Intergenerational factors in cardiovascular health : a prospective cohort study

Background. Cardiovascular etiologies are complex and influenced by several exposures acting together across the life-course. This dissertation examined the life-course pathways of cardiovascular health in relation to three intergenerational factors. First, we examined associations of genetic risk of higher body mass index (BMI) with BMI development, and whether the genetic associations differ by socioeconomic position (SEP) at different life stages (Study I). Second, we assessed intergenerational associations of ideal cardiovascular health behaviors between parents and their adult offspring, and whether these associations differ by SEP at different life stages (Study II). Third, we evaluated the association of favorable childhood psychosocial environment with adult cardiac structure and function (left ventricular (LV) mass (g/m2.7) and LV diastolic function (E/e’ ratio) and to which extent ideal cardiovascular health behaviors mediate these associations (Study III). Methods. Study participants were three subsamples from the prospective intergenerational Cardiovascular Risk in Young Finns Study (n=2441 in Study I, n=1856 in Study II, n=880 in Study III). We used data from follow-ups in 1980, 1983, 1986, 2001, 2007 and 2011. BMI was recorded in all follow-ups across 1980-2011. Genetic risk of BMI was assessed in 2001 with 97 single-nucleotide polymorphisms identified in the most recent GWAS meta-analysis. Four cardiovascular health behaviors – smoking, BMI, physical activity and diet – were recorded following the American Heart Association definitions. SEP was measured as educational attainment, income and occupational status. Adult SEP and health behaviors were self-reported or measured during a study visit in 1986, 2001, 2007 and 2011. Parental SEP and health behaviors were self-reported by parents of study participants in 1980. Childhood environment was assessed with a cumulative score comprising data from six relevant psychosocial domains reported by parents of study participants in 1980. Echocardiographic examination was conducted in 2011. Statistical analysis. Genetic associations with BMI development were analyzed with multilevel linear regression with random intercepts and random slopes. Intergenerational associations of health behaviors were assessed using ordinal and linear multilevel models with random intercepts. Associations of childhood psychosocial environment with adult cardiac structure and function and mediation pathways through cardiovascular health behaviors were assessed with marginal structural models in the causal mediation analysis framework. Results. In Study I, mean BMI increased from 22.6 to 26.6 kg/m2 during the follow-up. In growth curve analyses, the genetic risk score was associated with faster BMI increase over time (b=0.02, 95% CI, 0.01, 0.02 for genetic risk score x age interaction). The genetic associations with BMI were weaker among those with higher (vs lower) educational attainment in adulthood (b=-0.12, 95% CI, -0.23, 0.01 for genetic risk score x adult education interaction). At age 49, compared with those at the 10th percentile of the genetic risk score, those at the 90th percentile had 3.3 units higher predicted BMI at the lowest level of educational attainment and 2.4 units higher BMI at the highest level educational attainment. No interaction effect was observed between the genetic risk score and parental education (b=0.05, 95% CI, -0.09, 0.18 for genetic risk score x parental education interaction). In Study II, one additional ideal cardiovascular health behavior among parents was associated with 28% higher odds of one additional ideal behavior among offspring (odds ratio (OR)=1.28, 95% CI, 1.17,1.39). Furthermore, ORs for these intergenerational associations were greater among offspring who had higher own adult educational attainment or whose parents had higher educational attainment (OR=1.32 for high vs OR=1.04 for low offspring education; p=0.02 for interaction, OR=1.56 for high vs OR=1.19 for low parental education; p=0.01 for interaction). Similar trends were seen with parental income and offspring occupation. Results from linear regression analyses were similar. In Study III, favorable psychosocial environment in childhood was associated with more optimal cardiac structure and function in adulthood. Those above the median of the childhood score versus below the median had 1.28 g/m2.7 lower LV mass (95% CI=-2.63, 0.07) and 0.18 lower E/e’ ratio (95% CI=-0.39, 0.03). There was no evidence of indirect effects from childhood environment to LV outcomes through adult cardiovascular health behaviors after controlling for time-dependent confounding by adult SEP (indirect effect b=-0.30, 95% CI=-1.22, 0.63 for LV mass, b=-0.04, 95% CI=-0.18, 0.11 for E/e’ ratio). Results after multiple imputation were similar. Conclusions. These findings highlight the importance of intergenerational and early-life exposures in initiating pathways of long-term cardiovascular health and suggest these pathways may be shaped by socioeconomic circumstances at different life stages