher MedDiet and HRQoL values compared with those of other ethnicities (p < 0.05). We obtained a total of 16.8% who reported poor/fair, 56.5% good and 26.6% very good/excellent HRQoL. Respondents with very good/excellent HRQoL showed lower BMI, greater adherence to a Mediterranean diet (MedDiet) and higher physical activity. The results suggest the presence of interactions between the mental and physical components of HRQoL with obesity, sedentarism and dietary intake, which were dependent on disease status and ethnicity. Online HRQoL assessment could contribute to wider implementation of precision public health strategies to promote health targeted interventions with policy implications to community health promotion.


Introduction
Public Health Nutrition (PHN) involves the study of the environment, sociodemographic characteristics, diet, lifestyle and health, which affects the design, implementation and evaluation of nutritional interventions at the community level in order to improve the health status of specific groups [1,2]. In the new era of availability of information and big data, the Precision Public Health (PPH) concept is emerging, based on applying "the right intervention at the right time, every time to the right population". In this way, considering ered to enroll participants with age over 18 years with internet access, with the only requirement that they understood Spanish to complete the survey and not necessarily being from a Spanish-speaking country. The questionnaire was freely online accessible at https://nutrimdea2020.questionpro.com/ (accessed on 8 July 2020). Sources of information were an open survey and a rewarded survey. The first one was advertised in different communication/media channels as Spanish national newspapers or radio programs, and in the second one, audiences were purchased by omnibus companies in Spanish-speaking countries. Responders that completed the open survey obtained a personalized report based on their habits and health. Self-reported answers from multiethnic participants from Spanish-speaking countries were analyzed, where the Spanish National Health Survey 2017 (SNHS 2017) was used as reference for comparisons [21]. All those individuals who showed interest to be part of the study were properly informed about all the procedures before they entered the study. The questionnaire was delivered after asking conditions to IMDEA-CEI and the external companies that performed the surveys, which confirmed that filling the questionnaires is a proof of acceptance to participate and contribute to the NutrIMDEA study with own anonymized data. In our case, a disclaimer was incorporated to the survey to inform about these matters.

Questionnaire and Measurements
The survey was based on the questionnaire of de Cuevillas et al. [22], where quality of life phenotypical and lifestyle factors (diet/physical activity) were recorded to categorize individuals with a nutritional quantitative score or nutrimeter according to their nutritional well-being in order to discriminate nutritypes.
The baseline questionnaire included sociodemographic data (age, sex and educational level/occupation), self-reported anthropometric data (weight and height), cardiometabolic diseases prevalence (obesity, diabetes, hypertension, dyslipidemia), family history of cardiometabolic diseases (obesity, diabetes, hypertension, dyslipidemia), dietary habits (Mediterranean Adherence Score, number of meals per day, snacking habit, servings of vegetables per day, servings of legumes per day, servings of fish per day), lifestyle (nap habit, physical activity, smoking status) and quality of life features (SF12 Health Survey Item 1) or calculated physical and mental component scores of SF12 Health Survey (PCS12/MCS12). All of these variables were self-reported by the participants or calculated from the responses.
Among sociodemographic data, the age was analyzed within three categories (18-40 years, 40-70 years and >70 years). Sex consisted of two categories (male and female), while the category "other" was avoided in the current analyses. Educational level was classified into two categories: high school or less (primary education or less, low or intermediate secondary education, higher secondary education) and more than high school (intermediate vocational education, higher vocational education or university). Occupation was defined in three main groups: unemployed/retired, worker and student.
Concerning anthropometric data, BMI was calculated using data on self-declared weight and height, and individuals were stratified according to their Body Mass Index (BMI). Cut-off points were established according to the World Health Organization (WHO) as normal weight (BMI < 25.0 kg/m 2 ), overweight (BMI 25.0-29.9 kg/m 2 ) or obesity BMI ≥ 30.0 kg/m 2 ) [23].
Cardiometabolic disease prevalence was self-reported by participants with the following question: "Have you been diagnosed or are you currently undergoing treatment for any of the following conditions?" The options were diabetes, high blood pressure, dyslipidemia and obesity, and possible answers were yes/no.
As nutritional quality estimation, the Mediterranean Adherence Diet Score was assessed by using the PREDIMED questionnaire, known as MEDAS-14 [24]. In addition, dietary habits such as number of meals a day, snacking habit and servings of vegetables, legumes and fish per week were recorded. In relation to lifestyle, nap habit, which is defined as a short period of sleep typically taken during daytime hours as an adjunct to the usual nocturnal sleep period, was categorized as yes/no. Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ) using the Spanish version [25]. Activities were categorized as light, moderate or intense based on the metabolic equivalent value (MET), which was converted to hours/week units. Smoking status was categorized as current or former smoker.
Quality of Life Features were assessed with Item 1 of SF12 Health Survey (In general, would you say your health is?), and physical and mental component scores of SF12 Health Survey (PCS12/MCS12), were computed, which ranged from 0 to 100. High scores indicated a better quality of life [26]. Participants were categorized into three groups (poor/fair health, good health and very good/excellent health) according to self-reported answer SF12 Health Survey Item 1 "In general, would you say your health is?" This item was transformed: poor and fair were pooled into the group of "poor/fair" HRQoL, good into the group "good" HRQoL and very good and excellent made up the "very good/excellent" group, a third category. Moreover, participants were stratified by cardiometabolic health status into two groups (healthy cardiometabolic status and diseased cardiometabolic status). A diseased group was assigned if a participant had one of these cardiometabolic diseases (obesity, diabetes, hypertension or dyslipidemia). The last stratification was performed by ethnicity into two groups (European/Caucasians and other ethnicities). Other ethnicities included Africans, Asians, Hispanic/Latinos, mestizos and other ethnicities.

Statistical Analysis
For the Spanish population, the survey showed a margin of error considered as the degree of error in results received from random sampling surveys [27,28] of 0.9% with a confidence level of 95% (46,940,000 population and 11,883 sample). For the Hispanic sample, we took into account the countries of those with +20 observations. We calculated the same 0.9% margin of error with a 95% confidence level (1,055,910,000 population and 11,883 sample).
Characteristics of the study sample were presented using descriptive statistics such as mean and standard deviation for continuous variables or proportions for categorical variables. Differences in sociodemographic data, dietary patterns and lifestyle features according to HRQoL, cardiometabolic health status and ethnicity were assessed using either chi-squared tests (χ 2 test), two-sided student's t-tests or one-way analysis of variance (ANOVA). Significance threshold for the obtained p-values was set to p < 0.05. All descriptive and statistical analyses were performed using the R programming software (version 3.6.0; R Foundation (RStudio, PBC, Boston, MA, USA)).

Sample Characteristics
Participants (n = 17,333) of the NutrIMDEA 2020 Study (Table 1) were mainly females (62.7%) and had a higher educational level than the ones included in the Spanish National Health Survey 2017 (SNHS 2017). Our study population had better cardiometabolic health with less prevalence of obesity, HBP, diabetes and dyslipidemia, fewer current smokers, better self-perception of their health, lower BMI, greater consumption of vegetables per day and no significant differences in light physical activity. The SNHS 2017 population was older, with 22% of participants over 70 years of age compared to 5% of this age range in the NutrIMDEA Study. Regarding occupation, the large percentage of retirees in the SNHS 2017 is also noteworthy, and in terms of education level, NutrIMDEA mostly reaches university studies. Results after categorizing the sample by cardiometabolic health status are reported ( Table 2). The diseased cardiometabolic group had at least one of the following cardiometabolic diseases: obesity, diabetes, hypertension or dyslipidemia. There were significant differences in all variables except for snaking habit, servings of legumes per day and moderate/intense and total physical activity. The analyzed sample had mostly healthy cardiometabolic status (71%), and the majority age group in diseased participants was 40-70 years and included more diseased women than men. Most of the sample was university collective and had a higher percentage of workers among the healthy participants (75.4% vs. 68.8%). The diseased group (29%) showed more family history of cardiometabolic diseases, included more current smokers and presented a worse HRQoL than the healthy group (  The main characteristics of the participants according to their ethnicity (European/ Caucasian and other ethnicities) are also displayed (Table 3). There were no significant differences in diabetes prevalence, family history of HBP, light physical activity and MCS12. European/Caucasian individuals included in the study were younger compared to other ethnicities, less obese (9.9% vs. 12.3% with p < 0.05), were less likely to be current smokers and had less family history of obesity and family history of dyslipidemia compared to other groups. Europeans/Caucasian reported eating more meals a day and having less snacking habit than other ethnicities. BMI mean and Total PA was higher in other ethnicities and showed significant differences with Europeans/Caucasians (Figure 1), while high rates of MEDAS-14 were achieved in such group. A higher PCS12 for the European group (53.7 (95% CI, 53.6-53.9) points vs. 53.0 (95% CI, 52.8-53.2) points, p < 0.05) was found, while no significant differences in MCS12 (43.9 (95% CI, 53.6-44.1) points in both groups, p = 0.817) where noted ( Figure 2).  Regarding HRQoL, 16.8% of the sample of NutrIMDEA Study reported a poor/fair HRQoL, 56.5% of the surveyed subjects reported good self-perception of health and 26.6% reported a very good/excellent HRQoL (Table 4). There were significant differences in all variables when they were categorized by Item 1 of SF12 Health Survey. The participants who reported a very good/excellent HRQoL were mostly 40-70 years old, had a higher educational level, were workers and had a lower prevalence of cardiometabolic diseases and less family history of cardiometabolic diseases. A higher percentage of responders reported very good/excellent HRQoL, never smoked and showed higher scores in MCS12 (46.8 (95% CI, 46.5-47.1) points vs. 39.2 (95% CI, 38.7-39.7) points p < 0.05) and PCS12 (56.9 (95% CI, 56.8-57.1) points vs. 45.2 (95% CI, 44.8-46.5) points in poor/fair HRQoL, p < 0.05). Moreover, they reported lower weight, less obesity, more consumption of vegetables per day, more servings of legumes per day and more total physical activity than poor/fair HRQoL responders.   As a measure of global health, MCS12 and PCS12 were assessed (Figure 2). MCS12 showed significative differences between the three categories HRQoL (poor/fair, good and very good/excellent) (

Discussion
Public health strategies have considered nation-wide surveys as NHANES (National Health and Nutrition Examination Survey) to assess the health and nutritional status of adults and children in the United States [29] or EHIS (European Health Interview Survey) that aims at measuring on a harmonized basis and comparability among Member States the health status, health determinants and access to health of the EU citizens [30]. In Spain, the Spanish National Health Survey (SNHS) is carried out by the Spanish Ministry of Health, Consumption and Social Welfare with the collaboration of the Spanish National Institute of Statistics (INE) and collects health information related to the resident population in Spain [21]. This approach is a five-year study that allows researchers to know health aspects from citizens at national and regional levels by collecting data to plan, implement and evaluate actions in health matters, which was used to analyze our web-based population as the closest customary standard. Global trends of main inputs and outcomes are illustrated (Figure 1). Thus, in HRQoL, there is a negative association with BMI and a positive association with MEDAS14 and total physical activity. Regarding diseased cardiometabolic status, it was established a direct association with BMI and a negative association with MEDAS14 and total physical activity, but in this last variable with statistically marginal differences (12.0 (95% CI, 11.4-12.1) hours/week vs. 11.7 (95% CI, 11.8-12.2) hours/week, p = 0.075). Regarding ethnic issues, other ethnicities showed a positive association with BMI, a negative association with MEDAS-14 and higher total physical activity than European/Caucasian respondents.

Discussion
Public health strategies have considered nation-wide surveys as NHANES (National Health and Nutrition Examination Survey) to assess the health and nutritional status of adults and children in the United States [29] or EHIS (European Health Interview Survey) that aims at measuring on a harmonized basis and comparability among Member States the health status, health determinants and access to health of the EU citizens [30]. In Spain, the Spanish National Health Survey (SNHS) is carried out by the Spanish Ministry of Health, Consumption and Social Welfare with the collaboration of the Spanish National Institute of Statistics (INE) and collects health information related to the resident population in Spain [21]. This approach is a five-year study that allows researchers to know health aspects from citizens at national and regional levels by collecting data to plan, implement and evaluate actions in health matters, which was used to analyze our web-based population as the closest customary standard.
In this sense, we compared the latest SNHS 2017 reference population survey with the NutrIMDEA Study sample, aiming to find potential similarities/differences on several health indicators and health determinants in order to contextualize the health situation of the study population based on two different models. A wider proportion from the SNHS 2017 participants were older than 70 years, with a higher percentage of retirees compared to the NutrIMDEA Study, which was not unexpected given the data origin: traditional collection method vs. online, respectively. Accordingly, as a consequence of age distribution, most variables evidenced better health outcomes in the NutrIMDEA Study. This fact may be also be because the NutrIMDEA Study, as an online directed health survey, may imply that the participants are more aware and interested in health-related issues, as seen in previous web-based studies such as Food4Me [31]. In any case, the online health data collection method has been validated [31], where important variables such as reported and collected BMI are correlated [32]. In this direction, we can confirm that SNHS 2017 and NutrIMDEA Study involve noncomparable populations due to the data collection method, different periods of data inclusion (prior to 2017 and in 2020, respectively) and assumably different internet knowledge in an older Spanish population, which may explain the outcome heterogeneity and some input diversity.
Sociodemographic features, dietary habits and lifestyle factors are related to health and quality of life [33]. Our population showed expected trends concerning the categories sex, education, occupation, anthropometric data and lifestyle (nap habits, physical activity and smoking status), as previously published in comparable populations [11,34] Participants who reported a high self-referred HRQoL value showed better health outcomes, as found in previous studies in Spain [35,36]. Thus, dietary habits were more balanced as HRQoL increases and higher consumption of vegetables and fish were associated with superior HRQoL as found in a previous study by Sayón-Orea et al. [37]. Moreover, obesity was significantly related to a lower HRQoL [38]. There is an increasing trend of PCS12 and MC12 values as the HRQoL category rises. This direction also remains in the MCS12 value on both healthy/diseased groups. However, this tendency is reversed in the PCS12 score, which could be explained by a reduced physical activity. There were no significant differences in MCS12 between European/Caucasians and other ethnicities. An independent effect of BMI on HRQoL among racial and ethnic groups was previously identified in an investigation involving Blacks, Whites and Hispanics [39].
Cardiometabolic health status depends on sociodemographic characteristics, dietary habits and lifestyle [11,40]. In this sense, metabolic syndrome has been associated with poor overall health and poor physical mental health status in American adults [41], which is consistent with our results. Furthermore, a higher education level among the healthy group was detected, and twice, unemployed/retired were found in the diseased group, which may be due to older aged subjects in this category. Data collection has some influence since hypertension rates in the current cohort were lower than data provided by other studies [42,43]. Indeed, in an online population, it may be found that only those with antihypertension treatment may respond as suffering cardiometabolic manifestation, which also applies for the diabetes, obesity and dyslipidemia questions/answers. The higher percentage of family history of cardiometabolic diseases in the diseased group can be explained by genetic factors and worse inherited and shared domestic habits [44].
According to previous studies, normal weight participants had a lower probability to report poor HRQoL than overweight/obese participants with any chronic disease [37]. It was unexpected to not find significant differences in moderate, intense and total physical activity despite previous references in the literature [37]. Perhaps this finding may be partly due to the data collection period during the COVID-19 pandemic, which may have led changes in lifestyle.
In Europe, previous ethnicity research has reported inconsistent findings regarding health status and HRQoL [45]. Our study adds to the current evidence some multifactorial findings since more HBP prevalence in other ethnicities was found and higher normal weight prevalence in European group. On the other hand, the "Healthy migrant effect" is a situation where migrants often have a better health status than the remaining population in the native country, but not concerning BMI data, as contrasted in another study [46]. This same study showed other ethnicities had the highest odds of HRQoL, with worse health than White British adults [46]. In contrast, a higher PCS12 score in Europeans/Caucasians than in other ethnicities was found, while MCS12 was similar in both groups.
A negative correlation among very good/excellent HRQoL and high BMI was featured, which is consistent with previous research [11,47]. Moreover, a positive association among very good/excellent HRQoL with MEDAS14 was identified, which is in agreement with data from the PREDIMED-PLUS trial, where MEDAS14 has been associated with better self-reported quality of life and attributed to the fact that Mediterranean diet reduces the risk of chronic diseases and is protective against cardiovascular diseases [48,49]. In addition, total physical activity levels are associated with better HRQoL and dietary patterns, as reported in a previous studies [50,51]. These observations confirm the suitability of a web-based collection and provide support for actions focused on improving lifestyle and control obesity in the population based on online captured information. Previous studies have noted the protective effect of a high-quality diet on self-perceived health [11,47,48] as well as similar results observed in another European cohort that showed that higher adherence to a Mediterranean diet was associated with higher HRQoL, as assessed by other means [52]. In other studies, lower HRQoL scores were reported by people who were older, with elementary education or less and with fair or poor subjective health status, factors which are affected by ethnicity [53].
An overview of systematic reviews evidenced that sedentary behaviors are related with adverse health outcomes, which can benefit from computer and internet use [54]. Furthermore, an online approach may facilitate counseling in adults without known cardiovascular disease [55]. The comparison of data from subjects with different cardiometabolic status in our cohort demonstrated the reliability of the online instrument concerning the collection of anthropometrics, dietary habits, lifestyle and quality of life data. However, despite the plausibleness of the outcomes, this cross-sectional research was not able to address the issue if such relationships are causal for disease or are a consequence of a poor cardiometabolic health.
Concomitantly, the analyses of the survey responses concerning ethnic backgrounds revealed differences in personal and phenotypical features, as previously reported in Canadian immigrants and native-born individuals [56], which may be of help to implement future culturally based interventions for chronic disease management, as also reviewed in Chinese-Americans [57]. As other research has evidenced, a sense of belonging contributes positively to subjective well-being, particularly in countries with high ethnic heterogeneity [58]. Similarly, some studies have tried to prove that Europeans have a particularly negative relationship between the sense of pride and belonging concerning the compo-nent of identification, which may play a role in more homogeneous cultures, reducing the well-being status [59].
These findings are important for research and clinical applications in adult populations, because it may help to identify at-risk populations for nutritional screening by self-reported quality of life, self-reported cardiometabolic status and by ethnicity. Moreover, findings from this study can support targeted efforts to guide precision public health and to make aware healthcare professionals. From this approach, a prediction model for risk stratification is to identify potential risk groups to screening cardiometabolic health status and analyze the information on health and nutritional status to personalize individualized nutritional advice.

Strengths and Limitations
Some methodological considerations concerning this research are required since the self-reported online data collection comprise a free open survey and rewarded survey, which should be taken into account when interpreting the data. In any case, both web-based surveys have been proven to be useful, easy and sustainable compared with traditional presential surveys. Indeed, given the design of this study, it cannot establish causal inferences from the associations found but can generate hypothesis that could be assessed in future prospective trials. Therefore, the current results need to be interpreted with caution. The descriptive analyses included comparisons concerning age, sex and BMI to understand the role of these potential confounders in relation to quality of life, cardiometabolic status and ethnicity. Furthermore, current data were collected during the COVID-19 pandemic, so there might be several influences on habits and lifestyle and the inherent limits of prediction for individuals. The online collection and the need that participants have some web knowledge as internet users need to be also accounted for. Thus, some surveys such as the Nurses Health study [60] and the Health Professionals Follow-up [61] have reached useful conclusions, based on face-to-face interviews, while online collection has been validated in other surveys such as the SUN cohort [62] and Food4Me [9], supporting the validity or our findings involving an online recruited population.
A major strength of this study is the relatively large sample size, which backs webbased population representativity as well as an easy and reliable web-based data collection, where dietary behavior records can be as effective as a conventional approach [32]. Another important strength is the wide number and diversity of variables collected in the survey with validated tools that allowed a precision personalized analysis, while the important number of participants and origin diversity contribute to the representativity of the screened sample population.

Conclusions
In conclusion, in the NutrIMDEA web-based population, we found that age, high education, high PCS12 and MCS12, high MEDAS-14 and a healthy lifestyle were positively associated with very good/excellent HRQoL. The relationships between life quality and several lifestyle modifiable factors were found to be different in healthy and diseased groups. Interestingly, analyses by ethnicity concerning online self-reported health information is a pioneer manner to study sociodemographic and quality of life interactions, which revealed inequalities and inadequacies in wellbeing depending on ethnic backgrounds. Institutional Review Board Statement: The NUTRiMDEA survey included a disclaimer informing the participants about this information and acquiescent that submitting the questionnaire constitutes acceptance of the use of their anonymity data for scientific purposes. This statement was positively accepted by the IMDEA CEI and the companies in charge of the rewarded surveys.

Informed Consent Statement: Not required.
Data Availability Statement: Not required.