While intuitively and theoretically sound, the empirical support for acute stress-reducing effects of immersion in natural environments is tentative due to small sample sizes and methodological weaknesses in the studies

Effects of Public Green Space on Acute Psychophysiological Stress Response: A Systematic Review and Meta-Analysis of the Experimental and Quasi-Experimental Evidence. Lærke Mygind et al. Environment and Behavior, September 9, 2019. https://doi.org/10.1177/0013916519873376

Abstract: Contact with nature is widely considered to ameliorate psychological stress, but the empirical support for a causal link is limited. We conducted a systematic review to synthesize and critically assess the evidence. Six electronic databases were searched. Twenty-six studies evaluated the difference between the effect of natural environments and that of a suitable control on the acute psychophysiological stress response. Eighteen studies were rated as being of moderate quality, four studies of low quality, and four studies of high quality. Meta-analyses indicated that seated relaxation (g = .5, p = .06) and walking (g = .3, p = .02) in natural environments enhanced heart rate variability more than the same activities in control conditions. Cortisol concentration measures were inconsistent. While intuitively and theoretically sound, the empirical support for acute stress-reducing effects of immersion in natural environments is tentative due to small sample sizes and methodological weaknesses in the studies. We provide guidelines for future research.

Keywords: biomarker, green exercise, mental health, relaxation, restorative environments, social ecology/human ecology

---
 Strengths and Limitations of This Review 
A notable weakness in these meta-analyses is related to small sample sizes in the included studies and that at least two of the meta-analyses were at risk of being influenced by small-study effects, indicated by asymmetric funnel plots. Small-study effects, which can encompass publication bias where small studies are more frequently published when they report treatment effects, distort the estimated pooled effect sizes, rendering them a less accurate representation of the “true” effect sizes. Although the number of studies included in the funnel plots was small and indications of publication bias were tentative, we recommend that researchers as well as journals publish null findings and results that counter hypotheses. 

As described in the “Method” section, the review was nested within a larger review with a broader set of inclusion criteria. While the comprehensiveness of the literature search might have elicited a high retrieval rate by covering diverse fields of research and search terms, the approach would be immensely time-consuming to reproduce. Furthermore, the search strategy was based on the assumption that a broad, thematic search relating to health, well-being, and psychological stress would include studies utilizing psycho-physiological outcomes. As such, the dependent variables were not pre-defined or included in the search strategy. While this could be speculated to result in the omission of relevant papers, the retrieval rate within the specific field of this review was higher than previously seen (Bowler et al., 2010; Haluza et al., 2014; Twohig-Bennett & Jones, 2018). In comparison to the most recent review by Twohig-Bennett and Jones (2018), we retrieved nine studies not included in their review (Aspinall et al., 2015; Brown et al., 2014; Dettweiler et al., 2017; Gidlow, Randall, Gillman, Smith, & Jones, 2016; Hohashi & Kobayashi, 2013; W. Kim et al., 2009; Lee et al., 2009; Matsuura et al., 2011; Park et al., 2008), while we missed five studies included in theirs (Grazuleviciene et al., 2016; Jia et al., 2016; Song et al., 2013; Song, Ikei, Igarashi, et al., 2015; Tsunetsugu et al., 2013). To acknowledge the contributions made by these five additional studies, we performed post hoc quality assessments and included study characteristics and quality in Supplementary Material C. The studies generally reported positive findings but did not alter the overall conclusions of our review. The studies were rated as being of low-to-moderate quality and shared the limitations observed in the body of evidence discussed above. 

No one in the author group was proficient in Asian languages, and some studies that could potentially have been relevant were excluded from this review (e.g., Joung et al., 2015; Park et al., 2014; Song, Lee, Ikei, et al., 2015). In addition, studies exploring the effects of contact with nature through, for example, gardening, views through windows, or virtual nature were not included in this review.

Conclusion 
While intuitively and theoretically sound, the empirical support for a stress-reducing impact of natural environments is tentative. The majority of the studies reported positive effects, but small-study effects might bias the body of evidence. Where possible, random-effect meta-analyses were performed to calculate pooled effect sizes. Meta-analyses indicated that seated relaxation (g= .5, p= .06) and walking (g = .3, p= .02) in natural environments enhanced vagally mediated HRV more than the same activi-ties in control conditions. Cortisol concentration measures were inconsistent. Future research would benefit from including larger sample sizes, increased population diversity (in terms of sociodemographic factors, medical conditions and diagnoses, age, and sex), blinding of outcome assessors (for group or condition assignment) and participants (for research question and aims), and thorough descriptions of natural and control environments and conditions, as well as participant recruitment and inclusion criteria. Further attention to quantitative assessment and control for potential confounding factors, such as temperature and physical activity, as well as inconsistent baseline levels, is warranted. Last, we recommend that researchers preregister trials to enhance transparency and accountability in the research field.