Effects of Exercise Training on Handgrip Strength in Older Adults: A Meta-Analytical Review

Handgrip strength testing evaluates static force that is squeezed around the dynamometer, given in kilograms, pounds, or Newtons [1]. Handgrip strength assessment serves as a feasible, quick, and reliable tool to examine vitality and physical function in the elderly population. A standardized and calibrated equipment is required, especially when different assessors [2] or different brands of dynamometers are employed [3]. The test-retest reliability is considered high: The ICC (intraclass correlation coefficient) ranges between 0.91 and 0.95 for healthy older adults [4]. According to recommendations of the American Society of Hand Therapists [5], participants should be seated, with the shoulder adducted and neutrally rotated, the elbow flexed at 90°, and the forearm and the wrist in a neutral position. This society recommends three trials of grip strength and suggests using the best trial for further analysis [5].

Beside other declines of physical function (endurance, strength, balance capacity), aging people show a decrease in handgrip strength [6]. Women have weaker handgrip strength compared to men [6-8]. The handgrip strength reflects a variety of physical function indices and is therefore regarded as an important indicator of health-related quality of life of older adults [9]. Hence, handgrip strength predicts disability, morbidity, and mortality [10], accelerated dependency in activities of daily living and cognitive decline in the most elderly (over 85 years of age) [11]. A weak handgrip strength is also associated with hypertension, coronary heart disease, peripheral arterial disease, heart failure, stroke, or chronic obstructive pulmonary disease [8]. Furthermore, handgrip strength is negatively associated with physical frailty even when the effects of body mass index and arm muscle circumference are considered [12].

Strength, endurance, and balance training have been shown to attenuate numerous adverse effects of biological aging [13]. The occurrence and magnitude of potential transfer effects of these interventions with exercise training on handgrip strength as a well-established and feasible surrogate parameter in community-dwelling older adults have not yet been addressed systematically. Therefore, the objectives of the present systematic review and meta-analysis were: (1) to calculate the effects of numerous interventions with exercise training on handgrip strength in healthy community dwellers, (2) to describe training characteristics for older adults, and (3) to provide recommendations for future studies.

The PRISMA guidelines were used for reporting this meta-analytical review [14]. A literature search was independently conducted by two researchers (B.K.L. and H.B.) in three health- and sport-related databases (PubMed, Web of Science, SPORTDiscus). The search was conducted from November 25, 2018 until December 1, 2018. The relevant search terms were divided into three search levels and were combined with Boolean conjunctions (AND/OR). On the first search level “hand grip* OR grip strength OR grip power,” on the second level “AND (sport* OR train* OR exercis* OR strength OR intervention OR endurance OR resistance OR balance OR aerob*)” and on the third level “AND (old* OR elder* OR seniors*)” were used as search terms. The screening process consisted of title, abstract, and full-text screening. First, all resulting titles of the search were screened manually by the two researchers (B.K.L. and H.B.) and discussed for eligibility. All chosen articles were then transferred to a reference management software (Citavi 6.3, Swiss Academic Software GmbH, Waedenswil, Switzerland). Both researchers screened abstracts separately and their inclusion was discussed afterwards. Consent has been found upon the independent search procedure and through discussion. Irrelevant articles were excluded and the full texts were screened for the final inclusion/exclusion of the studies. The inclusion and exclusion criteria were defined based on the PICOS approach [population (P), intervention (I), comparators (C), main outcome (O) and study design (S)].

Only full-text articles published in English and with a target group of healthy, community-dwelling older adults of 60 years and older (P) were included in the meta-analysis. Further inclusion criteria were that studies had to be intervention studies (I) with an exercise intervention in the form of a controlled trial (C). There had to be at least one control group per study. Handgrip strength (O) had to be the main outcome and pre- and post-testing had to be performed (S).

The exclusion criteria were additional diet changes or nutrition supplements as part of the intervention. Furthermore, the authors aimed at investigating community-dwelling, healthy older adults, so that older adults living in a nursing home or receiving long-term care as well as older adults with mental decline, chronic diseases of the heart or vascular system, orthopaedic conditions, diabetes, sarcopenia, frailty, or obesity were excluded from the analysis.

To evaluate the methodological quality of the included studies, a scale by the Physiotherapy Evidence Database (PEDro scale) with 11 dichotomous items was used [15]. The two researchers independently rated all included studies and came to a consensus on every item after completing the individual rating (online suppl. table 1; see www.karger.com/doi/10.1159/000501203 for all online suppl. material). Researchers were not blinded to authors, place of publication, or results of the studies.

Handgrip strength outcomes (in kilograms, pounds, or Newtons) were extracted and transferred to a separate excel sheet that contained the equations for all relevant calculations that are described in the statistical analysis section. Furthermore, other relevant study information (author, year, number of participants, intervention groups, and details about the intervention) were extracted and compiled in Table 1.

Standardized mean differences (SMD, with 95% confidence intervals [CIs]) were calculated for each study according to Hedges’ (adjusted) g, which can be used for diverged and small sample sizes [16]. Negative values (–) reflect negative effects. The Cochrane Review Manager Software (RevMan 5.3.5, Cochrane Collaboration, Oxford, UK) was used to conduct the inverse-variance method [17]. Analyses were conducted using the random effects model [18]. The forest plot was generated for the respective outcome measure. The following scale was used to evaluate the standardized mean differences (SMDs) [19]: 0–0.19 = negligible effect, 0.20–0.49 = small effect, 0.50–0.79 = moderate effect, and 0.80 and higher = large effect. Study heterogeneity was assessed using I². A funnel plot was created to examine potential publication bias by comparing the effect sizes direction with the standardized mean differences. All statistical analyses were performed within the Cochrane Review Manager Software.

In the three databases, 5,752 relevant articles were initially detected (Fig. 1). After removing all irrelevant studies based on the title and removing duplicates, 137 articles remained for abstract screening. Ninety-eight studies did not meet the inclusion criteria. Thirty-nine articles were left for full-text screening. Three of them were excluded during data extraction because of lacking post-values for handgrip measurements. The authors were contacted in order to get the data for post-measurements which were not reported in the published version of the article. However, only one of the respective authors [20] reacted to the request and sent the missing post-values. Finally, 15 studies were excluded based on the defined exclusion criteria. For the final meta-analysis, 24 studies were included. All studies were published between 1995 and 2018.

In the 24 trials, 3,018 healthy community-dwelling older adults with a mean age of 73.3 ± 6.0 years were enrolled. The mean sample size was 81 ± 263, with a range from 22 [21] to 1,635 [22] participants.

Seventeen studies used a two-armed design [21-37], six studies used a three-armed design [38-43], and one study [20] used a four-armed design. The mean study quality was 5.8 (±0.9), ranging from 4 [37, 40] to 7 [23-25, 29, 39, 43]. None of the included studies blinded participants or instructors.

Training types of all 24 studies included aquatic exercise, walking, flexibility, TRX-training, home-trainer exercise, strength training in different forms, training on a vibration platform, dance, Tai Chi, exergames balance training, calisthenics, and multi-dimensional training regimes. Intervention durations ranged from 4 weeks [30] to 36 months (mean 22 ± 30 weeks) [22], whereas the majority of the interventions lasted for 8–15 weeks. Session duration varied between 15 [29, 43] and 72 [20] min per session (mean 51 ± 15 min per session) with most of the sessions lasting 60 min. Training frequency varied between one session per week [34] and two sessions per day [29, 43] (mean 3 ± 2 sessions per week) with mostly 2–3 sessions per week. The training intensity was measured differently with the Borg scale, 1 repetition maximum (1 RM), or maximum heart frequency (Hf max). Using the Borg rating of perceived exertion (RPE) scale the intensity ranged from 11 to 16 [22, 41, 42], and using the RPE ratio scale the intensity varied from 5 to 8 [30, 39]. Five studies [20, 27, 31, 40, 41] measured the training intensity with 1 RM ranging from 45 to 80%. The intensity measured with heart frequency varied from 60 to 80% of the maximum heart frequency [20, 21, 27, 34, 36, 37, 41]. Two studies did not report the intensity level [28, 35] and one study [23] used the Borg scale and heart rate monitor for assessing intensity but did not report values. One study reported not to have monitored training intensity [29]. Wolf et al. [43] reported modest training intensity in Tai Chi, and Campa et al. [24] encouraged the participants to increase intensity individually by changing position towards the anchor point of TRX. Englund et al. [26] increased the training intensity by using heavier dumbbells and by doing more complex and faster movements. Dondzila et al. [25] increased the walking intensity by 10% of baseline steps each week and the resistance training by increasing repetition and sets of exercises. Alexander et al. [38] used water exercise equipment for the aquatics exercise group to increase the intensity while the walking group exercised at a comfortable pace. Exercises increased progressively in the vibration platform training of Santin-Medeiros et al. [32]. The focus of the majority of the 24 included studies was on multimodal training, including aerobic and resistance exercises [20, 22, 23, 27, 28, 30, 36, 41].

The funnel plot (see online suppl. Fig. 1) shows a triangular funnel-shape with few outliers. The number of studies on the left and right side of the dashed SMD line seems equally distributed. Normally, studies with small sample sizes build the basis of the triangle. There is only one study [21] with a small sample size at the right bottom of the funnel plot. This two-armed controlled trial included 22 participants, 12 in the intervention group, and analysed the effects of a 24-week aquatic aerobic training. There are more studies with a bigger sample size of 30–70 participants, located in the middle of the funnel shape. The study with the biggest sample size [22] is clearly identified to the top left, outside of the funnel shape. This two-armed controlled trial included 1,635 participants, with 818 in the intervention group, and analysed the effects of a combined walking training the intervention group performed twice a week over 36 months.

Small effects were observed for handgrip strength outcome (p < 0.001, SMD 0.28, 95% CI 0.13–0.44, I² = 56%; Fig. 2) in favour of exercise training groups compared to control groups.

To the best of the authors’ knowledge, no previous systematic review with meta-analysis has addressed potential transfer effects of interventions with exercise training on handgrip strength serving as a well-established surrogate parameter of vitality and physical function in healthy community-dwelling older adults. Measuring handgrip strength is an inexpensive, quick, and objective means to evaluate physical frailty [44] or change in performance [45] in small and large samples. Furthermore, handgrip strength measurements can be used to identify older adults who are sarcopenic, frail, and malnourished [46]. This meta-analytical review provides a solid estimate of exercise-induced transfer effects of various interventions with exercise training on handgrip strength. The results of this review displayed small effects of a variety of interventions on handgrip strength measures.

In order to evaluate the effects of interventions with a measurement tool, i.e. handgrip strength dynamometer, besides reliability and validity, the standard error of measurement (SEM) and minimal detectable change (MDC) have to be known. In a recent systematic review, Bohannon [46] included 17 studies about the reliability and MDC of handgrip strength measurements in different older target groups. Four of the included studies included older community-dwelling adults without any disease. The review demonstrated a relative reliability with ICC >0.91 in these four studies with 211 participants with a mean age of 75.2 years [46]. Regarding absolute reliability, these four studies used the variables SEM, technical error of measurement (TEM) and least significant change (LSC). The SEM was 3.9 and 1.9 kg, TEM was 1.6 (left hand) and 2.2 (right hand), and LSC was 6.0 kg. These four relevant studies were evaluated by Bohannon [46] regarding their study quality with scores between 9 and 13 on a 14-point scale. If one takes the best results for absolute reliability for older community-dwelling older adults from Bohannon’s review [46], a change (for example via an intervention) above 1.6 kg can be interpreted as a real change.

Measuring handgrip strength with a dynamometer can be a useful assessment tool for determining the efficacy of different treatments and for an overall fitness assessment [45]. The studies included in this meta-analysis measured handgrip strength as a general parameter of the state of physical function but they did not include handgrip-specific exercises. In fact, the variety in exercise mode, duration, and intensity was large. For most of the studies, the “task-specificity principle” of neuromuscular adaptation, which was manifested for balance training, seems to apply: if the positions that are tested in pre- and post-testing are not trained during the intervention, no improvement compared to the control group is revealed [47, 48]. The outcome of this meta-analytical review supports the transfer of this idea of task specificity on handgrip strength training and measurements. Effects are more likely to appear when the tested tasks are closely related to the training contents [47-49]. Only small effects in favour of the exercise training groups, which were training with different training modes not specifically addressing handgrip strength, were found. In one of the included studies with the largest effect sizes in favour of the intervention group, participants trained with dumbbells, which included gripping of the weights [26]. Corroboratively, in another study with large effects, participants did Nordic walking [42], which included a repetitive movement similar to a gripping task by holding the pole and executing the correct Nordic walking technique. In contrast to handgrip strength assessment, Nordic walking improves the muscular endurance component and capacity instead of maximum strength, necessary for handgrip measurements. In most of the other studies, showing positive effect without task-specific training, transfer effects might account for positive changes. Additionally, it has to be taken into account that measurements of handgrip strength are usually performed isometrically, whereas exercises were mostly performed dynamically. To summarise, handgrip strength would benefit from training that mainly targets upper extremity exercises in which handgrip movements are performed.

As was mentioned, exercise mode, duration, and intensity varied among the studies included in this meta-analysis, and so did the changes in handgrip strength. As for training mode, improvements in handgrip strength indices were not only seen in interventions with pure strength/resistance training, but also in multimodal interventions as well as in a Nordic walking intervention. Interestingly, the majority of the studies that featured large effects in favour of the intervention group followed a multimodal approach to exercise training. It is assumed that handgrip strength capacity benefits more from incorporating different training focuses (strength, balance, flexibility, endurance). This finding should be taken into consideration when evaluating the efficacy of interventions with exercise training [48].

Almost all studies in this review met the recommendations of more than 12 weeks of intervention, except two studies that lasted less than 12 weeks [29, 30]. No clear relationship is apparent between the duration of the intervention and the changes in handgrip strength. One study with only 6 weeks of daily rocking chair exercise found a slightly significant change in handgrip strength in favour of the intervention group [29]. Greater changes in handgrip strength were measured after a 9-week calisthenics intervention [37]. On the other hand, the study with the longest intervention duration (36 month) did not show any effects in favour of the intervention group [22]. This finding can be attributed to the control conditions that were requested to complete home exercises. Additionally, the training frequency differed a lot between the studies, so that a statement on the effect of the study duration is hardly possible. A further factor potentially influencing effects on handgrip strength might be training intensity. When the intensity of the intervention was reported as between 60 and 80% of 1 RM, handgrip strength increased. Of those measuring intensity with 1 RM as reference, the intervention of Gudlaugsson et al. [27], where the participants trained with 6 RM, had the greatest changes in favour of the intervention group. In the studies using the Borg scale as the reference, handgrip strength increased when the perceived exertion was reported as “somewhat hard” or “hard” (12–16) [41, 42] with the exception of the study by Santanasto et al. [22]. Regarding Hf max, interventions with a variation of 15–20% between minimum and maximum heart frequency in training seem to be the most effective [21, 27]. Continuous training with 80% Hf max [20] is not more effective than training with 60% Hf max [41].

Due to the high variability of the training designs of all studies, it is not possible to tell the influence of specific training modalities (training mode, duration, or intensity) on changes in handgrip strength. The confidence intervals of four studies that did not touch the zero-effect line indicate a large effect in favour of the intervention groups [21, 26, 27, 42]. These studies included weight-bearing training over 12 months [26], endurance training combined with strength training over 6 months [27], aquatic aerobic training over 3 months [21], and Nordic walking over 3 months [42], and therefore obviously differed in terms of exercise mode, duration, and intensity despite comparable effects. Only five of the included studies trended towards a favourable effect for the control group [22, 25, 28, 33, 39]. These studies consisted of seated strength training [25], resistance and aerobic exercises [28] and dance-based aerobic exercises [33]. Additionally, the participants of one study did either low-loaded pedalling or unspecific exercises using elastic bands [39], and the remaining comprised workshops, strength, or balance training [22]. Again, no prototype for mode, duration, or intensity of the training can be found. The reasons for unchanged handgrip strength might be attributed to the age of the participants [39], a short intervention period [25], the lack of weekly training sessions [28], or short duration of single training sessions [22, 33].

Older adults who do not train beside the normal activities of daily living lose their body strength and arm strength [50]. Therefore, based on the present meta-analytical review, the authors recommend exercise training for community-dwelling older adults in order to improve or maintain functions of activities of daily living and to remain independent. The overall effect of the studies analysed in this meta-analysis and the overlapping intervals indicate that there is neither a clear effect in favour of the control nor the intervention group after an unspecific training intervention. Thus, if the aim is to increase handgrip strength, the authors recommend at least 9 weeks of task-specific training with variable heart frequency, 75% 1 RM, and perceived “somewhat hard” or “hard” exertion. Whether the same is true for frail older adults or for older adults living in care facilities remains unclear and needs to be addressed in future analyses. Illness and frailty could have a strong impact on the results so that these target groups need to be investigated separately. This meta-analysis therefore focused on the effects of training on handgrip strength of a healthy population only. According to the results, handgrip strength cannot clearly be recommended as an alternative outcome to assess general functional performance for all kinds of exercise programs. It can be assumed though that handgrip strength changes after multimodal interventions and thus can be used as an outcome measure for this training modality. However, it is questionable whether pure lower extremity exercise or endurance exercise programs, for example, substantially alter handgrip performance, even if they cause other performance changes.

This meta-analytical review was performed in accordance with the PRISMA statement [14]. Despite notable but acceptable heterogeneity of the included studies (intervention programmes, number of study arms, sample sizes), our findings provide a structured view on the transfer effects of interventions with exercise training on handgrip strength, independent of the underlying exercise regimen. It needs to be kept in mind that although handgrip strength is a feasible and easily applicable tool of vitality and functioning assessment in older adults, few interventions with exercise training include this outcome even if muscle strength has been targeted. In most of the included studies the training procedure and performed exercises were clearly described. The different interventions might bias the effects as none primarily promoted handgrip strength explicitly. The relative reliability of handgrip strength measurements is very good; however, the measurement tool showed weakness in absolute reliability. Furthermore, the achieved intervention effects have to be interpreted allowing for SEM and MDC.

The present meta-analysis focused on healthy, non-institutionalized older adults and revealed evidence that several interventions provide an appropriate training for improving handgrip strength, especially those with multimodal training. Small transfer effects can be explained due to the lack of task-specific exercises, the lack of weekly training sessions, or short duration of single training sessions. For meaningful conclusions, substantial changes in handgrip strength may be required to confidently conclude that real changes have occurred over time. Still, the effectiveness of exercise training is not to be underestimated. Beyond that, handgrip strength cannot clearly be recommended as an alternative outcome to assess general functional performance for all kinds of exercise programs, possibly apart from multimodal training modes. Furthermore, the effects of interventions with exercise training on handgrip strength in specific target groups, such as institutionalized or frail older adults, must be investigated in the future.

This meta-analytical review was prepared within the module “SBG11 – Review” in the master study program “sport and movement gerontology” (Master of Science) at the German Sport University Cologne.

The authors have no ethical conflicts to disclose.

The authors have no conflicts of interest to declare.