In this study no differences were found in the amount of protein intake between frail, pre-frail and non-frail community-dwelling older adults. With regard to the distribution of protein intake, frail subjects showed a different and more uneven distribution of their protein intake over the day with lower intake at breakfast and higher intake at lunch. To our knowledge this is the first study investigating the association between the distribution of protein intake and frailty.
Contrary to Bartali et al. , who found an association between low protein intake (lowest quintile) and frailty in a cross-sectional analysis of the InCHIANTI study, in our sample the risk of frailty was not reduced in the quartiles of higher intakes compared to the quartile of the lowest protein intake. Furthermore, there is some evidence on the association of protein intake with walking speed  as well as handgrip strength , which we could also not identify in our analyses. This might be due to the relatively high protein intake even in the lowest quartile, where the cut off was ≥0.9 g/kg BW protein/day. This is above the value of 0.8 g/kg which is the present recommendation for protein intake in younger as well as in older adults , and has been found to be a threshold for negative nitrogen balance , low muscle mass  and more health problems after 10 years .
Our study is the first to investigate the association between the distribution of protein over all the meals and frailty. Regarding evenness of this distribution we found a more even protein eating pattern in non-frail participants than in frail and pre-frail (Figure 1) and in those reporting exhaustion and slow walking speed (Table 5). These results are in line with the recommendations of Paddon-Jones et al.  to equally distribute the daily protein intake of older adults over breakfast, lunch and dinner. Arnal et al. , in contrast, found one large serving of protein (80% of daily protein intake in one of three meals) to be more effective to promote a positive nitrogen balance and muscle synthesis in older adults than protein intake evenly spread over four meals. This discrepancy may be explained by the low absolute amount of daily protein administered to Arnal et al.’s participants (approx. 65 g). That means that in the spread feeding group a single meal contained hardly 20 g of protein, the minimum needed for stimulating muscle protein synthesis according to Paddon-Jones et al. , whereas with a higher total daily protein intake, as found in our sample, an even distribution results in more meals containing at least 20 g protein. At the same time, fasting losses between meals, which increase with the amount of protein of the meal , are reduced in a more even distribution. Finally, comparing our data with trials on nitrogen balance and muscle protein synthesis we must be aware of the fact that these results may not be totally transferable to issues of physical performance and frailty.
For the assessment of usual dietary habits, we used the FFQ of the German part of the EPIC study, which is well validated . A limitation of this questionnaire is the proven underreporting of energy and protein intake of about 20% . Thus, actual protein intake in our participants is supposedly higher. The validity of the results on distribution of protein intake are limited by the fact that although our 46 a priori set of main protein sources obviously covered all important sources of actual dietary protein, they still left an average of 26.2% of dietary protein undocumented. This may be due to some of our participants consuming very small amounts of protein, mainly from vegetable foods that were not considered as main protein sources, e. g. fruit (especially bananas) and vegetables (especially tomato sauce).
We are aware of the limitation that we only assessed protein intake at the main meals and only for selected foods. This was decided to avoid an undue length of the FFQ, which would probably have overstressed our aged participants. On the other hand Roussett et al.  found in a cross-sectional study in healthy older adults, that snacks only contributed to 1.4% in men and 2.3% in women to protein intake. Therefore snacks were seen as negligible in our distribution analysis. Up to now, there is no common method for evaluating protein distribution over all meals. Using the CV as measure for unevenness of distribution can be seen as a first approach. Certainly, assessment tools have to be developed to enable a more detailed evaluation of meal habits and statistical methods have to be adapted to further clarify the impact of protein distribution on sarcopenia, physical performance and frailty.
A major limitation of the study is its cross-sectional design which does not allow any statements on causal relationship. It is also plausible that frailty vice versa affects protein intake, for example by impairments in going shopping, chewing or swallowing (Table 1). Another critical point is that we did not consider the influence of protein quality on frailty, although it is an important regulator of protein metabolism. The small sample size, especially in the group of frail participants limits the study’s statistical power. Nevertheless, we detected significant associations. Furthermore, the sample consists of volunteers and may therefore lack in representation of the older German population in general.
A strong advantage of our study is that all FFQ have been conducted in personal interviews by a single experienced nutritionist, and for all assessments well validated tools have been used. This is the first study investigating the distribution of protein intake in German older adults and the association between this distribution and frailty. Unique is the development of a parameter of unevenness of protein distribution, which to our knowledge has not been used before in this setting.
In summary, in our study group of very old independently living senior citizens only few were frail and even the lowest quartile of protein intake was above the recommendation of 0.8 g/kg BW. There was no significant difference in the amount of protein ingested between frailty groups and we found no reduced risk for frailty in the quartiles with a higher protein intake compared to the lowest quartile. Our results also showed a relation between frailty and the distribution of daily protein intake over the main meals. Studies on the effect of protein intake on functional and clinical outcomes are still scarce. Therefore we recommend further investigation on this topic on the basis of the results of our study.