Cardiometabolic Risk Determinants Across Adulthood

CARDIA is an ongoing, prospective study of the determinants and evolution of cardiometabolic risk starting in young adulthood. Participants for the baseline examination were randomly selected and recruited by telephone or door to door from census tracts in the four field centers: Minneapolis, MN and Chicago, IL, by telephone exchanges within the Birmingham, AL city limits, and from lists of the Kaiser-Permanente Health Plan membership in Berkeley, CA. Although the source populations and methods of recruitment varied slightly, all centers adhered to the following basic eligibility criteria: 18–30 years of age, Black or white race, and permanent residential address in the target areas. Individuals living with a long-term illness or disability that would prevent participation and pregnant women and those who were up to 3 months postpartum were excluded from recruitment [38 (link),39 (link)]. A total of 5,115 young adults aged 18–30 years were enrolled at baseline in 1985–1986 (Exam Year 0) with balance according to race (Black and white), sex, education (≤high school and >high school), and age (18–24 and 25–30 years) from the population in each of the four metropolitan areas. Follow-up examinations occurred in 1987–1988 (Exam Year 2), 1990–1991 (Exam Year 5), 1992–1993 (Exam Year 7), 1995–1996 (Exam Year 10), 2000–2001 (Exam Year 15), and 2005–2006 (Exam Year 20) and 2010–2011 (Exam Year 25); retention at each exam year was 91%, 90%, 86%, 81%, 79%, 74%, 72% and 72%, respectively. The CARDIA study methods are described in detail elsewhere [38 (link),39 (link)]. Each study participant provided written informed consent, and data were collected under protocols approved by the Institutional Review Boards at each study center and at the University of North Carolina at Chapel Hill [8 ]. The current study included CARDIA exam years 0, 5, 10, 15, 20 and 25. We excluded participants with only one of the six exams used in this analysis. As has been done in previous studies, to minimize bias resulting from illness that may affect body weight, we excluded participants with hypertension, diabetes or cancer at exam year 0 [32 (link),40 (link)]. We further restricted the analytic sample to those with data on diabetes, hypertension, or self-reported cancer diagnoses at each exam and those with waist circumference and BMI data at exam year 0. For individuals included in the primary analytic sample, observations were excluded at given exam years if participants were pregnant or breastfeeding or had implausible energy intakes (<600 kcal/d or >6000/d kcal for women and <800 kcal/d or >8000 kcal/d for men) at any exam or if they were missing exposure, outcome, or covariate data at a given exam year. We censored observations for participants with diabetes, hypertension, or self-reported cancer during follow-up at the year in which the disease was reported. To facilitate sensitivity analyses, a secondary analytic dataset was created in accordance with the primary analytic dataset exclusion criteria except for censoring on diabetes, hypertension, or self-reported cancer. The secondary dataset is described in the (S1 File) [8 ,32 (link),40 (link)].