Moderators of cognitive outcomes from an exercise program in people with mild to moderate dementia

Background/objectives Our aim was to estimate whether baseline participant variables were able to moderate the effect of an exercise intervention on cognition in patients with mild to moderate dementia. Design Subgroup analysis of a multicenter pragmatic randomized controlled trial. Setting Community-based gym/rehabilitation centers. Participants A total of 494 community-dwelling participants with mild to moderate dementia. Intervention Participants were randomized to a moderate- to high-intensity aerobic and strength exercise program or a usual care control group. Experimental group participants attended twice weekly 60- to 90-minute gym sessions for 4 months. Participants were prescribed home exercises for an additional hour per week during the supervised period and 150 minutes each week after the supervised period. Measurements Multilevel regression model analyses were undertaken to identify individual moderators of cognitive function measured through the Alzheimer Disease Assessment Scale-Cognitive Subscale score at 12 months. Results When tested for a formal interaction effect, only cognitive function assessed by the baseline number cancellation test demonstrated a statistically significant interaction effect (−2.7 points; 95% confidence interval = −5.14 to −0.21). Conclusion People with worse number cancellation test scores may experience greater progression of cognitive decline in response to a moderate- to high-intensity exercise program. Further analyses to examine whether these findings can be replicated in planned sufficiently powered analyses are indicated.