Identifying Key Patterns of Social and Structural Determinants of Health and Their Association with Dementia Risk in a Population-Based Study of Cognitive Aging
Maria Vassilaki1, Jeremiah Aakre1, Anna Castillo1, Timothy Lesnick1, Jonathan Graff-Radford2, David Knopman2, Clifford Jack3, Ronald Petersen2, Prashanthi Vemuri3
1Quantitative Health Sciences, 2Neurology, 3Radiology, Mayo Clinic
Objective:
To examine patterns of social and structural determinants of health (SSDoH) and their association with dementia risk in Mayo Clinic Study of Aging participants.
Background:
Alzheimer’s disease and Alzheimer’s disease-related dementias (AD/ADRD) risk, diagnosis, and disease prognosis are associated with SSDoH, but have not been studied with similar rigor as biological or behavioral risk factors.
Design/Methods:
We identified 2,880 participants (mean age (SD): 68.8 (9.7); 50.2% female) with complete SSDoH data at baseline. Using the recently proposed multidimensional SSDoH framework for AD/ADRD studies [Microsystem: social stressors, social support, health literacy (using "contact with health care and multivitamin use"); Exosystem: neighborhood; Macrosystem: social identity], we performed factor analysis and principal component analyses to determine the key components of SSDoH (along with education-occupation scores). Resulting composite scores were considered as covariates for incident dementia analysis using Cox Proportional Hazards models with age as the time scale.
Results:
Principal component (PC) analyses on the SSDoH sub-dimensions reduced variables down to ten and was further reduced to three PCs accounting for 39.2% variance in SSDoH data. The three components were: (PC1) limitations (walk/balance, visual difficulties, and health rating), education-occupation score, anxiety, and neighborhood deprivation; (PC2) Cognitive/physical activities; and (PC3) instrumental activities of daily living (iADLs), health literacy, and Rural-Urban Commuting Area (RUCA) Codes. The model estimates for the association of components with dementia risk were: (PC1) [HR=1.13 95%CI (0.95-1.35); p=0.15], (PC2) [HR= 0.85 95%CI (0.72-1.01); p=0.06] and (PC3) [HR= 1.53 95%CI (1.36-1.72); p<0.001]. Even after adjusting for cognitive status, PC3 kept statistical significance (HR=1.36, 95%CI=1.21-1.54; p<0.001).
Conclusions:
We reduced the high dimensionality of SSDoH variables and found that different groups of variables were associated differentially with dementia risk. “iADLs, health literacy, and RUCA codes” was an important aspect of SSDoH. Studies are warranted to delineate these associations further, facilitate our understanding, inform policy and intervention strategies.