Data were obtained from the National Health and Nutrition Examination Survey (NHANES), which consists of serial cross-sectional surveys conducted by the National Center for Health Statistic of the US Centers for Disease Control and Prevention to evaluate the health and nutritional status in US adults and children. Participants were selected using a multistage cluster probability sampling method, and all eligible participants were invited to complete a household interview and a physical examination. The questionnaire includes information on demographic, socioeconomic, lifestyle, and other health-related variables, and the physical examination consists of anthropometric and biological measurements. Data from the NHANES are publicly available [12]. The NHANES began in 1999-2000, and subsequent surveys were carried out every 2 years. In this study, we used the data from the 7 survey waves that were conducted from 2005 to 2018 since information on sleep health (1 component of the LE8) was initially collected since 2005. A total of 39,749 adults aged ≥20 years were initially included. We excluded individuals aged >79 years (n=2776), those with missing values on CVH metrics (total of 10,355, including 7996 due to missing data on diet [missing values on all two 24-hour dietary recalls: n=3876; missing on a 24-hour dietary recall: n=4120]) or demographic variables (n=2037), pregnant women (n=481), women who were breastfeeding (n=198), or individuals with self-reported CVD history (n=2235; Figure S1 in Multimedia Appendix 1). A total of 21,667 adults aged 20-79 years were included in this analysis.

We performed analyses in consideration of several demographic characteristics: age, sex, race or ethnicity, educational level, marital status, and household income. We classified participants into three age groups: 20-39, 40-64, and 65-79 years [13]. Self-reported race or ethnicity was classified as Hispanic, non-Hispanic White, non-Hispanic Black, and “Other” (including mainly Asian participants). To characterize an individual’s socioeconomic status, we used the ratio of family income to poverty by dividing family income by the federal poverty threshold for the survey year, adjusting for household size, categorized as <1.30 (low income), 1.30-2.99 (middle income), and ≥3.00 (high income) [14]. Educational level was divided into 4 groups: <high school graduate, high school graduate, some college or associate degree, and college graduate or above. Marital status was categorized as married, divorced/separated/widowed, and unmarried/cohabitation status.

The NHANES was approved by the National Center for Health Statistics Research Ethics Review Board. Written informed consent was obtained from all participants. This study was exempted from ethical approval by the institutional review board of Shandong University given the use of deidentified and open access data.

The LE8 score is based on 4 “behavioral factors” (diet, PA, smoking, and sleep health) and 4 “health factors” (BMI, blood lipids, blood glucose, and BP). Diet quality was assessed using the Healthy Eating Index 2015 (HEI-2015) based on data from 2 interviewer-administered 24-hour dietary recalls. Diet was assessed using the “Food Patterns Equivalents Database 14” from the US Department of Agriculture [15]. PA was assessed through self-reported frequency and duration of moderate (resulting in light sweating or a small increase in breathing or heart rate) and vigorous activity (resulting in heavy sweating or a large increase in breathing or heart rate) over the past 30 days. The PA duration was calculated by the frequency of PA in a week multiplied by the duration of PA each time. Nicotine exposure was assessed based on self-reported consumption of combustible cigarettes and use of e-cigarettes and other tobacco products in the previous 30 days, as well as self-reported secondhand smoke exposure in a participant’s household. Sleep health was assessed from the question on usual sleep duration per day. Height and weight were objectively measured using a stadiometer and digital weight scale, respectively, and BMI was calculated as weight (kg)/height squared (m²). BP was measured for three consecutive times after 5 minutes of seated rest, and the average of the second and third readings was used (or the average of the first and second measurements for participants who only had 2 readings or using the first measurement for participants who only had 1 reading). Information on the use of antihypertensive drugs was obtained from the interviewed questionnaire at home. Blood samples were collected at a mobile examination center and sent to a central laboratory for the assessment of blood lipids, fasting glucose, and hemoglobin A_1c. Data on the use of lipid-lowering medications, insulin, or oral hypoglycemic agents were obtained from the interviewed questionnaire at home. Changes for some biochemical indexes in measurement methods and used instruments over time were adjusted according to the official recommendation to make them comparable across different survey years [16]. The 8 behavioral and health factors and the thresholds used to allocate scores for each component in detail are shown in Table S2 in Multimedia Appendix 1. Scores for each component ranged from 0 to 100 points according to the AHA scoring algorithm [6], with a higher score meaning a better health status. The overall CVH score was calculated as the mean of the sum of all 8 metrics and similarly ranged from 0 (if the mean score of all components was 0) to 100 (optimal CVH). In this study, scores for the overall CVH and individual components were categorized into poor (0-50 points), intermediate (50-79 points), and high (80-100 points) status according to the AHA recommendation [6].

Differences in percentages of demographic characteristics across the 7 survey waves were assessed with the chi-square test. We calculated the age-standardized mean scores and their 95% CI for the overall CVH score and for each of the 8 CVH components. We performed stratified analysis by sex, age group, educational level, marital status, race or ethnicity, and family income category. We used linear regression model to estimate the linear trends between 2005 and 2018 with adjustments for sex, age group, educational level, marital status, and family income category if appropriate. Data were standardized for age using a direct standardization method based on the age distribution of the 2018 US population (20-39 years: 38.6%; 40-64 years: 44.6%; 65-79 years: 16.8%) [17]. Since oversampling was done in some particular subgroups of the total population in the NHANES to increase the reliability and precision of health indicators in the specific population, the appropriate sample weights, as well as strata and primary sampling units provided by the NHANES were used to make the data national representative of the US population. All statistical analyses were performed using Stata software (version 16.0, Stata Corp), and 2-sided P values of <.05 were considered statistically significant.

To our knowledge, this study is the first to examine the secular trends in CVH among US adults based on the new LE8 scoring algorithm proposed by the AHA. Using the 7 cross-sectional surveys between 2005 and 2018, the overall CVH score did not significantly change over time. However, the overall CVH status significantly worsened between 2005 and 2018 in older adults, Hispanic participants, and non-Hispanic White participants. There was no significant change over time for 3 CVH components (diet, PA, and BP) but 3 components (nicotine exposure, sleep health, and blood lipids) significantly improved, whereas 2 components (BMI and blood glucose) significantly, and markedly, worsened over time.

Although a number of public health and health care policies and programs have been implemented to prevent CVD since 2010 [18-20], the decline in age-standardized CVD mortality rate has slowed down or even stagnated considerably in recent years among US adults [2,21]. For example, the age-standardized mortality rate for heart disease declined between 1999 and 2010 by 8.3%, but only by 1.7% between 2010 and 2017 [21]. One important reason for this deceleration of the CVD decline may be that some CVD risk factors have not significantly improved or even deteriorated. This is consistent with other studies, which showed that the mean number of ideal CVH metrics based on the LS7 had not significantly changed in the past years [22,23]. For example, the Behavioral Risk Factor Surveillance System data showed a nonsignificant change in the mean number of ideal CVH metrics from 3.89 in 2011 to 3.96 in 2017 [22]. Another NHANES study among US adults showed that the mean number of ideal CVH components did not significantly change from 2007 to 2018 in both female participants (4.40 to 4.48) and male participants (3.97 to 3.93) [23]. In addition, previous studies among adults based on the LS7 also showed that there was no significant improvement in CVH among US adults [24,25]. Results of our study, which are based on the new LE8 scoring algorithm, are consistent with no change in CVH reported by the abovementioned studies based on the LS7, highlighting that there is still a long way to go to improve CVH among US adults.

The stable trends in the overall CVH score may mask disparities in trends in CVH components. We found that the mean scores of 4 CVH metrics including nicotine exposure (increase), BMI (decrease), non–high-density lipoprotein cholesterol (increase), and blood glucose (decrease) changed over time but not all in the same direction, and some (diet) did not change over time. Unhealthy diet is a major risk factor for CVD [26], but unfortunately, the dietary component gathered the lowest mean score [24]. Despite some improvement in some dietary components, the overall dietary quality did not significantly improve [27]. This emphasizes the need for interventions on those dietary components that tend to decline or do not improve over time. Actually, a number of interventions have already been implemented during the past years in the United States [28]. The smoking rate decreased from 24.8% in 1999-2000 to 18.1% in 2017-2018 [29], and this translated into improved mean LS7 scores of smoking from 1.53 in 2007-2010 to 1.60 in 2015-2018 [24], as well as in our study. Data from the NHANES showed that mean BMI increased from 28.0 kg/m² in 1999-2000 to 29.8 kg/m² in 2017-2018 [29], with the corresponding obesity prevalence dramatically increasing from 27.5% to 43.0% [30]. He et al [29] reported that the mean total cholesterol decreased from 203.3 mg/dL in 1999-2000 to 188.5 mg/dL in 2017-2018, which is consistent with our finding that the LE8 blood lipid component improved over time. Previous studies showed that the estimated prevalence of diabetes increased from 9.8% in 1999-2000 to 14.3% in 2017-2018 [31], consistent with the decreasing LS7 mean score of blood glucose from 1.42 in 2007-2010 to 1.29 in 2015-2018 [24], as well as the decreasing LE8 mean score observed in our study.

However, there were also several inconsistent findings in our study compared with previous studies regarding the secular trends in 2 CVH components (PA and BP) [24,32,33]. It is reported that the estimated prevalence of meeting the PA guideline target increased from 26.0% in 1998 to 37.4% in 2018 [32]. However, the overall mean score of PA in our study still remained unchanged and at a low level. As for hypertension, the prevalence as defined by the 2017 American College of Cardiology/AHA decreased (suggesting an improvement) from 48.1% in 1999-2000 to 44.1% in 2015-2016 [33]. However, another study reported that the mean BP score based on the LS7 slightly decreased (suggesting a deterioration) from 1.28 in 2007-2010 to 1.24 in 2015-2018 [24]. Inconsistent with the above 2 studies, we did not observe a significant change in the mean LE8 component score of BP between 2005 and 2018. In addition, the national data showed that more than half of the US population did not reach the sufficient sleep duration (20-64 years: 7-9 hours per day, 65-79 years: 7-8 hours per day) recommended by the National Sleep Foundation [34]. Inappropriate sleep duration (<6 hours or ≥9 hours) has been associated with an increased risk of CVD and related mortality [35]. Although previous studies reported a persistent deterioration of sleep quality in the US population from 2005 to 2018 [36,37], we observed an opposite trend with a significant increase (improvement) in the mean LE8 component score of sleep health from 2005 to 2018. These disparities might be partly due to methodological issues in assessing sleep quality across studies.

Consistent with previous studies [13,14], the overall LE8 score of CVH and its components differed markedly across the subgroups of sex, age, race or ethnicity, and socioeconomic status, with individuals who were female, younger, non-Hispanic White, and those with better socioeconomic status being more likely to have a better CVH. Although trends in the CVH metrics over time in these subgroups were mostly consistent with those in the overall population, there are some interesting findings. First, non-Hispanic White participants and the “Other” group (including mainly Asian participants) had a higher overall CVH score than non-Hispanic Black participants, but the difference decreased over time, partly due to deterioration in CVH over time among non-Hispanic White participants. The deterioration of 2 CVH components (diet and PA) among non-Hispanic White participants may be the reason for the worsening trend in overall CVH score in this subgroup. This highlights the need to consider LE8 trends according to racial or ethnic groups when designing CVD prevention and control policy and programs. Second, none of the CVH metrics significantly improved among older participants (65-79 years) over time. CVH usually markedly decreases with age [38], largely driven by the strong relation among hypertension, dyslipidemia, and type 2 diabetes with age. The CVH in older adults is a main challenge, as most hard CVD outcomes develop at an older age. In general, the less ideal CVH at all ages found in this study stresses the need for life course approaches to CVH and the need for high-risk individual-level approaches (clinical care) at middle and older ages. Third, the mean LE8 component of BMI (obesity) markedly deteriorated in all subgroups. This trend is particularly worrying, given the major role of adiposity on glucose metabolism, including insulin resistance, hypertension, and more generally poor CVH [39,40]. This highlights the need to strengthen preventive and medical approaches to weight gain prevention and control, including a broader use of novel effective clinical approaches [41]. Fourth, socioeconomic status such as educational level and family income are important social determinants of CVD [42]. Consistent with abundant literature on this topic, our results show that participants with higher socioeconomic status had much better CVH and the CVH gap (as assessed with the LE8 metrics) between individuals with low versus high socioeconomic status did not decrease over time, highlighting the need to further address social disparities in CVD prevention and control.

Our study is the first to examine the secular trends in CVH from 2005 to 2018 among US adults based on the nationally representative NHANES data using the new AHA LE8 metrics. However, several limitations should be mentioned. First, our study was based on data from 7 surveys over a rather short time period (<15 years). Second, data on diet, PA, nicotine exposure, and sleep health were obtained by self-report, which may be inaccurate due to recall and social desirability biases. Third, the 8 behavioral and health components are given equal weight (implying the same predictive performance) in calculating the LE8 overall CVH score. Fourth, although the 8 components constitute major or important risk/preventive factors for CVH, a number of other factors (eg, familial history/genetics, other conditions, or comorbidities) can largely alter CVD risk, particularly at the individual level. Fifth, missing values on relevant variables, especially for dietary variables, might have affected the representativeness of the study population. However, the inclusion of data on one 24-hour dietary recall showed similar results (age-standardized score did not significantly change from 40.46 in 2005-2006 to 40.47 in 2017-2018, P for trend=.77) with the primary ones.

In summary, based on the LE8 metrics, the overall CVH score did not significantly change from 2005 to 2018 among US adults. Very few individuals had CVH scores in good or optimal ranges, implying a large scope for improvement for most adults. Three CVH components remained unchanged over time (diet, PA, and BP), 3 improved (nicotine exposure, sleep health, and blood lipids), and 2 worsened (BMI and blood glucose), suggesting avenues for improving CVH status among US adults.