University of Birmingham Interventions to ameliorate reductions in muscle quantity and function in hospitalised older adults: a systematic review towards acute sarcopenia treatment

Objective Assimilate evidence for interventions to ameliorate negative changes in physical performance, muscle strength and muscle quantity in hospitalised older adults. Methods We searched for articles using MEDLINE, Embase, CINAHL and Cochrane library using terms for randomised controlled trials, older adults, hospitalisation and change in muscle quantity, strength or physical performance. Two independent reviewers extracted data and assessed risk of bias. We calculated standardised mean diﬀerences for changes in muscle function/quantity pre- and post-intervention. Results We identiﬁed 9,805 articles; 9,614 were excluded on title/abstract; 147 full texts were excluded. We included 44 studies including 4,522 participants; mean age 79.1. Twenty-seven studies ( n = 3,417) involved physical activity interventions; a variety were trialled. Eleven studies involved nutritional interventions ( n = 676). One trial involved testosterone ( n = 39), two involved Growth Hormone ( n = 53), one involved nandrolone ( n = 29), and another involved erythropoietin ( n = 141). Three studies ( n = 206) tested Neuromuscular Electrical Stimulation. Evidence for eﬀectiveness/eﬃcacy was limited. Strongest evidence was for multi-component physical activity interventions. However, all studies exhibited at least some concerns for overall risk of bias, and considering inconsistencies of eﬀect sizes across studies, certainty around true eﬀect sizes is limited. Conclusion There is currently insuﬃcient evidence for eﬀective interventions to ameliorate changes in muscle function/quan-tity in hospitalised older adults. Multiple interventions have been safely trialled in heterogeneous populations across diﬀerent settings. Treatment may need to be stratiﬁed to individual need. Larger scale studies testing combinations of interventions are warranted. Research aimed at understanding pathophysiology of acute sarcopenia will enable careful risk stratiﬁcation and targeted interventions.


Introduction
Sarcopenia is defined by low muscle strength with low muscle quantity/quality; additionally demonstrated low physical performance defines severe sarcopenia. Cut-offs are two standard deviations (SDs) below means of young healthy reference populations [1]. Acute sarcopenia (acute muscle insufficiency) particularly affects hospitalised older adults [2,3]. Normally proceeded by stressor events, it is defined by acute declines in muscle quantity/quality and/or function (strength or physical performance) producing incident sarcopenia [1,3]. Previous reviews considered chronic sarcopenia treatment/prevention [4][5][6]; strongest evidence exists for physical activity. Resistance training improves muscle quantity, strength and physical performance in community-dwelling populations [7]. Some trials demonstrated enhanced benefit of nutritional supplementation alongside [8]. Large studies are underway evaluating combined nutritional and exercise interventions for chronic sarcopenia [9].
It is unknown whether chronic sarcopenia interventions can treat acute sarcopenia. Mechanisms differ, which may affect treatment efficacy. Acute sarcopenia is associated with greater systemic inflammation and immune-endocrine dysregulation. Inflammation (acute or chronic) may blunt response to exercise or protein challenges (anabolic resistance), but this may be acutely/severely upregulated in acute sarcopenia [10]. Acute sarcopenia follows an accelerated course [3]; traditional treatments may not work fast enough. Additionally, community interventions may be unfeasible in hospital. This review aimed to identify trialled interventions for ameliorating negative changes in muscle quantity, strength or physical performance in hospitalised older adults, and to summarise/synthesise findings.

Protocol and registration
Protocol was agreed by all researchers and registered with Prospective Register of Systematic Reviews-CRD42018112 021. Reporting is consistent with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidance.

Eligibility criteria
We included randomised controlled trials (RCTs) and quasi-RCTs involving hospitalised patients ≥65 years-old, where pre-and post-intervention measurements of muscle quantity, strength or physical performance were available. Post-intervention measures until 28 days post-intervention were included. We included physical activity, nutritional, pharmaceutical or Neuromuscular Electrical Stimulation (NMES) trials. Exclusion criteria were: degenerative neuromuscular disorders; acute stroke; trials of parenteral nutrition, surgical technique/invasive procedure, chemotherapy/radiotherapy, or anaesthetic agents/techniques; no control group; lengths of stay less than 2 days. We included studies that measured muscle quantity using computed tomography (CT), magnetic resonance imaging (MRI), dual energy X-ray absorptiometry (DXA), bioelectrical impedance analysis (BIA), or ultrasound, muscle strength using handgrip strength, knee flexion, or knee extension, or physical performance using short physical performance battery (SPPB), gait speed, timed up and go (TUG), or 6-Minute Walking Test (6MWT). There were no date or language restrictions.

Information sources
We searched electronic databases (MEDLINE, Embase, CINAHL, CENTRAL) on 16 January 2019; search repeated on 3 April 2020. Grey literature was identified through Web of Science, Google Scholar, Clinicaltrials.gov, article references and protocol citations. We contacted authors for information where necessary, including requesting age breakdown of data. If no response was obtained, a decision was made to include studies where mean age was one SD > 65.

Search strategy
We used published and unpublished terms for study design (RCTs), population (older adults AND hospitalised) and outcome measures (muscle mass OR muscle strength OR physical performance) in our search. Full search strategy is available in the online supplement; this was reviewed and agreed with an information specialist.

Study selection
Citations were imported into Microsoft Excel 2016. Duplicates were removed automatically/manually. Two reviewers independently screened titles and abstracts for inclusion (CW, ZM). Disagreements were resolved through discussion. Full texts were reviewed independently by the same reviewers; disagreements were resolved through discussion or third review (TAJ).

Data extraction
Data were extracted independently by two reviewers (CW, ZM) using a template (Microsoft Excel 2016). Extracted data were country, study design, sample size and dropouts, sample characteristics (age, ethnicity, body mass index-BMI, sex), speciality, intervention description (type of intervention, how delivered), intervention characteristics (timing of intervention, dosage), control group, outcome data, length of stay and adverse events. Outcome data at baseline and follow-up to include muscle quantity, muscle strength and physical performance were extracted.

Risk of bias
Two reviewers (CW, ZM) independently assessed risk of bias using Cochrane risk of bias tool. Conflicts were resolved by discussion. Risk of bias was collated using RevMan version 5.3 [11].

Synthesis of results
We summarised study and participant characteristics, and outcome data at baseline and follow-up using means/SDs in text and tables. Interventions were grouped by subtype and outcomes. All studies were included in narrative synthesis. If sufficient information was available to estimate standardised mean differences (SMDs) of change scores, effect sizes were evaluated as described in statistical analysis section. Certainty of interventions with large effect sizes was evaluated using Grading of Recommendations, Assessment, Development and Evaluations [12].

Statistical analysis
Correlations for outcome measures were calculated from studies reporting SDs of change scores and baseline/ follow-up measures [13]. Mean correlation for each outcome was used to estimate SD of change in outcomes in studies where this was not available. We calculated SMDs of change scores by dividing difference in change score between comparison and intervention groups by SD of change score in comparison group [14]. Effect sizes were calculated to one decimal place and classified as no effect (≤0.1), small (0.2-0.4), medium (0.5-0.7) or large (0.8 or greater) [15]. If more than one effect size was available for a single trialled intervention and outcome type, the larger was included. Meta-analysis was not performed due to high heterogeneity of interventions and outcomes.

Study selection
We identified 9,805 articles after duplicates removal. We excluded 9,613 following title/abstract screening; 192 full texts assessed for eligibility. We excluded 148 full text articles due to mean age not more than one SD above 65 (n = 56), no control group (n = 10), follow-up over 28 days (n = 12), no baseline measures (n = 6), no measures meeting inclusion criteria (n = 20), duplicate data (n = 11), unable to obtain necessary data from authors (n = 24), other intervention type (n = 2), and non-hospitalised population (n = 6) ( Figure 1). We included 44 studies in narrative synthesis and 32 studies in effect size evaluation.

Study characteristics
This review included 4,522 participants (2,160 control, 2,362 intervention). Sample size per arm ranged from 7 to 232. Most studies were small; 52% (23/44)    Table 1 shows included studies' details; full study characteristics and results are available online. Table 2 shows effect sizes separated by interventions and outcomes.

Physical activity interventions
Most studies (61%, 27/44) reported physical activity interventions. Eighty-nine percent (24/27) included physical performance [17][18][19][20][21][22]34  Interventions that ameliorated reductions in physical performance in trial populations included backward walking [17] Figure 2 shows overall risk of bias across studies. Full risk of bias details is shown in the Supplementary Appendix. There were at least some concerns for overall risk of bias across most studies. Adherence to trial intervention was associated with lowest risk and selection of reported outcome with highest risk. Most common reason for high risk of bias related to randomisation processes. Over half of studies exhibited at least some concerns for selection of reported result. Table 3 shows assessment of certainty for two interventions (individualised physical training programmes and protein supplementation) across studies.

Interpretation of findings
Physical activity interventions were investigated more commonly than others. However, this mostly relates to studies 7 C. Welch et al.  . Conversely, many physical activity trials reported muscle strength change but only one measured muscle quantity change, a multi-arm study also involving nutritional/pharmaceutical interventions. Nutritional and pharmaceutical trials focused on muscle strength and quantity changes rather than physical performance. This suggests disconnect in how physical activity interventions are trialled compared to other interventions; physical performance declines may not be prioritised as organ insufficiency markers in need of urgent treatment.
Only nine trials reported muscle quantity change. This relates to historical reduced availability of feasible serial assessment tools; DXA, CT and MRI remain gold-standard, but ultrasound is increasingly utilised [1,16]. As sarcopenia definition has developed, measures of muscle function are considered more important than muscle quantity [1]. However, in acute sarcopenia, early muscle quantity declines may not be associated with muscle strength declines [3]; preventing this may be important to prevent longer term deteriorations. Additionally, muscle strength may be affected by fatigue/effort during acute illness making testing of efficacy/effectiveness challenging [17]. Muscle quantity may be an appropriate treatment target in hospitalised patients; future trials of interventions for acute sarcopenia should consider incorporating in outcomes. Measurement of muscle quantity is also important to show biological effectiveness/mechanistic action.
We identified several physical activity interventions that stratified treatment protocols individually (e.g. by frailty) [37,43,45,53,55]. Most substantial and significant effects on muscle strength and physical performance were demonstrated in the highest reported adherence trial [43]. Although this demonstrates high adherence of hospitalised older adults to complex trial designs is possible, effectiveness is expected to be reduced in clinical environments with limited compliance. Increasing mobilisation alone may be insufficient to prevent/treat acute sarcopenia [17,20,22,24,41], although this is safe to do when possible and should be commended [17,20,24,25,41]. Physical activity interventions can be multidimensional and include resistance exercise [43,44]; it is safe and feasible to use machines/weights during acute phase of illness in hospitalised older patients [43,44,57]. Pedal exercises [23,36] and seated side-tapping [47] are simple, cheap, feasible and potentially effective; these may be implemented as part of multidimensional stratified interventions. Group exercise may be as effective as individual exercise but more cost-effective [42]. Group exercise has additional benefits of improving social interaction, and potentially improving motivation [18] and adherence [43].
Several nutritional interventions were trialled. Although few trials showed statistically significant results, all trials were small and may have been under-powered for efficacy. Three trials combined nutritional intervention with physical activity [27,38,51]. Research in chronic sarcopenia suggested additional protein supplementation may be most effective when combined with targeted physical activity i.e. resistance exercise [19]. As inflammation and anabolic resistance are heightened with acute illness [3], greater doses (i.e. greater protein/amino acid intake) may be warranted in hospitalised older adults.
Few studies tested pharmaceuticals. There is suggestion from GH trials that this may be effective in ameliorating reductions in muscle quantity and strength [31,32]. Further research is needed, including longer term outcomes. Benefits of GH supplementation need to be balanced against adverse effects, although supplementation was safe in dosages used in these small studies. Research is ongoing into novel pharmaceutical agents for use in acute and chronic sarcopenia [20]. Studies assessing correlations between immuneendocrine biomarkers and phenotypic changes in muscle quantity, quality or function will enable stratified treatments and direct potential drug pathways.
Trials of NMES showed conflicting results. NMES involves delivery of controlled electrical stimuli to superficial muscles via self-adhesive skin electrodes. These stimuli evoke muscle contractions, recruiting motor units and activating muscle fibres [21]. NMES has been shown to ameliorate reductions in muscle quantity and function in healthy young volunteers during bed rest [22]. It is plausible that NMES may treat acute sarcopenia in hospitalised older adults. However, in establishing effectiveness in clinical practice, adherence, physical activity impact, and which muscle groups to stimulate should be considered.

What are the limitations of this review?
This review included hospitalised adults over 65 years-old. We excluded younger adults to focus towards most vulnerable patients, who are most likely to benefit from targeted interventions. More studies were excluded for participant age than were included (56 versus 44). This suggests persistent bias against involvement of older people in clinical trials, particularly those with frailty. Considering we included search terms for older people in our search, it is likely more trials involving younger adults were not identified, as well as trials excluded through abstract screening. Trials conducted in younger adults may be useful when developing interventions for acute sarcopenia in older adults, but caution should be taken extrapolating results from younger less heterogeneous populations.
It is important to consider only three studies reported frailty status in both control and intervention arms [37,45,50]. Frailty was measured in intervention arms but rates were not reported in studies that stratified by frailty [55,57]. Although important measures, handgrip strength and gait speed alone may be insufficient to diagnose premorbid frailty during acute illness [42]. Recording levels of frailty prior to hospitalisation can ensure control and intervention arms are matched and enable sub-group analysis assessing treatment effect in individuals with and without frailty [45]. Only one study reported ethnicity amongst participants [38]. Normative values of muscle quantity may vary according to ethnicity [23], and muscle echotexture may differ [24]. Further research is needed to assess effects of genetics and environment on ethnic differences, and how these relate to differences in muscle function and responsiveness to interventions. Without information on ethnicity within published trials, it is not possible to assess for between group differences.
As described, majority of trials were small; many may have been underpowered to detect changes. Due to high heterogeneity in populations, interventions, and outcome measures, it was not possible to conduct meta-analyses. Some interventions that were not shown to be effective in small individual trials may be effective in larger powered studies. Additionally, most studies exhibited some concerns for risk of bias overall, and due to inconsistencies in effect sizes across different studies, there is limited certainty around true effect sizes. Many different outcome measures were also assessed across different RCTs. We consider that standardisation of assessment and outcome measures within geriatric medicine research will enable greater ease of knowledge transfer, sharing of datasets and future meta-analyses of RCTs in ageing.
It is important to consider that none of the included trials specifically included the presence of (acute or chronic) sarcopenia as inclusion criteria, or stratified treatment by sarcopenia. However, we consider that results of identified RCTs identified will be pivotal towards designing trials for prevention and/or treatment of acute sarcopenia. Acute sarcopenia is a rapidly progressing research area and therapeutic target. Twenty-two percent of studies (10/44) included in this review were published in the last 18 months. This demonstrates how rapidly progressive this area is, with increasing numbers of studies measuring muscle quantity and function as outcome measures.

Conclusion
Deteriorations in muscle quantity, strength and physical performance are problematic in older adults following hospitalisation. However, insufficient evidence exists to enable targeted prevention/treatment strategies. A number of interventions have been trialled and shown to be safe for heterogeneous populations across various settings. Multidimensional physical activity interventions which are individually tailored (e.g. for frailty) have been trialled [43,45,55]; the trial with most substantial effect size reported excellent adherence [43]. Large scale multi-arm studies assessing effectiveness of combined interventions including physical activity [23,43,45,47,55], NMES [34], nutrition [59] and pharmaceuticals [31, 32] are warranted. Treatment may be most effective when stratified according to individual need. Treatment is likely to be guided by a combination of clinical and biological factors (e.g. immune-endocrine markers). Further research aimed at understanding pathophysiology of acute sarcopenia will enable risk stratification and targeted interventions.
Supplementary data: Supplementary data mentioned in the text are available to subscribers in Age and Ageing online.