Sample sizes were small (between 10 and 50), and the results need to be interpreted with caution. Detailed results can be found in Supplementary Appendix B, but these are summarized as follows grouped by outcome and then by intervention (garden

Galunisertib cost or horticulture therapy). Seven studies reported on dementia-related behaviors in response to time in a garden or engaged in horticultural activities. Agitation was reported in 6 studies, and other dementia-related outcomes, such as pacing, exit seeking, and violence, were reported less frequently and with mixed results. Only one study reported a negative trend of increased aggression over a 3-month period.20 Three garden studies measured agitation before and after exposure to a garden environment and

all used the Cohen-Mansfield Agitation Inventory (CMAI). All studies reported a positive trend18, 19 and 24 with CMAI scores, indicating reduced agitation associated with visiting the garden (P < .01); for example, Detweiler and colleagues 24 indicate an effect size of d = 0.64. Three studies measured dementia-related behaviors before and after horticultural therapy. 28, 30 and 32 Two studies 30 and 32 used an RCT design and report mixed results on the effectiveness of horticultural therapy in reducing physical and nonphysical GDC-0068 purchase aggression (also using CMAI). A positive trend was seen in the verbal agitation scores in both studies. Vuolo 28 also found a positive trend in the effect of horticultural therapy on physical and verbal aggression and a reduction in physically nonaggressive behaviors in a pre-post study of 50 residents with dementia, but the positive P-type ATPase changes were not statistically significant ( Supplementary Appendix B). Pacing or walking behaviors (including exit seeking and trespassing) were measured in 2 studies by observation.19 and 20 Both studies showed a positive trend in reduced pacing, trespassing, and exit seeking,

but also a decrease in walking (directed walking), which may be seen as a negative trend. Mooney and Lenore Nicell21 compared behaviors in 5 residential sites, 2 of which had gardens and 3 of which did not. Substantial differences between the residential sites with and without gardens were noted, with the rate of violence decreasing by 19% in the garden sites and increasing sevenfold in the nongarden sites over 1 year of observation. Similarly, the total rate of incidents decreased by 3.5% in the garden sites and increased threefold in the nongarden sites over the same period. However, it is difficult to know if these differences relate directly to the impact of the garden or if they are due to other aspects of the individual residential sites. In 2 studies,19 and 31 emotional outcomes, including pleasure, anxiety, interest, anger, sadness, and contentment, were measured by trained researchers using the Affect Rating Scale.