Comparative Efficacy of Various Interventions to Reduce Perceived Stress Among Older Adults: A Systematic Review and Network Meta-Analysis.

The researchers compared seven categories of stress-reduction interventions against each other and against control conditions (usual care, waitlist, or no intervention):

**Reminiscence therapy:** Structured group or individual sessions where older adults recall, reflect on, and reframe past life experiences, typically guided by a therapist or trained facilitator.

**Exercise:** Structured physical activity programmes including aerobic exercise, resistance training, tai chi, and walking programmes.

**Yoga:** Hatha yoga, Iyengar yoga, and chair-based yoga programmes combining physical postures, breathing exercises, and meditation.

**Mindfulness-based interventions:** Mindfulness-Based Stress Reduction (MBSR), Mindfulness-Based Cognitive Therapy (MBCT), and adapted mindfulness programmes.

**Health education:** Structured educational sessions covering stress management, healthy lifestyle, chronic disease self-management, and relaxation techniques.

**Cognitive-behavioural therapy (CBT):** Structured psychological interventions targeting maladaptive thoughts and behaviours related to stress.

**Music therapy:** Active or passive music interventions including listening, singing, or playing instruments.

The primary outcome was **perceived stress** measured by validated scales. Secondary outcomes were **anxiety**, **depression**, and **cortisol levels** (a physiological stress biomarker measured from saliva, blood, or hair samples).

The meta-analysis included 23 randomised controlled trials with a total of 1,847 participants. All participants were adults aged 60 years or older (mean ages across studies ranged from 65 to 82 years). Participants were community-dwelling older adults, residents of nursing homes or assisted living facilities, and outpatients from geriatric clinics. Studies were conducted across multiple countries including China, the United States, the United Kingdom, Australia, Japan, South Korea, and several European nations. Exclusion criteria across the individual trials typically included: diagnosed psychiatric disorders (severe depression, psychosis, dementia), cognitive impairment (Mini-Mental State Examination scores below 24), acute medical illness, and regular practice of the intervention being studied (e.g., current yoga practitioners were excluded from yoga trials).

**Perceived stress** was measured using the Perceived Stress Scale (PSS), a 10-item or 14-item questionnaire where participants rate how often they felt their lives were unpredictable, uncontrollable, or overwhelming over the past month. Scores range from 0 to 40 (10-item version) or 0 to 56 (14-item version), with higher scores indicating greater perceived stress. The PSS is the most widely used psychological instrument for measuring stress perception and has good reliability and validity in older adult populations.

**Anxiety** was measured using several validated instruments including the State-Trait Anxiety Inventory (STAI), the Geriatric Anxiety Inventory (GAI), the Hospital Anxiety and Depression Scale-Anxiety subscale (HADS-A), and the Beck Anxiety Inventory (BAI). These are self-report questionnaires with established cut-off scores for clinically significant anxiety.

**Depression** was measured using the Geriatric Depression Scale (GDS), the Center for Epidemiologic Studies Depression Scale (CES-D), the Patient Health Questionnaire-9 (PHQ-9), and the Hamilton Depression Rating Scale (HAM-D). The GDS is specifically designed for older adults and avoids somatic items that might be confounded with physical ageing.

**Cortisol levels** were measured from saliva samples (most common), blood serum, or hair samples. Salivary cortisol was typically collected at multiple time points (morning, evening, or diurnal slope) to capture circadian patterns. Hair cortisol provides a retrospective measure of cumulative cortisol exposure over several months.

**Study design:** This is a systematic review with pairwise meta-analysis and network meta-analysis (NMA) of randomised controlled trials (RCTs). A network meta-analysis is a statistical technique that allows simultaneous comparison of multiple interventions by combining direct evidence (head-to-head comparisons within individual trials) with indirect evidence (comparisons across trials through a common comparator, usually the control condition). This is particularly useful when few studies directly compare two active interventions against each other.

**Search strategy:** The researchers searched 10 electronic databases (PubMed, EMBASE, Web of Science, Cochrane Central Register of Controlled Trials, CINAHL, PsycINFO, CNKI, SinoMed, VIP, and WanFang) on January 9, 2024. They also manually searched reference lists of included studies and relevant review articles. The search covered publications from database inception through January 2024, with no language restrictions.

**Inclusion criteria:** Studies had to be RCTs (including cluster-RCTs and crossover RCTs) examining any intervention aimed at reducing perceived stress in adults aged 60 years or older. The intervention had to be compared with a control condition (usual care, waitlist, attention control, or no intervention) or another active intervention. Studies had to report at least one outcome measure of perceived stress using a validated scale.

**Risk of bias assessment:** Two reviewers independently assessed methodological quality using the Cochrane Risk of Bias tool (RoB 2) for RCTs. This evaluates random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias. Disagreements were resolved through discussion with a third reviewer.

**Statistical analysis:** Pairwise meta-analyses were conducted using Review Manager v.5.4 with random-effects models (DerSimonian and Laird method) to account for expected heterogeneity between studies. Network meta-analyses were conducted using Stata v.16.0 with a frequentist approach using multivariate meta-analysis. Effect sizes were reported as standardised mean differences (SMD) with 95% confidence intervals (CI) because different studies used different measurement scales for the same outcomes. SMD values of 0.2, 0.5, and 0.8 are conventionally interpreted as small, medium, and large effects, respectively. The researchers assessed inconsistency between direct and indirect evidence using node-splitting methods and evaluated publication bias using funnel plots and Egger's test.

**What this design can and cannot prove:**

**Can prove:** The NMA can rank interventions by their relative effectiveness for reducing perceived stress, anxiety, depression, and cortisol levels. The inclusion of only RCTs provides strong internal validity for causal claims about intervention effects. The comprehensive search across 10 databases reduces the risk of missing relevant studies.

**Cannot prove:** The NMA cannot establish optimal dose, frequency, or duration of interventions because individual studies varied widely in these parameters. The analysis cannot determine whether effects persist long-term because most included studies had follow-up periods of 8-16 weeks, with very few extending beyond 6 months. The NMA cannot identify which specific components of complex interventions (e.g., which aspect of yoga: postures, breathing, or meditation) drive the effects. The analysis cannot determine whether results generalise to older adults with specific medical conditions, cognitive impairment, or those living in different cultural contexts than those studied.

**Major methodological weaknesses:**

**Small number of studies per comparison:** Many intervention categories included only 2-4 studies, making network estimates imprecise and potentially unreliable.

**High risk of bias in included studies:** Many individual RCTs had unclear or high risk of bias due to lack of blinding (participants and personnel cannot be blinded to exercise, yoga, or reminiscence therapy), incomplete outcome data, and selective outcome reporting.

**Heterogeneity within intervention categories:** "Exercise" included everything from walking programmes to resistance training to tai chi, making it difficult to know which specific exercise modality works best.

**Publication bias:** Funnel plot asymmetry suggested possible publication bias favouring positive results, particularly for exercise interventions.

**Limited physiological data:** Only 8 of 23 studies measured cortisol, and measurement protocols varied substantially (different times of day, different sample types, different assays).

**Primary outcome - Perceived stress:**

Reminiscence therapy had the highest probability (98.6%) of being the most effective intervention for reducing perceived stress, with a large effect size compared to control (SMD = -1.12; 95% CI [-1.68, -0.56]; p < 0.001).

Exercise ranked second with a 68.1% probability of being among the most effective interventions (SMD = -0.67; 95% CI [-1.08, -0.26]; p = 0.001).

Yoga ranked third with a 56.1% probability (SMD = -0.54; 95% CI [-0.97, -0.11]; p = 0.01).

Mindfulness-based interventions showed a moderate effect (SMD = -0.48; 95% CI [-0.89, -0.07]; p = 0.02).

Health education showed a small-to-moderate effect (SMD = -0.35; 95% CI [-0.72, 0.02]; p = 0.06, not statistically significant).

CBT showed a moderate effect (SMD = -0.52; 95% CI [-1.13, 0.09]; p = 0.09, not statistically significant).

Music therapy showed a small effect (SMD = -0.28; 95% CI [-0.79, 0.23]; p = 0.28, not statistically significant).

**Secondary outcome - Cortisol levels:**

Across all interventions, there was a statistically significant reduction in cortisol levels compared to control (SMD = -0.30; 95% CI [-0.54, -0.06]; p = 0.01). This is a small-to-moderate effect.

Exercise showed the largest reduction in cortisol (SMD = -0.45; 95% CI [-0.82, -0.08]; p = 0.02).

Yoga also showed significant cortisol reduction (SMD = -0.38; 95% CI [-0.71, -0.05]; p = 0.03).

Reminiscence therapy showed a non-significant trend toward cortisol reduction (SMD = -0.25; 95% CI [-0.63, 0.13]; p = 0.20).

**Secondary outcome - Depression:**

Exercise showed the largest effect on reducing depression symptoms (SMD = -1.84; 95% CI [-3.69, 0.01]; p = 0.05). This is a very large effect but the confidence interval is extremely wide, crossing zero, meaning the true effect could range from negligible to very large.

Reminiscence therapy showed a moderate effect on depression (SMD = -0.56; 95% CI [-1.12, 0.00]; p = 0.05, borderline significant).

Yoga showed a moderate effect (SMD = -0.42; 95% CI [-0.89, 0.05]; p = 0.08, not significant).

Mindfulness showed a small effect (SMD = -0.31; 95% CI [-0.72, 0.10]; p = 0.14, not significant).

**Secondary outcome - Anxiety:**

Health education showed the largest and only statistically significant effect on reducing anxiety (SMD = -0.77; 95% CI [-1.27, -0.26]; p = 0.03).

Exercise showed a moderate effect (SMD = -0.52; 95% CI [-1.08, 0.04]; p = 0.07, not significant).

Yoga showed a small-to-moderate effect (SMD = -0.38; 95% CI [-0.82, 0.06]; p = 0.09, not significant).

Reminiscence therapy showed a small effect (SMD = -0.29; 95% CI [-0.71, 0.13]; p = 0.18, not significant).

**Network rankings (Surface Under the Cumulative Ranking curve - SUCRA values):**

For perceived stress: Reminiscence therapy (SUCRA = 98.6%), Exercise (68.1%), Yoga (56.1%), CBT (52.3%), Mindfulness (48.7%), Health education (38.2%), Music therapy (28.0%).

For depression: Exercise (SUCRA = 85.2%), Reminiscence therapy (72.1%), Yoga (58.3%), CBT (45.6%), Mindfulness (38.9%).

For anxiety: Health education (SUCRA = 82.4%), Exercise (68.7%), Yoga (55.2%), Mindfulness (43.8%), Reminiscence therapy (39.9%).

To translate these statistical effects into practical terms:

**Reminiscence therapy for perceived stress:** The SMD of -1.12 means that the average person in the reminiscence therapy group scored about 1.1 standard deviations lower on the Perceived Stress Scale than the average person in the control group. On the 10-item PSS (0-40 scale), this translates to roughly a 5-7 point reduction. To put this in context, a person who "often" felt stressed would shift to "sometimes" or "almost never" feeling stressed. This is a large enough effect that most people would notice a meaningful difference in their daily experience of stress.

**Exercise for depression:** The SMD of -1.84 is unusually large and likely inflated by the small number of studies (only 4 studies contributed to this analysis) and the wide confidence interval. A more realistic interpretation is that exercise probably has a moderate-to-large antidepressant effect in older adults, comparable to what has been found in larger meta-analyses of exercise for depression in general adult populations (typically SMD = -0.5 to -0.8).

**Cortisol reduction:** The overall SMD of -0.30 across all interventions means that the average person in an intervention group had cortisol levels about 0.3 standard deviations lower than controls. For morning salivary cortisol (typical range 10-20 nmol/L), this translates to roughly a 2-4 nmol/L reduction. This is a modest effect but biologically meaningful, as chronic cortisol elevation is linked to hypertension, impaired immune function, and cognitive decline.

**Health education for anxiety:** The SMD of -0.77 means the average person receiving health education scored about 0.77 standard deviations lower on anxiety measures than controls. On the Geriatric Anxiety Inventory (0-20 scale), this translates to roughly a 3-4 point reduction, which could move someone from above the clinical cut-off (typically 8-10) to below it.

**What the authors acknowledge:**

The small number of studies for many intervention categories limits the reliability of network rankings.

High heterogeneity within intervention categories (different types of exercise, different formats of reminiscence therapy) makes it difficult to identify optimal protocols.

Most included studies had short follow-up periods (8-16 weeks), so long-term effects are unknown.

Many studies had high or unclear risk of bias, particularly regarding blinding of participants and personnel.

Publication bias may have inflated effect estimates, especially for exercise interventions.

The network meta-analysis assumes transitivity (that studies are sufficiently similar in terms of participants, settings, and outcome measures to allow indirect comparisons), which may not hold across all comparisons.

**Additional critical concerns:**

**Cultural specificity:** Most studies were conducted in China, the United States, and Europe. Reminiscence therapy may be particularly effective in cultures that value intergenerational storytelling and collective memory, and results may not generalise to all cultural contexts.

**Self-report bias:** All primary outcomes (perceived stress, anxiety, depression) were measured by self-report questionnaires. Participants who volunteer for stress-reduction studies may be particularly motivated to report improvement, especially when they have invested time in an intervention.

**No blinding:** It is impossible to blind participants to whether they are doing yoga, reminiscence therapy, or exercise. This creates expectation effects where participants who believe an intervention will work may report greater benefits.

**Control group