Table 1 Observational studies on nutrition and frailty (associations)
From: Nutritional determinants of frailty in older adults: A systematic review
Authors and Year | Study Design | Country and Sample Characteristics | Setting | Operational Definition of Frailty | Nutritional Measurement Tools | Incidence/Prevalence | Main Findings |
|---|---|---|---|---|---|---|---|
Bartali et al., 2006 [9] | Cross-sectional | Europe (Italy); n = 802; age, mean ± SD: 74.1 ± 6.5 years; 56% women | Community-dwelling participants aged 65 years or older who participated in the InCHIANTI study | Modified Fried frailty phenotype [3]: 4 of the 5 criteria of the Fried frailty phenotype (“weight loss” no included) | Dietary intake was assessed by a food-frequency questionnaire that was created for the EPIC study [47]. Data on food consumption were transformed into daily intake of energy, macronutrients, and micronutrients | 20% frailty | Daily energy intake ≤21 kcal/kg and low intake of more than 3 nutrients were significantly and independently associated with frailty (OR 1.24, 95% CI: 1.02–1.5; OR 2.12, 95% CI: 1.29–3.50). Specifically, low energy intake of protein, vitamins D, E, and C and folate was related to frailty (OR 1.98, 95% CI 1.18–3.31; OR 2.35, 95% CI 1.48–3.73; OR 2.06, 95% CI 1.28–3.33; OR 2.15, 95% CI 1.34–3.45; and OR 1.84, 95% CI 1.14–2.98, respectively). Small Hasselblad & Hedges’ d ES, ranged from 0.05 to 0.2 |
Bollwein et al., 2013a [27] | Cross-sectional | Europe (Germany); n = 192; age, mean ± SD: 83.0 ± 4.0 years; 64.6% women | Community-dwelling older adults (>75 years) | Fried frailty phenotype [3] | Dietary quality was assessed by the alternate MDS of Fung et al. [48], who adapted the original score from Trichopoulou et al. [49]. MDS was calculated using a slightly modified version of the FFQ of the German part of the EPIC study (EPIC-FFQ) [50] | 15.1% frailty and 41.1% pre-frailty | A healthy diet significantly decreased the risk of being frail (Q4 of the MDS, OR 0.26, 95% CI 0.07–0.98; Hasselblad & Hedges’ d ES 0.32) |
Bollwein et al., 2013b [28] | Cross-sectional | Europe (Germany); n = 194; age, mean (range): 83.0 (75–96 years); 66.0% women | Community-dwelling older adults (>75 years) | Fried frailty phenotype [3] | Usual food intake was estimated using a slightly modified version of the FFQ of the German part of the EPIC study; (EPIC-FFQ) [50] | 15.4% frailty and 40.5% pre-frailty | No significant differences were observed in the daily amount of protein intake between frailty groups. However, the distribution of protein intake was significantly different. Low morning protein intake values in frail, pre-frail and non-frail elders were 11.9, 14.9, and 17.4%, respectively). Small Cohen’s h ES, ranged from 0.05 to 0.14. Higher midday protein intake in frail, pre-frail and non-frail elders was 61.4, 60.8, and 55.3% respectively. Small Cohen’s h ES, ranged from 0.02 to 0.12 |
Bollwein et al., 2013c [29] | Cross-sectional | Europe (Germany); n = 206; age, mean (range): 83.0 (75–96 years); 66.0% women | Community-dwelling older adults aged 75 years or older | Fried frailty phenotype [3] | MNA® [41] | 15.5% frailty and 39.8% pre-frailty; 15.1% at risk of malnutrition | A significant association between MNA total score and risk of frailty was found (2.2% of the non-frail, 12.2% of the pre-frail and 46.9% of the frail participants were at risk of malnutrition. Small Cohen’s h = 0.42 between non-frail and pre-frail but large ES between non-frail and pre-frail and frail (0.8 and 1.2, respectively). A significant association between 12 of the 18 MNA items and frailty was also observed |
Boulos et al., 2016 [33] | Cross-sectional | Asia (Lebanon); n = 1200; age, mean ± SD: 75.7 ± 7.1 years; 42.3% women | Community-dwelling older adults (≥65 years) living in a rural setting | MNA® [41] | 36.4% frailty and 30.4% pre-frailty; 8.0% malnourishment and 29.1% at risk of malnutrition | Malnutrition and risk of malnutrition (i.e., poor nutritional status) were related to a significantly increased risk of frailty (OR 3.72, 95% CI 1.40–9.94; OR 3.66, 95% CI 2.32–5.76, respectively) but with small Hasselblad & Hedges’ d ES. Fourteen of the 18 MNA items were associated with frailty in age-adjusted analyses. | |
Chan et al., 2015 [34] | Longitudinal | Asia (China); n = 2724; age, mean ± SD: 71.8 ± 4.8 years; 50.3% women | Community-dwelling older adults (≥65 years) | FRAIL scale [53] | Assessment of baseline dietary intake through the FFQ. assessment of diet quality with the DQI-I [54]., and assessment of the adherence to the MDS using the revised method of Trichopoulou et al. [49] | 1.1% frailty | Higher score of the “snacks-drinks milk products” significantly decreased the risk of being frail in a sex-age-adjusted model over a 4-year follow-up (adjusted OR 0.58, 95% CI 0.36–0.91). Better diet quality (higher DQI-I scores) significantly decreased the risk of being frail in both crude and sex-age-adjusted models over a 4-year follow-up (crude OR 0.61, 95% CI 0.43–0.86; adjusted OR 0.59, 95% CI 0.42–0.85, respectively). Nevertheless, no Hasselblad & Hedges’ d ES was observed. There was no association of MDS, “vegetables-fruits” pattern, or “meat-fish” pattern with incident frailty |
Chang, 2017 [11] | Cross-sectional | Asia (Taiwan); n = 432; age, mean ± SD: 72.3 ± 10.0; 64.3% women | Community-dwelling older adults (≥65 years) | MNA-SF® [41] | 10.4% frailty and 23.6% pre-frailty; 30.6% at risk of malnutrition | Frailty was more prevalent in the group at risk of malnutrition. Frail status was a related risk factor at risk of malnutrition (OR = 8.78, with small Hasselblad & Hedges’ d ES, 0.32). Lower body mass index and lower skeletal mass indices were related to a higher risk of malnutrition. Frail people had a particularly high risk of malnutrition. | |
Chang & Lin, 2016 [35] | Cross-sectional | Asia (Taiwan); n = 152; age, mean ± SD: 80.8 ± 7.2 years; 18.4% women | Community-dwelling older adults (≥65 years) | MNA® [41] | 40.1% pre-frailty; 8.2% malnourishment and 34.9% at risk of malnutrition | The pre-frail group had a lower total MNA score than the non-frail group (β = −0.36, p < 0.001). Cohen’s d ES for mean scores was 0.75 | |
El Zoghbi et al., 2014 [36] | Cross-sectional | Asia (Lebanon); n = 111; ≥65 years; 50.4% women | Institutionalized older adults | MNA® [41] | 37.9% frailty and 36.9% pre-frailty; 12.6% malnourishment and 48.7% at risk of malnutrition | The MNA score was inversely associated with the SOF Frailty Index (standardized beta coefficient − 0.18, 95% CI -1.46- -0.13). Mean scores comparison: small Cohen’s d ES (0.24) in malnutrition vs. risk of malnutrition, medium ES (0.65) in malnutrition vs. normal, and large ES (0.89) in risk of malnutrition vs. normal | |
Eyigor et al., 2015 [26] | Cross-sectional | Europe (Turkey); n = 1126; age range (65–85 years); 65.7% women | Community-dwelling older adults (living on their own or in a family house) and those living in nursing homes | Fried frailty phenotype [3] | MNA® [41] | 39.2% frailty and 43.3% pre-frailty; 5% malnutrition and 27.5% at risk of malnutrition | Malnutrition increased the risk of frailty (OR 48.545, 95% CI 6.647–354.554). Large Hasselblad & Hedges’ d ES (0.89). |
Jürschik et al., 2014 [30] | Longitudinal | Europe (Spain); n = 640; age, mean ± SD: 81.3 ± 5.0 years; 60.3% women | Community-dwelling older adults from the FRALLE survey | Slightly modified Fried frailty phenotype [3]: changes in metrics to characterize frailty [55] | MNA®, MNA-SF® [41] | 9.6% frailty and 47% pre-frailty; 1.9% malnutrition and 19.8% at risk of malnutrition | Both the MNA (0.75, p < 0.001) and the MNA-SF (0.80, p < 0.001) were accurate in identifying frailty. Nevertheless, no Cohen’s h ES was observed |
Kobayashi et al., 2013 [37] | Cross-sectional | Asia (Japan); n = 2108; age, mean ± SD: 74.7 ± 5.0 years; 100% women | Community-dwelling old women (≥65 years) | Modified Fried frailty phenotype [3]: includes only 4 components- slowness and weakness, exhaustion, low physical activity and unintentional weight loss [56] | Dietary protein intake source (animal or plant) and protein quality (amino acid components) were assessed by the BDHQ [57, 58] | 22.8% frailty | Higher total, animal, and plant protein intake was inversely associated with frailty (adjusted OR for Q5 vs. Q1 0.66, 95% CI 0.46–0.96; 0.73, 95% CI 0.50–1.06; and 0.66, 95% CI 0.45–0.95, respectively). A higher intake of amino acids was associated with a lower prevalence of frailty (range of adjusted ORs for Q5 vs. Q1 0.67 for cysteine to 0.74 for valine). No Hasselblad & Hedges’ d ES were observed for any protein or amino acid intake |
Kobayashi et al., 2014 [38] | Cross-sectional | Asia (Japan); n = 2121; age, mean ± SD: 74.7 ± 5.0 years; 100% women | Community-dwelling old women (≥65 years) | Modified Fried frailty phenotype [3]: includes only 4 components- slowness and weakness, exhaustion, low physical activity and unintentional weight loss [56] | Dietary TAC and food intake were assessed by the DHQ [58] | 22.9% frailty | Higher intake of dietary TAC (FRAP, ORAC, TEAC, and TRAP assays) was inversely associated with frailty (adjusted OR for Q5 vs. Q1 0.35, 95% CI 0.24–0.53; 0.35, 95% CI 0.23–0.52; 0.40, 95% CI 0.27–0.60; and 0.41, 95% CI 0.28–0.62, respectively), with small Hasselblad & Hedges’ d ES. Higher food intake (coffee, vegetables, and fruit) was inversely associated with frailty (adjusted OR for Q5 vs. Q1 0.48, 95% CI 0.32–0.72; 0.47, 95% CI 0.33–0.69; and 0.71, 95% CI 0.49–1.03, respectively), without Hasselblad & Hedges’ d ES. Higher nutrient intake (vitamin A, α-carotene, β-carotene, β-carotene equivalent, cryptoxanthin, vitamin D, α-tocopherol, vitamin B6, folate, and vitamin C) was inversely associated with frailty (adjusted OR for Q5 vs. Q1 0.72, 95% CI 0.49–1.04; 0.68, 95% CI 0.47–0.98; 0.53, 95% CI 0.36–0.76; 0.47, 95% CI 0.33–0.66; 0.78, 95% CI 0.54–1.12; 0.67, 95% CI 0.46–0.98; 0.51, 95% CI 0.36–0.74; a 0.50, 95% CI 0.34–0.72; 0.52, 95% CI 0.36–0.76; and 0.61, 95% CI 0.42–0.88, respectively), without Hasselblad & Hedges’ d ES |
Matteini et al., 2008 [39] | Cross-sectional | USA (Maryland); n = 703; age, range: 70–79 years; 100% women | Community-dwelling older women from the WHAS I and II | Fried frailty phenotype [3] | MMA, tHcy and cystathionine were assayed through stable isotope dilution capillary gas chromatography mass spectrometry with selected ion monitoring. Vitamin B6 was measured as pyridoxal 5-phosphate using high-performance liquid chromatography. Serum vitamin B12 and folate were measured by radiodilution assay | 13.7% frailty | Increased concentrations of MMA (a marker of vitamin B12 deficiency) were related to greater odds of pre-frailty and frailty (OR 1.59, 95% CI 0.95–2.65, no Hasselblad & Hedges’ d ES; OR 2.33, 95% CI 1.14–4.77, small Hasselblad & Hedges’ d ES) |
Michelon et al., 2006 [17] | Cross-sectional | USA (Maryland); n = 754; age, mean (range): 74.7 (70–80 years); 100% women | Community-dwelling older women from the WHAS I and II | Fried frailty phenotype [3] | Plasma carotenoids, retinol, and α-tocopherol were determined by high-performance liquid chromatography. Total carotenoids were calculated as the sum of α-carotene, β-carotene, β-cryptoxanthin, lutein/zeaxanthin, and lycopene (in μmoles/l). 25(OH)D was measured using a radioreceptor assay. Vitamin B6 status was assessed by pyridoxal 5-phosphate measurements using high-performance liquid chromatography. Serum vitamin B12 and folate were measured using RIA | 11.4% frailty and 44.7% pre-frailty | Lower serum levels of total carotenoids, α-tocopherol, 25-hydroxyvitamin D, and vitamin B6 significantly increase the risk of becoming frail in age-adjusted regression models (age-adjusted OR for Q1 vs. Q2-Q3-Q4 2.50, 95% CI 1.51–4.14; 1.64, 95% CI 0.95–2.84, 1.71, 95% CI 1.00–2.94; and 1.79, 95% CI 0.99–3.24, respectively), with only small Hasselblad & Hedges’ d ES for lower serum levels of total carotenoids. Lower serum levels of β-carotene, lutein/zeaxanthin, and total carotenoids significantly increase the risk of becoming frail with advancing age, sociodemographic status, smoking status, and body mass index models (OR ranging from 1.82 to 2.45, p = 0.05), with small or no Hasselblad & Hedges’ d ES |
Rabassa et al., 2015 [31] | Longitudinal | Europe (Italy); n = 769; age, mean ± SD: 72.7 ± 5.8 years; 55.4% women | Community-dwelling older adults from the Invecchiare in Chianti study | Fried frailty phenotype [3] | Habitual dietary resveratrol exposure was assessed. TDR was assessed through the Italian version of the FFQ developed and validated in the EPIC [47] and an ad hoc food-composition database on resveratrol [59, 60]. TUR was analyzed with the use of liquid chromatography-tandem mass spectrometry with a previous solid-phase extraction at baseline. TDR + TUD was measured and computed using Howe’s method [61] | 4.4% frailty and 37.4% pre-frailty | TDR, TUR, and TDR + TUR concentrations were inversely associated with frailty risk over 3-years of follow-up but not after 6- and 9-years of follow-up (OR for comparison of extreme tertiles: 0.17, 95% CI 0.05–0.63; 0.32, 95% CI 0.09–1.11; and 0.11, 95% CI 0.03–0.45, respectively), with small and medium Hasselblad & Hedges’ d ES (0.42, 0.27 and 0.53, respectively) |
Rahi et al., 2016 [32] | Cross-sectional | Europe (Paris); n = 1345; age, mean ± SD: 75.6 ± 5.1 years; 60.4% women | Community dwellers aged 65 years and above | Modified Fried frailty phenotype [3]: handgrip strength was replaced by the chair standing method; since walking speed was missed for many elders, the Rosow-Breslaw test was used [62] | Daily intakes of energy and protein were set at ≥30 kcal/kg body weight/d and ≥1 g/kg body weight/d, | 4.1% frailty; 57.7% protein intake ≥1 g/kg body weight/d | A higher protein intake was associated with a lower prevalence of frailty because the association with a slow walking speed (OR = 0.63; with no Hasselblad & Hedges’ d ES |
Semba et al., 2006 [10] | Longitudinal | USA (Maryland); n = 766; age, mean ± SD: 78.2 ± 7.6 years; 100% women | Community-dwelling older women from the WHAS-I | Fried frailty phenotype [3] | Nutrient concentrations were measured through blood analysis. Serum samples for total carotenoids, retinol, and α-tocopherol, levels were determined by high-performance liquid chromatography. Serum selenium and zinc levels were measured by graphite furnace atomic absorption spectrometry using a PerkinElmer AAnalyst 600 (Norwalk, CT) with Zeeman background correction. 25(OH)D was measured using a radioreceptor assay. Serum vitamin B12 and folate were measured using RIA | 32.6% frailty | Lower levels of serum carotenoids and α-tocopherol significantly increase the risk of becoming frail over a period of 3 years (HR for Q1 vs. Q2-Q3-Q4 1.30, 95% CI 1.01–1.92; and 1.39, 95% CI 1.02–1.89, respectively). The number of nutrient deficiencies was associated with an increased risk of becoming frail over a period of 3 years (adjusted HR 1.10; 95% CI 1.01–1.209). No Hasselblad & Hedges’ d ES were observed. |
Shikany et al., 2014 [40] | Longitudinal | USA; n = 5925; age, mean ± SD: 75.0 ± 5.7 years; 100% men | Community-dwelling from the Osteoporotic Fractures in Men (MrOS) study | Slightly modified Fried frailty phenotype [3]: due to the lack of data on body weight prior to enrollment, appendicular lean mass in the lowest quintile was used for the shrinkage component | Food intake was assessed through Block 98 of the FFQ [63, 64]. Diet quality was assessed with the DQI-R [65, 66] | 8.4% frailty and 45.2% pre-frailty | At baseline: higher intake of fiber significantly decreased the risk of intermediate or frail status relative to a robust status (OR for Q5 vs. Q1 0.83, 95% CI 0.69–1.00; and 0.51; 95% CI 0.36–0.73, respectively). Higher intake of carbohydrate was significantly associated with reduced odds of frailty relative to a robust status (OR for Q5 vs. Q1 0.65; 95% CI 0.45–0.94). Higher intake of fat was significantly associated with greater odds of frailty relative to a robust status (OR for Q5 vs. Q1 1.61; 95% CI 1.12–2.31). DQI-R was inversely associated with frailty relative to a robust status (OR for Q5 vs. Q1 0.44, 95% CI 0.30–0.63). No Hasselblad & Hedges’ d ES were observed. Prospective analysis: DQI-R was inversely associated with the risk of intermediate or frailty status relative to a robust status (OR for Q5 vs. Q1 0.82, 95% CI 0.60–1.11; and 0.18, 95% CI 0.03–0.97, respectively) with no Hasselblad & Hedges’ d ES for the risk of frailty but medium d ES in frailty |