Abstract
to explore whether sarcopenia is associated with depression.
electronic literature databases from PubMed, Scopus, Embase and Google Scholar were searched. A systematic review and meta-analysis of observational studies was conducted.
community and outpatient clinic.
people with and without diagnoses of sarcopenia.
outcome measures of depression.
about 15 articles were included, 5 of which were retrieved for narrative review. The crude odds ratios (ORs) between sarcopenia and depression were extracted from the remaining 10 studies, 6 of which also included adjusted ORs. Sarcopenia was associated with depression without adjusting covariates (crude OR, 1.640; 95% confidence interval (CI), 1.247–2.155). After adjusting for potential confounders such as age, gender, cognitive performance and physical activity, sarcopenia still demonstrated a significant positive association with depression (adjusted OR, 1.821; 95% CI, 1.160–2.859). A stratified analysis showed that the studies that used bioelectrical impedance analysis for measurement of body composition tended to have an elevated association between sarcopenia and depression compared with those that used dual-energy X-ray absorptiometry or equation estimation.
sarcopenia was independently associated with depression. The causal relationship between the two clinical conditions requires future validation with cohort studies.
Introduction
Sarcopenia, which is characterised by a reduction in muscle mass and strength, is mainly seen in the elderly and develops secondarily to chronic diseases such as cancer, heart failure, chronic obstructive lung disease, rheumatoid arthritis and various neurodegenerative disorders [1]. The clinical condition has been recognised as a serious geriatric syndrome, causing disability and increasing health care costs. The pathophysiology of sarcopenia includes inactivity, bed rest, malnutrition, hormone deficiency and chronic inflammation [2]. Resistance exercise and a supplement of androgenic hormone were reported to be effective treatments for sarcopenia, as well as for the maintenance of psychological homoeostasis [3].
In recent years, there has been an observed association between sarcopenia and depression, a mental health disorder [4]. Characterised by low mood, loss of interest and psychomotor retardation, depression is emerging as a major contributor to the global burden of disease. The aetiology of depression is multifactorial, consisting of psychosocial distress, sedentary life style, pro-inflammatory status, sex hormone depletion and deficiency of certain nutrients like vitamin D [5, 6]. Pharmacological therapy, psychological support and exercise intervention are useful in alleviating symptoms of depression [5]. Sarcopenia and depression seem to share several common risk factors, such as physical inactivity, upregulation of inflammatory cytokines and dysregulation of hormones in the hypothalamic–pituitary–adrenal axis [2, 5]. However, the association between sarcopenia and depression appeared inconsistent from the data of the latest observational studies. In 2014, Hsu et al. reported that sarcopenia was associated with depressive symptoms [7], whereas in 2016, Byeon found that patients with sarcopenia did not show a higher prevalence of depression [8]. However, the distinct direction of association between various studies might be derived from difference in assessing sarcopenia, such as using the muscle mass alone or additional inclusion of physical performance. Therefore, this meta-analysis aims to explore whether an independent association exists between sarcopenia and depression.
Method
Search strategy and inclusion and exclusion criteria
The present meta-analysis was compiled using cross-sectional, case-control and cohort studies investigating factors associated with sarcopenia. A comprehensive search was conducted in four electronic databases, including PubMed, Scopus, Embase and Google Scholar [9, 10]. Relevant research was also identified by reading reference lists of associated reviews. The key words were sarcopenia, fragility, grip strength, gait speed, depression and depressive symptoms. Additionally, only articles published in English were included. The combination of key terms for literature search was listed as follows: (i) (sarcopenia OR fragility) AND (depression OR depressive symptom) and (ii) [sarcopenia and (grip strength OR gait speed)] AND (depression OR depressive symptom). The targeted population was adults undergoing the survey of sarcopenia, comprising measurement of body composition, muscle strength and physical performance. Depression, defined by validated symptom scales or a subset in a standardised global measurement, was a required covariate in the enroled studies. We excluded studies lacking a clear definition of depression, a control group (subjects without sarcopenia) and proportion of participants with or without depression.
Data extraction and outcomes of interest
The extracted information included study type, participant characteristics, tools and cut-off points for evaluation of sarcopenia and depression, methods of measuring body composition and muscle strength and all the pertinent covariates modifying the relationship between sarcopenia and depression. The primary outcome of interest was the crude and adjusted associations of sarcopenia with depression, expressed by odds ratio (OR) and 95% confidence interval (CI). The crude OR was derived from the individual numbers of depressive patients in the sarcopenia and control groups, which was usually presented in the tables describing the demographic of study populations.
Study quality assessment
The quality of the included citations, which was assessed by the Newcastle-Ottawa Scale, was independently scored by two authors [9–11]. The scale was widely used for the evaluation of non-randomised controlled trials with respect to selection, comparability and outcome/exposure of the enroled studies. The highest score was 9 points for cohort or case-control studies and 6 points for cross-sectional studies [11]. A higher score indicated a methodology of better quality. Discrepancies between both reviewers were resolved by discussion or evaluated by the corresponding author.
Statistical analysis
The Comprehensive Meta-analysis Software version 3 (Biostat, Englewood, NJ, USA) statistical package was employed to analyse the OR as well as the 95% CI, using the random effect model to prevent overestimation of the effect size [12]. The heterogeneity across studies was examined by the chi square-based Cochran's Q statistic test and I2 statistic [13, 14]. Significant heterogeneity was defined as P < 0.1 or I2 > 50% [13, 14]. The subgroup analysis was conducted based on difference in recruited participants, diagnostic criteria of sarcopenia, depression scales and methods of measuring body composition. The funnel plot along with Begg's test was performed to plot the log OR against its standard error for evaluation of publication bias, while the extent of asymmetry was assessed by Egger's unweighted regression asymmetry test [10]. We used two-tailed P-values and P < 0.05 was regarded as statistical significance except that for determining heterogeneity (P < 0.1).
Results
Methodological quality and baseline characteristics
We retrieved 461 articles by searching electronic databases and retained 22 of relevance after screening their titles and abstracts. We further excluded 7 articles because they lacked a clear measurement of sarcopenia or depression. The final quantitative analysis included 15 observational studies, comprising 33,030 participants (Figure 1).
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for the study selection process.
In terms of the targeted population, ten studies enroled elders or middle-aged persons (>50 years of age) [4, 7, 15–22], four studies enroled adults without specifying the age range [8, 23–25] and one study enroled patients with end-stage renal disease [26]. The average age of the observed population ranged from 43.3 to 86.7 years, and women accounted for 56.31% of all participants. Regarding the measurement of body composition, six studies used dual-energy X-ray absorptiometry (DEXA) [4, 8, 15, 20, 21, 23], three used bioelectrical impedance analysis [7, 17, 26], two used the estimate from equations [16, 18] and four used muscle strength or mobility without measuring muscle mass [19, 22, 24, 25]. The diagnostic algorithm by the European Working Group on Sarcopenia in Older People (EWGSOP) or relevant protocols were widely used in selected studies [7, 16–20, 26], while two employed the modified version of a joint document elaborated upon by Special Interest Groups (SIG) [8, 23], two employed the diagnostic criteria defined by the Asian Working Group for Sarcopenia (AWGS) [15, 22] and four did not mention the specific diagnostic criteria they used [4, 21, 24, 25]. Besides using EWGSOP as the suggested diagnostic algorithm, the study conducted by Sugimoto et al. also used AWGS to define low skeletal muscle mass [17]. Regarding the assessment tool for depression, eight studies adopted the Geriatric Depression Scale (GDS) [4, 7, 16, 17, 19–22], one adopted the Beck Depression Inventory (BDI)-II [26], one adopted the depression subscale in Minimal Data Set (MDS) for Home Care [18], two adopted the Centers for Epidemiologic Studies Depression Scale (CESD) [15, 24], one used 5-item version of the Mental Health Inventory [25] and two adopted self-designed questionnaires (SDQ) (Supplementary data, Appendix Table 1 available in Age and Ageing online) [8, 23]. The result of the quality assessment for the included studies is listed in (Supplementary data, Appendix Table 2 available in Age and Ageing online).
Outcome
Due to lack of dichotomous classification of sarcopenia or depression, five articles were retrieved for narrative synthesis [4, 21, 22, 24, 25]. Participants with lower physical performance [21], with lower grip strength [25] and with sarcopenia defined by the AWGS diagnostic algorithm [22] were shown to have higher scores in GDS than those without the abovementioned medical conditions. One study showed an increase in grip strength was associated with lower risks of depressive symptoms [24] and the other revealed that participants with depression had a lower appendicular skeletal muscle mass than those without depression [4]. All the five studies were in favour of a positive association between sarcopenia and depression.
The crude ORs between sarcopenia and depression were extracted from the remaining 10 studies [7, 8, 15–20, 23, 26], 6 of which also included adjusted ORs [7, 8, 19, 20, 23, 26]. In a study dividing the participants into four groups based on the presence of muscle weakness and mobility limitation, we mainly used the comparison between those with neither low strength nor slow gait speed and those with both for analysis [19]. As shown in Figure 2, sarcopenia was associated with depression without adjusting for covariates (crude OR, 1.640; 95% CI, 1.247–2.155). After adjusting for potential confounders such as age, gender, cognitive performance and physical activity, sarcopenia still demonstrated a significant positive association with depression (adjusted OR, 1.821; 95% CI, 1.160–2.590). Moderate heterogeneity was noted both for the crude ORs (P = 0.003 and I2 = 64.38%) and the adjusted ORs across studies (P = 0.002 and I2 = 74.11%).
Forest plot of the (A) crude and (B) adjusted associations between sarcopenia and depression.
The subgroup analysis showed that the group using bioelectrical impedance analysis tended to have a stronger crude association (OR, 2.802; 95% CI, 1.318–5.954) than those employing DEXA (OR, 1.419; 95% CI, 1.018–1.979) or equation estimation (OR, 1.129; 95% CI, 0.661–1.927) (Figure 3) and the trend remained following adjustment for confounders. Regarding different evaluation tools for depression, although the group using the BDI-II showed higher crude and adjusted ORs than others, the estimates resulted from merely one study enroling patients with end-stage renal disease (Figure 4) [26]. In addition, the group incorporating maximal grip strength as part of the diagnostic criteria for sarcopenia tended to have a higher adjusted association (OR, 3.134; 95% CI, 1.197–8.208) than those without measuring grip strength (OR, 1.226; 95% CI, 0.860–1.748) (Supplementary data, Appendix Figure 1 available in Age and Ageing online). Visual inspection of the funnel plots determined that they appeared asymmetrical, indicating existence of publication bias; the findings were also supported by the results of the Egger test, which were significant in all analyses (P-values = 0.008 for crude OR and 0.019 for adjusted OR) (Supplementary data, Appendix Figure 2 available in Age and Ageing online).
Forest plot of the subgroup analysis based on measurements for body composition: (A) crude OR; (B) adjusted OR. Abbreviation: BIA, bioimpedance analysis; EES, equation estimate; NA, not applicable.
Forest plot of the subgroup analysis based on assessment tools of depression: (A) crude OR; (B) adjusted OR.
Additional sensitivity analyses were performed to confirm the association between sarcopenia and depression. First, a study targeting patients with end-stage renal disease was removed [26]; it did not mitigate the positive association between sarcopenia and depression (crude OR, 1.470; 95% CI, 1.177–1.838; adjusted OR, 1.530; 95% CI, 1.039–2.253). Second, we eliminated two studies only measuring individual components of sarcopenia (Supplementary data, Appendix Figure 3 available in Age and Ageing online) and the effect sizes did not change significantly. Third, regarding the study conducted by Vasconcelos et al. [19], the comparisons between the other two groups and the controls were employed instead of the primary comparison between those with neither low strength nor slow walking speed and those with both (Supplementary data, Appendix Figures 4 and 5 available in Age and Ageing online). We did not identify a significant deviation of the associations between sarcopenia and depression following the replacement of the inputted values.
Discussion
The present study aimed to integrate the most updated evidence to evaluate the association between sarcopenia and depression. The overall results indicated that sarcopenia was positively associated with depression both in the narrative review and quantitative analysis; this relationship was not influenced by adjusting for relevant covariates. The stratified analysis further showed that the studies using bioelectrical impedance analysis for measurement of body composition tended to have an elevated association compared with those using DEXA or equation estimation.
Sarcopenia has been linked to many adverse health conditions such as diabetes, metabolic syndrome, cardiovascular diseases, physical inactivity and higher mortality [27, 28]. The association between psychological health and sarcopenia has been recently recognised, with an increasing number of studies documenting the symptoms of stress, anxiety, suicide ideation and depression while investigating muscle strength and mass [29]. Kim et al. first found a lower appendicular skeletal muscle mass in depressive elderly participants [4], and Hsu et al. later reported an independent positive association between sarcopenia and depressive symptoms [7]. Nevertheless, the latest report from Byeon et al. did not identify a correlation between the two clinical conditions [8]. Thus, based on the conflicting results across studies and the significance of both clinical conditions, a meta-analysis was conducted to investigate whether sarcopenia was associated with depression.
Our results indicated that sarcopenia was associated with depression, as was evident from the crude and adjusted ORs. It is not surprising that sarcopenia is concomitant with depression, because there are many known factors relevant to both medical conditions. Gender plays an important role in hormone regulation and muscle metabolism and influences the prevalence of depression in the general population [29, 30]. Moreover, diabetes mellitus leads to chronic stress and inflammation, which can increase the risk of sarcopenia and depression [6, 25]. Patients with a depressive mood are likely to have physical inactivity, which is a well-recognised cause of sarcopenia [2, 5]. Therefore, whether sarcopenia is independently associated with depression, apart from the confounders, will be of great interest and clinical significance.
We found that age, gender and body mass index are three covariates adjusted in all the retrieved studies implementing multivariable regression analysis. Other commonly included confounders were cognitive function, physical performance, activities of daily living, smoking and drinking habits, diabetes mellitus and cardiovascular disease. Although the point estimate of the adjusted association (OR, 1.821) seemed to be higher than that of the crude association (OR, 1.640), there was a significant overlap of the 95% CI between both estimated values. Therefore, the pooled OR did not increase following adjustment. Our result only revealed that the association between sarcopenia and depression remained after adjustment and was not mediated by the aforementioned factors. In the sensitivity analysis, we discarded a study that recruited patients with end-stage renal disease with an extraordinarily high OR. We did not notice significant deviations of the values pooled from remaining studies. The result further strengthened the possible application of our observation to the general population and implied that the association between sarcopenia and depression might vary in certain disease groups.
Our results indicated that different analysis tools of body composition might modify the association between sarcopenia and depression. Estimation based on skin fold thickness and mid-arm circumference is applicable for ambulatory settings, but is susceptible to age-related fatty change and the underestimation of loss of fat-free muscle mass [31]. Bioelectrical impedance analysis has the advantage of low cost and portability and is widely used in epidemiological research [31]. DEXA is capable of differentiating fat, bone mineral, and lean tissues and is considered an alternative ‘gold standard’ of muscle mass measurement with respect to traditional modalities such as computed tomography and magnetic resonance imaging [31]. The main drawback of DEXA is its unsuitability for massive screening due to its lack of portability. Some studies have reported that substantial inconsistencies in lean body mass measurements exist between DEXA and bioelectrical impedance analysis, and the disagreement may vary by gender [32, 33]. Our study found that the OR pooled from studies using bioelectrical impedance analysis tended to be higher than that derived from studies employing DEXA. The result reflected that the distribution of sarcopenia could be influenced by measuring tools, which further modified its association with depression. Another explanation is that the skeletal mass index estimated by bioelectrical impedance analysis is though calculation from an equation, which usually includes gender, body mass index and electric resistance of different body component [34]. In the equation, the skeletal muscle mass is proportional to body mass index. Therefore, the patients with a lower body mass index tend to have a decreased skeletal muscle mass estimated by bioimpedance analysis (BIA) and are prone to be categorised as sarcopenia. The association between sarcopenia and depression was strengthened by BIA-estimated body mass index, a low value of which is also a feature of depression. Therefore, the specification of assessment methods is a prerequisite for describing the relationship between sarcopenia and depression to prevent over- or underestimation.
Among all selected studies, the shortened version of the GDS is the most commonly used depression evaluation tool. The scale is designed for community-dwelling elderly and elders living in nursing facilities, who are also at an increased risk of developing sarcopenia compared with young adults. The cut-off points varied from 6 to 8 for our enroled articles, corresponding to the range of mild depression (5–8) suggested by literature [35, 36]. Although the scale cannot substitute for a diagnostic interview by psychological professionals, we believe it is sensitive enough to screen depressive symptoms in the elderly. Treating the group employing the GDS as a reference, we did not find significant differences of ORs regarding those utilising the Epidemiologic Studies Depression Scale, the MDS subscale and SDQ. One study employing the BDI-II had extraordinarily high crude and adjusted ORs [26]. However, we believe that the BDI-II was not the main cause of the outlier value because the research also used bioelectric impedance analysis and enroled patients with end-stage renal disease.
In our result, there was no difference in the crude ORs between the groups with and without using maximal grip strength for diagnosing sarcopenia. However, the group incorporating grip strength as part of the diagnostic criteria for sarcopenia tended to have a higher adjusted OR than those without measuring grip strength. Our result was consistent with a recent large cohort study, showing lower hand-grip strength, adjusted by age and gender was associated with depression [25]. The finding might be derived a bidirectional association between lower physical performance and presence of depressive symptoms.
The main strength of our meta-analysis is the analysis of the adjusted OR, which is further evidence of the existence of an independent association between sarcopenia and depression. Although lower body mass index, physical inactivity and chronic morbidities were previously considered to be the causal factors bridging sarcopenia and depression [2, 5, 6], our results indicated that sarcopenia might directly cause depression and that the reverse pathogenic pathway was possible. Implementation of a prospective cohort study examining the incidence of sarcopenia and depression with inclusion of more pertinent serum and nutrition markers is warranted to clarify the true biological mechanism.
The meta-analysis had several limitations. First, all selected articles used a cross-sectional design, which could not provide a causal relationship of the analysed factors. Second, the severity and heterogeneity of diagnostic criteria for sarcopenia and depression might lead to underestimation or overestimation of their association. Regarding sarcopenia, difference in the tools measuring skeletal muscle mass had shown its influence on the strength of association between both medical conditions in our subgroup analysis. In terms of depression, the definition used in most retrieved studies fell in the category of mild severity; thus, whether the association in patients with major depression differs requires further research to explore. Third, although malnutrition was a major cause of sarcopenia and a possible cause of depression [37], there exists little research providing a detailed nutritional profile of the patients. Fourth, the variables adjusted differed in the retrieved studies. A standard set of confounders was suggested to use in future relevant research to increase the comparability between different studies.
In conclusion, our meta-analysis indicated that patients with sarcopenia were likely to present with depression and that the positive association was independent of common covariates like age, gender, body mass index, physical performance and chronic comorbidities. The causal relationship between both clinical conditions requires a future cohort study for clarification.
Sarcopenia was independently associated with depression.
Sarcopenia and depression seem to share several common risk factors.
These results indicate that sarcopenia might directly cause depression and that the reverse pathogenic pathway is possible.
The causal relationship between depression and sarcopenia requires future validation with cohort studies.
Supplementary data
Supplementary data are available at Age and Ageing online.
Conflicts of interest
None declared.
Funding
The study was made possible by the research funding of the Community and Geriatric Research Center, National Taiwan University Hospital, Bei-Hu Branch.
Comments