This study was conducted in urban garden spaces in Songdo, Incheon, Republic of Korea. These gardens are located in commercial zones, apartments, shopping malls, and other urban structures. These gardens include individual garden plots, where participants can cultivate vegetables, such as lettuce, peppers, tomatoes, and eggplants individually, and communal garden plots, where groups of five or more individuals can gather to cultivate crops collectively.

This study targeted individuals with sleep disorders and poor sleep quality, who had no experience in urban gardening. Specifically, the study focused on healthy men and women aged 40 and older who reported impaired sleep quality (Pittsburgh Sleep Quality Index [PSQI] >5 and Insomnia Severity Index [ISI] ≥8 and <22) and were able to engage in gardening activities and attend a 12-week gardening program.

The urban gardening activity program consisted of 12 sessions aimed at reducing stress and improving sleep quality through regular gardening activities and program-based interactions among participants. These sessions were conducted once a week for 2 hours each from April to July 2023.

Individual garden plots primarily featured sleep-promoting vegetables, such as lettuce (Heukharang) [21,22], celery [23], and tomatoes, whereas communal garden plots included herbs, such as lavender, dill, and thyme, along with staple crops, such as potatoes and sweet potatoes. The 12-session program was structured into two sequential phases. The first six sessions primarily focused on acquiring basic horticultural skills and engaging in physical gardening activities, such as soil preparation, planting, and cultivation. The latter six sessions maintained comparable levels of physical activity while additionally incorporating activities utilizing harvested produce (e.g., cooking, tasting, and sharing) to promote dietary awareness and psychosocial well-being.

Because outcome measures were only collected at baseline and after the full 12-session program, differences between the two phases were not statistically analyzed, but were described to provide context for the intervention design.

All outcome measures were collected at three standardized assessment time points: baseline (within one week before the start of the intervention), post-intervention (within one week after the completion of the 12-week intervention), and 12-week follow-up (12 weeks after intervention completion).

Statistical significance was tested using the latest versions of R Statistics (version 4.0, R Foundation for Statistical Computing) and SPSS Statistics (version 28; IBM Corp.). Statistical significance testing was conducted, with p<0.05 considered significant. Given the small sample size and non-normal distribution of the data, non-parametric statistical methods were used.

For three-time-point comparisons (pretreatment, posttreatment, and 12-week follow-up), we used the Friedman test (nonparametric equivalent of repeated measures ANOVA) to assess the overall differences across time points. When the Friedman test indicated significance (p<0.05), post-hoc Wilcoxon signed-rank tests were performed for pairwise comparisons. To address multiple comparisons and control for Type I errors, the Bonferroni correction was applied to the post-hoc analyses. For primary outcomes (PSQI and ISI), we maintained α=0.05, while secondary outcomes were evaluated with Bonferroni-adjusted alpha levels.

Effect sizes for nonparametric tests were calculated using r=z/√n, where z is the standardized test statistic and n is the total sample size. Effect sizes were interpreted as small (r=0.1), medium (r=0.3), and large (r=0.5) according to Cohen’s conventions to facilitate clinical interpretation and statistical significance.

The z- and p-values from the statistical tests are reported in Table 1. Negative z-values indicate a decrease in the variables from pretreatment to posttreatment or follow-up. Significant p-values (p<0.05) are marked with an asterisk (*).

The analysis was focused on two groups: all participants (n=19) and a subgroup with objectively poor sleep quality, defined as having sleep efficiency below 85% as established by the American Academy of Sleep Medicine criteria for sleep efficiency impairment, as determined by polysomnography (n=12). This enabled the examination of the effects of the intervention on both the overall sample and specific subgroups with impaired sleep. Subgroup analysis was conducted in participants with objectively poor sleep efficiency (<85%) because this population represents individuals with clinically significant sleep impairment who may benefit the most from non-pharmacological interventions. This study aimed to explore the magnitudes and patterns of changes in this clinically relevant group. However, no direct statistical comparison was performed between participants with and without poor sleep efficiency.

This study was conducted on March 14, 2023, after receiving approval from the Institutional Review Committee of Catholic Kwandong University International St. Mary’s Hospital (IS23ONSI0006). Written informed consent was obtained from all participants after a detailed explanation of the study protocol, including sleep quality measurements, stress-level assessments, and urban gardening activities. This study was conducted in accordance with the ethical standards of our institution’s research committee and the 1964 Helsinki Declaration and its later amendments.

A total of 19 participants (18 women and one man) who reported subjective sleep disturbance completed the study with a mean age of 53.3±7.6 years (range: 32–67 years) (Table 1). The Wilcoxon signed-rank test revealed significant improvements in depression and perceived stress scores after gardening. The depression VAS scores decreased from 37.57±28.88 to 25.84±22.63 (p=0.05, r=0.45), while perceived stress (PSS total) decreased from 19.31±4.85 to 16.57±3.76 (p=0.03, r=0.49). Specifically, improvements were observed in detailed PSS items, including feeling unable to control important matters (p=0.01, r=0.57), feelings of restlessness or stress buildup (p=0.01, r=0.56), anger from uncontrollable circumstances (p=0.02, r=0.53), and feeling overwhelmed by accumulated difficulties (p=0.03, r=0.49). Differences were considered statistically significant. Importantly, these statistically significant improvements persisted in the re-evaluation conducted 12 weeks later for the item “feeling unable to control important matters” (p=0.05), while other PSS items did not maintain sta-tistical significance at follow-up. The PGA results also revealed significant improvements in sleep quality (p<0.001, r=0.75) and enjoyment of life activities (p=0.01, r=0.60) after participating in the gardening activities.

According to the Wilcoxon signed-rank test, significant improvements were observed in overall sleep quality, as measured by the PSQI and insomnia severity evaluation after the intervention (Table 1). Evaluation of the PSQI scores revealed that participants showed improvement from 9.05±2.19 points before the activity to 5.42±2.36 points after the activity (p< 0.001, r=0.81). The score was 5.89±3.03 points at the 12-week follow-up. Although the improvement was not statistically significant compared to baseline (p=0.73), the scores remained numerically lower than pretreatment levels.

Upon examination of the z-values for variables with significant differences, all z-values were negative. This indicates that, compared to the start of the study (pretreatment), these values decreased following participation in gardening activity and remained relatively low for up to 12 weeks afterward. This suggests that the improvements observed for the indicators in this study were maintained over a relatively long period.

When comparing the results of assessments conducted before the gardening activities, immediately after the gardening activities, and during the follow-up visits 12 weeks later, the severity of insomnia was found to decrease from 13.36±3.40 points prior to the treatment to 6.78±3.22 points immediately after the treatment (p<0.001, r=0.80). This improvement was maintained at 7.73±4.48 points at the 12-week follow-up, though the difference compared to baseline was not statistically significant (p=0.32).

All sub-items of insomnia severity, including difficulty falling asleep, difficulty maintaining sleep, and early morning awakening, were found to improve at the end of the treatment and were maintained at the 12-week follow-up. Satisfaction with sleep patterns showed significant improvement immediately after the gardening activities (p<0.001, r=0.80); although some decline was observed at the 12-week follow-up, the scores remained significantly better than those at the baseline (p=0.01, r=0.57). An evaluation of concerns regarding insomnia indicated significant improvement after gardening activities (p=0.02, r=0.56), though the difference was not statistically significant at the 12-week follow-up (p=0.53).

The cortisol levels changed from 0.27±0.26 μg/dL before treatment to 0.27±0.18 μg/dL immediately after treatment (p=0.754), increasing to 83.61±288.27 μg/dL after 12 weeks (p=0.05). An unexpected increase in salivary cortisol observed at the 12-week follow-up may reflect external stressors, seasonal variation, or menopausal-related hormonal fluctuations rather than a direct effect of the intervention. This finding highlights the limitations of using cortisol as a single physiological marker of stress in lifestyle intervention studies with small sample sizes.

An objective and scientific evaluation of sleep before and after gardening activities was conducted using polysomnography, which provided quantitative data on sleep quality (Table 2). According to the Wilcoxon signed-rank test, sleep latency decreased from 15.83±15.38 minutes before gardening activities to 9.54±8.65 minutes after these activities (p=0.04, r=0.47). Rapid eye movement sleep latency also decreased from 111.73±48.01 minutes to 88.13±31.67 minutes (p=0.03, r=0.51). The total sleep time showed an increasing trend from 356.64±61.32 minutes to 373.88±37.37 minutes; however, this change was not statistically significant (p=0.24, r=0.27). Upon examination of the z-values for the variables showing significant differences, all z-values were confirmed to be negative.

This study specifically targeted individuals with poor sleep efficiency who reported decreased sleep quality and examined the changes after gardening activities (n=19). However, although seven participants subjectively reported poor sleep quality in the sleep questionnaires, objective laboratory measures, including polysomnography, indicated good sleep quality with a sleep efficiency of 85% or higher. This discrepancy suggests that subjective insomnia may have been due to sleep misperception. Therefore, further analyses were repeated on the remaining 12 participants who exhibited objectively impaired sleep quality.

Those participants who had objective insomnia, as indicated by a sleep efficiency of less than 85%, as measured by polysomnography, were reanalyzed. These further analyses revealed that the perception of stress regarding the inability to control important matters improved following engagement in gardening activities (p=0.03, r=0.61) (Table 3). Upon examining the z-values of the variables that showed significant differences, all z-values were found to be negative. This indicated a reduction in the values of these variables between the pre- and post-gardening activity assessments. The global assessment of overall daily activities showed a reduction in dysfunction caused by poor sleep quality after gardening activities, although this change was not statistically significant (p=0.57).

The Wilcoxon signed-rank test results revealed improvements in both PSQI scores and insomnia severity following the gardening activity (Table 3). Specifically, the PSQI scores improved from 9.33±2.01 points before the treatment to 5.08± 2.15 points (p<0.001, r=0.85) after the treatment. At the 12-week follow-up assessment, the score was 6.25±3.07 points; although numerically lower than baseline, this difference was not statistically significant (p=0.31).

The severity of insomnia, as assessed using the ISI, also improved. The score decreased from 14.00±3.71 points before the gardening activity treatment to 6.66±3.33 points after participation in the treatment (p<0.001, r=0.85). This improvement was sustained at the 12-week follow-up assessment; however, the difference was not statistically significant (p=0.31 compared to baseline). Although the score increased slightly compared to that immediately after the treatment, it remained numerically lower than the score before the gardening activities, though this difference was not statistically significant (p=0.32).

Furthermore, improvements were observed in all the evaluation items of insomnia severity, including difficulty falling asleep, difficulty maintaining sleep, and early morning awakening. Satisfaction with sleep patterns improved (p=0.01, r=0.74), and there was also an improvement in the impact of insomnia on quality of life (p=0.04, r=0.59) after gardening activities.

Analysis of the 12 participants with a sleep efficiency of <85% revealed significant improvements in several sleep parameters, as shown in Table 4. Sleep efficiency improved from 75.47%±10.56% to 85.20%±7.50% (p=0.01, r=0.74), indicating a statistically significant enhancement. Additionally, sleep latency decreased from 21.69±16.59 minutes to 10.60±9.49 minutes, showing a significant reduction (p=0.002, r=0.78). The total sleep time increased from 323.02±50.58 minutes to 370.47±36.80 minutes (p=0.01, r=0.74).

The total arousal time (i.e., the cumulative duration of awakenings after sleep onset) showed a statistically significant decrease from 102.47±41.81 minutes to 59.65±36.28 minutes (p=0.015, r=0.70). This indicates a reduction in awakening during sleep. The percentage of total sleep time spent in arousals, reflecting interruptions in sleep maintenance, decreased significantly from 19.12%±10.70% to 8.85%±5.13% (p=0.01, r=0.74), indicating improved sleep continuity. While the total wake time and percentage of wake time showed significant improvements, these measures should be interpreted alongside standard metrics, such as the arousal index and wake after sleep onset (WASO), which did not show significant changes in our study. Upon examining the z-values of the variables that showed significant differences, all z-values were negative. These results, along with other polysomnography findings, are presented in Table 4.

The observed improvement in sleep efficiency from 75.47% to 85.20% is particularly noteworthy, as a sleep efficiency above 85% is generally considered indicative of good sleep quality in clinical practice. This magnitude of this improvement is comparable to that reported for CBT-I, the current gold standard non-pharmacological treatment, suggesting that urban gardening may offer similar therapeutic benefits through different mechanisms [24,25]. This reduction in sleep latency further suggests that urban gardening may facilitate the transition from wakefulness to sleep, possibly through increased physical fatigue and reduced pre-sleep cognitive arousal. These findings extend previous research on nature-based interventions by providing objective polysomnographic evidence, which was lacking in earlier studies that relied solely on self-reported measures [26,27]. Notably, in participants with objectively poor sleep efficiency (<85%), clinically meaningful improvements in sleep efficiency, sleep latency, and total sleep time were observed following the intervention. This differential response pattern suggests that individuals with more severe sleep impairment may derive greater benefits from the intervention, a finding consistent with the broader literature on sleep interventions. However, as no direct statistical comparison was performed between the subgroups, these findings should be interpreted as exploratory and hypothesis-generating, rather than as evidence of differential effectiveness. Future studies with larger sample sizes are needed to formally test this potential moderating effect.

This significant reduction in perceived stress is consistent with previous studies demonstrating that gardening activities and exposure to green spaces can effectively alleviate psychological stress in urban populations [28,29]. This stress-reducing effect may be attributed to multiple factors inherent in gardening activities: the rhythmic and repetitive nature of gardening tasks may promote a meditative state, whereas the outdoor setting provides a temporary escape from daily stressors and screen-based activities. The trend toward improvement in depressive symptoms (p=0.05), although at a threshold of statistical significance, highlights potential antidepressant effects that warrant further investigation. Gardening activities may address several known risk factors for depression, including physical inactivity, social isolation, and a lack of purposeful engagement [30]. The improvements observed in both the quality of life and enjoyment of daily activities indicate that the benefits of urban gardening extend beyond specific symptom reduction to broader aspects of well-being. Importantly, these psychological improvements may serve as mediating pathways through which urban gardening enhances sleep quality, given the well-established bidirectional relationship between sleep disturbance and psychological distress. Therefore, addressing psychological factors through gardening activities may create a virtuous cycle wherein improved mood and reduced stress facilitate better sleep, which in turn reinforces psychological well-being.

The discordance between subjective stress improvements and salivary cortisol levels warrants careful interpretation of the results. The trend toward increased cortisol levels at the 12-week follow-up (p=0.05) may reflect several confounding factors rather than a true intervention effect. First, the timing of the follow-up assessment may have coincided with seasonal stressors or environmental changes unrelated to the intervention. Second, since most participants were middle-aged women, menopausal hormonal fluctuations may have introduced substantial variability in cortisol measurements. Third, the single-timepoint cortisol sampling used in this study may not have adequately captured the dynamic nature of the hypothalamic– pituitary–adrenal axis response to chronic stress reduction. Dissociation between subjective and objective stress measures is common in behavioral intervention research and highlights the complexity of stress physiology [31,32]. Future studies should consider multiple daily cortisol samplings, cortisol awakening responses, and hair cortisol levels as more stable biomarkers of chronic stress.

The beneficial effects of urban gardening on sleep are likely mediated by multiple interacting mechanisms operating through distinct but complementary pathways. According to the two-process model of sleep regulation, moderate physical activity during gardening may enhance the homeostatic sleep drive (Process S) by increasing adenosine accumulation and promoting deeper and more restorative sleep [33-36]. Regular exposure to natural light during outdoor activities, particularly morning light exposure, may help entrain circadian rhythms (Process C) and optimize the timing of melatonin secretion [37]. The combination of these effects may be particularly beneficial for middle-aged and older adults who often experience age-related weakening of circadian signals. Additionally, psychosocial factors, such as reduced stress, improved mood, and enhanced social connectedness, likely play important roles in the observed improvements in sleep. The group-based nature of urban gardening programs may address social isolation, which is increasingly recognized as a contributor to poor sleep in urban populations. Furthermore, the sense of purpose and accomplishment derived from nurturing plants and harvesting produce may reduce rumination and pre-sleep cognitive arousal, which are common barriers to sleep initiation. The multifactorial nature of urban gardening, which integrates physical, environmental, psychological, and social elements, may explain why its effects on sleep appear to be comparable to those of more targeted single-modality interventions.

From clinical and public health perspectives, urban gardening offers several unique advantages over conventional sleep interventions. Unlike pharmacological treatments, it carries no risk of dependency, tolerance, or adverse effects, such as next-day sedation or cognitive impairment. Although herbal medicines have been investigated as alternative treatments for insomnia [38-41], urban gardening may offer broader therapeutic benefits by simultaneously addressing physical, psychological, and social factors contributing to sleep disturbance. Unlike CBT-I, which is the current gold-standard behavioral treatment, urban gardening does not require trained therapists, structured clinical settings, or significant healthcare resources, thus making it more accessible and scalable [24,25]. The community-based nature of urban gardening programs may also address the social determinants of health, including social isolation and limited access to green spaces, which disproportionately affect urban populations with poor sleep. Healthcare providers may consider recommending urban gardening as an adjunct or complementary strategy for patients with mildto-moderate sleep disturbances, particularly those who prefer non-pharmacological approaches or have contraindications to sleep medications. Furthermore, public health policymakers should consider integrating urban gardening programs into community health initiatives, particularly in underserved urban areas where sleep problems and limited green space access are prevalent.

This study had several limitations. First, the single-arm design without a control group limited causal inference. Second, the small sample size and female predominance restricted the generalizability of the results. Third, we did not control for additional physical activities or lifestyle changes during the study period. Fourth, polysomnography was not conducted during the follow-up assessment and melatonin could not be measured owing to technical limitations. To address these limitations, future studies should employ randomized controlled designs with larger and more balanced populations, including objective circadian biomarkers such as melatonin, and explore the relative contributions of specific components of urban gardening (e.g., physical activity, social interaction, and dietary use) to sleep outcomes.

This study found urban gardening to be a promising multifaceted approach for improving sleep quality and promoting psychosocial well-being in urban populations. By integrating light physical activity, exposure to natural environments, social interactions, and potential circadian rhythm regulation through sunlight, gardening activities demonstrated benefits in terms of both subjective sleep measures and objective polysomnographic parameters. However, due to several experimental limitations, including the absence of a control group, small sample size, female-dominant participants, and lack of control over confounding physical activities, these findings should be interpreted cautiously. Since more robust evidence is required before definitive conclusions can be drawn regarding therapeutic efficacy, future studies should employ randomized controlled designs with larger and more diverse populations, systematically control potential confounders, and investigate the optimal duration, frequency, and components of urban garden programs. Given these initial promising results, it is clear that the integration of urban gardening with established treatments for sleep disturbance, including digital therapeutics and other evidence-based interventions, shows great potential and warrants further investigation [42-44].