Fig. 2
Comparison of the association estimates of the main models (single-exposure to green space) to those of the multi-exposure models as represented by the change in the outcome of cognitive parameters for the IQR increment in green space within several radii around the residence. A Change in mean reaction time (msec) for the Continuous Performance Test (n = 307); B Change in mean reaction time (msec) for the Stroop Test (n = 305); C Change in the number of digits for the Digit Span Forward Test (n = 305) and D. for the Digit Span Backward Test (n = 306); E Change in latency (sec) for the Pattern Comparison Test (n = 299) and F. for the Digit-Symbol Test (n = 307). The main models were adjusted for sex, age (linear and quadratic term), BMI, education of the mother, highest occupation of either parent, passive smoking, day of the week, season of examination, neighbourhood household income, and the random effect of the study area and subject. The multi-exposure models were additionally adjusted for exposure to PM2.5, NO2 during the year before examination, distance to major roads, or traffic noise. Statistically significant estimates (p < 0.05) at a 95% confidence level are marked with an asterisk, and with †for borderline significance (0.05 < p < 0.10)