JAMDA 15 (2014) 899e903
JAMDA journal homepage: www.jamda.com
Original Study
Mediterranean Diet and Risk of Frailty in Community-Dwelling Older Adults Luz M. León-Muñoz PhD a, b, Pilar Guallar-Castillón MD a, b, Esther López-García PhD a, b, Fernando Rodríguez-Artalejo MD a, b, * a b
CIBER of Epidemiology and Public Health (CIBERESP), Madrid, Spain Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autónoma de Madrid/IdiPaz, Madrid, Spain
a b s t r a c t Keywords: Mediterranean diet frailty older adults Spain
Background and Objective: Low intake of certain micronutrients and protein has been associated with higher risk of frailty. However, very few studies have assessed the effect of global dietary patterns on frailty. This study examined the association between adherence to the Mediterranean diet (MD) and the risk of frailty in older adults. Design, Setting, and Participants: Prospective cohort study with 1815 community-dwelling individuals aged 60 years recruited in 2008e2010 in Spain. Measurements: At baseline, the degree of MD adherence was measured with the Mediterranean Diet Adherence Screener (MEDAS) score and the Mediterranean Diet Score, also known as the Trichopoulou index. In 2012, individuals were reassessed to detect incident frailty, defined as having at least 3 of the following criteria: exhaustion, muscle weakness, low physical activity, slow walking speed, and weight loss. The study associations were summarized with odds ratios (OR) and their 95% confidence interval (CI) obtained from logistic regression, with adjustment for the main confounders. Results: Over a mean follow-up of 3.5 years, 137 persons with incident frailty were identified. Compared with individuals in the lowest tertile of the MEDAS score (lowest MD adherence), the OR (95% CI) of frailty was 0.85 (0.54e1.36) in those in the second tertile, and 0.65 (0.40e1.04; P for trend ¼ .07) in the third tertile. Corresponding figures for the Mediterranean Diet Score were 0.59 (0.37e0.95) and 0.48 (0.30e0.77; P for trend ¼ .002). Being in the highest tertile of MEDAS was associated with reduced risk of slow walking (OR 0.53; 95% CI 0.35e0.79) and of weight loss (OR 0.53; 95% CI 0.36e0.80). Lastly, the risk of frailty was inversely associated with consumption of fish (OR 0.66; 95% CI 0.45e0.97) and fruit (OR 0.59; 95% CI 0.39e0.91). Conclusions: Among community-dwelling older adults, an increasing adherence to the MD was associated with decreasing risk of frailty. Ó 2014 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Frailty is a medical syndrome with multiple causes and contributors that is characterized by diminished strength and endurance, and reduced physiological function, which increases an individual’s vulnerability of developing dependence and/or of death.1,2 Given the accelerated aging of the population in most countries and the expected increase in the number of individuals with disability and
This work has been supported by FIS grants 11/01379, 12/1166, and 13/00288 (Ministry of Health of Spain) and by FP7-HEALTH-2012-Proposal No: 305483-2 (FRAILOMIC Initiative). All authors declare that they have no conflict of interest. * Address correspondence to Fernando Rodríguez-Artalejo, MD, Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autónoma de Madrid, Arzobispo Morcillo s/n, 28029 Madrid, Spain. E-mail address: [email protected] (F. Rodríguez-Artalejo).
dependence, prevention of frailty may be a key policy to reduce the social and healthcare burden associated with disability. There is evidence that poor nutrition is related to frailty.3 Specifically, low intake of certain micronutrients and protein,4e6 or an inadequate distribution of protein intake,7 have been associated with higher risk of frailty. However, very few studies have assessed the effect of dietary patterns on frailty. Instead of looking at individual nutrients or foods, pattern analysis examines the effects of overall diet. Conceptually, dietary patterns represent a broader picture of food and nutrient consumption and, thus, may be more predictive of disease risk than individual foods or nutrients.8 To our knowledge, only 3 studies have explored the association between dietary patterns and frailty.6,9,10 One study reported that a higher score in the Diet Quality Index Revised was inversely associated with prevalent and future frailty status in a cohort of older men.6
1525-8610/$ - see front matter Ó 2014 AMDA e The Society for Post-Acute and Long-Term Care Medicine. http://dx.doi.org/10.1016/j.jamda.2014.06.013
900
L.M. León-Muñoz et al. / JAMDA 15 (2014) 899e903
The other 2 studies focused on the Mediterranean Diet (MD) pattern9,10; Bollwein et al9 found a cross-sectional association, and Talegawkar et al10 a prospective association between adherence to the MD and lower risk of frailty. In both studies, however, the MD was defined from sample-specific scores; thus, only the relative but not the absolute effect of MD was assessed, and the results of these studies are difficult to compare across populations. Lastly, no information was available on the specific components of the MD, which were more closely related to frailty. Accordingly, we examined the prospective association between the adherence to MD, as assessed by normative and sample-specific scores, and the risk of frailty in older adults from the general population in Spain. Moreover, we assessed the association between the individual components of the MD and incident frailty. Methods Study Design and Population Data were taken from the Seniors-ENRICA (Estudio de Nutrición y Riesgo Cardiovascular) cohort. Briefly, the cohort was established in 2008e2010 with 2519 individuals selected through stratified random sampling from the population aged 60 years and older in Spain.11 At baseline, data were collected in 3 stages. First, a phone interview to obtain information on health status, lifestyle, morbidity, and health services use; second, a home visit to collect blood and urine samples; and third, another home visit to perform a physical examination and to record habitual diet and prescribed medication.11 In 2012, a second wave of data collection was performed; in total, 2037 participants provided updated information for the phone interview, the physical examination, and diet. The sociodemographic, lifestyle, and clinical characteristics at baseline were similar in those who provided updated information at 2012 and those who did not. Study participants gave written informed consent. The Clinical Research Ethics Committee of the ‘La Paz’ University Hospital in Madrid approved both the baseline and follow-up studies.
5e25 g/day in women. The range in this score is 0 (lowest) to 9 (highest MD adherence). Frailty We used a slight modification of the operational definition of frailty developed by Fried and colleagues in the Cardiovascular Health Study.17 Specifically, frailty was defined as having 3 or more of the following 5 criteria as follows. (1) Exhaustion was evaluated as any of the following responses to 2 questions taken from the Center for Epidemiologic Studies Depression Scale18: “I felt that anything I did was a big effort” and “I felt that I could not keep on doing things” at least 3 to 4 days a week.” (2) Weakness, defined as the lowest quintile in the Cardiovascular Health Study of maximum strength on the dominant hand adjusted for sex and body mass index (BMI). Strength was measured with a Jamar dynamometer, and we selected the highest value in 2 consecutive measures.19,20 (3) Low physical activity, defined as walking 2.5 hours/week in men and 2 hours/week in women. (4) Slow walking speed, defined as the lowest quintile in our study sample for the three-meter walking speed test, adjusted for sex and height.20,21 (5) Weight loss, defined as involuntary loss of 1 kg of body weight in the preceding 3 months. Other Variables At baseline we collected information on sociodemographic variables, lifestyle, and diseases, which could be related to both diet and frailty. Specifically, study participants reported their sex, age, educational level, tobacco consumption, number of medications used, and energy intake (calculated with standard food composition tables). They also reported if they suffered from any of the following physician-diagnosed diseases: cardiovascular disease (myocardial infarction, stroke, heart failure), diabetes mellitus, cancer at any site, asthma or chronic bronchitis, osteomuscular disease (osteo-arthritis, arthritis, hip fracture), or depression requiring drug treatment. Lastly, weight and height were measured in standardized conditions,22 and the BMI was calculated as weight in kg divided by square height in m. Statistical Analysis
Study Variables Diet Food consumption was assessed with a validated computerized diet history developed from that used in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort study in Spain.12,13 The diet history collected data on the consumption of 880 foods in the preceding year and used a set of photographs to help in estimating the amount of intake. Adherence to the MD was summarized with 2 instruments: the Mediterranean Diet Adherence Screener (MEDAS)14 and the Mediterranean Diet Score (MDS),15 also known as the Trichopoulou index. The MEDAS was developed to assess compliance with the dietary intervention of the Prevención con Dieta Mediterránea (PREDIMED) study, a clinical trial of the effect of the MD on primary cardiovascular prevention.16 MEDAS consists of 12 items with targets for food consumption and another 2 items with targets for food intake habits characteristic of the MD in Spain. One point is given for each target achieved. The total MEDAS score ranged from 0 to 14, with a higher score indicating better MD adherence. In the MDS, the intake of vegetables, legumes, fruits and nuts, grains, and fish is considered beneficial, thus, a value of 1 is assigned to consumption above the sex-specific median in the study sample; in contrast, intake of red meat and poultry, and dairy products is considered detrimental, and a value of 0 is assigned to consumption above the median. Moderate alcohol consumption is also considered beneficial: 1 point is assigned for intake of 10e50 g/day in men and
Among the 2037 study participants, we excluded 112 who lacked complete data on diet and frailty, and 31 with missing information on potential confounders. Of the remaining 1894 persons, we excluded 79 who had frailty at baseline. Therefore, the analyses were conducted with 1815 individuals. The association between MD adherence and risk of frailty was summarized with odds ratio (OR) and 95% confidence interval (CI), obtained from logistic regression. MD adherence, assessed with the MEDAS or the MDS, was classified into tertiles and the lowest tertile was used as reference. Two logistic models were built: the first one adjusted for sex, age, and education; and the second one additionally adjusted for the rest of variables described above. The reason for building 2 models is that model 2 may be considered as overadjusted, since BMI, chronic diseases, and the number of drug treatments could be in the pathway from diet to frailty. However, as will be shown below, both models yielded similar results. We ran several sensitivity analyses to assess the robustness of the main results. First, we re-estimated the study associations using a >3 kg-loss in the preceding 3 months as the weight loss criterion. This may reflect a severe frail status more accurately than simply losing 1 kg. Second, given that the main aim of the study was to evaluate the association between diet and frailty, we repeated the analysis excluding the weight loss criteria from the definition of frailty.10 Thus, in this analysis frailty was defined as having 2 out of the 4 remaining Fried criteria. Third, we replicated the analyses using the lowest quintile of grip strength in our population to define weakness.
L.M. León-Muñoz et al. / JAMDA 15 (2014) 899e903
Fourth, we ran analyses after excluding those who reported severe or substantial difficulty in chewing or eating, and those with cognitive impairment (score 22 on the Mini-Mental State Examination23), at the end of follow-up. Finally, the study association was assessed after excluding individuals with severe diagnosed disease (cardiovascular disease, diabetes, cancer, chronic lung disease, or depression requiring treatment). Finally, we used the same type of modeling to assess the association between MD adherence and the onset of each frailty criterion in robust individuals (free of all 5 criteria) at baseline. Likewise, we used logistic regression to study the association between each component of the MEDAS or MDS and the risk of frailty; these analyses were further adjusted for all components of the MEDAS or MDS except the component of interest. Statistical significance was set at 2-sided P < .05. The analyses were performed with Stata v 11.1 (StataCorp, College Station, TX). Results Among the 1815 study participants at baseline, the mean (range) adherence to the MD was 7.18 (1e13) using the MEDAS, and 4.93 (0e9) using the MDS. Supplementary Table 1 shows the sociodemographic and clinical characteristics of participants according to
901
MD adherence. Compared with those in the lowest tertile of the MEDAS score, those in the highest tertile were more frequently women and never smokers, and showed a lower frequency of current smoking, obesity, cardiovascular disease, diabetes and depression, and less energy intake. Individuals showed a similar distribution when MD adherence was defined with the MDS. Over a mean follow-up of 3.5 years, we identified 137 persons with incident frailty. Compared with individuals in the lowest tertile of the MEDAS score, those in the highest tertile showed a lower risk of frailty after adjustment for sex, age, and education (OR 0.55; 95% CI 0.35e0.86; P for trend ¼ .009); in the fully adjusted analyses, the association was somewhat attenuated and only reached marginal statistical significance (OR 0.65; 95% CI 0.40e1.04; P for trend ¼ .07). With regard to the MDS, an increasing MD adherence was associated with a progressively reduced risk of frailty in the analyses adjusted for all confounders; compared with the lowest tertile of MDS, the OR (95% IC) of frailty was 0.59 (0.37e0.95) in the second quartile, and 0.48 (0.30e0.77) in the highest quartile; P for trend ¼ .002 (Table 1). The results were similar when the analyses were repeated using >3 kg-loss as the weight loss criterion, when weight loss was not considered among the frailty criteria, when the cut-off point for grip strength was based on the study sample, and when individuals with eating difficulty and cognitive impairment were excluded from the
Table 1 Association Between Adherence to the MD and Risk of Frailty During a 3.5-Year Follow-Up of Older Adults (N ¼ 1815) MEDAS
MDS
Tertile 1 Tertile 2 OR (95% CI) Main analyses Number of frailty events Model 1 Model 2 Sensitivity analyses Defining weight loss as losing >3 kg in the last three months Number of frailty events Model 1 Model 2 Excluding weight loss from the definition of frailty Number of frailty events Model 1 Model 2 Defining weakness as lowest quintile of grip strength in the study population Number of frailty events Model 1 Model 2 Excluding 52 individuals with eating difficulty Number of frailty events Model 1 Model 2 Excluding 47 individuals with Mini-Mental Examination 22 Number of frailty events Model 1 Model 2 Excluding 527 individuals with diagnosed severe diseases Number of frailty events Model 1 Model 2x
Tertile 3
P Trend Tertile 1 Tertile 2
OR (95% CI)
OR (95% CI)
Tertile 3
P Trend
OR (95% CI)
61 Ref. Ref.
41 35 0.76 (0.49e1.18) 0.55 (0.35e0.86)y .009 0.85 (0.54e1.36) 0.65 (0.40e1.04) .07
67 Ref. Ref.
36 34 0.58 (0.37e0.91)* 0.43 (0.27e0.67)z
JAMDA journal homepage: www.jamda.com
Original Study
Mediterranean Diet and Risk of Frailty in Community-Dwelling Older Adults Luz M. León-Muñoz PhD a, b, Pilar Guallar-Castillón MD a, b, Esther López-García PhD a, b, Fernando Rodríguez-Artalejo MD a, b, * a b
CIBER of Epidemiology and Public Health (CIBERESP), Madrid, Spain Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autónoma de Madrid/IdiPaz, Madrid, Spain
a b s t r a c t Keywords: Mediterranean diet frailty older adults Spain
Background and Objective: Low intake of certain micronutrients and protein has been associated with higher risk of frailty. However, very few studies have assessed the effect of global dietary patterns on frailty. This study examined the association between adherence to the Mediterranean diet (MD) and the risk of frailty in older adults. Design, Setting, and Participants: Prospective cohort study with 1815 community-dwelling individuals aged 60 years recruited in 2008e2010 in Spain. Measurements: At baseline, the degree of MD adherence was measured with the Mediterranean Diet Adherence Screener (MEDAS) score and the Mediterranean Diet Score, also known as the Trichopoulou index. In 2012, individuals were reassessed to detect incident frailty, defined as having at least 3 of the following criteria: exhaustion, muscle weakness, low physical activity, slow walking speed, and weight loss. The study associations were summarized with odds ratios (OR) and their 95% confidence interval (CI) obtained from logistic regression, with adjustment for the main confounders. Results: Over a mean follow-up of 3.5 years, 137 persons with incident frailty were identified. Compared with individuals in the lowest tertile of the MEDAS score (lowest MD adherence), the OR (95% CI) of frailty was 0.85 (0.54e1.36) in those in the second tertile, and 0.65 (0.40e1.04; P for trend ¼ .07) in the third tertile. Corresponding figures for the Mediterranean Diet Score were 0.59 (0.37e0.95) and 0.48 (0.30e0.77; P for trend ¼ .002). Being in the highest tertile of MEDAS was associated with reduced risk of slow walking (OR 0.53; 95% CI 0.35e0.79) and of weight loss (OR 0.53; 95% CI 0.36e0.80). Lastly, the risk of frailty was inversely associated with consumption of fish (OR 0.66; 95% CI 0.45e0.97) and fruit (OR 0.59; 95% CI 0.39e0.91). Conclusions: Among community-dwelling older adults, an increasing adherence to the MD was associated with decreasing risk of frailty. Ó 2014 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Frailty is a medical syndrome with multiple causes and contributors that is characterized by diminished strength and endurance, and reduced physiological function, which increases an individual’s vulnerability of developing dependence and/or of death.1,2 Given the accelerated aging of the population in most countries and the expected increase in the number of individuals with disability and
This work has been supported by FIS grants 11/01379, 12/1166, and 13/00288 (Ministry of Health of Spain) and by FP7-HEALTH-2012-Proposal No: 305483-2 (FRAILOMIC Initiative). All authors declare that they have no conflict of interest. * Address correspondence to Fernando Rodríguez-Artalejo, MD, Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autónoma de Madrid, Arzobispo Morcillo s/n, 28029 Madrid, Spain. E-mail address: [email protected] (F. Rodríguez-Artalejo).
dependence, prevention of frailty may be a key policy to reduce the social and healthcare burden associated with disability. There is evidence that poor nutrition is related to frailty.3 Specifically, low intake of certain micronutrients and protein,4e6 or an inadequate distribution of protein intake,7 have been associated with higher risk of frailty. However, very few studies have assessed the effect of dietary patterns on frailty. Instead of looking at individual nutrients or foods, pattern analysis examines the effects of overall diet. Conceptually, dietary patterns represent a broader picture of food and nutrient consumption and, thus, may be more predictive of disease risk than individual foods or nutrients.8 To our knowledge, only 3 studies have explored the association between dietary patterns and frailty.6,9,10 One study reported that a higher score in the Diet Quality Index Revised was inversely associated with prevalent and future frailty status in a cohort of older men.6
1525-8610/$ - see front matter Ó 2014 AMDA e The Society for Post-Acute and Long-Term Care Medicine. http://dx.doi.org/10.1016/j.jamda.2014.06.013
900
L.M. León-Muñoz et al. / JAMDA 15 (2014) 899e903
The other 2 studies focused on the Mediterranean Diet (MD) pattern9,10; Bollwein et al9 found a cross-sectional association, and Talegawkar et al10 a prospective association between adherence to the MD and lower risk of frailty. In both studies, however, the MD was defined from sample-specific scores; thus, only the relative but not the absolute effect of MD was assessed, and the results of these studies are difficult to compare across populations. Lastly, no information was available on the specific components of the MD, which were more closely related to frailty. Accordingly, we examined the prospective association between the adherence to MD, as assessed by normative and sample-specific scores, and the risk of frailty in older adults from the general population in Spain. Moreover, we assessed the association between the individual components of the MD and incident frailty. Methods Study Design and Population Data were taken from the Seniors-ENRICA (Estudio de Nutrición y Riesgo Cardiovascular) cohort. Briefly, the cohort was established in 2008e2010 with 2519 individuals selected through stratified random sampling from the population aged 60 years and older in Spain.11 At baseline, data were collected in 3 stages. First, a phone interview to obtain information on health status, lifestyle, morbidity, and health services use; second, a home visit to collect blood and urine samples; and third, another home visit to perform a physical examination and to record habitual diet and prescribed medication.11 In 2012, a second wave of data collection was performed; in total, 2037 participants provided updated information for the phone interview, the physical examination, and diet. The sociodemographic, lifestyle, and clinical characteristics at baseline were similar in those who provided updated information at 2012 and those who did not. Study participants gave written informed consent. The Clinical Research Ethics Committee of the ‘La Paz’ University Hospital in Madrid approved both the baseline and follow-up studies.
5e25 g/day in women. The range in this score is 0 (lowest) to 9 (highest MD adherence). Frailty We used a slight modification of the operational definition of frailty developed by Fried and colleagues in the Cardiovascular Health Study.17 Specifically, frailty was defined as having 3 or more of the following 5 criteria as follows. (1) Exhaustion was evaluated as any of the following responses to 2 questions taken from the Center for Epidemiologic Studies Depression Scale18: “I felt that anything I did was a big effort” and “I felt that I could not keep on doing things” at least 3 to 4 days a week.” (2) Weakness, defined as the lowest quintile in the Cardiovascular Health Study of maximum strength on the dominant hand adjusted for sex and body mass index (BMI). Strength was measured with a Jamar dynamometer, and we selected the highest value in 2 consecutive measures.19,20 (3) Low physical activity, defined as walking 2.5 hours/week in men and 2 hours/week in women. (4) Slow walking speed, defined as the lowest quintile in our study sample for the three-meter walking speed test, adjusted for sex and height.20,21 (5) Weight loss, defined as involuntary loss of 1 kg of body weight in the preceding 3 months. Other Variables At baseline we collected information on sociodemographic variables, lifestyle, and diseases, which could be related to both diet and frailty. Specifically, study participants reported their sex, age, educational level, tobacco consumption, number of medications used, and energy intake (calculated with standard food composition tables). They also reported if they suffered from any of the following physician-diagnosed diseases: cardiovascular disease (myocardial infarction, stroke, heart failure), diabetes mellitus, cancer at any site, asthma or chronic bronchitis, osteomuscular disease (osteo-arthritis, arthritis, hip fracture), or depression requiring drug treatment. Lastly, weight and height were measured in standardized conditions,22 and the BMI was calculated as weight in kg divided by square height in m. Statistical Analysis
Study Variables Diet Food consumption was assessed with a validated computerized diet history developed from that used in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort study in Spain.12,13 The diet history collected data on the consumption of 880 foods in the preceding year and used a set of photographs to help in estimating the amount of intake. Adherence to the MD was summarized with 2 instruments: the Mediterranean Diet Adherence Screener (MEDAS)14 and the Mediterranean Diet Score (MDS),15 also known as the Trichopoulou index. The MEDAS was developed to assess compliance with the dietary intervention of the Prevención con Dieta Mediterránea (PREDIMED) study, a clinical trial of the effect of the MD on primary cardiovascular prevention.16 MEDAS consists of 12 items with targets for food consumption and another 2 items with targets for food intake habits characteristic of the MD in Spain. One point is given for each target achieved. The total MEDAS score ranged from 0 to 14, with a higher score indicating better MD adherence. In the MDS, the intake of vegetables, legumes, fruits and nuts, grains, and fish is considered beneficial, thus, a value of 1 is assigned to consumption above the sex-specific median in the study sample; in contrast, intake of red meat and poultry, and dairy products is considered detrimental, and a value of 0 is assigned to consumption above the median. Moderate alcohol consumption is also considered beneficial: 1 point is assigned for intake of 10e50 g/day in men and
Among the 2037 study participants, we excluded 112 who lacked complete data on diet and frailty, and 31 with missing information on potential confounders. Of the remaining 1894 persons, we excluded 79 who had frailty at baseline. Therefore, the analyses were conducted with 1815 individuals. The association between MD adherence and risk of frailty was summarized with odds ratio (OR) and 95% confidence interval (CI), obtained from logistic regression. MD adherence, assessed with the MEDAS or the MDS, was classified into tertiles and the lowest tertile was used as reference. Two logistic models were built: the first one adjusted for sex, age, and education; and the second one additionally adjusted for the rest of variables described above. The reason for building 2 models is that model 2 may be considered as overadjusted, since BMI, chronic diseases, and the number of drug treatments could be in the pathway from diet to frailty. However, as will be shown below, both models yielded similar results. We ran several sensitivity analyses to assess the robustness of the main results. First, we re-estimated the study associations using a >3 kg-loss in the preceding 3 months as the weight loss criterion. This may reflect a severe frail status more accurately than simply losing 1 kg. Second, given that the main aim of the study was to evaluate the association between diet and frailty, we repeated the analysis excluding the weight loss criteria from the definition of frailty.10 Thus, in this analysis frailty was defined as having 2 out of the 4 remaining Fried criteria. Third, we replicated the analyses using the lowest quintile of grip strength in our population to define weakness.
L.M. León-Muñoz et al. / JAMDA 15 (2014) 899e903
Fourth, we ran analyses after excluding those who reported severe or substantial difficulty in chewing or eating, and those with cognitive impairment (score 22 on the Mini-Mental State Examination23), at the end of follow-up. Finally, the study association was assessed after excluding individuals with severe diagnosed disease (cardiovascular disease, diabetes, cancer, chronic lung disease, or depression requiring treatment). Finally, we used the same type of modeling to assess the association between MD adherence and the onset of each frailty criterion in robust individuals (free of all 5 criteria) at baseline. Likewise, we used logistic regression to study the association between each component of the MEDAS or MDS and the risk of frailty; these analyses were further adjusted for all components of the MEDAS or MDS except the component of interest. Statistical significance was set at 2-sided P < .05. The analyses were performed with Stata v 11.1 (StataCorp, College Station, TX). Results Among the 1815 study participants at baseline, the mean (range) adherence to the MD was 7.18 (1e13) using the MEDAS, and 4.93 (0e9) using the MDS. Supplementary Table 1 shows the sociodemographic and clinical characteristics of participants according to
901
MD adherence. Compared with those in the lowest tertile of the MEDAS score, those in the highest tertile were more frequently women and never smokers, and showed a lower frequency of current smoking, obesity, cardiovascular disease, diabetes and depression, and less energy intake. Individuals showed a similar distribution when MD adherence was defined with the MDS. Over a mean follow-up of 3.5 years, we identified 137 persons with incident frailty. Compared with individuals in the lowest tertile of the MEDAS score, those in the highest tertile showed a lower risk of frailty after adjustment for sex, age, and education (OR 0.55; 95% CI 0.35e0.86; P for trend ¼ .009); in the fully adjusted analyses, the association was somewhat attenuated and only reached marginal statistical significance (OR 0.65; 95% CI 0.40e1.04; P for trend ¼ .07). With regard to the MDS, an increasing MD adherence was associated with a progressively reduced risk of frailty in the analyses adjusted for all confounders; compared with the lowest tertile of MDS, the OR (95% IC) of frailty was 0.59 (0.37e0.95) in the second quartile, and 0.48 (0.30e0.77) in the highest quartile; P for trend ¼ .002 (Table 1). The results were similar when the analyses were repeated using >3 kg-loss as the weight loss criterion, when weight loss was not considered among the frailty criteria, when the cut-off point for grip strength was based on the study sample, and when individuals with eating difficulty and cognitive impairment were excluded from the
Table 1 Association Between Adherence to the MD and Risk of Frailty During a 3.5-Year Follow-Up of Older Adults (N ¼ 1815) MEDAS
MDS
Tertile 1 Tertile 2 OR (95% CI) Main analyses Number of frailty events Model 1 Model 2 Sensitivity analyses Defining weight loss as losing >3 kg in the last three months Number of frailty events Model 1 Model 2 Excluding weight loss from the definition of frailty Number of frailty events Model 1 Model 2 Defining weakness as lowest quintile of grip strength in the study population Number of frailty events Model 1 Model 2 Excluding 52 individuals with eating difficulty Number of frailty events Model 1 Model 2 Excluding 47 individuals with Mini-Mental Examination 22 Number of frailty events Model 1 Model 2 Excluding 527 individuals with diagnosed severe diseases Number of frailty events Model 1 Model 2x
Tertile 3
P Trend Tertile 1 Tertile 2
OR (95% CI)
OR (95% CI)
Tertile 3
P Trend
OR (95% CI)
61 Ref. Ref.
41 35 0.76 (0.49e1.18) 0.55 (0.35e0.86)y .009 0.85 (0.54e1.36) 0.65 (0.40e1.04) .07
67 Ref. Ref.
36 34 0.58 (0.37e0.91)* 0.43 (0.27e0.67)z
