USING RACIAL/ETHNIC COMPARISONS TO EVALUATE UNMEASURED CONFOUNDING: CHILDHOOD DISADVANTAGE AND COGNITIVE RESERVE IN LATE LIFE IN THE MULTI-ETHNIC KHANDLE COHORT

Abstract

Numerous indicators of childhood disadvantage predict dementia risk and cognitive reserve. Evaluating whether specific types of disadvantage are causal for late life cognitive outcomes is challenging due to numerous unmeasured potential confounders. Triangulating associations across racial/ethnic groups with different confounding patterns may help evaluate whether any specific measure is causal or merely proxies for general disadvantage. We hypothesized that childhood experience of a traumatic event, specifically death of a child in the immediate family (e.g., a sibling), would predict cognitive reserve similarly for four racial/ethnic groups, despite different patterning of this social exposure. At baseline, 250 Asian, 189 Latino, 251 African-American, and 348 white Kaiser Healthy Aging and Diverse Life Experiences (KHANDLE) cohort participants (average age=70.3 years; standard deviation=6.8) reported on childhood social experiences, including whether a child in their immediate family died (child death). We Z-scored a composite of verbal memory, episodic memory, and executive function as an indicator of cognitive reserve and estimated linear regressions for whether child death predicted cognitive reserve differentially by race/ethnicity. Child death was common for Asian (22.4%), African-American (17.5%) and Latino (19.1%) participants, but less frequent for white (8.6%) participants (p<.01 for racial/ethnic differences). Child death was associated with lower maternal (bwhite=−1.8 years difference; bLatino=-1.4) and paternal (bwhite=−1.4; bLatino=-1.5) education for whites and Latinos, but not associated with paternal education for Asians (bAsian=0.1) or maternal education for African-Americans (bAfrican-American=0.1). In age and sex adjusted models, child death predicted lower cognitive reserve similarly for whites (bwhite=−.44; 95% CI: −.76, −.13) and Latinos (bLatino=−.49; 95% CI: −.78, −.19); the association was weaker among African-American (bAfrican-American =−.22; 95% CI: −.48, .03) and not present among Asian (bAsian=−.02; 95% CI: −.28, .24) participants. Contrary to our hypothesis, childhood exposure to death of a child was associated with lower cognitive reserve only in racial/ethnic groups for whom exposure was correlated with other indicators of childhood disadvantage. Childhood exposure to death of a child may be relevant as a marker of general disadvantage rather than causal in and of itself. Multi-ethnic research studies provide opportunities to triangulate evidence on risk factors with complex, unmeasured confounders.