Effectiveness of Horticultural Therapy in Older Patients With Dementia: A Meta-Analysis Systemic Review.

This was a meta-analysis that pooled data from 9 randomised controlled trials (RCTs) to evaluate horticultural therapy (HT) compared to conventional care (standard dementia care without gardening activities) in older patients with dementia.

**Intervention:** Horticultural therapy – defined as structured or semi-structured gardening activities led by a trained therapist or healthcare professional. Activities included planting, watering, pruning, harvesting, flower arranging, and sensory garden experiences. Sessions ranged from 30–60 minutes, with frequencies from once weekly to daily.

**Comparator:** Conventional care – standard dementia care including medication management, basic nursing care, social activities, and routine daily living support. No studies used a sham or placebo control.

**Primary outcomes measured:**

Cognitive function (assessed via standardised scales)

Depression symptoms

Activities of daily living (ADL)

Quality of life (QoL)

**Secondary outcomes examined in subgroup analyses:**

Effect of intervention frequency (≥2 times/week vs. <2 times/week)

Effect of intervention duration (<6 months vs. ≥6 months)

Effect of setting (outdoor vs. indoor)

Effect of structure (structured vs. non-structured sessions)

**Total sample:** 655 older patients diagnosed with dementia across 9 RCTs.

**Population specifics:**

All participants were aged 60 years or older (mean ages ranged from 72–86 years across studies)

Diagnosed with dementia (types included Alzheimer's disease, vascular dementia, and mixed dementia)

Mild to moderate severity (most studies excluded severe dementia)

Recruited from: nursing homes, long-term care facilities, dementia care units, and hospital geriatric wards

Countries represented: China (5 studies), Taiwan (2 studies), South Korea (1 study), and the United States (1 study)

Both male and female participants (approximately 60–70% female, consistent with dementia epidemiology)

Exclusion criteria common across studies: severe hearing/vision impairment, acute medical illness, severe behavioural disturbances, and inability to stand or walk with assistance

**Important note for generalisability:** All studies were conducted in East Asian or Western care settings. No studies included community-dwelling older adults living independently. Results may not apply to home-based settings or to individuals with severe dementia.

The meta-analysis used standardised mean differences (SMD) to combine results from different measurement tools across studies. Here are the specific instruments used:

**Cognitive function:**

Mini-Mental State Examination (MMSE, 0–30 scale, higher = better cognition)

Montreal Cognitive Assessment (MoCA, 0–30 scale, higher = better cognition)

Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog, 0–70 scale, lower = better cognition)

Clock Drawing Test (CDT, 0–5 scale, higher = better)

**Depression:**

Geriatric Depression Scale (GDS, 0–15 or 0–30 scale, higher = more depressed)

Cornell Scale for Depression in Dementia (CSDD, 0–38 scale, higher = more depressed)

Hamilton Depression Rating Scale (HDRS, 0–52 scale, higher = more depressed)

**Activities of Daily Living (ADL):**

Barthel Index (0–100 scale, higher = more independent)

Katz Index of Independence in ADL (0–6 scale, higher = more independent)

Functional Independence Measure (FIM, 18–126 scale, higher = more independent)

**Quality of Life:**

Quality of Life in Alzheimer's Disease (QoL-AD, 13–52 scale, higher = better QoL)

Dementia Quality of Life Instrument (DQoL, 0–100 scale, higher = better QoL)

World Health Organization Quality of Life-BREF (WHOQOL-BREF, 0–100 scale, higher = better QoL)

**Risk of bias assessment:** Two independent reviewers used the Cochrane Risk of Bias tool (RoB 2.0) for RCTs. Publication bias was assessed using funnel plots and Egger's test.

**Study design:** Systematic review and meta-analysis of randomised controlled trials (RCTs). This is the highest level of evidence when well-conducted, as it pools data from multiple trials to increase statistical power and estimate overall effect sizes.

**Search strategy:** The authors searched 6 databases (PubMed, Web of Science, Cochrane Library, Embase, CNKI, and Wanfang) in November 2023. They included both English and Chinese language publications with no date restrictions. Two independent reviewers screened titles/abstracts, then full texts, with disagreements resolved by a third reviewer.

**Inclusion criteria:**

RCT design only

Participants aged ≥60 years with a formal dementia diagnosis

Intervention explicitly described as horticultural therapy (not just passive garden exposure)

Comparator was conventional care or standard treatment

Reported at least one of the four outcomes (cognition, depression, ADL, QoL)

Published in English or Chinese

**Exclusion criteria:**

Quasi-experimental designs, case studies, observational studies

Participants with other major psychiatric disorders (e.g., schizophrenia)

Interventions combining HT with other therapies (e.g., music therapy, pet therapy) where HT effects couldn't be isolated

**Data extraction and synthesis:**

Two independent reviewers extracted data on: sample size, participant characteristics, intervention details (frequency, duration, setting, structure), outcome measures, and results

Standardised mean differences (SMD) with 95% confidence intervals were calculated using random-effects models (DerSimonian and Laird method)

Heterogeneity was assessed using I² statistics (I² > 50% considered substantial heterogeneity)

Subgroup analyses were pre-planned to explore sources of heterogeneity

**Risk of bias assessment:**

Using RoB 2.0, studies were rated as "low risk," "some concerns," or "high risk" across five domains: randomisation process, deviations from intended interventions, missing outcome data, measurement of outcome, and selection of reported result

Overall, the included studies were rated as "low quality" – most had "some concerns" or "high risk" due to lack of blinding (participants and personnel knew they were receiving HT), small sample sizes, and incomplete outcome reporting

**What this design can prove:**

A well-conducted meta-analysis of RCTs can establish that HT causes improvements in cognition, mood, ADL, and QoL compared to usual care

The random-effects model accounts for between-study variability, making results more generalisable

Subgroup analyses can identify which intervention parameters (frequency, duration, setting) are most effective

**What this design cannot prove:**

Cannot determine the optimal "dose" of HT (exact minutes per session, specific activities) – only broad categories (≥2 vs. <2 sessions/week; <6 vs. ≥6 months)

Cannot establish mechanisms – why HT works (sensory stimulation, physical activity, social interaction, sunlight exposure, or a combination)

Cannot rule out placebo effects – since participants and staff knew who received HT, expectation effects may inflate results

Cannot compare HT to other active interventions (e.g., exercise therapy, music therapy) – only to usual care

Publication bias is possible – studies with null or negative results may be unpublished, especially in Chinese databases

The "low quality" rating of included studies means conclusions are tentative

**Major methodological weaknesses:**

No blinding of participants or personnel in any included study (impossible to blind a gardening intervention)

Small sample sizes in individual studies (range: 30–120 participants)

Short follow-up periods (most studies lasted 8–16 weeks; none followed participants after intervention ended)

High heterogeneity for some outcomes (I² > 60% for depression and QoL), meaning results varied considerably across studies

Potential for selective outcome reporting (some studies didn't report all four outcomes)

All studies conducted in institutional settings – no community-based or home-based trials

All results are reported as standardised mean differences (SMD) with 95% confidence intervals (CI). SMD is used because different studies used different scales. An SMD of 0.2 is considered small, 0.5 moderate, and 0.8 large.

**Primary outcomes (HT vs. conventional care):**

**Cognitive function:** Significant improvement favouring HT (SMD = 0.48, 95% CI: 0.28 to 0.68, p < 0.001, I² = 34%). Based on 8 studies with 587 participants. This is a moderate effect size. Heterogeneity was low (I² = 34%), meaning results were fairly consistent across studies.

**Depression:** Significant reduction in depressive symptoms favouring HT (SMD = -0.52, 95% CI: -0.82 to -0.22, p < 0.001, I² = 62%). Based on 6 studies with 412 participants. This is a moderate-to-large effect. Heterogeneity was substantial (I² = 62%), meaning some studies showed much larger effects than others.

**Activities of Daily Living:** Significant improvement favouring HT (SMD = 0.41, 95% CI: 0.16 to 0.66, p = 0.001, I² = 45%). Based on 5 studies with 348 participants. This is a small-to-moderate effect. Heterogeneity was moderate.

**Quality of Life:** Significant improvement favouring HT (SMD = 0.55, 95% CI: 0.22 to 0.88, p = 0.001, I² = 65%). Based on 4 studies with 276 participants. This is a moderate-to-large effect. Heterogeneity was substantial (I² = 65%).

**Subgroup analyses (for cognitive function only):**

**Frequency:** Interventions with ≥2 sessions per week showed significant cognitive improvement (SMD = 0.56, 95% CI: 0.32 to 0.80, p < 0.001), while <2 sessions per week did not reach significance (SMD = 0.28, 95% CI: -0.08 to 0.64, p = 0.13). This suggests a minimum dose of twice weekly is needed.

**Duration:** Interventions lasting <6 months showed significant cognitive improvement (SMD = 0.55, 95% CI: 0.30 to 0.80, p < 0.001), while those lasting ≥6 months showed a smaller, non-significant effect (SMD = 0.31, 95% CI: -0.05 to 0.67, p = 0.09). This suggests diminishing returns with longer interventions.

**Setting:** Outdoor HT showed significant cognitive improvement (SMD = 0.62, 95% CI: 0.34 to 0.90, p < 0.001), while indoor HT showed a smaller but still significant effect (SMD = 0.34, 95% CI: 0.06 to 0.62, p = 0.02). Outdoor settings were superior.

**Structure:** Structured HT (planned activities with specific goals) showed significant cognitive improvement (SMD = 0.56, 95% CI: 0.32 to 0.80, p < 0.001), while non-structured HT (free gardening with minimal guidance) did not reach significance (SMD = 0.28, 95% CI: -0.12 to 0.68, p = 0.17). Structured sessions are essential.

**Publication bias:** Funnel plot asymmetry was detected for depression outcomes (Egger's test p = 0.04), suggesting possible publication bias – small studies with negative results may be missing.

To translate these statistical effects into real-world terms:

**Cognitive function (SMD = 0.48):** This is roughly equivalent to a 2–3 point improvement on the MMSE (0–30 scale) or a 3–4 point improvement on the MoCA. In practical terms, this might mean a person with dementia who could recall 2 out of 3 words at baseline might recall all 3 after HT, or someone who could not draw a clock correctly might be able to place numbers and hands. This is a clinically meaningful change – comparable to what some dementia medications achieve, but without side effects.

**Depression (SMD = -0.52):** This translates to approximately a 3–4 point reduction on the Geriatric Depression Scale (15-item version) or a 4–6 point reduction on the Cornell Scale. In practical terms, a person who reported feeling "often" depressed, hopeless, or withdrawn might shift to feeling "sometimes" or "rarely" depressed. This is roughly equivalent to the effect of a low-dose antidepressant, but achieved through a non-pharmacological activity.

**Activities of Daily Living (SMD = 0.41):** This translates to approximately a 10–15 point improvement on the Barthel Index (0–100). In practical terms, this might mean a person who needed help with bathing or dressing might become able to perform these tasks with minimal assistance, or someone who was incontinent might regain some bladder control. This is a modest but meaningful improvement in functional independence.

**Quality of Life (SMD = 0.55):** This translates to approximately a 4–6 point improvement on the QoL-AD scale (13–52). In practical terms, this might mean a person who rated their life as "poor" might shift to "fair" or "good" across domains like mood, memory, relationships, and overall wellbeing.

**Subgroup effects:** The cognitive benefit of outdoor HT (SMD = 0.62) is roughly 80% larger than indoor HT (SMD = 0.34). The benefit of twice-weekly sessions (SMD = 0.56) is roughly double that of once-weekly sessions (SMD = 0.28, non-significant). These are large differences in practical effectiveness.

**Acknowledged by authors:**

Low quality of included studies (most rated as "some concerns" or "high risk of bias" on RoB 2.0)

Small number of studies (only 9 RCTs) and small total sample (655 participants)

High heterogeneity for depression and QoL outcomes, limiting confidence in those findings

Publication bias detected for depression outcomes

Inability to perform subgroup analyses for depression, ADL, and QoL due to insufficient studies

All studies conducted in institutional settings, limiting generalisability to community-dwelling older adults

**Additional critical limitations:**

**No blinding whatsoever:** Participants, family members, and staff all knew who received HT. This introduces substantial placebo and expectancy effects. The "improvements" could partly reflect increased attention, social interaction, or staff enthusiasm rather than the gardening itself.

**No active control group:** HT was compared to "usual care" which typically involves minimal structured activity. A more rigorous test would compare HT to another structured group activity (e.g., art therapy, music therapy, exercise class) to isolate the specific effect of gardening.

**Short-term outcomes only:** No study followed participants after the intervention ended. We don't know if benefits persist for weeks or months after stopping HT.

**No dose-response data:** While subgroup analyses suggest ≥2 sessions/week and <6 months are optimal, we don't know the optimal session length (30 vs. 60 minutes) or the specific activities that drive benefits.

**Measurement limitations:** Most cognitive tests (MMSE, MoCA) are designed for screening, not for detecting subtle changes over short periods. Practice effects (improvement from taking the same test twice) could inflate results.

**Confounding by sunlight and physical activity:** Outdoor HT involves sunlight exposure (which improves mood and vitamin D levels) and light physical activity (which benefits cognition). The "gardening" component may be less important than these co-occurring factors.

**