Fig 1.
Baseline measures of enrolled participants are provided in Table 1. Of the intent-to-treat (ITT) cohort (n = 101), 8 individuals withdrew due to time or travel concerns, 8 withdrew due to medical issues, 1 was dissatisfied with his group allocation, and 1 was lost to follow-up. Another 6 individuals were non-adherent to the exercise prescription. Those who did not adhere had slightly more education (17.8yrs [3.2] vs 16.1yrs [2.4]) otherwise there were no significant differences. The remaining 77 individuals were included in per-protocol (PP) analyses: control (n = 23), 75min/wk (n = 18), 150min/wk (n = 21), and 225min/wk (n = 15).
Table 1.
Demographic and descriptive baseline data.
Table 2.
Mean fitness and physical function change from baseline in both the intent-to-treat and per protocol cohorts.
Fig 2.
Visuospatial processing but not attention increases with increasing aerobic exercise dose.
Percent change in VO2 peak (blue bars) increases in a dose-response fashion across the PP exercise groups. The best fitting model of Visuospatial Processing (red bars) follows a similar dose-response pattern. The best fitting model of Simple Attention (green bars) shows that any exercise results in improvement.
Fig 3.
Cardiorespiratory fitness change mediates exercise duration effects on visuospatial performance.
In the basic model without cardiorespiratory fitness change (%change in VO2 peak over 26 weeks) as a mediator, total number of minutes exercised (Exercise Duration) was associated with change in Visuospatial Processing. When change in cardiorespiratory fitness was added to the model as a potential mediator, it fully mediated the relationship of Exercise Duration and Visuospatial Processing improvement. Betas (Standard Error) are reported as the product of simultaneous regression with bootstrap replacement.