Forest therapy can prevent and treat depression: Evidence from meta-analyses

1 Forest therapy can prevent and treat depression: evidence from meta-analyses 1 2 Forest therapy involves engaging in a combination of forest-based activities to improve 3 one’s health or wellbeing. We conducted an overview of systematic reviews (SRs) and 4 meta-analyses of primary studies to provide the most comprehensive summary of the 5 effect of forest therapy on depression. We included 13 primary studies that matched our 6 eligibility criteria - all were included in four recent SRs and were conducted in the 7 Republic of Korea. We carried out meta-analyses with data extracted directly from these 8 13 studies and assessed their risk of bias. Outcomes of interest were depressive 9 symptoms, temporary recovery from depression (i.e. remission), response to treatment 10 (i.e. ≥ 50% reduction on depressive symptoms from baseline), adherence to treatment, 11 and adverse effects. Considering pooled estimates from randomized controlled trials 12 with adults, we found that compared to no intervention/usual care, forest therapy 13 produced a greater reduction of depressive symptoms (Hedges’g = 1.18, 95% CI [0.86, 14 1.50], p < .00001). Also compared to no intervention/usual care, participants in the 15 forest therapy group were 17 times as likely to achieve remission (Risk Ratio = 17.02, 16 95% CI [3.40, 85.21], p = .0006) and three times as likely to have a ≥ 50% reduction on 17 depressive symptoms (Risk Ratio = 3.18, 95% CI [1.94, 5.21], p < .00001). Forest 18 therapy, on average, reduced depressive symptoms more than engaging in similar 19 activities in a hospital or non-forested urban area, or participating in an intervention 20 focused on diet plus forest-based exercise. We did not find evidence that adherence to 21 forest therapy is different from the adherence to alternative interventions and the 22 adverse effects of forest therapy appear to be rare. These results indicate that, relative to 23 many more conventional alternatives, forest therapy is a more effective short-term 24 intervention for the prevention and treatment of depression in adults. 25 2 Keywords: contact with nature; dysthymia; forest bathing; mental health; mood 1 disorder; shinrin-yoku 2 3 1. Introduction 4 Depression is considered one of the most important global health challenges (Cipriani et 5 al., 2018). More than 300 million people worldwide suffer from this disorder, which can 6 harm many aspects of life (e.g. affective relationships and work) and, in the worst-case 7 scenario, lead to suicide (World Health Organization, 2017). Common symptoms of 8 depression are sad mood, anxiety, insomnia, loss of vitality, and lack of interest in life 9 (Fried, 2017). Depression is best conceptualized in a continuum ranging from the 10 presence of a few/mild/rare symptoms to the presence of several/severe/frequent 11 symptoms that lead to a debilitating life condition (Fried, 2017; van de Leemput et al., 12 2014). The diagnosis of a person as depressed is based on specific criteria and should 13 ideally be done through a structured or semi-structured interview (Nordgaard et al., 14 2013). Considering that depression is understood as a combination of symptoms (Fried, 15 2017; van de Leemput et al., 2014), interventions designed to prevent or treat 16 depression should not focus solely on one symptom. By preventing depression, we refer 17 to a process in which a non-depressed person achieves a reduction in depressive 18 symptoms. Such a reduction helps this person to remain non-depressed. By treating 19 depression, we refer to a process in which a depressed person achieves a reduction in 20 depressive symptoms. 21 Common treatments for depression are psychotherapy and antidepressants 22 (Cuijpers, 2018). These have advantages, such as accessibility to treatment, and have 23 been proven to be efficient in ameliorating depressive symptoms (Cuijpers et al., 2020), 24 but they also present some disadvantages. For example, the use of antidepressants can 25 3 have secondary effects like gastrointestinal symptoms (e.g. nausea and diarrhea), weight 1 gain, and metabolic abnormalities (Carvalho et al., 2016), and both antidepressants and 2 psychotherapy lack effectiveness in the short-term (Cuijpers, 2018). Considering these 3 disadvantages of psychotherapy and antidepressants, researchers have called for 4 complementary or even alternative treatments for depression (Lopresti, 2019; 5 Munkholm et al., 2019). One of these alternatives may be direct contact with nature 6 (Lee et al., 2017; Rajoo et al., 2020; van Tulleken et al., 2018). Specifically, previous 7 studies have found encouraging results regarding the potential of forest therapy to 8 prevent and treat depression (e.g. Chun et al., 2017; Kim et al., 2009). 9 The human health benefits of exposure to trees and forests abound and include 10 restorative capacities such as stress reduction as well as improvement in clinical mental 11 health outcomes (Wolf et al., 2020). To maximize these capacities, forest therapy is 12 often recommended as a form of preventive medicine (Park et al., 2010). Forest therapy, 13 also known as “shinrin-yoku”(Oh et al., 2017; Park et al., 2010; Rajoo et al., 2020), 14 involves engaging in a combination of activities in a forest environment to improve 15 one’s health or wellbeing (Han et al., 2016; Lee et al., 2017; Yu and Hsieh, 2020). 16 Forest therapy may include forms of forest-based exercise but should involve more than 17 just physical activity - typically incorporating other activities that foster positive mental 18 health such as meditation, games using forest elements, and/or group activities (Bang et 19 al., 2018; Chun et al., 2017; Djernis et al., 2019; Han et al., 2016; Lee et al., 2017; 20 Rajoo et al., 2019). In urban environments that increase stress and hinder psychological 21 restoration, direct immersion in forests can help people to calm down and reflect 22 (Collado et al., 2017; Kaplan, 1995; Kim et al., 2009; Kotera et al., 2020; Mayer et al., 23 2009; Rajoo et al., 2019). There is evidence that even a short period spent in a forest can 24 help people reduce stress, recover their attentional capabilities, and shift towards more 25 4 positive emotions (Djernis et al., 2019; Kaplan, 1995; Kotera et al., 2020; Lee et al., 1 2017; Rajoo et al., 2020; Wen et al., 2019; Yu and Hsieh, 2020). These benefits of 2 exposure to forests provide insights regarding the value of forest environments and the 3 relevance of conserving and utilizing these settings to improve human health (Bratman 4 et al., 2019). 5 Yet, while an abundance of research suggests forest-based activities produce 6 positive health outcomes (Hansen et al., 2017; Park et al., 2010), less research has 7 explored direct links between forest therapy and depression (Wen et al., 2019). For 8 example, in Wen et al.’s (2019) study of the effects of forest activities on health, only 9 three out of the 28 studies included in their analyses reported the effect of forest therapy 10 on depression. Furthermore, because previous evidence synthesis combined results from 11 depression measures with other constructs, it is difficult to discern if forest therapy is an 12 effective way of preventing and treating depression, specifically (Djernis et al., 2019; 13 Kotera et al., 2020). It is also unknown if forest therapy is safe (i.e. are adverse effects 14 rare?) and acceptable (i.e. do people adhere well to forest therapy?). Such knowledge 15 gaps limit the development of guidelines for practitioners who might be willing to 16 employ forest therapy to prevent and treat depression. 17 We therefore conducted an overview of systematic reviews (SRs) and meta18 analyses to answer the following research question: Is there sufficient evidence 19 supporting forest therapy as an effective intervention to prevent and treat depression? 20 Our literature review offers four novel contributions. First, to our knowledge, this is the 21 first review to focus exclusively on the effects of forest therapy on depression. Second, 22 we included more primary studies that reported depression outcomes than previous SRs, 23 and used these studies to provide estimates of the effect of forest therapy on depression. 24 Third, in addition to previous reviews that primarily focused on the alleviation of 25 5 depressive symptoms based on statistical significance or standardized effect estimates, 1 we considered more interpretable outcomes (Riedel et al., 2010) such as temporary 2 recovery from depression (i.e. remission), response to treatment (≥ 50% reduction in 3 depressive symptoms from baseline), and treatment acceptability or adherence (i.e. 4 drop-outs for any reason). Fourth, different from previous SRs, we assessed the risk of 5 bias of primary studies using tools that favor the identification of all potential sources of 6 bias (Sterne et al., 2019, 2016). 7 8 2. Methods 9 Our overview of SRs and meta-analyses of the effects of forest therapy on depression 10 was based on guidance from the latest edition of the Cochrane Handbook for Systematic 11 Reviews of Interventions (Higgins et al., 2019). We began the study by selecting SRs 12 most relevant to our research question centered on forest therapy and depression. We 13 located SRs by testing the utility of several databases (e.g., MEDLINE and PsycINFO) 14 and search terms such as forest therapy, forest bathing, and shinrin-yoku (see 15 Supplementary File p.1). Through this process, we found three SRs that met our criteria, 16 and a fourth was later identified via social media for researchers. However, if we were 17 to interpret only the results of these four SRs (rather than of the primary studies), we 18 would emerge with a limited answer to our research question. Thus, we decided to 19 analyze all eligible primary studies included by these four SRs. We did this by 20 developing our eligibility criteria (Table 1) and synthesis plan (see Supplementary File 21 p. 1 to 4) based on recent guidance from meta-analysis experts (e.g. Bender et al., 2018; 22 Higgins et al., 2019). 23 Regarding the eligibility criteria, we made important distinctions between forest 24 therapy, forest exposure, forest exercise, and forest walking. Forest exposure refers to 25 6 be in a forest. Forest exercise involves doing physical exercise in a forest, which can 1 include walking. We operationalized forest therapy to include engagement in a 2 combination of forest-based activities to improve one’s health or wellbeing. Thus, 3 forest-based exercise combined with other forest-based activities (e.g. meditation, 4 psychotherapy, group activities) met our definition of forest therapy. Nonetheless, just 5 being in a forest or exercising in a forest was not enough to satisfy this definition and 6 qualify as a forest therapy intervention for this study. 7 < Table 1 about here > 8 The first author of the present study collected relevant data from the four SRs, 9 assessed the eligibility of primary studies (Table 1), gathered relevant data on eligible 10 primary studies, and assessed the risk of bias of SRs and primary studies. All primary 11 studies included in our meta-analyses were identified on the four SRs, so the first author 12 screened primary studies for eligibility based on the information provided by the four 13 SRs. Decisions regarding the eligibility of primary studies at full-text and their risk of 14 bias assessment were checked by at least one co-author. Most information describing 15 primary studies (e.g. sample size) was collected from the four SRs and then checked 16 within the primary studies (Saldanha et al., 2019). The information used in our meta17 analyses was extracted directly from primary studies. The first author checked the 18 information from primary studies at least once after finishing the data extraction phase. 19 The risk of bias of the four SRs was assessed using the ROBIS tool (Whiting et 20 al., 2016). The risk of bias of the primary studies was assessed using the RoB 2 for 21 randomized controlled trials (RCTs) and cross-over trials (Sterne et al., 2019), and 22 ROBINS-I for non-randomized controlled trials (NRCT) (Sterne et al., 2016). These are 23 the most comprehensive tools available to assess potential bias in SRs, RCTs, cross24 overs, and NRCTs. Studies that assigned participants to groups based on a random or 25 7 quasi-random process were classified as RCTs (Sterne et al., 2019). Studies that did not 1 describe the randomization process or assigned participants to interventions based on a 2 non-random criterion (e.g. participants’ preference) were classified as NRCTs (Sterne et 3 al., 2016). 4 5 2.1. Synthesizing Data from Primary Studies 6 Following recommendations from the Cochrane Handbook, we focused on 7 results from RCTs and analyzed them separately from cross-overs trials, and NRCTs 8 (Higgins et al., 2019). Our main outcome was the standardized mean difference (SMD) 9 between the post-intervention depressive symptoms of two intervention groups. 10 Depressive symptoms scores are usually calculated by summing the score of items on a 11 depression rating scale for an individual. These items often cover a specific symptom 12 frequency and, sometimes, symptom intensity. The mean we used in our analysis was 13 the average score for the sample group in an intervention (e.g. the post-intervention 14 average score of the forest therapy group). When primary studies met all criteria needed 15 to be included in a meta-analysis (see Supplementary File p.1 to 2), we pooled their 16 SMDs because pooled SMDs are more precise than estimates of effect from single 17 studies (Higgins et al., 2019). Following Sawilowky (2009), we interpreted SMDs as: 18 very small = 0.01, small = 0.2, medium = 0.5, large = 0.8, very large = 1.2, and huge = 19 2.0. 20 To improve the interpretability of the effect of forest therapy on depression 21 compared to other interventions, we also considered reductions in depressive symptoms 22 based on dichotomous outcomes such as remission from depression and response to 23 treatment (Riedel et al., 2010). Remission refers to a temporary recovery from 24 depression and is often assessed as “the number of patients with a score for depressive 25 8 symptoms below a specific cut-off on a validated rating scale” (Cuijpers et al., 2020, p. 1 93). Response to treatment is usually registered as the number of people who exhibit ≥ 2 50% reduction of depressive symptoms from baseline following treatment. This 3 threshold is appropriate for the most commonly used scales to register depression: the 4 Hamilton Depression Rating Scale (HDRS or HAMD), Montgomery-Asberg 5 Depression Rating Scale (MADRS), and the Beck Depression Inventory (BDI), but 6 might not be appropriate for other scales (Riedel et al., 2010). None of the primary 7 studies we analyzed reported the number of people who responded to treatment (i.e. 8 responders). Thus, the number of responders in studies using one of these three scales 9 was estimated using the formula described by Furukawa et al. (2005). We do not report 10 the number of responders for primary studies that did not use one of these three scales. 11 We used drop-out for any reason as a proxy for treatment acceptability or adherence 12 (Cipriani et al., 2018; Cuijpers et al., 2020). For dichotomous outcomes, we calculated 13 risk ratios as they are easier to interpret than odds ratios (Higgins et al., 2019). We 14 reported the percentage of reduction in depressive symptoms from baseline in forest 15 therapy and comparison groups as a descriptive statistic (Vickers, 2001). We describe in 16 Equation 1 how this percentage was calculated. We also collected and reported 17 information about any adverse effects of forest therapy treatment described by primary 18 studies’ authors. 19 20 ((Post-intervention mean score - Baseline mean score)/Baseline mean score)*100 21 22 Equation 1 23 24 Statistical analyses were performed using RevMan 5.3 (“Review Manager 25 (RevMan) [Computer program],” 2014), and figures illustrating the risk of bias of 26 9 primary studies were created using robvis (McGuinness, 2019). Data are publicly 1 available at: (inserting link when published). 2 3 3. Results 4 3.1 Results of systematic reviews 5 Selected characteristics (e.g. research question, eligibility criteria, search strategy, risk 6 of bias assessment, and main results) of the four SRs that we reviewed are described in 7 Supplementary File p. 6. By analyzing the eligibility criteria of these SRs, we noted that 8 they were not able to include: (1) unpublished studies; (2) studies in languages other 9 than English and Korean; (3) studies published after October 2019; and, (4) within 10 Korean studies: studies with children or adolescents (< 18 years old), without a 11 comparison group, or published after 2016. 12 The four SRs were deemed as at high risk of bias because of limitations that 13 could hinder the adequate identification, selection, data extraction, appraisal, or 14 synthesis of relevant primary studies (Table 2). For example, the synthesis methods 15 used in the four SRs conducted before our review were limited. Moreover, no SR 16 focused exclusively on the effect of forest therapy on depression. It is possible, 17 however, to extract some information about the effect of forest therapy on depression 18 because three of the four SRs reported results for each primary study. Djernis et al.’s 19 (2019) SR was the only one that did not report results for each primary study. Instead, 20 the authors provided, for example, a pooled estimate of the effect of forest activities on 21 a combination of psychological constructs. Kotera et al.’s (2020) SR provided estimates 22 of the effect of forest therapy on depression, but only for three of the 13 studies 23 analyzed in our study. The meta-analysis performed by Kotera et al. (2020) combined 24 results from measures of depression with results collected using a measure of mood 25 16 Only one of the meta-analyses we conducted (Supplementary Fig. 1 in 1 Supplementary File p. 7) produced substantial statistical heterogeneity, but we did not 2 try to explain this heterogeneity (e.g. using meta-regression) due to the small number of 3 studies included in this meta-analysis (Higgins et al., 2019). 4 5 4. Discussion 6 . In the present study, we analyzed four recent SRs that explored connections 7 between forest therapy and depression, focusing on forest therapy’s capacity to reduce 8 depressive symptoms in comparison to no intervention/usual care and four alternative 9 interventions. Our findings show the effect of forest therapy on depression is greater 10 than the effect of any alternative intervention. People in forest therapy groups had a 11 higher reduction in depressive symptoms than people in the other groups. 12 13 When compared to similar interventions in non-forest settings, the benefits of 14 forest therapy were clear. Being involved in therapeutic activities in a forest appears to 15 be more effective than participating in such activities in a hospital or in an urban (non16 forested) area. Assuming the only difference between the forest therapy group and the 17 other groups was the intervention setting, these findings suggest that exposure to a 18 forest environment may provide additional benefits beyond the therapeutic activities 19 themselves. This is in line with research showing the benefits of exposure to nature, and 20 forests specifically (Wen et al., 2019; Wolf et al., 2020). For example, Bowler et al. 21 (2010) conducted meta-analyses of studies comparing the effect of the same activity 22 conducted in a natural versus a synthetic environment. They found that individuals who 23 conducted the activities in natural environments expressed less anger, fatigue, and 24 sadness than individuals in synthetic environments. Our results are also in line with 25 17 research conducted under stress reduction theory (Ulrich et al., 1991) and attention 1 restoration theory (Kaplan, 1995), which have consistently shown that exposure to 2 natural environments favors stress reduction, mood improvement, and the recovery of 3 attentional capabilities more than non-natural environments. 4 Our findings also revealed that forest therapy was more effective than depression 5 treatment regimens focused on diet plus forest-based exercise. Whereas diet, exercise, 6 and forest exposure may reduce depressive symptoms, the greater effect of forest 7 therapy on depression may be explained by the activities in which the forest therapy 8 groups participated (Djernis et al., 2019; Kim et al., 2009). The distinction between 9 forest therapy (a combination of activities positive for mental health) and forest exercise 10 is important. Whereas green exercise can yield a variety of positive health outcomes 11 (Bowler et al., 2010; Gladwell et al., 2013), forest therapy, which includes other 12 activities positive for mental health, may be more effective to prevent and treat 13 depression. Similar conclusions can be drawn for the comparison between forest 14 therapy and walking in a forest, as the forest therapy group generally had a higher 15 reduction in depressive symptoms and greater response to treatment than the forest 16 walking group (although the confidence interval for the estimate of the difference 17 between these interventions overlaps zero). 18 Analyzing the number of drop-outs in the intervention groups across the studies 19 synthesized, we found no evidence that forest therapy was a less acceptable treatment 20 than other alternatives (Supplementary Fig. 4 in the Supplementary File p. 8). We also 21 found that the adverse effects of forest therapy may be rare. 22 23 4.1. Limitations 24 18 Several limitations should be considered when interpreting the results of our 1 study. First, we selected only four relevant SRs. Nonetheless, as the last SR we included 2 (Kotera et al., 2020) did not provide any new primary study, it seems that our approach 3 was sufficiently comprehensive. Also supporting this view, no new eligible primary 4 study was found in a recently published SR about the effect of forest activities on 5 physiological and psychological outcomes (Rajoo et al., 2020). 6 Second, the screening process of primary studies and subsequent data extraction 7 was not checked by another reviewer. We judged that it was not necessary to have 8 another reviewer involved in the screening because this process was relatively simple. 9 Regarding data extraction, the first author checked the information describing primary 10 studies and information used in meta-analyses at least once after finishing the data 11 extraction phase. 12 Third, although the findings from our meta-analyses are encouraging and 13 underscore the potential of forest therapy to prevent and treat depression, we should 14 interpret these findings with some caution due to the limitations of existing primary 15 studies. Overall, the primary studies included a considerable diversity of participants 16 (e.g. health and unhealthy people), interventions, comparison groups, outcome 17 measures, and settings. Nonetheless, young adults were underrepresented, as the mean 18 age/age range of adults was above 39 years and only one study included children (and it 19 yielded inconclusive results). Thus, it remains unclear if the effect of forest therapy on 20 depression is higher or lower on children, adolescents, and young adults. 21 Fourth, caution should also be executed when interpreting our results because all 22 the studies included in our review were conducted in the Republic of Korea, as these 23 were the only studies we located from anywhere in the world that evaluated forest 24 therapy interventions and reported results for depression outcome measures. 25 19 Considering that forest activities are popular elsewhere in Asia (Yu and Hsieh, 2020), 1 future research could systematically search for primary studies conducted in countries 2 such as Japan and Taiwan. This also highlights the need for wider geographic coverage 3 in research about the forest therapy effect on depression. 4 Fifth, our inability to incorporate unpublished research also casts some doubt 5 about the true effectiveness of forest therapy, as there is evidence that studies are more 6 likely to be published if they reveal statistically significant treatment effects (Higgins et 7 al., 2019). Although the effect of forest therapy might have been overestimated (or 8 underestimated), the publication of forest therapy findings is unlikely to be heavily 9 influenced by financial interests, which favor the publication of positive results, as is the 10 case for antidepressant studies (Munkholm et al., 2019). We examined the reported 11 funding and conflict of interest of primary studies and found no evidence of any conflict 12 of interest. 13 Finally, methodological limitations of the primary studies in our analyses 14 increase uncertainty about the true effect of forest therapy on depression. Only four 15 RCTs and two cross-over trials were included in our meta-analyses, and all had a high 16 risk of bias. The main limitations of these studies were the lack of blinding, which is 17 infeasible in forest therapy interventions because people experience the treatment 18 environment, and the lack of a protocol describing the analysis plan. The former 19 increases the risk of deviations from the intended intervention and can introduce bias in 20 the assessment of treatment-related outcomes. The latter hinders the possibility of 21 assessing selective reporting. Limitations were also identified in the randomization 22 process and in the way some studies dealt with missing outcome data. Whereas most of 23 these limitations are typically associated with an exaggeration of experimental 24 intervention effects (in our case, forest therapy), in some circumstances the effect can be 25 20 underestimated (Sterne et al., 2019, 2016). Moreover, a NRCT was the only source of 1 evidence for the impact of forest therapy on children’s depressive symptoms, and other 2 NRCTs provided additional information for two comparisons (i.e. forest therapy versus 3 no intervention/usual care, and forest therapy versus similar activities in a hospital). 4 Similar to RCTs and crossover trials, the NRCTs were not blinded and did not publish 5 an analysis plan. Besides these limitations, NRCTs tend to have a higher risk of bias 6 than RCTs due to confounding, as the assignment of participants to the intervention can 7 be related to baseline variables that influence the outcome (i.e. prognostic factors). 8 9 5. Future research opportunities and conclusions 10 In summary, future literature reviews may extend the evidence we have 11 synthesized by systematically searching for unpublished studies, studies in other 12 languages than English and Korean, and studies with children and adolescents. Future 13 primary studies should document if (and what) adverse effects occurred in the forest 14 therapy and comparison group(s) and report remission from depression and response to 15 treatment. Future research could also examine whether certain aspects of forest therapy 16 (e.g., meditation vs. green exercise) are more beneficial than others when it comes to 17 treating depression. It includes assessing if or how different frequency and duration of 18 forest therapy interventions may influence the reduction in depressive symptoms. The 19 adoption of relevant Consolidated Standards of Reporting Trials (CONSORT) should 20 facilitate the assessment of primary studies' risk of bias and the interpretation of their 21 results (Moher et al., 2010). Our findings also highlight the need for more 22 methodologically rigorous RCTs examining the effects of forest therapy on depression. 23 Despite some limitations, our review of SRs and primary studies examining the 24 effectiveness of forest therapy as a preventive measure and treatment for depression 25 21 yielded the most conclusive evidence to date. Compared to antidepressants, similar 1 activities in a hospital or non-forested urban settings, or even diet and forest-based 2 exercise, forest therapy appears to be more likely to produce outcomes like remission 3 and response to treatment, with adequate acceptability or adherence. Thus, while more 4 studies are needed, we believe practitioners should consider the use of forest therapy as 5 both a preventive measure and a treatment for depression in adults. This is in line with 6 the growing support for the incorporation of therapeutical activities in contact with 7 nature as a mainstream intervention for the prevention and treatment of mental health 8 problems (Buckley et al., 2018). 9 10 References 11 Bang, K.-S., Kim, S., Song, M., Kang, K., Jeong, Y., 2018. The effects of a health 12 promotion program using urban forests and nursing student mentors on the 13 perceived and psychological health of elementary school children in vulnerable 14 populations. Int. J. Environ. Res. Public Health 15. 15 https://doi.org/10.3390/ijerph15091977 16 Bender, R., Friede, T., Koch, A., Kuss, O., Schlattmann, P., Schwarzer, G., Skipka, G., 17 2018. Methods for evidence synthesis in the case of very few studies. Res. Synth. 18 Methods. https://doi.org/10.1002/jrsm.1297 19 Bowler, D.E., Buyung-Ali, L.M., Knight, T.M., Pullin, A.S., 2010. A systematic review 20 of evidence for the added benefits to health of exposure to natural environments. 21 BMC Public Health 10, 456. https://doi.org/10.1186/1471-2458-10-456 22 Bratman, G.N., Anderson, C.B., Berman, M.G., Cochran, B., de Vries, S., Flanders, J., 23 Folke, C., Frumkin, H., Gross, J.J., Hartig, T., Kahn, P.H., Kuo, M., Lawler, J.J., 24 Levin, P.S., Lindahl, T., Meyer-Lindenberg, A., Mitchell, R., Ouyang, Z., Roe, J., 25 22 Scarlett, L., Smith, J.R., van den Bosch, M., Wheeler, B.W., White, M.P., Zheng, 1 H., Daily, G.C., 2019. Nature and mental health: An ecosystem service 2 perspective. Sci. Adv. 5, eaax0903. https://doi.org/10.1126/sciadv.aax0903 3 Buckley, R.C., Brough, P., Westaway, D., 2018. Bringing outdoor therapies into 4 mainstream mental health. Front. Public Heal. 6. 5 https://doi.org/10.3389/fpubh.2018.00119 6 Carvalho, A.F., Sharma, M.S., Brunoni, A.R., Vieta, E., Fava, G.A., 2016. The safety, 7 tolerability and risks associated with the use of newer generation antidepressant 8 drugs: A critical review of the literature. Psychother. Psychosom. 85, 270–288. 9 https://doi.org/10.1159/000447034 10 Choi, Y.H., Ha, Y.S., 2014. The effectiveness of a forest-experience-integration 11 intervention for community dwelling cancer patients’ depression and resilience. J. 12 Korean Acad. Community Heal. Nurs. 25, 109. 13 https://doi.org/10.12799/jkachn.2014.25.2.109 14 Chun, M.H., Chang, M.C., Lee, S.-J., 2017. The effects of forest therapy on depression 15 and anxiety in patients with chronic stroke. Int. J. Neurosci. 127, 199–203. 16 https://doi.org/10.3109/00207454.2016.1170015 17 Cipriani, A., Furukawa, T.A., Salanti, G., Chaimani, A., Atkinson, L.Z., Ogawa, Y., 18 Leucht, S., Ruhe, H.G., Turner, E.H., Higgins, J.P.T., Egger, M., Takeshima, N., 19 Hayasaka, Y., Imai, H., Shinohara, K., Tajika, A., Ioannidis, J.P.A., Geddes, J.R., 20 2018. Comparative efficacy and acceptability of 21 antidepressant drugs for the 21 acute treatment of adults with major depressive disorder: A systematic review and 22 network meta-analysis. Lancet 391, 1357–1366. https://doi.org/10.1016/S014023 6736(17)32802-7 24 Collado, S., Staats, H., Corraliza, J.A., Hartig, T., 2017. Restorative environments and 25 23 health, in: Navarro, O., Fleury-Bahi, G., Pol, E. (Eds.), Handbook of 1 Environmental Psychology and Quality of Life Research. Springer, New York, pp. 2 127–148. https://doi.org/10.1007/978-3-319-31416-7_7 3 Cuijpers, P., 2018. The challenges of improving treatments for depression. JAMA 320, 4 2529–2530. https://doi.org/10.1001/jama.2018.17824 5 Cuijpers, P., Noma, H., Karyotaki, E., Vinkers, C.H., Cipriani, A., Furukawa, T.A., 6 2020. A network meta‐analysis of the effects of psychotherapies, 7 pharmacotherapies and their combination in the treatment of adult depression. 8 World Psychiatry 19, 92–107. https://doi.org/10.1002/wps.20701 9 Djernis, Lerstrup, Poulsen, Stigsdotter, Dahlgaard, O’Toole, 2019. A Systematic review 10 and meta-analysis of nature-based mindfulness: effects of moving mindfulness 11 training into an outdoor natural setting. Int. J. Environ. Res. Public Health 16, 12 3202. https://doi.org/10.3390/ijerph16173202 13 Fried, E.I., 2017. The 52 symptoms of major depression: Lack of content overlap 14 among seven common depression scales. J. Affect. Disord. 208, 191–197. 15 https://doi.org/10.1016/j.jad.2016.10.019 16 Furukawa, T.A., Cipriani, A., Barbui, C., Brambilla, P., Watanabe, N., 2005. Imputing 17 response rates from means and standard deviations in meta-analyses. Int. Clin. 18 Psychopharmacol. 20, 49–52. https://doi.org/10.1097/00004850-200501000-00010 19 Gladwell, V.F., Brown, D.K., Wood, C., Sandercock, G.R., Barton, J.L., 2013. The 20 great outdoors: How a green exercise environment can benefit all. Extrem. Physiol. 21 Med. 2, 1–7. https://doi.org/10.1186/2046-7648-2-3 22 Han, J.-W., Choi, H., Jeon, Y.-H., Yoon, C.-H., Woo, J.-M., Kim, W., 2016. The 23 Effects of Forest Therapy on Coping with Chronic Widespread Pain: Physiological 24 and Psychological Differences between Participants in a Forest Therapy Program 25 24 and a Control Group. Int. J. Environ. Res. Public Health 13. 1 https://doi.org/10.3390/ijerph13030255 2 Hansen, M.M., Jones, R., Tocchini, K., 2017. Shinrin-yoku (forest bathing) and nature 3 therapy: A state-of-the-art review. Int. J. Environ. Res. Public Health 14. 4 https://doi.org/10.3390/ijerph14080851 5 Higgins, J.P.T., Thomas, J., Chandler, J., Cumpston, M., Li, T., Page, M.J., Welch, V.A. 6 (Eds.), 2019. Cochrane Handbook for Systematic Reviews of Interventions, 2nd 7 ed. John Wiley & Sons, Chichester. 8 Hong, S.-S., Kim, H.-C., Cho, S.-H., 2013. The Effects of Forests Healing for Cognitive 9 Function. J. Orient. Neuropsychiatry 24, 63–74. 10 https://doi.org/10.7231/jon.2013.24.1.063 11 Hong, S.-S., Lee, J.-E., Kim, H.-C., Cho, S.-H., 2012. The Effects of Forests Healing 12 for Hwa-Byung. J. Orient. Neuropsychiatry 23, 169–182. 13 https://doi.org/10.7231/jon.2012.23.4.169 14 Kaplan, S., 1995. The restorative benefits of nature: Toward an integrative framework. 15 J. Environ. Psychol. 15, 169–182. https://doi.org/10.1016/0272-4944(95)90001-2 16 Kim, W., Lim, S.-K., Chung, E.-J., Woo, J.-M., 2009. The effect of cognitive behavior 17 therapy-based psychotherapy applied in a forest environment on physiological 18 changes and remission of major depressive disorder. Psychiatry Investig. 6, 245. 19 https://doi.org/10.4306/pi.2009.6.4.245 20 Kotera, Y., Richardson, M., Sheffield, D., 2020. Effects of Shinrin-Yoku (Forest 21 Bathing) and Nature Therapy on Mental Health: A Systematic Review and Meta22 Analysis. 23 Lee, I., Choi, H., Bang, K.-S., Kim, S., Song, M., Lee, B., 2017. Effects of forest 24 therapy on depressive symptoms among adults: A systematic review. Int. J. 25 25 Environ. Res. Public Health 14. https://doi.org/10.3390/ijerph14030321 1 Lim, Y., Kim, D., Yeoun, P., 2014. Changes in depression degree and self-esteem of 2 senior sitizens in a nursing home according to forest therapy program. J. Korean 3 Inst. For. Recreat. 18, 1–11. https://doi.org/10.34272/forest.2014.18.1.001 4 Lopresti, A.L., 2019. It is time to investigate integrative approaches to enhance 5 treatment outcomes for depression? Med. Hypotheses 126, 82–94. 6 https://doi.org/10.1016/j.mehy.2019.03.008 7 Mayer, F.S., Frantz, C.M.P., Bruehlman-Senecal, E., Dolliver, K., 2009. Why is nature 8 beneficial? The role of connectedness to nature. Environ. Behav. 41, 607–643. 9 https://doi.org/10.1177/0013916508319745 10 McGuinness, L., 2019. robvis: An R package and web application for visualising risk11 of-bias assessments. 12 Moher, D., Hopewell, S., Schulz, K.F., Montori, V., Gøtzsche, P.C., Devereaux, P.J., 13 Elbourne, D., Egger, M., Altman, D.G., 2010. CONSORT 2010 explanation and 14 elaboration: Updated guidelines for reporting parallel group randomised trials. Int. 15 J. Surg. 10, 28–55. https://doi.org/https://doi.org/10.1136/bmj.c869 16 Munkholm, K., Paludan-Müller, A.S., Boesen, K., 2019. Considering the 17 methodological limitations in the evidence base of antidepressants for depression: 18 A reanalysis of a network meta-analysis. BMJ Open 9, e024886. 19 https://doi.org/10.1136/bmjopen-2018-024886 20 Nordgaard, J., Sass, L.A., Parnas, J., 2013. The psychiatric interview: Validity, 21 structure, and subjectivity. Eur. Arch. Psychiatry Clin. Neurosci. 263, 353–364. 22 https://doi.org/10.1007/s00406-012-0366-z 23 Oh, B., Lee, K.J., Zaslawski, C., Yeung, A., Rosenthal, D., Larkey, L., Back, M., 2017. 24 Health and well-being benefits of spending time in forests: systematic review. 25 Woo 2012 Patients with major depression taking antidepressants 45.68 ---- Forest activities included cognitive-behavioral therapy, meditation, and relaxation training. Four weeks Once a week (for 3 hours) 28 1: (N = 21) Similar activities in a hospital; 2: (N = 15) Treatment as usual; 3: (N = 17) Walking in a forest. HDRS-17, MADRS, and BDI Randomized controlled trial Seoul Arboretum T1: Just before treatment; T2: just after treatment Bang 2018 Elementaryschool students in grades 4 to 6 at five community centers 11.79 55.56 Forest activities included five senses experience, walking, and games. Ten weeks Once a week (for 60 min) 24 (N = 28) Normal daily routines Children’s Depression Inventory Nonrandomized controlled trial Urban forest T1: Just before treatment; T2: just after treatment Kim 2009 Patients with major depression taking antidepressants 46.2 85.7 Forest activities included cognitive-behavioral therapy, positive psychology tools, and mindfulness meditation on breath, wind, forest, and sounds. Four weeks Once a week (three hours/session) 23 1: (N = 19) Similar activities in a hospital; 2: (N = 21) Treatment as usual. BDI, HDRS, MADRS Randomized controlled trial Hong-Reung; 44-ha arboretum T1: Just before treatment; T2: T1 + 1 week; T3: T1 + 2 weeks; T4: T1 + 3 weeks Note: *When results for more than one outcome measure were available, we gave preference to results from one outcome measure based on specific criteria (see Supplementary File p. 2 for more information). The chosen measures are in bold. Underlined studies were written in Korean. BDI = Beck Depression Inventory; HDRS = Hamilton Depression Rating Scale; MADRS = Montgomery-Asberg Depression Rating Scale. 33 Table 4. Percentage of reduction in depressive symptoms from baseline for forest therapy and other interventions, listed by primary study First author and year Groups Forest therapy Similar activities in a hospital Similar activities in an urban area No intervention/usual care Diet plus exercise in the forest Walking in a forest Randomized controlled trials Chun 2017 -77.46 NA -1.39 NA NA NA Kim 2009 -50.08 -19.91 NA -7.57 NA NA Shin 2012 -64.04 NA NA 0.20 NA NA Woo 2012 -50.27 -34.81 NA -10.10 NA -32.49 Mean (SD) -60.46 (13.08) -27.36 (10.54) -1.39 -5.82 (5.37) NA -32.49 Median -57.16 -27.36 -1.39 -7.57 NA -32.49 Cross-over trials Hong 2013 -46.43 NA NA NA -5.10 NA Hong 2012 -56.04 NA NA NA -9.91 NA Mean (SD) -51.24 (6.80) NA NA NA -7.51 (3.40) NA Median -51.24 NA NA NA -7.51 NA Non-randomized controlled trials Han 2016 -46.08 NA NA -15.85 NA NA Lim 2014 -27.71 -16.76 NA 2.18 NA NA You 2014 -27.62 NA NA -0.58 NA NA Choi 2014 -9.61 NA NA 3.43 NA NA M.-H. Kim 2015 -13.03 NA NA 6.99 NA NA Y. G. Kim 2015 -9.79 NA NA 3.93 NA NA Mean (SD) -22.31 (14.33) -16.76 NA 0.02 (8.15) NA NA Median -20.33 -16.76 NA 2.81 NA NA Non-randomized controlled trial with children Bang 2018 -21.13 NA NA -8.73 NA NA Table 5. Aggregated sample size (N) and studies providing data for each comparison Comparison N Primary studies’ first author and year Forest therapy versus no intervention/usual care 525 Kim 2009; Woo 2012; Shin 2012; You 2014; Lim 2014; Choi 2014; M-H. Kim 2015; Y-G. Kim 2015; Bang 2018 Forest therapy versus similar activities in a hospital 134 Kim 2009; Woo 2012; Lim 2014 Forest therapy versus similar activities in an urban area 59 Chun 2017 Forest therapy versus diet plus forest-based exercise 31 Hong 2013; Hong 2012 Forest therapy versus walking in a forest 43 Woo 2012 Note: Some forest therapy groups were counted in more than one comparison. Randomized controlled trials are in bold and cross-over trials are underlined. The other studies are nonrandomized controlled trials. Fig. 1. Flow diagram illustrating the selection process of primary studies. Records identified through the four systematic reviews (N = 101) Records excluded because they are duplicates or did not measure depression (N = 82) Full-text articles assessed for eligibility (N = 19) Screening Included Eligibility Identification Full-text articles excluded with reasons (N = 6) # Studies that involved just exercising in forest (N = 3); # Studies in settings other than forest (N = 2); # We could not have access to one study. Studies synthesized (N = 13) Records screened (N = 101) Fig. 2. Results of randomized controlled trials: (a) Comparison of the post-intervention mean score of forest therapy groups versus no intervention/usual care using the inverse variance fixed-effect meta-analysis. (b) Comparison of the risk of temporary recovery from depression (i.e. remission) between forest therapy groups and usual care for depression groups, using the Mantel-Haenszel fixed-effect meta-analysis. (c) Comparison of the risk of response to treatment (i.e. ≥ 50% reduction on depressive symptoms) between forest therapy groups and no intervention/usual care groups, using the Mantel-Haenszel fixed-effect meta-analysis. Events refer to cases of remission (b) or response (c). Green squares refer to standardized mean differences and blues squares to risk ratios. Bigger squares indicated more participants in a study or more events and a bigger diamond indicates greater uncertainty in the estimate. Crossovers and non-randomized controlled trials were analyzed separately. Fig. 3. Results of randomized controlled trials: (a) Comparison of the post-intervention mean score of forest therapy groups versus similar activities in hospital groups using the inverse variance fixed-effect meta-analysis. (b) Comparison of the risk of temporary recovery from depression (i.e. remission) between forest therapy groups and similar activities in hospital groups, using the Mantel-Haenszel fixed-effect meta-analysis. (c) Comparison of the risk of response to treatment (i.e. ≥ 50% reduction on depressive symptoms) between forest therapy groups and similar activities in hospital groups, using the Mantel-Haenszel fixed-effect meta-analysis. Events refer to cases of remission (b) or response (c). Green squares refer to standardized mean differences and blues squares to risk ratios. Bigger squares indicated more participants in a study or more events and a bigger diamond indicates greater uncertainty in the estimate. Cross-over and non-randomized controlled trials were analyzed separately.