Medicinal-Plant Healing Gardens in Oncology Care: A Systematic Review of Clinical Evidence, Spatial Design Parameters and Pharmacovigilance Constraints

Healing gardens are a routine component of cancer-center architecture, and medicinal-plant collections are increasingly proposed as their organizing principle. We searched PubMed/MEDLINE and ClinicalTrials.gov, supplemented by citation searching and grey literature, and included 16 primary studies in cancer populations (four randomized trials, two controlled non-randomized studies, six single-arm studies, one post-occupancy evaluation, one comparative device study and two qualitative programs), together with eight contextual entries. Garden, forest-therapy and nature-exposure interventions were consistently associated with short-term improvements in mood, anxiety and perceived stress, with inconsistent biomarker findings; one controlled study found increased cortisol and unchanged natural killer cell activity. Two randomized trials in early breast cancer found improved directed-attention capacity with prescribed nature exposure of about 120 min/week. Horticultural therapy on palliative wards improved existential distress but not total quality of life, with high attrition. No study in any design has evaluated a hard oncological endpoint. Individual species carry genuine randomized supportive-care evidence—Zingiber officinale for chemotherapy-induced nausea, Panax quinquefolius for fatigue, Calendula officinalis for radiodermatitis—but that evidence concerns standardized preparations, not garden presence. Conversely, the safety literature directly addresses gardens: ingested Hypericum perforatum preparations reduce exposure to the active irinotecan metabolite SN-38 by 42% through CYP3A4 and P-glycoprotein induction; furanocoumarin species cause treatment-related photosensitivity; and environmental infection-control guidance restricts the use of plants and cut flowers for immunosuppressed patients. As expert-consensus recommendations informed by this evidence, we propose a planting plan, a list of species to avoid, a zoned garden layout, a risk-stratified access system and a core set of outcomes to measure. The certainty of evidence is low or very low for all domains except cognition, the review protocol was registered only retrospectively, the garden registry rests mainly on institutional records, and the design framework has not been validated in practice.