Dietary protein intake is associated with better physical function and muscle strength among elderly women

Abstract

Dietary protein intake might be beneficial to physical function (PF) in the elderly. We examined the cross-sectional and prospective associations of protein intake of g/kg body weight (BW), fat mass (FM) and lean mass (LM) with PF in 554 women aged 65·3–71·6 years belonging to the Osteoporosis Risk Factor and Prevention Fracture Prevention Study. Participants filled a questionnaire on lifestyle factors and 3-d food record in 2002. Body composition was measured by dual-energy X-ray absorptiometry, and PF measures were performed at baseline and at 3-year follow-up. Sarcopaenia was defined using European Working Group on Sarcopenia in Older People criteria. At the baseline, women with higher protein intake (≥1·2 g/kg BW) had better performance in hand-grip strength/body mass (GS/BM) (P=0·001), knee extension/BM (P=0·003), one-leg stance (P=0·047), chair rise (P=0·043), squat (P=0·019), squat to the ground (P=0·001), faster walking speed for 10 m (P=0·005) and higher short physical performance battery score (P=0·004) compared with those with moderate and lower intakes (0·81–1·19 and ≤0·8 g/kg BW, respectively). In follow-up results, higher protein intake was associated with less decline in GS/BM, one-leg stance and tandem walk for 6 m over 3 years. Overall, results were no longer significant after controlling for FM. Associations were detected between protein intake and PF in non-sarcopaenic women but not in sarcopaenic women, except for change of GS (P=0·037). Further, FM but not LM was negatively associated with PF measures (P<0·050). This study suggests that higher protein intake and lower FM might be positively associated with PF in elderly women.

The aetiology of sarcopaenia is multifactorial. The European Working Group on Sarcopenia in Older People (EWGSOP) has provided a working definition of sarcopaenia^{( 1 , 2 )}. EWGSOP proposed that sarcopaenia is diagnosed using the criteria of low lean mass (LM) and low physical performance, either low muscle strength (MS) and/or low physical function (PF) in elderly^{( 1 , 2 )}. It is known that decline in MS and PF is an important contributing factor of the quality of life, and it increases the risk of frailty, fracture and falls in older individuals^{( 2 – 4 )}. Although the aetiology of the decline in physical performance is not fully understood, poor nutrition may contribute to its development and progression^{( 5 )}. Therefore, measurement of MS and PF as indicators of physical performance status, as well as nutritional status, gained considerable attention in the past years^{( 6 )}.

Indeed, new evidence shows that adequate dietary protein is beneficial to support good health, promote recovery from illness and maintain LM in older adults^{( 7 – 11 )}. It also has positive association with MS and PF^{( 12 – 15 )}. However, the adequacy of current RDA^{( 16 )} for protein of 0·8 g/kg body weight (BW) has been questioned recently regarding that it might not be enough to maintain the LM and to prevent functional decline among the elderly^{( 5 , 17 , 18 )}. To this end, recent reviews and consensus statements have suggested that a protein intake between 1·0 and 1·5 g/kg per d may confer health benefits beyond those afforded by simply meeting the minimum^{( 5 , 19 )}. It might be inappropriate also to generalise the protein intake requirements based on healthy young men to older adults^{( 18 )}. International Study Group, and including 11 other members, to review dietary protein needs with aging (PROT-AGE Study Group) recommendations for dietary protein intake in healthy older adults is an average in the range of 1·0–1·2 g/kg BW^{( 11 )}. Further, Nordic Nutrition Recommendation 2012 (NNR) for elderly also suggested protein intake in the range of 1·1–1·3 g/kg BW (1·2 g/kg BW for planning purposes at the population level)^{( 20 – 22 )}.

Ageing is accompanied with changes in body composition with a gradual increase in the proportion of fat mass (FM) and decline in LM^{( 23 )}. LM is the main reservoir of protein in the human body, and it has a significant role in movement and posture, regulation of metabolism, and storage of energy and N^{( 24 )}. Previous studies supported the correlation between decreased LM and impaired physical performance^{( 25 )}. In a study by Pedrero-Chamizo et al.^{( 26 )}, elderly men and women with sarcopaenic obesity showed lower physical fitness levels compared with non-sarcopaenic subjects^{( 27 )}. Notably, older individuals have an attenuated muscle protein synthetic response after the ingestion of dietary protein and amino acids. This resistance to the usually anabolic effect of protein on myofibrillar protein synthesis (MPS) may partially contribute to the age-related decline in LM^{( 28 )}. Because of metabolic changes associated with ageing, elderly persons may produce less LM than younger people from the same amount of ingested protein^{( 29 )}. It is recommended, therefore, that in cases of acute illness or psychological stress or sarcopaenia higher protein intake is required^{( 30 )}.

The primary aim of this study was to evaluate the differences in MS and PF in elderly women with higher protein intake than current daily allowance as compared with those with lower intake at the baseline and over the 3-year follow-up. We hypothesised that a positive association of protein intake with PF measures is more pronounced in non-sarcopaenic women as compared with those with diagnosed sarcopaenia based on EWGSOP criteria^{( 2 )}. Further, the associations of total body FM and LM with PF and MS measures were examined at the baseline and at 3 years of follow-up.

Methods

Study design and participants

Data of the present study were collected from the Osteoporosis Risk Factor and Prevention Fracture Prevention Study (OSTPRE-FPS), which was a 3-year intervention to investigate the effect of Ca and vitamin D supplementation on the incidence of falls and fractures among elderly women^{( 31 )}. Inclusion criteria were being older than 65 years of age by the end of November 2002, residing in Kuopio region and no previous participation in OSTPRE bone densitometry sample^{( 31 )}. Supplementation group received daily 800 IU (20μg) of cholecalciferol and 1000 mg of Ca for 3 years, whereas the control group received neither supplementation nor placebo with the aim to study the effects of vitamin D and Ca supplementation on bone mineral density. In total, 750 women were randomly taken into this subsample for participating in detailed examinations including measurement of body composition, physical performance tests and food records^{( 32 )}. Out of those, 554 women returned valid food record and had valid body composition and physical performance measurements for both at the baseline and at the 3-year follow-up. All clinical measurements were performed in the Kuopio Musculoskeletal Research Unit of the Clinical Research Center of the University of Kuopio. All participants provided written permission for participation. The study was approved in October 2001 by the ethical committee of Kuopio University Hospital. The study was registered in Clinical trials.gov by the identification no. NCT00592917.

Body composition measurements

The height and weight of participants were measured in light indoor clothing without shoes, and BMI was calculated as weight (kg) divided by height squared (m²). FM and LM were measured by dual-energy X-ray absorptiometry (DXA) by specially trained nurses. The DXA measurements were carried out using the same Lunar Prodigy adhering to the imaging and analysis protocols provided by the manufacturer (Lunar Co.)^{( 32 )}. DXA is currently a common tool suitable for estimation of body composition in terms of evaluating the ratio between fat, muscle and bone in different parts of the body^{( 33 )}. DXA also has been showed to be superior to bioimpedance for estimation of the body composition^{( 34 )}.

Physical performance measurements

Physical performance measures were assessed by trained nurses at baseline and at year 3, consisting of three main domains – (1) MS: hand-grip strength (GS, kPa), number of chair rises in 30 s, ability to squat, ability to squat to the ground and knee extension (kPa); (2) mobility test: walking speed (WS) for 10 m (m/s) and tandem walk for 6 m (m/s); and (3) balance ability: standing with closed eyes for 10 s and one-leg stance performance for 30 s. GS was measured in a controlled sitting position with a pneumatic hand-held dynamometer (Martin Vigorimeter; Gebruder Martin GmbH & Co., KG) by calculating the mean of three successive measurements from the dominant hand. To standardise, GS and knee extension were further expressed as a ratio to body mass (BM) (FM+LM), which have been suggested to be better predictors of GS and knee extension alone^{( 35 , 36 )}. The chair rise test was conducted if the participant was able to stand at least once without using arms from a straight-backed, non-padded, armless chair. Any measurement errors were excluded from the statistical analysis^{( 37 )}. The follow-up variable of knee extension was excluded from analysis because of an unexpected increase in measured extension force and/or possible data entry errors. Further, based on EWGSOP definition, short physical performance battery (SPPB) score was calculated using three individual measures of physical performance including WS for 10 m (m/s), chair rises in 30 s and one-leg stance performance categorised in quartiles^{( 38 )}. Each quartile was scored on a scale of 1–4 points, with the total score ranging to 12; higher scores of SPPB indicate better performance. Further, absolute changes in PF and MS measures were calculated by subtracting the baseline measures from those measured at year 3. The magnitude of meaningful changes in physical performance measures, as well as SPPB, have been evaluated previously, and these measures are consistently used as preferred indicators of physical performance in older adults^{( 2 , 38 , 39 )}.

Diagnosis of sarcopaenia

Relative skeletal muscle index (RSMI) was calculated as the sum of the non-fat, non-bone skeletal muscle in arms and legs divided by height squared (m²). Women were subdivided into quartiles according to their RSMI values: (1) 5·3–6·3 kg/m², (2) 6·3–6·7 kg/m², (3) 6·7–7·2 kg/m² and (4) 7·2–9·3 kg/m². Baumgartner^{( 23 )} reported that the sarcopaenia cutoff point was 5·45 kg/m², which was calculated as 2 sd below the mean in the young reference population. However, in our study, there were only six women whose RSMI was <5·45 kg/m². Accordingly, we decided to use the lowest quartile below 6·3 kg/m² as cutoff in the present study^{( 37 )}. The study population was divided into quartiles also for their GS: (1) <22·3 kPa, (2) 22·3–25·7 kPa, (3) 25·7–28·7 kPa and (4) 28·7–40 kPa. Physical performance test was assessed by measuring WS by a 10-m WS test in a controlled situation and the WS was divided into quartiles: (1) <0·51 m/s, (2) 1·42–1·63 m/s, (3) 1·64–1·85 m/s and (4) >1·85 m/s. The women who were not able to walk were allocated into the group of the lowest quartile. A woman was classified as sarcopaenic if she belonged to the lowest quartile of RSMI and the lowest quartile of either GS or WS or both. A non-sarcopaenic woman did not belong to the lowest quartile of any measurement (RSMI, GS or WS), whereas pre-sarcopaenic women were in the lowest quartile of RSMI but not in the lowest quartile of any other outcome measure. Non-classified women belonged to the lowest quartile of either GS or WS or both, but not to that of RSMI.

Dietary intakes

Dietary intake was collected by using 3-d food record at baseline. A questionnaire and instructions were sent to the participants beforehand, and they were returned on the visiting day. Participants were advised to fill the questionnaire for 3 consecutive days, including 2 d during the week and 1 d in the weekend (Saturday or Sunday). Participants were instructed to write down everything they ate and drank and to evaluate the amount of food consumed using household measures. In case of uncertainties in the food record, a nutritionist called the participant for additional information^{( 40 )}. To assess the under-reporting, the ratio of energy intake:estimated BMR was calculated based on BW according to equations given by the Department of Health^{( 41 )} in the UK. The ratio of energy intake:BMR cutoff value for under-reporting was chosen to be 1·49, as derived from Goldberg et al.^{( 42 )} and Black^{( 43 )}, and none of the participants was excluded from the analyses. Nutritional intake from food was calculated using Nutrica program (version 2.5; Finnish Social Insurance Institute). Collected data provided calculations of animal and plant sources of protein in addition to total protein intake.

Potential confounders

All lifestyle-related information was gathered by the self-administered questionnaire. The questionnaire included questions on age, hormone therapy use (never used or used), time since menopause (years), smoking status (never, former and current), self-reported Ca and vitamin D supplementation, and alcohol consumption (portions/week). Total physical activity was based on self-reported amounts of sports, recreation and miscellaneous activities, including walking, jogging, skiing, cycling, swimming, aerobic exercise, ball sports and other more strenuous activities. Women were asked how many days they performed each activity per month. The sum of each activity days during all 12 months was divided by 12 in order to obtain the mean activity level per month. Furthermore, the mean activity level was multiplied by self-reported strenuousness of the exercise (the scale was 1 (low) to 4 (strenuous))^{( 37 )}.

Statistical analysis

Protein intake was reported as crude protein intake per BW (g/kg BW). Protein intake was categorised based on three different nutrition recommendations, RDA^{( 16 )} (≤0·8 g/kg BW), PROT-AGE Study Group recommendation^{( 11 )} (0·81–1·19 g/kg BW) and NNR(≥1·2 g/kg BW)^{( 20 )}. For the purpose of this study, these three categories were referred to as lower, moderate and higher intake, respectively. Continuous variables were compared across the protein intake categories using ANOVA and ANCOVA, and categorical variables were compared using χ ² test.

Mean values and standard deviations of PF and MS measures at the baseline and absolute changes in them were tested in the ANCOVA across the categories of protein intake. Multiple linear regression or logistic regression models were used to calculate β-coefficients and 95 % CI of PF and MS measures at the baseline and changes in them across categories of protein intake. Tests for a linear trend across categories of protein intake were conducted by using the median value in each category as a continuous variable in the linear and logistic regression models. Pairwise comparisons of the group means were performed with Tukey’s post hoc test. Linear and logistic regression analyses evaluated the association of FM and LM with PF and MS measures at baseline and over the 3-year follow-up. We examined further the association of protein intake of g/kg BW with PF measures at baseline and over the 3-year follow-up according to sarcopaenia status. To achieve balanced numbers of participants in the stratified analysis and to evaluate our secondary hypothesis, women were classified as sarcopaenic if they belonged to the pre-sarcopaenia, sarcopaenia and severe sarcopaenia (lowest quartile of RSMI) group, and non-sarcopaenic group was compiled from normal and non-classified groups (normal RSMI).

We initially assessed known covariates of frailty, including age, total energy intake, smoking status, alcohol consumption (portions/week), physical activity (h/week), hormone therapy use, osteoporosis and self-reported history of medical conditions (fall in the past 12 months, depression, diabetes mellitus, hypertension and rheumatoid arthritis) and also for baseline height, FM and LM. Further, covariates were selected on the basis of their multicollinearity and their predictive values alone, which led to selection of the following models. Model 1 presents the unadjusted results controlling only for age and energy intake. Model 2 was adjusted for variables in model 1 plus smoking status, alcohol consumption, physical activity, hormone therapy use, osteoporosis, LM and height. Model 3 was adjusted for variables in model 2, but LM was replaced by FM. Longitudinal analyses were adjusted for vitamin D and Ca supplementation (study group) to control for plausible vitamin D effect on physical performance, as well as PF and MS baseline measures to account for differential subsequent changes in physical performance depending on the initial physical performance measures. Comparing model 2 and 3 provided the opportunity to evaluate whether LM and FM differently associate with PF and MS, as suggested by previous studies^{( 4 , 44 , 45 )}.

All statistical analyses were executed using the SPSS software version 21 for Windows (IBM Corp.). The result was considered significant if a P value was <0·05.

Results

The participants were 65·3–71·6 years old (mean age was 68 (sd 1·9) years), and mean energy intake was 6560 (sd 1556) kJ/d (Table 1). Total protein intake was 68·2 g/d, which constituted to 17 % of total energy intake and corresponded to 0·96 g/kg BW. The minimum protein intake reported was 0·24 g/kg BW and the maximum was 2·25 g/kg BW. In addition, 30 % of women had protein intake ≤0·8 g/kg BW, 48 % were in the moderate range of 0·8–1·19 g/kg BW and 22 % consumed protein ≥1·2 g/kg BW. Higher protein intake was significantly associated with higher energy intake and lower carbohydrate intake as percentage of energy, but higher carbohydrate intake as g/d.

Table 1 Baseline characteristics of the participants in different protein intake categories (Mean values and standard deviations)

^a Mean values with unlike superscript letters of the lowest category was significantly different from the middle and highest categories after Tukey’s post hoc test.

^b Mean values with unlike superscript letters of the middle category was significantly different than from the highest category after Tukey’s post hoc test.

* ANCOVA and χ² tests were used to evaluate the differences between participants’ characteristics and dietary intake with protein intake categories as expressed per body weight according to different recommendations.

† Includes walking, gardening, cycling, cross-country skiing and other more strenuous activities (times/month×strenuousness).

In total, 8 % of women had osteoporosis, 42 % had hypertension, 3 % had diabetes, 6 % had rheumatoid arthritis, 3 % had depression, 12 % had hip arthrosis, 28 % had knee arthrosis and 21·8 % reported fall accident in the past 12 months. However, no significant associations between reported diseases and protein intake of g/kg BW were observed. Mean duration of hormone therapy was 11 years, and time passed after menopause was 18 years. Women with higher protein intake reported more frequent use of hormone therapy, weighed less and had lower BMI as compared with moderate and lower intake. Among body composition measurements, FM index, LM index and LM index (LMI) were significantly lower for higher protein intake. Women with higher protein intake had significantly higher RSMI than those with lower protein intake.

In Table 2, differences of baseline characteristics between non-sarcopaenic and sarcopaenic participants are presented. The sarcopaenic group (n 127) had significantly lower mean weight (−13·2 %), BMI (−12·7 %), FM (−16·0 %) and LM (−12·0 %) as compared with the non-sarcopaenic group (n 369). Average protein intake was similar in the sarcopaenic and non-sarcopaenic groups: 17·6 (sd 2·9) and 17·9 (sd 3·1) % of energy, respectively.

Table 2 Baseline characteristics of the participants according to sarcopaenia status (Mean values and standard deviations)

* Independent sample t test and χ² test were used to evaluate the differences between participant’s characteristics according to sarcopaenia status.

† Includes walking, gardening, cycling, cross-country skiing and other more strenuous activities (times/month×strenuousness).

Significant differences in physical performance measures between women with higher protein intake and those with lower protein intake at the baseline and over the 3-year follow-up were detected (Table 3). At the baseline after adjustment for selected factors previously described as associated with physical performance (model 2), those with higher protein intake as compared with those with moderate and lower intake had greater GS/BM (P=0·001), knee extension/BM (P=0·003), longer one-leg stance performance (P=0·047), better chair rise performance (P=0·043), faster WS for 10 m (P=0·005), squat completion (P=0·019), squat to the ground completion (P=0·001) and higher SPPB score (P=0·004). Overall results were no longer significant after controlling for FM (model 3). Results for the prospective analysis showed that those with higher protein intake had less decline in GS/BM (P=0·027), one-leg stance performance duration (P=0·024) and had increased tandem walk speed (P=0·024), which were no longer significant after controlling for FM.

Table 3 Physical performance measures in protein intake categories at the baseline and over the 3-year follow-up* (Mean values and standard deviations)

BW, body weight.

^a Mean values with unlike superscript letters of the lowest category was significantly different from the middle and highest categories after Tukey’s post hoc test.

^b Mean values with unlike superscript letters of the middle category was significantly different from the highest category after Tukey’s post hoc test.

* Tests for a linear trend across categories of protein intake were conducted by using the median value in each category as a continuous variable in the linear and logistic regression models. Median total protein intake for each category was 0·66, 0·98 and 1·34 g/kg BW, respectively.

† Model 1 was adjusted for age and total energy intake.

‡ Model 2 was adjusted for variables in model 1 plus smoking status, alcohol consumption (portions/week), physical activity level, hormone therapy use, osteoporosis, baseline height and lean mass.

§ Model 3 was adjusted for variables in model 2, but lean mass was replaced by fat mass.

|| Longitudinal analyses were adjusted also for physical performance baseline variables and Ca and vitamin D intervention.

In linear regression analyses with physical performance measures and SPPB as the dependent measures, results from models including energy-adjusted fat intake (g/d) or energy-adjusted carbohydrate intake (g/d) as determinant instead of protein showed no significant contribution for fat (g/d) and carbohydrate (g/d) (data not shown).

Further, we examined the association of protein intake with physical performance measures according to sarcopaenia status (Table 4). Results of model 2 showed that among non-sarcopaenic women protein intake was positively associated with GS/BM (β=0·35 and P=0·001), knee extension/BM (β=0·25 and P=0·008), one-leg stance performance (β=0·26 and P=0·001), chair rises (β=0·15 and P=0·039), WS for 10 m (β=0·30 and P<0·001), ability to squat (β=0·18 and P=0·003), ability to squat to the ground (β=0·29 and P=0·001) and also with SPPB score (β=0·32 and P<0·001) at the baseline. However, significant associations were lost after controlling for FM. Results of the prospective analysis indicated that higher protein intake in non-sarcopaenic women was in positive relationship with changes of one-leg stance performance (β=0·14 and P=0·037) and standing with eyes closed (β=0·23 and P=0·001). No significant associations between protein intake and physical performance measures were observed among sarcopaenic women, except for GS/BM change (β=0·23 and P=0·037) and a non-significant relation with chair rise change (β=0·27 and P=0·064), which were lost after controlling for selected confounders and FM.

Table 4 ·Effect of protein intake (g/kg body weight) and physical performance measures according to sarcopaenia status (Regression coefficients and 95 % confidence intervals)

* Model 1 was adjusted for age and total energy intake.

† Model 2 was adjusted for variables in model 1 plus smoking status, alcohol consumption (portions/week), physical activity level, hormone therapy use, osteoporosis, study group and baseline height.

‡ Model 3 was adjusted for variables in model 2 plus fat mass.

§ Longitudinal analyses were adjusted also for physical performance baseline variables and Ca and vitamin D intervention.

The associations between total body FM and LM with physical performance measures and changes in them are shown in Table 5. After adjustment for LM and factors previously described as associated with physical performance, FM was negatively correlated with GS/BM, GS, knee extension/BM (only at the baseline), one-leg stance, chair rises, WS for 10 m, squat, squat to the ground and SPPB score at the baseline and over the 3-year follow-up (β≥−0·07 and P≤0·050). FM was also negatively associated with change of standing with closed eyes for 10 s (β=−0·22 and P<0·001). Further, LM was positively associated with GS, knee extension and one-leg stance performance at the baseline, as well as with GS changes over the 3-year follow-up (β≥0·06 and P≤0·025). Results remained significant after controlling for FM.

Table 5 Association of total body fat mass and lean mass with physical performance measures at the baseline and over the 3-year follow-up (β-Coefficients with their standard errors)

* Model 1 was adjusted for age, total energy intake, smoking status, alcohol consumption (portions/week), physical activity level, hormone therapy use, osteoporosis and height.

† Model 2 adjusted for variables in model 1, and lean mass and fat mass were adjusted for each other.

‡ Longitudinal analyses were adjusted also for physical performance baseline variables and Ca and vitamin D intervention.

Discussion

This study examined cross-sectional and prospective associations of protein intake (g/kg BW) and body composition (FM and LM) with different PF and MS measures in 554 elderly women belonging to the OSTPRE-FPS study. Associations of protein intake with PF and MS were also evaluated according to sarcopaenia status. However, the significant associations were lost in the final models because of high collinearity of FM with physical performance. Our findings supported the hypothesis that protein intake higher than the current RDA (0·8 g/kg BW) might be associated with better PF and MS among elderly women. Further, the present study showed that the total body FM was negatively associated with physical performance tests, whereas total body LM was positively associated with GS, knee extension and one-leg stance.

In recent years, there has been increased support for the contention that the current daily allowance (0·8 g/kg BW) for protein is insufficient to promote optimal health and preserve physical performance in the elderly^{( 5 , 12 , 13 , 18 , 45 – 47 )}. Consistently, in our cross-sectional findings, those women with higher protein intake performed better in many of the physical performance measures as compared with those who had moderate and lower protein intakes. The higher protein intake category had greater GS/BM, knee extension/BM, longer one-leg stance, better chair rise performance, faster WS for 10 m, better squat and squat to the ground ability, and higher SPPB score. The prospective results also showed that women in the higher protein intake group had less decline in GS/BM and one-leg stance performance, and had the highest increased chair rise performance over the 3-year follow-up. No significant differences were observed between protein intake categories and WS for 10 m and tandem walk speed for 6 m prospectively. Thus, it might be that higher protein intake (g/kg BW) can be more related to preserving MS rather than mobility, which may partially explain the protein–frailty association. However, these associations were no longer significant after adjustment for FM.

Findings of the study by Gregorio et al.^{( 13 )} among 387 healthy women aged 60–90 years showed that those in the lower protein intake <0·8 g/kg BW category performed less well in the single-leg stance test than those in the higher protein intake ≥0·8 g/kg BW category. They also walked 8 feet at a slower pace, and their SPPB score was lower than in women in the higher protein category. Further, Lemieux et al.^{( 45 )} indicated that among seventy-two postmenopausal women higher protein intake ≥1·2 g/kg BW was positively correlated to GS and knee extension. Women’s Health Initiative clinical and observational study^{( 12 )} was conducted in 134 961 participants aged 50–79 years for an average of 7 years of follow-up. Results showed that mean GS at baseline was slightly higher among women with higher calibrated daily protein intake (using urinary N protocol to estimate protein consumption over 24-h period), and these women experienced a smaller decline in GS over time than those with low calibrated protein intake. In addition, women in the highest quintile of calibrated protein intake completed on average 0·5 more chair rises at baseline than women in the lowest quintile. In contrast, there was no significant association between calibrated protein intake and the timed 6-m walk in either cross-sectional or prospective analyses. Furthermore, the same results were shown when protein intake was expressed as g/kg BW.

A new finding was that among non-sarcopaenic women at the baseline protein intake (g/kg BW) was in positive relationship with GS/BM, knee extension/BM, one-leg stance ability, chair rise performance, WS for 10 m, ability to squat and squat to the ground, and SPPB. Protein intake in these women was also associated with preserving physical performance over the 3-year follow-up, including one-leg stance and standing with eyes closed for 10 s. No such association was observed in sarcopaenic women, except a positive relationship between protein intake and GS change. Thus, consistent with our hypothesis, the positive association of protein intake (g/kg BW) with PF was more pronounced in non-sarcopaenic than in sarcopaenic women. It has been suggested that older individuals suffering from illness, physiological stress or sarcopaenia are required to consume higher protein intake (1·2–1·5 g/kg BW) as compared with healthy older people (1–1·2 g/kg BW)^{( 30 )}. However, we could not explore this because of the threshold of protein intake in these data between sarcopaenic and non-sarcopaenic women.

A preponderance of evidence now suggests that ageing might result in the stimulation of MPS becoming resistant to the anabolic effect of hyperaminoacidemia, particularly at lower protein intakes^{( 24 , 30 , 48 – 50 )}. It was shown in the study by Moore et al.^{( 28 )} that the relative quantity of ingested protein required to maximise MPS is greater in older as compared with younger men^{( 18 )}. However, it is unestablished whether elderly individuals with greater LM have higher capacity of MPS as compared with those with lower LM. Besides, previous research indicates that protein from different sources (animal and plant protein) may have different effects on physical performance^{( 51 , 52 )}. However, this study did not find any significant association between animal and plant protein intake with PF and MS measures.

Declines in LM might predict a reduction in muscle force and performance^{( 1 , 48 )}. It has also been shown that FM is associated with functional decline and muscle weakness in elderly individuals^{( 35 , 44 , 53 )}. In this study, total body FM was in strong negative correlation with all PF and MS measures at baseline and changes in them at 3 years, except for knee extension, tandem walk and standing with eyes closed at the baseline, whereas LM was positively correlated with GS and the change in it, knee extension and one-leg stance. Therefore, these findings accompanied with the loss of significant associations between protein intake and physical performance measures after controlling for FM but not LM suggest that FM and LM may have opposite association with PF and MS in elderly women. There are different pathways through which fatness might be related to LM and MS^{( 54 )}. However, more studies are needed to disentangle the relationship between FM and physical performance.

It is well known that adequate energy intake is required to optimally utilise dietary protein to maintain physical performance rather than as an energy source^{( 13 )}. It was to our surprise that those with higher energy and protein intake had a lower weight. The actual cause is uncertain, but this might be because of the higher physical activity level in the higher protein category, and also because of the possible under-reporting of total energy and fat intake in those with higher BMI^{( 55 )}. Worthy of note is that LMI (LM/height (m²)) and RSMI are both used as indicators of muscle mass in the diagnosis of sarcopaenia^{( 2 )}. However, in this study, protein intake showed the same association with LMI and RSMI, and thus we used RSMI as a clinical indicator of sarcopaenia, as adapted by EWGSOP^{( 2 )}.

A limitation of this study was that the study population consists of only elderly women, and therefore caution should be taken when generalising the findings to elderly men. The 3-d food records method has been described as a suitable instrument for assessing energy and protein intake in elderly people^{( 56 , 57 )}. The latter study has also validated protein intake against urinary N studies in both community-dwelling and institutionalised elderly people^{( 57 )}. However, a single 3-d dietary record at the baseline might not be an appropriate method to capture the long-term effect of protein intake. Albeit we covered a wide selection for several known confounders that might influence physical performance, other factors such as health status, habitual physical activity level and/or dietary habits in participants in different protein intake categories might have affected the observed results. Last, based on the observational nature of our study, we cannot establish a causal association.

An additional analysis in the present data showed no significant effect of vitamin D (800 IU; 20μg) and Ca supplementation (1000 mg) on MS and PF, and longitudinal analysis was controlled for the study group receiving those. The availability of multiple standardised physical performance measures at baseline and over a 3-year period added significant strength to our study. Dynamometric measures of GS as a physical marker of lower limb strength and knee extension for a variety of functional tasks, such as walking, chair rising and stair climbing, particularly are predominate for the quantification of physical performance in older adults^{( 36 , 58 )}. The introduced protein intake categories in the present study took into account the newer intake recommendations for elderly, which have not been used in the previous studies.

Conclusion

It is appropriate to focus on the relationship between protein intake and MS and PF in the elderly, because this group is most vulnerable to nutritional deficiencies. This cohort study suggests that higher protein intake and lower FM might be positively associated with MS and PF in elderly women. However, further research is required to establish causal association.