Table 2.

Change in cognitive composite scores during the study

Composite	LMM coefficients (linear time specification)		LMM coefficients (factorial time specification)
	Parameter	Estimate [95% CI]	Parameter	Estimate [95% CI]
PACC5Abridged			Post-intervention visit	0.04 [− 0.01, 0.10]
	Time (weeks)	0.12 [0.05, 0.18]	Follow-up visit	0.16 [0.10, 0.22]
	Time × arm	− 0.02 [− 0.09, 0.04]	Post-intervention × arm	0.04 [− 0.02, 0.11]
	Practice	0.04 [− 0.02, 0.09]	Follow-up × arm	− 0.01 [− 0.08, 0.05]
Attention composite			Post-intervention visit	0.05 [− 0.01, 0.10]
	Time (weeks)	0.02 [− 0.04, 0.08]	Follow-up visit	0.06 [0.01, 0.11]
	Time × arm	− 0.01 [− 0.07, 0.05]	Post-intervention × arm	− 0.01 [− 0.07, 0.05]
	Practice	0.04 [− 0.01, 0.09]	Follow-up × arm	− 0.01 [− 0.07, 0.05]
Executive composite			Post-intervention visit	0.05 [− 0.03, 0.12]
	Time (weeks)	0.03 [− 0.06, 0.11]	Follow-up visit	0.07 [− 0.00, 0.15]
	Time × arm	0.04 [− 0.04, 0.13]	Post-intervention × arm	0.02 [− 0.06, 0.11]
	Practice	0.04 [− 0.03, 0.11]	Follow-up × arm	0.05 [− 0.04, 0.13]

The model fits presented in the table are PP analyses. Regression coefficients are standardized. The time metric for linear time models was weeks (continuous) and for factorial time models, visits (factor). For factorial time models, the reference visit is the baseline. The post-intervention visit was at week 8, and the follow-up visit was at week 24. For both types of model, the reference trial arm is HSMP; positive coefficients for the interaction terms thus represent a relatively greater improvement in the HSMP (vs. CMBAS) arm; negative coefficients indicate the converse. Coefficient estimates in bold had p-values < 0.05 in the initial models. All models were adjusted for sex, age, years of education, state anxiety, depressive symptoms, and trial site; models using the linear time specification were also adjusted for cognitive retest effects. None of the models was substantively altered in sensitivity analyses which only included participants who attended ≥ 4 intervention sessions

Abbreviations: PACC5_Abridged Abridged Preclinical Alzheimer Cognitive Composite 5, CMBAS Caring Mindfulness-Based Approach for Seniors, HSMP Health Self-Management Program, CI confidence interval, LMM linear mixed model, PP per-protocol