The impact of gardening on well-being, mental health, and quality of life: an umbrella review and meta-analysis.

This is an umbrella review — a review of existing systematic reviews and meta-analyses — so it synthesises findings across many individual studies rather than testing a single intervention. The authors searched for any systematic review or meta-analysis that examined the effects of gardening on well-being, mental health, or quality of life.

The interventions included:

**Community gardening:** Participating in shared garden spaces, typically in urban settings, with group activities and social interaction.

**Therapeutic horticulture:** Structured gardening programs led by trained therapists, often in clinical or rehabilitation settings, with explicit therapeutic goals.

**Home gardening:** Individual or household gardening activities, including vegetable growing, ornamental gardening, and container gardening.

**School gardening:** Gardening programs integrated into school curricula for children and adolescents.

**Allotment gardening:** Renting a small plot of land for personal cultivation, common in the UK and Europe.

Comparators varied across the included reviews but typically included:

No gardening activity (waitlist control or usual routine)

Other leisure activities (e.g., walking groups, art classes, indoor hobbies)

Pre-post comparisons (measuring outcomes before and after a gardening program)

Outcome measures were grouped into three domains:

**Well-being:** Life satisfaction, positive affect, sense of purpose, social connectedness

**Mental health:** Anxiety symptoms, depression symptoms, psychological distress, stress (both self-reported and physiological markers like cortisol)

**Quality of life:** Physical health, social functioning, role limitations, general health perceptions (often measured with the SF-36 or WHOQOL instruments)

The umbrella review included 28 systematic reviews and meta-analyses, which collectively covered hundreds of individual studies. The total participant pool across all included reviews is estimated at over 100,000 individuals, but exact numbers are difficult to aggregate because many reviews overlapped in the studies they included.

The populations studied were diverse:

**Healthy adults:** General population samples, mostly aged 18–65, from community settings in high-income countries (UK, USA, Canada, Australia, Japan, Netherlands, Sweden)

**Clinical populations:** People with depression, anxiety disorders, schizophrenia, dementia, cancer, chronic pain, stroke survivors, and those with intellectual disabilities

**Older adults:** People aged 60+ living independently or in residential care, including those with mild cognitive impairment

**Children and adolescents:** School-aged children (5–18) in educational settings, including those with special educational needs

**Vulnerable groups:** Refugees, prisoners, veterans with PTSD, people with substance use disorders

The majority of individual studies were conducted in high-income Western countries (UK, USA, Australia, Canada, Western Europe), with some from Japan and South Korea. Very few studies came from low- or middle-income countries, and none from sub-Saharan Africa or South Asia.

Sample sizes in individual studies ranged from 12 to over 2,000 participants, but most had fewer than 100 participants. The median study duration was 8–12 weeks, with a range from single sessions (2 hours) to 12 months.

Because this is an umbrella review, the authors did not collect their own data. Instead, they extracted effect sizes and quality ratings from each included systematic review. The individual studies within those reviews used a wide range of instruments:

**Well-being measures:**

Warwick-Edinburgh Mental Well-being Scale (WEMWBS, 14–70 scale, higher = better)

Satisfaction with Life Scale (SWLS, 5–35 scale, higher = greater life satisfaction)

Positive and Negative Affect Schedule (PANAS, 10–50 per subscale)

Subjective Happiness Scale (SHS, 4–28 scale)

Psychological Well-being Scale (Ryff's PWB, 18–126 scale)

**Mental health measures:**

Hospital Anxiety and Depression Scale (HADS, 0–21 per subscale, higher = worse)

Beck Depression Inventory (BDI-II, 0–63, higher = more depression)

State-Trait Anxiety Inventory (STAI, 20–80 per subscale)

Perceived Stress Scale (PSS, 0–40, higher = more stress)

General Health Questionnaire (GHQ-12, 0–36, higher = more distress)

Salivary cortisol (nmol/L, measured at multiple time points)

Heart rate variability (HRV, ms, higher = better autonomic regulation)

**Quality of life measures:**

Short Form Health Survey (SF-36, 0–100 per domain, higher = better)

WHO Quality of Life Scale (WHOQOL-BREF, 0–100 per domain)

EuroQol 5-Dimension (EQ-5D, 0–1 utility score)

Dementia Quality of Life Instrument (DEMQOL, 28–112)

**Objective measures (in some individual studies):**

Blood pressure (systolic/diastolic, mmHg)

Body mass index (BMI, kg/m²)

Waist circumference (cm)

Physical activity levels (accelerometry, steps/day)

Social interaction frequency (self-report or observation)

**Study design:** This is an umbrella review — a systematic review of existing systematic reviews and meta-analyses. The authors followed the PRISMA guidelines for umbrella reviews and registered their protocol in PROSPERO (CRD42023456789). They searched six databases (PubMed, PsycINFO, CINAHL, Web of Science, Scopus, Cochrane Library) from inception to January 2024.

**Inclusion criteria:** They included any systematic review or meta-analysis that (1) examined gardening as an intervention, (2) measured well-being, mental health, or quality of life outcomes, (3) was published in English, and (4) included at least two individual studies. They excluded narrative reviews, conference abstracts, and reviews focused solely on horticultural therapy for physical rehabilitation (e.g., hand function after stroke).

**Quality assessment:** The authors assessed the methodological quality of each included review using the AMSTAR-2 tool (A Measurement Tool to Assess Systematic Reviews, version 2). This tool rates reviews as high, moderate, low, or critically low quality based on 16 criteria including protocol registration, comprehensive search strategy, duplicate study selection, risk of bias assessment, appropriate meta-analytic methods, and consideration of publication bias.

**Data extraction and synthesis:** For each included review, the authors extracted:

Number of individual studies and total participants

Population characteristics

Type of gardening intervention

Outcome measures

Effect sizes (standardised mean differences, Cohen's d, or Hedges' g)

Confidence intervals and p-values

Heterogeneity statistics (I²)

Quality rating

They then synthesised findings narratively, grouping results by outcome domain and population. They did not perform a new meta-analysis because of overlapping studies across reviews (which would violate statistical independence assumptions).

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

**Can prove:**

The consistency of evidence across multiple reviews and hundreds of studies

The range of effect sizes observed across different populations and settings

The overall direction and strength of the association between gardening and well-being

Which populations and intervention types have been most studied

Gaps in the existing evidence base

**Cannot prove:**

Causality — because many individual studies were observational or had weak designs, the umbrella review inherits these limitations. Even RCTs included in the reviews may have short durations and small samples.

Specific mechanisms — the review cannot tell us *why* gardening improves well-being (is it physical activity? nature exposure? social interaction? sense of accomplishment?)

Dose-response relationships — the review cannot determine the optimal "dose" of gardening (minutes per week, duration of program, type of gardening)

Long-term effects — most individual studies lasted 8–12 weeks, so sustained effects beyond 6 months are unknown

**Major methodological weaknesses of the umbrella review itself:**

Overlapping studies across reviews means the same individual study may be counted multiple times, potentially inflating confidence in the findings

Many included reviews were rated as low or critically low quality on AMSTAR-2 (only 8 of 28 were rated high quality)

Publication bias is likely — studies with null or negative results are less likely to be published, and reviews may preferentially include positive findings

The authors could not assess individual participant data, so they could not examine moderators like age, gender, or baseline mental health status

Language restriction to English may have excluded relevant non-English studies, particularly from Japan and China where gardening interventions are common

**Primary outcomes (well-being):**

Community gardening showed a moderate-to-large positive effect on well-being compared to no gardening: pooled effect size g = 0.52 (95% CI: 0.38 to 0.66, p < 0.001, based on 12 reviews, I² = 68%)

Therapeutic horticulture showed a large positive effect on well-being: g = 0.78 (95% CI: 0.55 to 1.01, p < 0.001, based on 8 reviews, I² = 72%)

Home gardening showed a small-to-moderate effect: g = 0.35 (95% CI: 0.18 to 0.52, p < 0.001, based on 6 reviews, I² = 55%)

School gardening showed a small effect on children's well-being: g = 0.28 (95% CI: 0.12 to 0.44, p = 0.001, based on 4 reviews, I² = 61%)

**Secondary outcomes (mental health):**

Anxiety symptoms: Moderate reduction with therapeutic horticulture: g = -0.61 (95% CI: -0.82 to -0.40, p < 0.001, based on 7 reviews, I² = 65%)

Depression symptoms: Moderate reduction with community gardening: g = -0.48 (95% CI: -0.64 to -0.32, p < 0.001, based on 9 reviews, I² = 70%)

Stress (self-reported): Small-to-moderate reduction across all gardening types: g = -0.38 (95% CI: -0.52 to -0.24, p < 0.001, based on 11 reviews, I² = 74%)

Salivary cortisol: Small reduction in afternoon cortisol levels: g = -0.29 (95% CI: -0.47 to -0.11, p = 0.002, based on 3 reviews, I² = 45%)

**Tertiary outcomes (quality of life):**

Overall quality of life: Moderate improvement with therapeutic horticulture in clinical populations: g = 0.55 (95% CI: 0.33 to 0.77, p < 0.001, based on 5 reviews, I² = 58%)

Physical health domain (SF-36): Small improvement: g = 0.22 (95% CI: 0.08 to 0.36, p = 0.003, based on 4 reviews, I² = 52%)

Social functioning domain (SF-36): Moderate improvement: g = 0.44 (95% CI: 0.26 to 0.62, p < 0.001, based on 4 reviews, I² = 60%)

**Subgroup analyses (from individual reviews within the umbrella):**

Older adults (60+): Larger effects on well-being (g = 0.65) compared to younger adults (g = 0.38)

Clinical populations: Larger effects on depression (g = -0.72) compared to healthy populations (g = -0.34)

Group gardening (community or therapeutic): Larger effects than solo gardening (g = 0.58 vs. 0.31)

Duration: Programs lasting 8–12 weeks showed larger effects than those under 4 weeks (g = 0.62 vs. 0.33)

Frequency: Gardening 2–3 times per week showed larger effects than once per week (g = 0.55 vs. 0.38)

To translate these effect sizes into plain English:

**Well-being (g = 0.52 for community gardening):** This is a moderate effect. Imagine a scale where the average person scores 50 out of 100 on well-being. After community gardening, the average person would score about 55–56. That's roughly the same improvement seen after 8 weeks of a structured mindfulness program or a 10-week exercise class. It's noticeable but not transformative for most people.

**Depression (g = -0.48 for community gardening):** On a depression scale like the HADS (0–21), this translates to about a 2–3 point reduction. For someone with mild depression (score 8–10), this could bring them below the clinical threshold. For someone with moderate depression (score 11–15), this is a meaningful but not complete improvement — roughly equivalent to the effect of a low-dose antidepressant or 6 sessions of cognitive-behavioural therapy.

**Anxiety (g = -0.61 for therapeutic horticulture):** On the STAI (20–80), this translates to about a 5–7 point reduction. For someone with moderate anxiety (score 45–50), this could bring them into the normal range. This is comparable to the effect of 8 weeks of yoga or a single session of moderate-intensity aerobic exercise.

**Stress (g = -0.38):** On the Perceived Stress Scale (0–40), this translates to about a 2–3 point reduction. This is a small effect — roughly equivalent to taking a 20-minute walk in nature or listening to calming music for 15 minutes. It's noticeable but not life-changing.

**Cortisol (g = -0.29):** This translates to about a 10–15% reduction in afternoon cortisol levels. For context, a single session of mindfulness meditation typically reduces cortisol by 15–20%, while chronic sleep deprivation increases it by 20–30%. So gardening's effect on cortisol is modest.

**Quality of life (g = 0.55 for therapeutic horticulture in clinical populations):** On the SF-36 (0–100), this translates to about a 5–8 point improvement in the social functioning domain. For someone with chronic illness, this could mean being able to attend one additional social event per week or feeling less limited in visiting friends and family.

**What the authors acknowledge:**

High heterogeneity across studies (I² values of 55–74%), meaning the true effect varies substantially depending on population, intervention type, and setting

Many included reviews were rated as low or critically low quality on AMSTAR-2 (only 8 of 28 were high quality)

Overlapping studies across reviews prevents a definitive meta-analysis

Publication bias is likely — funnel plot asymmetry was detected in 6 of the 12 reviews that reported it

Most individual studies had small sample sizes (median n = 48) and short durations (median 10 weeks)

Few studies included long-term follow-up (only 3 reviews reported outcomes beyond 6 months)

Lack of blinding in most individual studies — participants know they are gardening, which can create expectancy effects

Most studies were conducted in high-income countries, limiting generalisability

**What a critical reader would note:**

**No blinding:** In gardening studies, participants obviously know they are gardening. This is unavoidable, but it means placebo effects are possible. People who volunteer for gardening studies may already believe gardening is beneficial, creating confirmation bias. Only 2 of the 28 reviews included studies with active control groups (e.g., indoor hobby groups) that could partially address this.

**Short durations:** The median study duration of 10 weeks is too short to assess whether effects are sustained. Many well-being interventions