Comment on ‘Effectiveness of Horticultural Therapy in Older Patients With Dementia: A Meta-Analysis Systematic Review’

Abstract

We read with great interest the article by Wang et al. (2024), titled ‘Effectiveness of Horticultural Therapy in Older Patients with Dementia: A Meta-Analysis Systematic Review’, recently published in the Journal of Clinical Nursing. The study's focus on evaluating the impact of horticultural therapy on cognitive function, depression, activities of daily living (ADL) and quality of life in older adults with dementia addresses an important gap in non-pharmacological interventions for this population. However, upon a closer examination of the methodology and scope of the study, several limitations and areas for improvement could be considered to enhance the robustness and generalisability of the findings.

First, the meta-analysis included a relatively small number of studies (n = 9) with a limited overall sample size of 655 participants, which diminishes the statistical power of the conclusions drawn. The authors would have benefited from expanding their search to include more diverse databases and unpublished studies to increase the pool of eligible trials. A larger sample size would also allow for more nuanced subgroup analyses and reduce the impact of potential outliers. Additionally, the geographic concentration of studies, predominantly conducted in China and Italy, limits the generalisability of the findings across diverse cultural contexts. Expanding the scope to include studies from other regions could provide a more comprehensive understanding of the effects of horticultural therapy across different settings.

Moreover, the heterogeneity observed in several outcomes, particularly cognitive function (I² = 91.1%), suggests significant variability in intervention protocols, participant characteristics and outcome measurement tools. The high degree of heterogeneity may undermine the reliability of the pooled estimates. The authors could have addressed this issue by conducting a more detailed subgroup analysis that considers the frequency, intensity and specific components of the horticultural therapy interventions. Furthermore, adopting a standardised approach to defining intervention types (e.g., structured vs. unstructured) and specifying consistent outcome measures across studies would have strengthened the internal validity of the findings.

The authors could have further strengthened their analysis by incorporating prediction intervals in the forest plots (Higgins, Thompson, and Spiegelhalter 2009). Unlike confidence intervals, prediction intervals would provide a range for expected outcomes in future studies, thereby offering clearer insights into potential variability across different settings. This addition would have been particularly valuable given the moderate heterogeneity observed (I² = 91.1% for the meta-analysis of the effectiveness of horticultural therapy on cognitive function in older patients with dementia). Moreover, applying the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach to assess the certainty of the evidence would offer a structured evaluation of the quality of the findings. This framework would enable readers to understand the level of confidence in the reported effects and guide clinical decision-making accordingly (Shi and Lin 2019).

The quality of the included studies was another concern. Although the authors used the RoB 2.0 tool to assess bias, many of the included trials exhibited risks related to randomisation, blinding and outcome reporting. This could introduce biases that affect the validity of the meta-analytic conclusions. We suggest that future meta-analyses include a sensitivity analysis that excludes high-risk studies to determine the robustness of the findings. In addition, the absence of a publication bias analysis, due to the small number of included studies, is a notable limitation. If possible, the authors could have utilised alternative methods such as a trim-and-fill analysis to evaluate potential biases more comprehensively (Shi and Lin 2019).

Furthermore, the scope of the analysis could have been expanded to include additional outcomes that are highly relevant for dementia care, such as agitation, caregiver burden and social engagement. Addressing these outcomes would provide a more holistic assessment of the effectiveness of horticultural therapy. The authors could also have considered stratifying results based on dementia subtype, as different types of dementia may respond differently to therapeutic interventions. This would have allowed for more targeted recommendations for clinical practice.

Another point of consideration is the duration and frequency of the interventions. The results indicated that shorter interventions (< 6 months) were more effective, but the underlying reasons for this pattern were not explored. It would have been valuable if the authors had discussed potential explanations, such as differences in participant adherence, motivation or therapeutic novelty. Longer follow-up periods and standardised durations across studies would also help in understanding the sustainability of the observed benefits.

Finally, the lack of consideration for potential confounding variables, such as baseline cognitive function, comorbidities and concurrent treatments, is a methodological limitation. Employing meta-regression techniques or multivariable regression analysis in future studies to adjust for these factors would provide a clearer picture of the true impact of horticultural therapy on the measured outcomes.

In conclusion, while the study provides important insights into the potential benefits of horticultural therapy for older patients with dementia, addressing the aforementioned methodological limitations would greatly enhance the validity and applicability of the findings. We commend the authors for their contribution to this field and hope that our suggestions will be considered for future research endeavours.

R.M., G.B. and A.K.B. critically reviewed and provided comments on methodological aspects, and S.S. and R.M. wrote the edited draft.

The authors have nothing to report.

The authors declare no conflicts of interest.