Correlation between Handgrip Strength and Depression in Older Adults—A Systematic Review and a Meta-Analysis

Background: Depression remains an important health problem among older adults and it may be correlated with the deterioration of physical fitness, whose chief indicator is hand grip strength (HGS). The aim of the study was to investigate the relationship between depression and HGS among older populations using the available literature. Methods: PubMed, Web of Science and Science Direct databases were searched. The inclusion criteria were as follows: written in English and published after 2009, subject age: ≥60 years, HGS measured using a hand dynamometer, assessment of the depressive symptoms using a validated tool. The following articles were excluded: studies conducted among institutionalized subjects and/or populations with a specific disease. Results: The total combined effect of 33 results presented in 16 studies included in the meta-analysis, converted to the correlation coefficient, was OEr = −0.148 (SE = 0.030, 95%CI: −0.206–−0.091), indicating a weak, negative correlation between HGS and depressive symptoms. Conclusions: The review of the literature and the meta-analysis demonstrated a relationship between low muscle strength and intensified depressive symptoms in older populations. Bearing in mind that depression is often unrecognized or underdiagnosed among older patients, lowered muscle strength should be an important sign for physicians and an incentive to screen them for depression.


Introduction
Depression remains an important health problem among older adults, resulting in lower cognitive ability and quality of life [1]. According to the World Health Organization (WHO), approximately 7% of the general older population were estimated to suffer from unipolar depression [2]. Depressive disorders in older age result from the accumulation of various factors, chief among them somatic diseases, stressful life events (e.g., loss of partner), loneliness, social isolation, unfavorable social attitudes towards older people, declined cognitive function, malnutrition, and polypharmacy. In many cases, depression may be the consequence of diseases which are typical for older populations, e.g., arteriosclerosis, hypertension, arrhythmia, diabetes, Parkinson's disease, or osteoporosis.
According to a number of studies, as many as 80% of older people are affected by at least one somatic condition [3,4], and 69% patients with depression report health complaints which are solely somatic in nature: pain, feeling of heaviness, fatigue, disturbed sleep, appetite loss, impaired gait and functional performance [5]. Depression may negatively affect the treatment of chronic diseases, thus worsening the prognosis, due to appetite loss, reluctance to comply with medication regimen, avoidance of social contacts, increasing social isolation, and unwillingness to engage in any forms of physical activity, including rehabilitation [6].
On the other hand, an equally impressive amount of data demonstrated that higher cardiorespiratory fitness and regular aerobic exercise prevent the development of depression among older individuals [7]. According to the WHO guidelines on minimal physical activity for older populations, 150 min/week of moderate-intensity (brisk marching, cycling, swimming) or 75 min/week of vigorous-intensity (jogging) aerobic exercises are recommended [8]. Physical activity contributes to maintaining physical and mental health and exerts beneficial effects on the treatment of depression.
According to the available literature, a relationship between low muscle strength and cognitive function impairment and the risk for developing neurodegenerative diseases such as Alzheimer's or Parkinson's disease has been confirmed [9]. Furthermore, decreased muscle strength may correlate with depressive symptoms among older subjects [10]. Handgrip strength (HGS) test is a simple method of measuring the extent of muscle power loss in clinical practice. The European Working Group on Sarcopenia in Older People 2 (EWGSOP 2) treats HGS as an index of global muscle power and as one of the criteria for the diagnosis of sarcopenia [11,12]. Sarcopenia, which may be defined as loss of skeletal muscle combined with decreased muscle strength and/or physical performance, is an increasingly common problem of older populations globally [12]. It is associated with higher risk for mortality, deteriorated functional performance, falls and hospitalization, as well as increased risk for depression in that age group [13]. In the elderly, the loss of muscle strength can also be caused by dynapenia, but there is no loss of muscle mass [14].
The impact of depressive symptoms on HGS has been investigated for various age groups. In a systematic review [10], summarized the studies on the relationship between muscle strength and depressive symptoms among middle-aged and older subjects, and emphasized the fact that lower risk for the development of depressive symptoms correlated with higher muscle strength.
The aim of the study was to investigate the relationship between depression and HGS among older populations using the available literature.

Materials and Methods
The review of the literature and the meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [15]. The procedures (search strategy, inclusion/exclusion criteria, and data extraction) were established and included in the protocol.

Search Strategy
PubMed, Web of Science and Science Direct databases were searched. An algorithm with key words ('handgrip strength' AND 'depression' AND 'older' OR 'elderly') was used to identify the publications. Additionally, the reference sections of the included articles were manually inspected to identify additional records. Two authors (EZ and TT) conducted their independent searches between 1 January 2010 and 17 January 2020. Only articles published in English were taken into consideration.

Inclusion and Exclusion Criteria
The inclusion criteria for the reports were as follows: (a) written in English and published after 2009, (b) subject age: ≥60 years, (c) HGS measured using a hand dynamometer, (d) assessment of the depressive symptoms using a validated tool.
The following articles were excluded: studies conducted among (a) institutionalized subjects and/or (b) populations with a specific disease, e.g., malignancy or carpal tunnel syndrome.

Data Collection and Analysis
All articles were independently analyzed by two authors (E.Z. and T.T.) to remove duplicates. The results were reviewed, and full versions were checked for compliance with the inclusion and exclusion criteria. The following data were extracted from each study: first author, year of publication, study population characteristics, study design, inclusion/exclusion criteria, method of assessing HGS and depressive symptoms, assessment of the outcome, and results. Next, methodological quality standards and risk of bias (using the Newcastle-Ottawa Scale (NOS) adapted for cross sectional studies were investigated [16]. NOS assesses three parameters (selection, comparability, and outcome) on eight items, with the maximum score of 10 ints. Studies with the score of <5 points (<5NOS) are considered to have high risk of bias. The questionnaire was adapted for the present study in the following way: (a) in the Selection section (Ascertainment of exposure), 1 point was awarded if the measuring tool was presented but the method was either not described or not compliant with the American Society of Hand Therapists (ASHT) guidelines [17], and 2 points were awarded if the tool was presented and the method was compliant with the ASHT guidelines; (b) in the Outcome section (Assessment of outcome), 2 points were awarded if HGS and depressive symptoms were assessed by two independent blind assessors. Methodological quality assessment was conducted independently by two authors (E.Z. and T.T.). All conflicts were discussed and if consensus could not be reached, the third author (A.P.) had the deciding vote.

Measures of Effect Sizes
The collected results had been presented in various ways and had to be converted to the correlation coefficients. The highest number of relationships between depressive symptoms and HGS was expressed as beta coefficients, obtained by the linear regression analysis (n = 12). The linear regression method allowed to control for the effects of mediator variables such as age, sex, education, and others, and that is why these values represent a partial effect of HGS on the level of depression. We converted these values to r_eq using the formula √ t 2 /t 2 + N − 2. Odds ratio (OR) calculated for cross tables was the next most frequent (n = 9) measure. The tables were prepared based on the threshold values for the number of depressive symptoms and according to the generally accepted thresholds for HGS or the thresholds calculated using the Receiver Operating Curve (ROC). We converted these values to r_eq using first the formula d = ln(OR) · √ 3/π. and then the formula d/ √ d 2 + 4. Other indices of the effect size expressed as r-equivalent were calculated using mean values in the study and control groups (n = 4), correlation (n = 2), contingency coefficient (n = 2, here we also used means and standard deviations), standardized beta value (n = 1, similarly to stated earlier formula for betas) and d Cohen (n = 1, similarly to stated earlier the second formula for converting OR). In two cases, the identified texts did not report the effect size but only the p-values. Given the sample size and assuming test power of the analyzed studies (1 − β = 0.80), the minimal effect was calculated (we have calculated this with the calculator on the website https://www. campbellcollaboration.org/escalc/html/EffectSizeCalculator-R7.php). The total effect for the relationship between depressive symptoms and HGS was not described in six studiesthe results were presented for groups with varying levels of depression and different sex groups, but it did not differentiate the total effect (Z = 0.715, p = 0.475).
Most studies did not present sex-stratified data as they used statistical measures which allowed to control for sex, but also other variables. Nevertheless, nine results were related to male subjects only, also nine to female subjects only, and eleven to male and female participants. No statistically significant differences in the effect size were found between these groups (F(2, 15.6) = 0.741, p = 0.493).

Statistical Analysis
Statistical analysis was conducted using the Jamovi software (2020), Jamovi project, Sydney, Australia, with the Viechtbauer metaphor package [18]. Publication bias was assessed visually by funnel plots and statistically by Egger's test. The τ index, calculated using the Maximum-Likelihood method, was τ = 0.147 and was statistically significantly different from zero (Q(32) = 280.4, p < 0.001) and revealed high heterogeneity of the results. The I 2 index was 92.3%, which justifies the use of the random effects model in the meta-analysis.

Results
Out of 473 records identified during the database search and 21 additional records identified after reference list search, a total of 16 entries [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] were included in the study. The flow-chart of the search process is presented in Figure 1. The methodological quality of all included studies was sufficient ( Table 1). The highest (9 points-NOS) quality was attributed to the studies by Vasconcelos et al. [32] and Han et al. [24] and the lowest (5 points-NOS) by Laredo-Aguilera et al. [27] and Olgun Yazar and Yazar [34]. The Sample size section was the weakest point of the studies-only three articles were awarded 1 point [20,24,32]. The strongest points proved to be the Representativeness of the sample-only Laredo-Aguilera et al. [27] did not receive a point based on design and analysis-Seino et al. [29] and Olgun Yazar and Yazar [34] received 1 out of 2 points.

Sensitivity Analysis
The Fail-Safe N coefficient, according to the Rosenberg algorithm, indicates that a 2932 of texts with null effect would need to be included in the selection for the total score to reach zero. Sequence analysis (Figure 4) of the results included in the meta-analysis confirmed a statistically significant correlation between depression and HGS in the study population. Based on the data, the chance for the effect in the population is 81-fold higher that the chance for null effect.

Discussion
This systematic review and meta-analysis of 16 studies involving 19,637 individuals aged ≥ 60 revealed a weak negative correlation between HGS and depression. The GDS-15 scale was the most commonly used tool for evaluating depression in the analyzed publications. According to Friedman et al. [35] GDS has robust internal reliability, construct validity, and operational characteristics for the screening of community-dwelling, cognitively intact older adults. Its usefulness for the evaluation of depression in older subjects has been demonstrated by numerous reports [36,37]. HGS was most often measured using the Takei or Jamar hand dynamometers. Studies show that both measuring devices are successfully used to assess HGS [38,39], but the scores may differ due to different shapes of the handles. The devices and the methods of measuring HGS should be made uniform. The use of standardized measuring tools for assessing HGS and depressive symptoms, in well-matched groups of older people, will allow us to achieve reliable results [40].
In recent years, HGS has been perceived as a reliable indicator of the whole-body muscle strength, physical function, and health status, as well as a predictor of the length of hospitalization and even mortality for older populations, and has been investigated by a number of authors [41][42][43]. Additionally, HGS is a diagnostic criterion for sarcopenia [11,12]. Thus, as many as four publications on the matter were found during our search and included in the analysis [19,21,22,34]. Olgun Yazar and Yazar [34] and Wang et al. [19] concluded that sarcopenia was more common in older people with depression and depressive symptoms and that HGS was lower in those individuals. In turn, M. Hamer et al. [22] demonstrated that reduced grip strength was associated with higher risk of depressive symptoms in obese participants only. The possible link between weight status and HGS was investigated by Smith et al. [30], who found that obese subjects with moderate to severe depressive symptoms had lower HGS. Brown et al. [21] who analyzed frailty and depression in older adults, demonstrated that older people with symptoms of depression had lower HGS.
HGS test is also used to assess physical and functional fitness in older people. Vasconcelos et al. [32] investigated the cut-off points of HGS to identify mobility limitation and calculated the following values: ≤17.4 kg for women and ≤25.8 kg for men in communitydwelling settings. HGS below these values was characteristic for the group with muscles weakness, who presented with depressive symptoms significantly more often.
Chen et al. [26] investigated the relationship between HGS and duration of sleep in older people, with depressive symptoms as an additional variable. These authors found lower HGS in subjects with depressive symptoms. Laredo-Aguilera et al. [27] also analyzed the link between HGS and quality of sleep in older populations, considering their mood and psychical functioning. They found a correlation between HGS and vigor, depression, insomnia, and sleep quality. Pearson correlation analysis adjusted for age showed significant correlations between HGS and depression (r = 0.379, p = 0.021).
HGS allows to predict mortality among older populations, especially the 'oldest' old [25,31]. Additionally, late-life depression could be associated with high risk mortality, as reported by Hamer et al. [23]. These authors suggest that the relationship may be the result of lack of physical activity and poor physical function, measured with HDS. Depressed patients had lower HGS scores as compared to non-depressed individuals.
Depression in older people is associated with more functional and cognitive impairment than in younger adults [44]. Holmquist et al. [33] investigated the risk factors for depression in the context of functional performance. They concluded that high risk for depressive symptoms in older people was associated with low levels of functional performance (including HGS) combined with low physical activity.
Depression is one of the multiple geriatric syndromes [29,45]. Seino et al. [29] analyzed different measures of physical performance in order to determine the indicators of geriatric syndromes. They demonstrated that lower HGS was found in all geriatric syndromes (including depression) apart from urinary incontinence and malnourishment.
In our meta-analysis, we also took into consideration research of Korean authors, who studied a relationship between blood cadmium levels and HGS and depressive symptoms.
Higher HGS values were associated with a lower number of depressive symptoms, assessed with the Korean Version of the GDS-short form [28].
Out of the 16 texts which were deemed eligible for the meta-analysis, only two aimed to establish the relationship between HGS and depression [20,24]. Han et al. [24] investigated that link in the context of socioeconomic status of the older subjects. These authors demonstrated a strong relationship between low HGS and intensified depression in socioeconomically deprived older people. Brooks et al. [20] concluded that reduced levels of combined HGS are independently associated with depression among U.S. adults aged 60 years and older.

Limitation
The fact that the modified NOS scale was used to analyze the quality of the studies, due to the lack of a more adequate tool to evaluate cross-studies, was a certain limitation of our study.

Conclusions
To sum up, the review of the literature and the meta-analysis demonstrated a relationship between low muscle strength measured with the HGS test and intensified depressive symptoms in older population, even though they were merely the additionally analyzed variables in the vast majority of the included texts.
Bearing in mind that depression is unrecognized or underdiagnosed in approximately 16% of the older patients [46], lowered muscle strength reported or found in older subjects should be an important sign for physicians and physiotherapists who work with older people and an incentive to screen them for depression, especially among older adults.
The mechanism linking HGS with depressive syndromes remain to be fully investigated. Their interdependence is a complex matter, indicating a strong two-way interconnection and the need for further studies to elucidate the matter.