Therapist-Guided Forest Therapy in Adolescents: Operational Feasibility and Anxiety-Symptom Signals from a Two-Classroom Quasi-Experimental Pilot Study

Adolescence is a sensitive period for anxiety symptoms, yet evidence on repeated forest therapy in secondary-school populations remains limited. This two-classroom quasi-experimental pilot examined operational delivery, acute affective responses, and follow-up anxiety-symptom signals after four therapist-guided forest therapy sessions in a coastal pine forest. Two intact, non-clinically selected classrooms were allocated without randomization. The principal all-classroom-assigned analysis included 26 intervention-classroom and 28 usual-activities control-classroom students aged 17–18 years. The sole principal outcome domain was the Spence Children’s Anxiety Scale (SCAS) total raw score, measured at baseline and four follow-ups through 10–11 weeks; acute mood was assessed with the Profile of Mood States—Adolescents (POMS-A) before and after each session. Descriptively larger SCAS reductions occurred in the intervention classroom at every follow-up. The largest baseline-adjusted difference occurred 8–9 weeks post-intervention (coefficient = −5.14 points, 95% confidence interval [−10.25, −0.04]; two-sided p = 0.048; false-discovery-rate-adjusted q = 0.135), and no principal SCAS comparison survived multiplicity correction. Acute total mood disturbance improved most clearly during the third scheduled session (n = 21; mean improvement = 9.71; q = 0.002; Cohen’s dz = 1.29), but POMS-A had no control-classroom comparator. Discordant exploratory findings included lower adjusted conduct/behavior (q = 0.028) and lower post-intervention connectedness to nature (q = 0.005) in the intervention classroom. Because condition was inseparable from classroom and no active comparator was included, favorable and unfavorable between-classroom differences cannot be attributed to forest therapy and do not establish efficacy. The study documents delivery in this setting and provides descriptive signals and design information for preregistered, adequately replicated cluster-randomized research.