Table 3. Mediation Model of the Association Between Most Urban Residency and Adolescent Psychotic Experiences via Air Pollutant Exposurea.
| Air Pollutant Mediator | Mediation Model 1b | Mediation Model 2c | ||||||
|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | % Mediated | OR (95% CI) | % Mediated | |||||
| Total | Direct | Indirect | Total | Direct | Indirect | |||
| NO2 | 1.91 (1.34-2.73)d | 1.43 (0.96-2.12)e | 1.34 (1.11-1.61)f | 45g | 1.49 (0.94-2.36)e | 1.19 (0.75-1.91) | 1.25 (1.07-1.45)f | 55g |
| NOx | 1.91 (1.34-2.72)d | 1.43 (0.96-2.12)e | 1.34 (1.12-1.61)f | 45g | 1.49 (0.94-2.36)e | 1.18 (0.74-1.89) | 1.26 (1.08-1.47)f | 58g |
| PM2.5 | 1.93 (1.35-2.75)d | 1.66 (1.13-2.43)h | 1.16 (1.00-1.35)e | 23 | 1.50 (0.95-2.37)e | 1.35 (0.85-2.15) | 1.11 (0.99-1.23)e | 25 |
| PM10 | 1.92 (1.35-2.74)d | 1.81 (1.26-2.60)f | 1.06 (0.99-1.14)e | 9 | 1.50 (0.95-2.37)e | 1.47 (0.93-2.32) | 1.02 (0.99-1.06) | 5 |
Abbreviations: NO2, nitrogen dioxide; NOx, nitrogen oxides; OR, odds ratio; PM2.5, particulate matter with aerodynamic diameter of less than 2.5 μm; PM10, particulate matter with aerodynamic diameter of less than 10 μm.
Mediation models were calculated separately for each air pollutant. This explains the very small differences in total ORs between models. The final mediation model simultaneously estimated the mediatory effects of NO2 and NOx. Analyses included participants with full data in model 2 (n = 1705). Mediatory percentages are rounded to whole numbers. Note that mediation analyses were conducted using the 3-level urbanicity variable. Only the results for most urban settings are reported. Mediatory pathways arising for the intermediate urban settings are shown in eTable 9 in the Supplement. Analyses account for the nonindependence of twin observations.
Indicates the unadjusted association between most urban (vs rural) residency at 18 years of age and adolescent psychotic experiences, split into the total effects (overall association between urbanicity and adolescent psychotic experiences), direct effects (the part of the association that is not explained by mediators in the model, plus measurement error), and the indirect effects (the part of the association that is statistically mediated via specified pollutants in the model).
Indicates total, direct, and indirect effects of most urban residency on adolescent psychotic experiences, adjusted simultaneously for family factors, childhood psychotic symptoms, adolescence substance use, and neighborhood factors.
P < .001.
P > .05 and P < .10.
P < .01.
Indicates significant indirect (mediation) pathways at P <.05.
P < .05.