In community-dwelling older adults, the geriatric syndrome of frailty is common [1]. Frailty is defined as a state of high vulnerability and is caused by malnutrition, chronic inflammation and sarcopenia [2], which is a progressive loss of muscle mass in combination with a decrease in muscle strength or physical performance [3].

The consequences of frailty are adverse health outcomes such as disability, dependency, hospitalisation and need for long-term care [2]. Furthermore, when compared to robust community-dwelling persons, frail adults demonstrate significantly lower quality of life (QoL) [47]. Since sufficient energy, freedom from pain and the ability to perform the activities of daily living are important factors influencing QoL [8], it can be assumed that disabilities, physical limitations and deterioration of psychological well-being are possible explanations for the poorer QoL of frail adults [4, 5].

There is evidence that low daily physical activity (DPA) is associated with poor QoL in older adults [9, 10]. Furthermore, previous studies of frail persons have demonstrated that muscle strength, as represented by handgrip strength [3], plays an important role regarding QoL [6, 7, 11]. Since muscle mass is an important prerequisite for muscle strength [12], it is clear that there is also an association between muscle mass and QoL. To the best of our knowledge, no study to date has observed this relationship in prefrail and frail adults. However, some studies have showed that not only loss of muscle mass but also muscle quality (e.g. muscle composition, metabolism, neural activation, fibrosis) contributes to the age-related decline in physical performance and mobility [1315]. The link between physical performance and QoL in frail adults has been demonstrated in previous research. Accordingly, an association between slowness (assessed by gait speed or the Timed Up and Go test) and QoL has been shown [6, 11]. Furthermore, Gobbens et al. [7] revealed that, in addition to handgrip strength, difficulties in maintaining balance and difficulties in walking are associated with poor QoL in frail adults living in nursing homes.

Since frailty is a public health challenge [16], and the number of frail persons is expected to increase in the future [1], it is of particular importance to better understand the factors associated with poor QoL. Thus, the aim of this analysis was to examine the associations between DPA, handgrip strength, appendicular skeletal muscle mass (ASMM), physical performance and the different QoL domains in prefrail and frail older persons still living in their own homes.


Study sample

Data for this cross-sectional analysis were derived from the baseline assessment of a randomised controlled intervention study, conducted between September 2013 and July 2015 in Vienna, Austria. The study protocol has been previously published [17]. In this study, persons older than 65 years, who were still living in their own homes, were included. These persons had to be prefrail or frail according to the Frailty Instrument for primary care of the Survey of Health, Ageing and Retirement in Europe (SHARE-FI) [18]. SHARE-FI is a sex-specific calculator that includes items concerning exhaustion, weight loss, handgrip strength, slowness and low activity. SHARE-FI is based on discrete factor scores, and it divides persons into robust (female <0.315; male <1.212 points), prefrail (female <2.103; male <3.005 points) and frail (female <6; male <7 points). As prefrail and frail persons were included, females had to score more than 0.315 points and males more than 1.212 points, respectively. In addition, adults at risk of malnutrition or persons who were malnourished according to the Mini Nutritional Assessment Short-Form (MNA®-SF ≤ 11 points) were included [19]. As only one participant in the main study was at risk of malnutrition without being at least prefrail, we excluded this person from the present cross-sectional study to harmonise the sample. Furthermore, since the data were baseline data from a randomised trial, participants had to be willing to be visited at home by trained lay volunteers twice a week to perform six strength exercises and talk about nutrition-related aspects [17]. Persons with impaired cognitive function according to the Mini-Mental State Examination (MMSE < 17 points), insufficient German language skills, chemo- or radiotherapy at the moment or planned, insulin-treated diabetes mellitus, chronic obstructive pulmonary disease stage III or IV and patients with chronic kidney insufficiency with protein restriction or on dialysis were excluded. Persons living in nursing homes or retirement housing were also not allowed to participate in the study.


The following measurements were taken at participants’ homes by members of the study team (sports and nutritional scientists). Due to impaired vision, all items of the questionnaires were read aloud to the participants.

Daily physical activity (DPA)

The Physical Activity Scale for the Elderly (PASE) [20] was used to assess DPA. This is a validated questionnaire for persons over 55 years [21], which includes items concerning: (1) time spent sitting; (2) time spent walking outdoors; and (3) time spent on light, (4) moderate and (5) strenuous sports [20]. In addition, the following yes or no questions concerning household activity were asked: (6) light household tasks; (7) exhausting household tasks; (8) repair work; (9) light gardening; (10) exhausting gardening; and (11) caregiving activities. In order to analyse the questionnaire, these 11 items were multiplied by a weight score dependent on the level of exhaustion. Finally, all the items were summed. The range of possible scores was from 0 (worst score) to 360 (best score) [20].

Appendicular skeletal muscle mass (ASMM)

Body composition was assessed with phase-sensitive bioelectrical impedance analysis (BIA 2000-S device; Data input®, Darmstadt, Germany). For this purpose, participants were placed in a supine position and four electrodes were attached to each person’s dominant hand and foot [22]. An alternating current was then passed through the body to measure resistance (R) and reactance (Xc) [23]. ASMM was calculated using the validated formula of Sergi et al. [24]:

$${\text{ASMM }}\left( {\text{kg}} \right) = - 3.964 + \left( {0.227 \times {\text{height}}^{2} /R} \right) + \left( {0.095 \times {\text{weight}}} \right) + \left( {1.384 \times {\text{sex}}} \right) + \left( {0.064 \times X_{c} } \right)$$

Handgrip strength

Handgrip strength was measured with a hydraulic dynamometer (Jamar®, Lafayette, Louisiana) following the standard procedure [25]. Accordingly, participants were placed in a sitting position on a chair, with their forearms on the arms of the chair and their wrists over the end. The thumb was placed facing upwards. After a short demonstration, each participant performed three attempts on each side, alternating between the right and left hand. Between each attempt, there was a break of 1 min. Finally, the highest value of all six measurements was taken and analysed.

Physical performance (balance skills, gait speed, chair stands)

Physical performance was assessed with the Short Physical Performance Battery (SPPB) test [26]. This test is subdivided into three categories, namely, balance skills, gait speed and chair stands. Balance was assessed using side-by-side, semi-tandem and tandem stands. If the first two tasks were possible, participants scored 1 point; if a tandem stand was possible for <3 s, 0 points were given; if a tandem stand was possible for >9 s, participants scored 2 points. Gait speed was tested with a single 4-m walk, with or without assistive devices such as a wheeled walker. Results were divided into four categories (not possible = 0 points; >8.7 s = 1 point; 8.70–6.21 s = 2 points; 6.20–4.82 s = 3 points; <4.82 s = 4 points). The ability to rise from a chair and return to seated position five times with arms crossed was also tested. These results were again divided into four categories (not possible or <60 s = 0 points; >16.7 s = 1 point; 16.69–13.70 s = 2 points; 13.69–11.20 s = 3 points; <11.19 s = 4 points). Finally, a performance score was calculated, summing all the results. The range of possible scores was from 0 (worst) to 12 (best performance).

Quality of life (QoL)

The German version of the World Health Organisation Quality of Life-BREF assessment (WHOQOL-BREF) [16], an abbreviated, cross-culturally validated version of WHOQOL-100 [27], was used to assess QoL. The assessment consists of 26 items with a five-point Likert scale response format. The first two questions assess the ‘overall QoL’ of the past 2 weeks, whereas the remaining questions assess QoL in four different domains: ‘physical health’ (seven items), ‘psychological health’ (six items), ‘social relationships’ (three items) and ‘environment’ (eight items). According to the standard procedure [28], all the domains were scored and transformed into a scale ranging from 0 to 100, where a lower value indicates a lower QoL. The ‘social relationships’ domain was calculated using two instead of three items, because only eight participants replied to the question ‘How satisfied are you with your sex life?’

In addition to WHOQOL-BREF, the following four domains of the German version of the World Health Organisation Quality of Life-OLD assessment (WHOQOL-OLD) [29] were added: ‘sensory abilities’ (four items), ‘autonomy’ (four items), ‘past, present and future activities’ (four items) and ‘social participation’ (four items). All the QoL domains used in this study and a brief description of their components are shown in Fig. 1.

Fig. 1
figure 1

Used domains of WHOQOL-BREF and WHOQOL-OLD and a brief description of their components

Further measurements

Age, education level (‘elementary school or no degree’, ‘secondary school’, ‘university entrance diploma or higher degree’) and comorbidities were recorded. In addition, each participant’s medication was also documented. Furthermore, body mass index (kg/m2) was calculated by dividing the body weight (kg) (as measured by a calibrated scale) by the squared body height (m2) (which was measured with a tape).

Statistical analysis

Based on a median split, study participants were divided into ‘low overall QoL’ (≤ 40 points) and ‘high overall QoL’ (>40 points). Group differences in the continuous variables were assessed by t tests or Mann–Whitney U tests, depending on the distribution. For group differences in the categorical variables, Chi-square tests were used, and in the case of the group being smaller than five persons, Fisher’s exact tests were applied. Whenever an item of WHOQOL-BREF or WHOQOL-OLD was missing, the mean of the other items belonging to this domain was calculated [28]. This was undertaken in all domains except for ‘social relationships’, which was calculated with two instead of three items. As a measure of reliability, the internal consistency was determined for each single domain of WHOQOL-BREF and WHOQOL-OLD using Cronbach’s alpha. Furthermore, the correlations between the included independent variables were analysed using Spearman’s or Pearson’s correlation coefficients.

Multiple linear regression analyses were performed to determine the associations between the included variables and QoL. In the first model, a single variable was included as an independent variable, adjusted for age and sex. In the second model, we wanted to identify the strongest indicator for each QoL domain. We also examined how these variables explained each QoL domain. Thus, we undertook a stepwise multiple linear regression analysis including all the variables (DPA, handgrip strength, ASMM, balance skills, gait speed, chair stands). However, we only included variables with a p value threshold of 0.20. As in model 1, model 2 was adjusted for sex and age by entering these variables in the models irrespective of their significance. For all the statistical analyses, IBM® SPSS® Statistics 20 software (IBM Corp., Armonk, NY, U.S.) was used. All the tests were two-sided, and a p value of <0.05 was considered to be statistically significant.


In total, 482 people were screened for eligibility, 285 of them in hospitals. Since 208 inpatient individuals close to discharge did not meet the inclusion criteria, 54 refused participation and 19 were excluded for other reasons, only four subjects were recruited in the hospitals. The remaining 80 subjects were recruited via the media: 197 people responded to two editorial features, 47 did not meet the inclusion criteria, 34 refused participation after being provided with detailed project information and 37 were not included for other reasons. Finally, 29 prefrail and 54 frail participants (86 % women) were included in this analysis. Characteristics of the study sample are provided in Table 1.

Table 1 Characteristics of the study sample based on a median split of the ‘overall quality of life’ variable

Using Cronbach’s alpha, the internal consistency was determined for each single domain: ‘overall QoL’ (α = 0.662), ‘physical health’ (α = 0.673), ‘psychological health’ (α = 0.658), ‘social relationships’ (α = 0.580), ‘environment’ (α = 0.624), ‘sensory ability’ (α = 0.919), ‘autonomy’ (α = 0.640), ‘past, present and future activities’ (α = 0.636) and ‘social participation’ (α = 0.491).

The correlation coefficients within the included independent variables are shown in Table 2. Accordingly, DPA was found to be associated with balance skills, gait speed and chair stand, but not with handgrip strength and ASMM. The strongest significant correlation was found between DPA and balance skills. Furthermore, a moderate association between handgrip strength and ASMM was identified along with a weak association between handgrip strength and chair stands. In Table 3, the associations between each independent variable and the QoL domains, adjusted for sex and age, are presented. In this regard, DPA was found to be significantly associated with ‘overall QoL’ as well as with ‘physical health’, ‘psychological health’, ‘autonomy’ and ‘social participation’. Handgrip strength was found to be significantly related to ‘overall QoL’. Balance skills was found to be associated with ‘overall QoL’ and the QoL domains of ‘physical health’, ‘psychological health’, ‘environment’, ‘autonomy’ and ‘social participation’. Gait speed and chair stands were found to be related to social participation’ only.

Table 2 Correlations between included independent variables
Table 3 Model—linear regression models including one independent variable (e.g. handgrip strength) and one QoL domain (dependent variable), adjusted for sex and age

According to the multiple linear regression analysis (Table 4), DPA, handgrip strength and balance skills together explained 31 % of the variance in ‘overall QoL’. Furthermore, balance skills alone explained 20 % of the QoL domain of ‘physical health’. DPA, handgrip strength and balance skills were independent indicators for the QoL domain of ‘autonomy’ (R 2 = 0.247), whereas balance was the strongest indicator. Moreover, DPA and balance skills together explained 36 % of the variance in the QoL domain of ‘social participation’, and balance again showed the strongest association.

Table 4 Model—multiple linear regression model including all independent variables (e.g. handgrip strength) and one QoL domain (dependent variable), adjusted for sex and age


The main findings indicated that there was no association between skeletal muscle mass and QoL, whereas balance skills, DPA and handgrip strength were associated with QoL. Furthermore, balance was the factor most strongly associated with the QoL domains of ‘physical health’, ‘autonomy’ and ‘social participation’.

Before discussing the associations, it ought to be mentioned that when compared to previous trials, our participants scored similar values in the QoL domains, except for ‘social relationships’, ‘environment’ and ‘physical health’ [30, 31]. The higher scores in the ‘social relationship’ domain might be explained by the fact that we excluded the question ‘How satisfied are you with your sex life?’ Higher scores in the ‘environment’ domain might be explained by the different environmental circumstances of the countries. Lower scores in the ‘physical health’ domain might be traced back to the fact that we only included prefrail and frail persons, i.e. persons with defined physical limitations.

The correlation between handgrip strength and QoL is in accordance with other studies [6, 32, 33]. As handgrip strength is an overall measurement of body strength in older adults [34, 35], our results indicate that muscle strength is an important factor for QoL, whereas muscle mass is not. Hence, muscle quality and factors such as muscle composition, neural activation, metabolism and fibrosis might be relevant [1315]. Our findings also revealed that balance was the variable most strongly associated with the various QoL domains. An association between balance and poorer QoL was also described by Gobbens et al. [7]. This relationship might be due to the fact that balance is the most important requirement in daily life [36], and problems in maintaining balance lead to a restriction of activities due to the fear of falling [37]. In this context, it is noteworthy that muscle strength and muscle mass are important biomechanical requirements for maintaining balance [38]. However, in our sample, neither muscle strength nor muscle mass was found to be associated with balance skills. The nonsignificant correlation is comparable to the findings of Visser et al. [34] and a British study [39]. A reduction in the association between muscle strength and balance over the lifespan has also been confirmed in the current literature [40], with a change in the neuromuscular components being identified as the underlying reason [40]. Misic et al. [41] showed that muscle quality and not muscle mass was the strongest independent factor for balance in older adults. However, apart from muscle quality, limitations in the sensory system (visual, vestibular, proprioceptive, tactile somatosensory), cognitive impairments and orthopaedic problems also influence balance [36, 38]. Hence, the data indicate that it is not muscle mass, but rather factors such as muscle quality, constraints in the sensory system and orthopaedic problems that are closely linked to QoL in prefrail and frail persons. Further research on this assumption is needed.

As previous studies have showed, ‘social participation’ and contact with neighbours are important factors for the well-being and mental health of older persons [42, 43]. As the recent study of Etman et al. [44] showed that limited ‘social participation’ is associated with further worsening of frailty symptoms, this QoL domain is of special interest. Our data showed that individuals with better DPA and balance skills have a better QoL in ‘social participation’, indicating that balance and DPA should be kept as high as possible. However, it could also be the other way around: QoL in the ‘social participation’ domain should be increased to enhance balance and DPA. Apart from this, the fact that good balance is an essential precondition for leaving the house and for participating in social activities might be the reason for the close association. The same considerations might also apply to the QoL domain of ‘autonomy’.

A major strength of our study was the inclusion of very old community-dwelling prefrail and frail subjects. We used reliable and valid measurements to assess variables such as muscle mass, DPA and QoL. One limitation to the study design was that a temporal and causal link between independent variables and QoL could not be proven. The small sample size was another limitation. Nevertheless, we were able to detect the effects of the physical training and nutritional intervention carried out by the trained lay volunteers. However, the internal consistency was lower than in other validation studies [45, 46]. Hence, an acceptable internal consistency of >0.70 [47] was only achieved in the ‘sensory ability’ domain. However, the domain scores were sufficient for the study purpose, as the correlation between the items in each domain was adequate. Nevertheless, these questionnaires should be validated for prefrail and frail persons in further research. Moreover, ASMM was calculated based on the results of the bioelectrical impedance analysis using the validated formula of Sergi et al. [24], who validated the ASMM calculation for individuals with a mean age of 71.4 years (SD: 5.4) without chronic comorbidities. Due to this fact, this formula might not be directly comparable to our study participants since our population was both older and had chronic comorbidities.


As skeletal muscle mass was neither associated with ‘overall QoL’ nor with any QoL domain, skeletal muscle mass can be considered as not playing a role in the QoL context of prefrail and frail older persons. However, balance skills and DPA are relevant factors. These parameters were particularly associated with the QoL domains of ‘social participation’ and ‘autonomy’. However, we do not know whether low balance skills and low DPA are the cause or the consequence of low QoL.