Alcohol consumption as a risk factor for sarcopenia - a meta-analysis

Although old age is not a disease, many diseases and syndromes are more prevalent among older adults. As the proportion of older adults is increasing, an understanding of age-related diseases and syndromes is critical. One syndrome particularly common among older adults is sarcopenia [1]. Described by the European Working Group on Sarcopenia in Older People (EWGSOP) as a loss of muscle mass and strength and decreased physical performance, sarcopenia is associated with untoward outcomes such as disability, falls, and mortality [2]. Although deterioration of muscle and functional limitations have many causes and accompany natural aging, they could be accelerated by modifiable behavioral factors such as inactivity, undernutrition, smoking, and alcohol consumption. The possible role of alcohol consumption is of particular interest as it is associated with a number of pathologies, including alcoholic liver disease, pancreatic disease, neurological problems, cancer, and immunosuppression [3]. Ethanol impairs skeletal muscle protein synthesis and muscle autophagy is increased by ethanol exposure [4]. Alcohol consumption is associated with cachexia [5]. Whether, however, alcohol consumption is a risk factor for sarcopenia is unclear. Although individual risk factors do not work independently, the role of alcohol consumption as a risk factor for sarcopenia should be clarified. The main objective of this study was to explore relationships between sarcopenia and alcohol consumption in people over 65 years of age. Following the recommendation of the Cochrane Handbook for Systematic Reviews [6], we pursued this objective by calculating odds ratios ORs. The OR compares the relative odds of an outcome occurring (in our case sarcopenia) based on exposure to a variable of interest (in our case alcohol). In our study an OR greater than 1 would suggest that sarcopenia is associated with higher odds of developing sarcopenia and an OR less than 1 would suggest that sarcopenia is associated with lower odds of having sarcopenia.

This review included cross-sectional studies and data from baselines of longitudinal cohort studies. For the sake of meta-analyses, all data were treated as if from a case–control study- where sarcopenia was considered as the case and alcohol consumption represented the exposure. Participants who were included in those studies were people over 65 years of age who lived primarily in their own homes in the community. Ethical approval and consent from participants were declared in all the studies which were analyzed.

The results of the meta-analyses show that alcohol drinking did not contribute to sarcopenia development. The proportion of people categorized as sarcopenic was less, albeit not always significantly, in alcohol drinking group in all the analyses. This finding was strengthened by sensitivity analysis. Our findings notwithstanding, there were several problems encountered in the meta-analysis.

There were huge ambiguities in scientific approaches across the studies, for example, in diagnosing sarcopenia. Currently the EWGSOP algorithm [12] is considered the best tool for diagnosing sarcopenia. For the purposes of this meta-analyses, the EWGSOP algorithm and muscle mass measurement by dual-energy X-ray absorptiometry (DEXA) or bioelectrical impedance analysis (BIA) were considered suitable methods of sarcopenia diagnosis. Even though the EWGSOP has established cut off points for DEXA determined appendicular skeletal muscle mass (ASM) (i.e., 7.26 kg/m² for men and 5.5 kg/m² for women) and skeletal muscle index (SMI) (i.e., 7.23 or 7.25 kg/m² for men and 5.67 kg/m² for women), different cut off points for ASM or SMI were used in almost every study. For example, for males SMI cut off points of <5.72 kg/m² [13] and <7.0 kg/m² [14] were used [13, 14], for females cut off points of <4.82 kg/m² < 5.9 kg/m² were used [13, 14]. The EWGSOP has also established cut off points for muscle mass determined by BIA. Nevertheless, the SMIs cut off points used in the articles included in this review varied considerably. The SMI cut off points for males ranged from <7.0 kg/m² [10] to <8.82 kg/m² [15]; for females they ranged from <5.67 kg/m² [16] to <6.42 kg/m² [15]. This may be one reason the prevalence of sarcopenia varied so widely- from 3.8 % in the study by Domiciano et al. [11] to 27.2 % in the study by Lau et al. [13].

Variability and a lack of objectivity in the description of alcohol exposure might also have affected our results. It was almost impossible to find precise cumulative data such as a total number of alcohol units over time (e.g., drink-years). Exposure status was described using incompatible categories according to the daily amount, exposure period in the subjects’ lifetimes or current habits. Therefore, it was difficult in this work to find and establish an optimal stratification for exposure. In any event, all approaches were based on the subjective summarization by the participants. Although in recent years self-reports has been shown repeatedly to have good concordance with other methods [17], they can be influenced by deliberate under- or overestimation of consumption and by failures of memory and other cognitive factors [18]. Nevertheless, some suitable methods have been created previously, for example, the alcohol use disorders identification test (AUDIT) [19], following the recommendations of the World Health Organization (WHO). Unfortunately, AUDIT does not divide people into the categories of drinkers and nondrinkers. Anyway, the subjective component of alcohol exposure probably contributed to the huge variability of percentage of alcohol drinkers described in individual studies.

Despite problems with diagnosing sarcopenia and ascertaining exposure we detected a statistically significant age-difference between sarcopenia and no sarcopenia groups. That difference could work as important confounder of results. Aging is known to play a significant role in sarcopenia development. The mean age differed by more than 5 years in some of the studies we included. Such a difference existed in males studied by Liu et al. [14] and Silva Alexandre et al. [20], in females studied by Akune et al. [21] and Silva Alexandre et al. [15], and in the overall population studied Volpato et al. [15] and Wu et al. [16].

Beyond the aforementioned limitations, we must note that no study we examined had a primary focus on the relationship between sarcopenia and alcohol consumption. In all studies, alcohol consumption was merely one of many observed variables. Also, the exclusion and inclusion criteria varied between studies and resulted in the elimination of participants with characteristics which could be interesting to explore. Among individuals excluded from participation in different studies were those with poor functional status or metal implants [14] and those with limb edema, a pacemaker, joint prosthesis, or severe varicosities [15]. Other studies excluded older adults who answered yes for the question “Do you currently have kidney failure?” [22] or who were unable to perform the handgrip strength test or the walking portion of the Short Physical Performance Battery (SPPB), or who were unable to stand for measurement of weight and height [20]. Some of these exclusions might not be necessary in the event that the main aim of study was to explore the relation between sarcopenia and alcohol consumption. There were thus excluded subjects who were older, had less education, drank less, reported more difficulties in activities of daily living (ADL) and instrumental activities of daily living (IADL), had more sever hypertension, diabetes, lung disease, heart disease, stroke, falls, instances of hospitalization, a more sedentary lifestyle, more cognitive impairment, undernutrition and risk for undernutrition according to the Mini-nutritional assessment (MNA) [20]. Nevertheless, the inclusion of those subjects in the meta-analysis could confound the results. In contrast, exclusion criteria were not an issue is some studies [13, 21, 23, 24]. All the data from these studies were clustered into the analyses without regard to whether participants were accepted according to the same exclusion and inclusion criteria. Therefore, the cluster of the included studies could bring a little bit different results if the criteria were the same for all of them.

We also need to note that our results may have been influenced due to different design between cohort studies and cross-sectional studies. Excepting the results of Castillo et al. [21], the results of individual cohort studies indicated that for the overall population alcohol consumption might increase the risk of sarcopenia, which was completely opposite to the results in cross-sectional studies. However, the design of those studies was classified by their authors. We accepted their classifications in good faith even though there could be some concerns that the classifications might not have been exactly correct. Epidemiological studies can be truly cross-sectional (present drinking and present sarcopenia), case control (drinking history and present sarcopenia), or cohort (present drinking and future sarcopenia). Moreover, out of cohort studies we only used the data from baselines, therefore they could be considered as case–control studies rather than cohort studies and as such they do provide a higher level of evidence than cross-sectional studies. Consolidation of the two study designs could be justified in this meta-analysis, because they both are observational studies and they all were based on a similar approach in extracting information about alcohol consumption and sarcopenia diagnostics.

The results of the study show that alcohol consumption is not associated with sarcopenia development. Boffetta and Garfinkel [25] came to a similar conclusion regarding alcohol consumption and mortality and coronary heart disease (CHD). There, moderate alcohol intake had a protective effect on CHD mortality. Other authors have partly confirmed the possible protective influence of alcohol consumption on CHD mortality [26].

To sum up, we found out that alcohol consumption was not a risk factor for the development of sarcopenia, even more, according to the results alcohol consumption could have protective character against sarcopenia. This does not mean, however, that the authors recommend drinking alcohol to prevent any diseases.

The implications for research would be summarized into three recommendations. First, it would be beneficial to diagnose sarcopenia more specifically using age, gender and ethnicity specific reference values. Second, valid tools for quantifying exposure need to be developed and employed. Finally, it would be useful to study groups more similar according to the age. Those groups’ comparison would be much more meaningful.