2.4. Study Data
Natural outdoor environments were defined as all public and private open spaces that contain “green” and/or “blue” natural elements. Green natural elements included, among others, roof gardens, forests, farmland, city parks, and nature reserves. Blue spaces encompassed water structures such as canals, ponds, creeks, rivers, lakes, and beaches [52
Childhood NOE exposure:
Participants were asked to retrospectively rate how often they spent time in NOE during their childhood. This included purposeful visits (e.g., hiking in natural parks with the parents) and non-purposeful visits (e.g., playing outside in the backyard). Answers were scored on a five-point scale, which was dichotomized (“never”, “sometimes” as low levels of childhood NOE exposure; “regularly”, “often”, “very often” as high levels of childhood NOE exposure) as done in previous studies [44
Socio-demographic characteristics: Age (in years), gender (male; female), birth in the country of residence (yes; no), smoking status (current; former; never), household composition (living alone; with partner but without children; with children <12 years; with children ≥12 years; other), educational achievement (primary school or no education; secondary school or further education; university degree or higher), perceived income situation (cannot make ends meet; just have enough to get along; comfortable), and SES of the neighborhood were included in the adjusted models as potential confounders.
Perceived amount of NOE: Respondents were asked to describe their close living environment in terms of green and blue. The question consisted of three items concerning the participants’ (a) street, (b) neighborhood, and (c) view from home, and each item was answered on a five-point scale ranging from “not at all” to “very”. A summary score was calculated with higher scores indicating higher perceived amount of NOE.
Use of NOE: It was assessed by combining two associated questions concerning the visits (frequency and duration per visit) of NOE close to home during the last four weeks. Answers were measured on five-point scales (Frequency: “n.a.”, “never”, “≤1”, “2–3”, “1–4 times weekly”, “(almost) daily; Duration: “n.a.”, “<1 hour [h]”, “1–2 h”, “3–5 h”, “6–10 h”). The middle values were determined (e.g., 3–5 h equaling 4) and the values of both questions were multiplied with each other to calculate the total hours spent in NOE during the last four weeks.
Residential surrounding greenness:
This was assessed by using the normalized difference vegetation index (NDVI), derived from satellite images from Landsat 8 at a resolution of 30 m × 30 m [54
]. The NDVI is based on the fact that healthy vegetation absorbs most visible light and at the same time reflects large parts of near-infrared light. It provides a number on a scale from −1 to +1, with higher values indicating higher density of green vegetation [54
]. Average NDVI values were calculated within a (Euclidean) buffer of 100 m, 300 m, and 500 m around participants’ dwellings [55
Satisfaction with NOE: The satisfaction with NOE in the surrounding living environment was assessed in terms of (a) quality, (b) amount, (c) maintenance, and (d) safety of the green/blue environment. Answers were scored on a five-point scale (“very dissatisfied” to “very satisfied”), with higher scores indicating higher satisfaction. A summary score of the four items was calculated by summing the points of all items together.
Importance of NOE: Participants were asked how important green/blue spaces near place of residence are for: (a) physical activity, (b) social activities, (c) relaxation, and (d) that there are green walking and biking paths to go to work, school and other destinations. Answers were measured on a five-point scale (“not important at all” to “very important”). Higher scores indicated higher importance followed by calculation of a summary score.
Mental health and vitality:
Mental health and vitality were assessed using two subscales from “The Medical Outcome Study 36-item Short-Form Health Survey (SF-36)”, which have previously been shown to be valid and reliable [56
]. The five-item mental health component assessed nervousness and feelings of depression during the past four weeks. The four-item vitality component concerned perceived level of energy and fatigue and is recommended in order to better capture differences in subjective well-being [56
]. All nine items were scored on a six-point scale, ranging from “All of the time” to “none of the time”. In line with previous studies, two items of the mental health and two of the vitality subscale were reverse-coded [10
]. In case a maximum of two items of the mental health subscale or a maximum of one item of the vitality subscale were missing, they were replaced by the average of the remaining items. By doing so, we followed the manual and interpretation guide for the SF-36 Health Survey [57
]. Accordingly, sum scores were calculated and transformed into continuous scales (0–100), with higher scores reflecting better mental health or vitality [10
2.5. Statistical Analysis
Equivalence tests with Benjamini–Hochberg adjustment for false discovery rates (5%) were used to test for differences between low and high level childhood NOE exposure groups [58
]. We analyzed the association between childhood NOE exposure and mental health/vitality as the main aim of the study. Then, following the steps as suggested by Baron and Kenny, we investigated the mediation roles of residential surrounding greenness, perceived amount, use, satisfaction, and importance of NOE in the association between childhood NOE exposure and mental health as well as vitality (see step 2 and 3 below) [59
]. All analyses were performed with pooled data and for the four cities separately. Random intercepts were used. In the analyses with pooled data, we included city and neighborhood as random intercepts. In the analyses per city, we only included neighborhood as a random intercept. For all analyses, first crude models were built, which were then adjusted for potential confounders (age, gender, native/foreign born, level of education, perceived income situation, neighborhood SES, household composition, smoking status). All analyses were performed using Stata 15 [60
Step 1—The association between childhood NOE exposure and mental health/vitality: The associations between the exposure variable childhood NOE exposure and the outcomes mental health and vitality were assessed using mixed effects regression analyses.
Step 2—The association between childhood NOE exposure and the potential mediators: Logistic mixed effects models were developed for the exposure variable childhood NOE exposure and the potential mediators (perceived amount of NOE, use of NOE, satisfaction with NOE, importance of NOE, and residential surrounding greenness). In the pooled analyses, the mediators were dichotomized based on the overall median. For analyses on the city level, city-specific cut points, based on the city-specific median, were used. The lower value was defined as the reference group and indicated lower rating.
Intermediate step—The association between the potential mediators and mental health/vitality: The associations between the potential mediators and mental health as well as vitality were assessed using mixed effects models.
Step 3—The association between childhood NOE exposure and mental health/vitality including the potential mediators: The potential mediators were added (separately and altogether) to the adjusted main model which has been described in step 1, in order to assess the association between childhood NOE exposure and mental health/vitality controlled for the potential confounders and for the potential mediators.