**1. Introduction**

The frailty phenotype is a multifactorial syndrome associated with aging, characterized by unintentional weight loss, self-reported exhaustion, muscular weakness, slow walking speed, and low physical activity [1]. Physical frailty is also recognized as a risk factor for mortality, increased morbidity [2], malnutrition [2], and falls [3,4]. If not managed, frailty can lead to disability and dependency [5,6], and become a burden to the individual, caregivers, and public health authorities. This process, which moves from robustness to frailty, disability, then dependency, is preventable, and could even improve in some aspects, if addressed in the early stages (i.e., in the prefrail state) [7–10].

Frailty, being a multifactorial condition, is related to several sociodemographic-, lifestyle-, and health-related factors. Associations were found between frailty risk and marital status [11], education [12], depression [13], polypharmacy [14], and nutritional status [2,15].

**Citation:** Yaghi, N.; Yaghi, C.; Abifadel, M.; Boulos, C.; Feart, C. Dietary Patterns and Risk Factors of Frailty in Lebanese Older Adults. *Nutrients* **2021**, *13*, 2188. https:// doi.org/10.3390/nu13072188

Academic Editor: Rosa Casas

Received: 16 May 2021 Accepted: 22 June 2021 Published: 25 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Gender discrepancies have also been reported regarding the risk of frailty. Although women tend to live longer, their health status is poorer than men and biological and socio-behavioral factors may contribute to a higher frailty predisposition [16,17]. In many studies, the prevalence and incidence of frailty in women was found to be greater than in men [18].

Among other risk factors for frailty, anthropometric, nutritional, and dietary risk factors are often mentioned [17]. Frailty is also closely related to body composition and nutritional status. It is well established that malnutrition and the risk of malnutrition, characterized by low mini-nutritional assessment (MNA) scores, increase the risk of frailty in older adults [19,20]. BMI was also associated with frailty in a U shape trend, with both low and high BMI being associated with higher incidence of frailty [21–25]. Furthermore, central obesity characterized by a high waist circumference (WC), has also been linked to frailty risk, and recent studies even reported that the association between BMI and frailty was in part mediated by waist to height ratio (WHTR) [26–28].

According to several authors, current recommended dietary allowances (RDA) for protein intake at 0.8 g/kg (BW) appears to be insufficient to prevent muscle loss with aging [29]. The protein intake that showed a better muscle preservation and was linked to lower frailty risk was proposed at protein intakes between 1 and 1.5 g/kg BW [29–32].

Apart from the nutritional status, dietary patterns were also linked with risk of frailty in older adults. Rashidi et al. showed that people adopting a healthy dietary pattern, characterized by a high consumption of fruits, vegetables, and whole grains, had 31% less chance of becoming frail [33]. Pooled results showed that a greater adherence to a Mediterranean diet, evaluated by the Mediterranean Diet Score (MDS), was associated with a significantly lower risk of frailty compared to poorer adherence [34,35]. Similar findings from the Hellenic Longitudinal Investigation on Aging and Diet (HELIAD) showed that each additional unit in the MDS was associated with a 5–7% decrease in the odds of frailty, depending on the tools used in the evaluation of frailty [36]. In Spain, a higher prevalence of frailty was observed in older adults adopting an unhealthy dietary pattern (DP) compared to individuals adopting a healthy dietary pattern [37]. In England, a longitudinal study showed that a dietary pattern characterized by a high fat intake and low fiber intake was associated with a higher risk of frailty in men. Following a prudent diet and having a higher adherence to a Mediterranean dietary pattern was also found to decrease the likelihood of frailty [38]. The Rotterdam cohort study showed that a higher adherence to national dietary guidelines was associated with lower risk of frailty over time, and a traditional dietary pattern characterized by a high consumption of legumes, eggs, and savory snacks was the only one protecting against frailty; health-conscious dietary patterns and high meat patterns failed to be associated with frailty [39]. Results from the 12-year follow-up, three-city study, showed that men following the dietary pattern characterized by a "pasta" pattern, and women adopting the "biscuits and snacking" pattern had a significantly higher risk of frailty compared with those following the "healthy" pattern (characterized by higher fish intake in men and higher fruits and vegetables intake in women) [40]. In the multicentric, 1 year NU-AGE interventional study, the administration of a Mediterranean diet for a 1year period, by modulating the microbiome ecosystem was linked with lower frailty incidence in older adults [41].

Several studies showed that prevalence of frailty and pre-frailty in low- to middleincome countries (LMIC) was higher than in high income countries (HIC) [18,42]. In Lebanon, one of the countries bordering the Mediterranean basin with an aging population, few studies have ye<sup>t</sup> been interested in the frailty status among its elderly population. In 2016, the prevalence of frailty (including pre-frailty) was estimated to 66.8% in rural living older individuals, aged 65 years and more, with a higher prevalence of frailty among individuals considered undernourished or at risk of malnutrition according to MNA [19,43]. However, no data is available on the association between nutritional status, dietary patterns, and frailty in this population.

The aims of the present study were (i) to determine the prevalence of frailty and its covariates among older Lebanese adults, and (ii) to explore the association between dietary patterns and frailty among these individuals.

#### **2. Materials and Methods**

#### *2.1. Study Design and Participants*

We conducted a national, cross-sectional study in seven of the eight governorates of Lebanon. Recruitment of participants and data collection were carried out in collaboration with the Ministry of Social Affairs (MOSA) through 77 medico-social centers serving low to middle class income families, from October 2017 to October 2019. The sample was weighed according to the proportion of older adults over 60 years in each governorate and was randomly selected based on the sample previously drawn for the validation of the Arabic version of Mini-Mental State Examination (MMSE) [44,45].

As shown in Figure 1, a sample of 600 individuals was initially targeted. Individuals were included in the present study if they were aged 60 years and above, community dwelling, and attending the MOSA socio-medical centers for medical care and social assistance. Non-inclusion criteria included artificial feeding, total dependency, major hearing and visual impairment, active cancer disease, end stage kidney disease with hemodialysis, and advanced liver disease, as diagnosed by the center medical team. Once contact was initiated, 159 individuals were either unable to be reached, refused to participate, or were deceased after randomization, and six were excluded because of newly discovered cancer or major disability. Data were finally retained for 401 participants, of whom, 352 (88%) had a valid Food Frequency Questionnaire (FFQ).

**Figure 1.** Study Flow chart.

Participants were contacted (via phone) by social workers or nurses working in the MOSA centers and were invited to attend the MOSA center. Information was collected by 10 dietitians living in the area surrounding the MOSA centers and who were familiar with the dietary habits of the local inhabitants. All investigators underwent several training sessions before starting the survey. To decrease investigator bias, all filled questionnaires were reviewed by the principal and field investigators before data entry. Each person was interviewed at the center near his/her home, and for participants unable to attend, the interview was performed by the research team at home. The interview with each participant or caretaker lasted around 30–45 min. Depending on literacy, the cognitive functions of each participant were assessed either through the Mini-Mental State Examination (MMSE) [44] or "Test des Neufs Images" (TNI) [46]. In case of significant cognitive decline, then the accompanying person was asked to fill the questionnaire on behalf of the participant.

The study was conducted in accordance with the Declaration of Helsinki, and the protocol of the study was approved by the Ethics Committee of Saint Joseph University of Beirut (USJ 2016-99). Written informed consent was signed by each participant, prior to completion of the interview.

#### *2.2. Data Collected*

#### 2.2.1. Frailty and FRAIL Scale

Frail status of the participants was assessed based on the five criteria of FRAIL scale including fatigue, resistance, ambulation, illness, and loss of weight [47]. Fatigue was measured by asking respondents how much time during the past 4 weeks they felt tired, with responses of "all of the time" or "most of the time" scoring 1 point. Resistance was assessed by asking participants if they had any difficulty walking up 10 steps alone without resting and without aids, and ambulation, by asking if they had any difficulty walking several hundred yards alone and without aids; "yes" responses were each scored as 1 point. Illness was scored 1 for respondents who reported 5 or more illnesses out of the 11 following illnesses: heart attack, congestive heart failure, angina, asthma, arthritis, stroke, kidney disease, hypertension, diabetes, cancer, and chronic lung disease. Weight loss was scored 1 for respondents with a self-reported weight decline of 5% or greater within the past 12 months. Frail scale scores range from 0–5 and represent frail (score = 3–5), pre-frail (score = 1–2), and robust (score = 0) health status [47]. For data analysis, the studied sample was further categorized to frail versus non-frail (robust or prefrail).

#### 2.2.2. Sociodemographic and Health-Related Data

Information regarding age and gender, living conditions, marital status, and economic situation were collected. Participants were asked to state whether they live alone or with any relative, spouse, family members, or friends. Educational level and literacy were evaluated and classified into two categories: >7 years and ≤7 years of education.

Number and type of diseases and health conditions were reported by each participant or its caregiver: cardiovascular diseases, diabetes, hypertension, heart failure and arrhythmia, renal disease, thyroid disorder, gastrointestinal diseases, anemia, osteoarthritis and arthritis, and osteoporosis. The presence of these diseases was ascertained based on: (a) self-reports linked to previous diagnoses, and/or (b) available medical records kept in the MOSA center, and/or (c) available medical prescriptions. The number of diseases was summed and categorized to multimorbidity defined as the simultaneous presence of two or more diseases in the same individual. Hearing or visual impairment, poor oral health, and sleep disorders were grouped as age-related conditions. Results were then dichotomized based on the presence of two or more of these age-related conditions. Current medications were recorded by caregiver, and polypharmacy was defined as the regular use of six or more prescription medications daily.

Since cognitive decline might influence reporting in the elderly population, cognitive impairment was evaluated depending on literacy, using population specific cut-offs based on age, educational level and sex, of MMSE for literate participants, and a score of TNI, Total Recall ≤ 9, for illiterate participants [44,46,48].

#### 2.2.3. Anthropometric and Functional Measurements

Data included height, weight, body mass index (BMI), waist circumference (WC), and waist to height ratio (WHR). BMI was categorized according to the Lipchitz classification (<22, 22–27, and >27 kg/m2) [49], Hand grip strength (HGS) was measured using the handgrip electronic dynamometer (Camry EH101). Handgrip strength was performed 3 times and the average of the three measurements was the reported result. We then used the BMI and gender specific cut-off values of HGS as classified by Fried phenotype, to classify as low or good HGS measurements [50].

Physical activity was evaluated using the Rapid Assessment of Physical Activity (RAPA) 2 scale. The nine-item questionnaire covered all ranges of activity, from sedentary to regular and vigorous physical activity in addition to strength training and flexibility adapted to the elderly [51]. Respondent's score was initially categorized into five levels of physical activity. For data analysis, these levels were then classified into three categories: sedentary, regular active and optimal active. Responses to the strength training and flexibility items were scored separately [51].

#### 2.2.4. Nutritional and Dietary Data

Nutritional status was estimated using the Mini-Nutritional Assessment Short Form (MNA-SF) [52], classifying individuals into three levels: normal nutritional status (score above 12), at risk of malnutrition (scores 8–11), and malnutrition (scores below 8). Participants were then classified into two categories: the first category named "poor nutritional status" included participants that were considered malnourished and at risk of malnutrition by the MNA-SF, and the second category named "normal nutritional status" included participants who had a normal nutritional status according to MNA-SF.

Dietary assessment and food consumption was measured using a population based FFQ, reporting the consumption of a list of 90 food items (validation in progress of publication) and representing all food groups, consumed the previous year: bread and cereals, milk and dairy, vegetables and fruits, meat, poultry and fish, fats and oils, sweets and desserts, and non-alcoholic beverages. Consumption of these items was reported as daily, weekly, or monthly, as usual portion size. A manual illustrating the usual servings and portions of foods listed in the FFQ was developed for the study to help investigators and participants better estimate quantities consumed. The portions consumed were then translated into daily consumption in grams. Daily consumption was later analyzed by Nutrilog software (Nutrilog, version 3.20, France) to extract daily nutrient intake. To extract dietary patterns, the 90 foods listed in the FFQ were then grouped into 20 predefined categories based on similarities in nutrient composition and consumption characteristics (Supplemental Material: Table S1). These categories were then entered in the K-mean cluster analysis to determine the dietary patterns of our population.

For all the categories of food, except for sweets, reported portions used to estimate consumption were based on standard portions adopted in the dietary guidelines 2015–2020 [53]. Sweetened soft drinks were added to sugars and jams after sugar content estimation was made. As for sweets and desserts, usual serving size was adopted for cluster analysis.

#### *2.3. Statistical Analysis*

Statistical analysis was performed using the SPSS program version 21.0. Difference between genders and frailty status for sociodemographic, nutritional, dietary, health-related, and anthropometric data, were compared using the non-parametric tests, Mann–Whitney and Kruskal–Wallis for numeric variables that are not normally distributed, and chi-square test for categorical variables. K-means clustering was used to regroup participants with similar dietary patterns. Differences in food intake between dietary patterns was calculated by Kruskal–Wallis test. Cut-offs for age, polypharmacy, multimorbidity, age-related

conditions, and waist to height ratio (WHTR) were determined through classification tree and used for subsequent multivariate analysis. In the multivariate analyses, we proceeded to several models of binary logistic regression using dichotomized frailty variable as the dependent variable. Odds-ratios with 95% confidence intervals were calculated. The main explanatory variables reaching significance level identified in each model were added to the following model. Age category, marital status, education level, and living conditions were entered first (model 1). From this model, we retained covariates significantly associated with frailty and then added anthropometric variables including MNA-SF, BMI, and waist to height ratio categories (model 2). From model 2, we retained all covariates significantly associated with frailty and then added health-related covariates (multimorbidity, polypharmacy, and age-related conditions (model 3). From model 3, we retained all covariates significantly associated with frailty and finally added food dietary patterns (model 4). Each Model was rerun with the food patterns groups in order to adjust for the variables identified respectively in each model. The multivariate analysis was rerun for each gender. The level of significance was fixed at a *p* = 0.05 for all analyses.
