Urbanization, loneliness and mental health model - A cross-sectional network analysis with a representative sample

Abstract

With increasing urbanization, more people are exposed to mental health risk factors stemming from the urban social or physical environment. However, research on the relationship between urbanization and mental health is lacking. This cross-sectional study aimed to explore the relationships of the physical environment (spatial cohesion and urban environment) and social factors (neighborhood cohesion) with mental health (stress, anxiety and depression symptoms) and physical health and the mediating role of loneliness based on the proposed theoretical model. The study was conducted in Metropolis GZM (Silesia, Poland) in a representative sample of 3296 residents (48% women). The measurements used were the PSS-10, GAD-7, PHQ-9, R-UCLA3 and neighborhood cohesion scale. ANOVA results showed that city residents had better mental health indices than residents of villages and small towns. The network approach revealed that urbanization was one of the most influential nodes and played the role of a bridge between all other nodes. The model was confirmed and showed that the relationships between the physical environment and mental health were consecutively mediated by neighborhood cohesion and loneliness. Spatial cohesion related to factors of the physical environment and physical health, while physical health was directly connected to sociodemographic factors and weakly to stress. Anxiety was the strongest risk factor. Mental health can be improved by social and architectural factors, such as strengthening neighborhood cohesion and improving neglected buildings.

Fig. 1

Theoretical model of mental health in the context of the physical environment at the social and individual levels with the mediating role of loneliness and neighborhood cohesion.

Fig. 2

Mean values with error bars for (a) perceived stress, (b) anxiety symptoms, and (c) depression symptoms across the place of residence among participants in the GZM metropolitan region, Poland (N = 3296).

Fig. 3

Pearson’s r heatmap. N = 3296. Purple represents positive correlations, while red represents negative correlations. The most intense shading represents a large effect size, while the lightest shading represents a small effect size.*p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 4

Network visualization of the estimated mixed graphical model (MGM). Nodes represent variables from various domains such as Sociodemographic, Physical Environment, Spatial Cohesion, Neighborhood Cohesion, Loneliness, Physical Health, and Mental Health (each distinguished by unique colors) among GZM inhabitants (N = 3296). Edges between nodes signify estimated relationships: green for positive, red for negative, and grey for relationships with categorical variables where the direction is undefined. The thickness of each edge is proportional to the magnitude of the corresponding edge parameter, reflecting the strength of the association. The dichotomized nodes are as follows: Sex (woman = 0, man = 1), City (village, small town, town = 0; city 100K−300K = 1), and Blocks (house = 0; block of flat, tenement = 1).

Fig. 5

Centrality indices plot for EBIC model characterized by strength, closeness, betweenness and expected influence measures. UCLA = loneliness; SocCoh = social cohesion; Size = personal space per m²; PNC = poor neighborhood condition; PhHealth = Physical health; PhAct – Physical activity; PbFreq = frequency of use of urban public area; PbDist = distance to urban public area; NghBel = Neighborhood belonging; GrFreq = frequency of use of the green public area; GrDist = distance to the green public area; Depr = Depression; Blocks = Blocks of flats; Anx – Anxiety.The dichotomized nodes are as follows: Sex (woman = 0, man = 1), City (village, small town, town = 0; city 100K−300K = 1), and Blocks (house = 0; block of flat, tenement = 1).