Data and material disclosure statement

Data documentation for CARDIA is publicly available online (cardia.dopm.uab.edu.). Data used in this article are available on reasonable request from the CARDIA coordinating Center.

Study population

Details of the CARDIA study design and examinations have been reported previously [9]. The CARDIA study is a longitudinal cohort study which began in 1985–1986 (Y0). A cohort sample of 5,115 healthy participants were recruited at four US sites: Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California, and were balanced for gender, race, age (18–24 years and 25–30 years), and educational level (high school or less education and greater than high school education). Seven follow-up examinations of the cohort were conducted at years 2, 5, 7, 15, 20 (Y20), and 25. At each study, all participants provided written informed consent and the institutional review boards at each study site and coordinating center have granted approval annually for all examinations.

The flow chart of this study is presented in supplementary figure 3. Of the initial 5,115 recruited participants at baseline, 1,587 did not return at Y20 and 593 had missing CVH data or covariates data, leaving 2,935 participants included in the basic analysis. For analysis of CAC, 123 participants who had not taken CT scan at either Y20 or Y25 were excluded, while for analysis of IMT 244 participants who had not taken carotid ultrasound at Y20 were excluded, leaving 2,812 participants included in the analysis of CAC and 2,691 participants included in the analysis of IMT.

Cardiovascular health

CVH, consistent with Life’s Simple 7 according to AHA recommendation, consists of four components of metabolic risk: blood pressure, blood glucose, total cholesterol, and body mass index, and three components of lifestyle: smoking, physical activity, and diet. All metrics were measured at baseline and Y20, obtained by trained and certified technicians. Blood pressure was measured three times after five minutes rest at seated and recorded as the average of second and third-time measurement. Participants were asked to fast for 12 hours before venous blood was drawn for blood glucose and total cholesterol analysis. Body mass index was calculated as weight in kilograms divided by height in meters, squared. Cigarette smoking history was self-reported. Physical activity scoring was assessed from a physical activity questionnaire and calculated with the frequency and duration of 13 kinds of activities. Dietary intakes were assessed from interviewer-administrated, quantitative food history [10]. Medication use for controlling blood pressure, cholesterol, or glucose was self-reported. Covariates such as age, gender, race, education level, and history of disease were self-reported by questionnaire.

CVH scores were calculated according to prior studies [11, 12, 13] (Table 1). According to prior study [11], we chose 40% as a cut-off point for physical activity. Participants with top 40% physical activity were given two points, and participants with lowest 20% were given zero points, the rest were given one point. Dietary intake measure in CARDIA is most useful for ranking individuals based on one-month consumption rather than absolutely quantifying food intake [14]. Therefore, we adjusted the dietary score based on traditional AHA standard and calculated separately for each sex with four food sources: potassium(mg), calcium(mg), fiber(g), and saturated fat(g). A higher score was assigned to participants with higher intake of potassium, calcium, and fiber and a lower intake of saturated fat (High = 5 to low = 1). Four scores summed up to a total score ranging from 4 to 20. The highest 40% get two points. This method was proven to be in concordance with other dietary eating measure patterns and the traditional AHA standard [11, 12]. Based on CVH scores, participants were assigned to 3 classes: high (12–14 points), moderate (8–11 points), and low (0–7 points). CVH scores at baseline subtracted from those at Y20 are the change of CVH scores from baseline to Y20.

Coronary artery calcium

The extent of CAC was quantified at years 20 and 25 using a standardized protocol [15]. At Y20, two sequential CT scans were provided using either a cardiac-gated electron beam CT scanner or a multidetector CT system. At Y25, only a single CT scan was performed, using a 64-channel multidetector CT system. Prior study suggested that two CT scans were comparable [16]. Computed tomography images were read by experienced technicians in a central reading center. Total CAC score was obtained by Agatston method [17], summing up all lesions within a given artery and across all arteries (left main, left anterior descending, left circumflex, and right coronary artery). A CAC score over zero at either Y20 or Y25 was defined as the presence of CAC.

Intima-media thickness

Images of distal common carotid artery, the carotid bulb, and the proximal internal carotid artery on both sides were obtained by high-resolution B-mode ultrasonography using a standardized protocol at Y20. Intima-media thickness was calculated from the average of the mean intima-media thickness for the internal, bulb, and common carotid near and far walls of the right and left sides [18]. Carotid intima-media thickness over one was defined as a presence of the thickening of carotid intima-media thickness, which is abnormal IMT.

Statistical analyses

Continuous data are expressed as means ± SD and categorical variables are presented as counts and proportions. Multivariable adjusted logistic regression analyses were conducted to estimate the association of baseline CVH and changes and presence of CAC and abnormal IMT. Multivariable adjusted linear regression analyses were used to estimate the association between changes of CVH and IMT. We adjusted for demographic variables (age at baseline, gender, race, and education), current drinker, history of hypertension, history of diabetes. We also performed subgroup analysis based on gender, race, alcohol use, history of hypertension, history of diabetes. We further adjusted for family income at Y5 to measure the socio-economic status affect.

A 2-sided P value of <0.05 was considered statistically significant. All analyses were performed using SPSS version 26 and Stata SE version 15.

Characteristic of participants

The characteristics of the participants at Y0 and Y20 were presented at Table 2. Of 2,935 participants at baseline, 43.8% were male, 46.7% were black, and the average (SD) age was 25 (3.59) years. The proportion of participants with high CVH class dropped from 38.3% at the baseline to 21.8% at Y20, while participants with low CVH class climbed from 5.55% to 21.8%. The incident rate of CAC was 26.5% (744/2,812). The incident rate of abnormal IMT was 3.6% (98/2,691). The average change of CVH scores from baseline to Y20 is –1.26 ± 2.13. The distribution of CVH points and CVH classes at both Y0 and Y20, and also the 20-year changes of both were presented at Figure 1 and Supplementary figure 4–6. Most of the CVH scores of participants were in the 10–12 range at Y0, while the majority were in the 9–11 range at Y20, showing the overall downward trend of CVH from Y0 to Y20.

Association between CVH and CAC

The association between CVH and incidence of CAC was presented in Table 3. After adjustment by age, gender, race, education, current drinker, history of hypertension, and history of diabetes, the odds ratio (OR) is 0.81 (95% CI 0.78, 0.86) for per 1 point increase in CVH at baseline. For the further-adjusted baseline CVH score, the OR is 0.86 (95% CI 0.82, 0.90) for per 1 point increase in CVH changes. An inverse association was also found between changes of CVH scores with incidence of CAC (P_trend < 0.001) (Supplementary Table 1).

In order to assess the specific risk of CAC in different situations, we separated the baseline CVH class into low, moderate, and high classes to see whether the change of CVH class altered the risk of CAC (Table 4). For the moderate class at baseline compared to the stable moderate class, participants who climbed to high class at Y20 had a lower risk of CAC (Adjusted OR 0.55, 95% CI 0.35, 0.89), while those who dropped to low class at Y20 had a higher risk of CAC (Adjusted OR 1.71, 95% CI 1.32, 2.23). Compared to stable high class, the high-to-low class had a higher risk of CAC (Adjusted OR 3.48, 95% CI 1.80, 6.73) and the high-to-moderate class had no significant different risk of CAC (Adjusted OR 1.18, 95% CI 0.81, 1.72). For low class at baseline, significant different risk of CAC was not observed between participants with stable low class and low-to-moderate/high class (Adjusted OR 0.77, 95% CI 0.36, 1.65).

Association between CVH and IMT

The association between CVH and incidence of abnormal IMT was presented in Table 5. After adjustment by age, gender, race, education, current drinker, history of hypertension, and history of diabetes, the OR is 0.85 (95% CI 0.76, 0.94) for per 1 point increase in CVH at baseline. For the further adjusted baseline CVH score, the OR is 0.81 (95% CI 0.73, 0.90) for per 1 point increase in CVH changes. An inverse association was also found between changes of CVH scores with incidence of abnormal IMT (P_trend < 0.001) (Supplementary Table 2). Linear association was also found between changes of CVH scores with IMT (Adjusted β (se) –0.012 (0.001), p-value < 0.001) (Supplementary Table 3).

In order to assess the specific risk of abnormal IMT in different situations, we separated baseline the CVH class into low, moderate, and high classes to see whether the change of CVH class altered the risk of abnormal IMT (Supplement Table 4). For the moderate class at baseline, compared to stable moderate class, moderate-to-low class had a higher risk of abnormal IMT (Adjusted OR 2.46, 95% CI 1.46, 4.12) while those in the moderate-to-high class did not observe a significant lower risk of abnormal IMT (Adjusted OR 0.70, 95% CI 0.20, 2.37). Compared to the stable high class, the high-to-low and high-to-moderate classes had no significant different risk of abnormal IMT (Adjusted OR 0.60, 95% CI 0.06, 5.64 and Adjusted OR 0.76, 95% CI 0.24, 2.41). For the low class at baseline, significant different risk of abnormal IMT was not observed between participants with stable low class and low-to-moderate/high class (Adjusted OR 0.13, 95% CI 0.01, 1.17).

Subgroup Analysis

Subgroup analyses based on gender, race, history of hypertension, history of diabetes, and alcohol use were done separately for CAC and abnormal IMT (Supplementary Figure 1 and Figure 2). No significant difference of risk was found between subgroups. After adjustment for family income at Y5, the association between CVH and incidence of CAC and abnormal IMT remained positive. (Supplementary Table 6).