Life's Essential 8 and Poor Brain Health Outcomes in Middle-Aged Adults

Abstract

Background and Objectives

Mounting evidence points to a strong connection between cardiovascular risk during middle age and brain health later in life. The American Heart Association's Life's Essential 8 (LE8) constitutes a research and public health construct capturing key determinants of cardiovascular health. However, the overall effect of the LE8 on global, clinically relevant metrics of brain health is still unknown. We tested the hypothesis that worse LE8 profiles are associated with higher composite risk of the most important clinical endpoints related to poor brain health.

Methods

We conducted a two-stage (discovery and replication) prospective study using data from the UK Biobank (UKB) and All of Us (AoU), 2 large population studies in the United Kingdom and the United States, respectively. The primary exposure was the LE8 score, a validated tool that captures 8 modifiable cardiovascular risk factors (blood pressure, glucose, cholesterol, body mass index, smoking, physical activity, diet, and sleep duration), organized in 3 categories (optimal, intermediate, and poor). The primary outcome was a composite of stroke, dementia, or late-life depression. We evaluated associations using multivariable Cox proportional hazard models.

Results

The discovery stage included 316,127 UKB participants (mean age 56, 52% female). Over a mean (SD) follow-up time of 4.9 (0.4) years, the unadjusted risk of the composite outcome was 0.7% (95% CI 0.61–0.74), 1.2% (95% CI 1.11–1.22), and 1.8% (95% CI 1.70–1.91) in participants with optimal, intermediate, and poor cardiovascular health, respectively (p < 0.001). This association remained significant in multivariable Cox models (intermediate vs optimal cardiovascular health hazard ratio [HR], 1.37; 95% CI 1.24–1.52, and poor vs optimal cardiovascular health HR, 2.11; 95% CI 1.88–2.36, p trend <0.001). The replication stage included 68,407 AoU participants (mean age 56, 60% female). Over a mean (SD) follow-up time of 2.9 (1.41) years, the unadjusted risk of the composite outcome was 2.8% (95% CI 2.49–3.05), 6% (95% CI 5.76–6.22), and 9.7% (95% CI 9.24–10.24) in participants with optimal, intermediate, and poor cardiovascular health, respectively (p < 0.001). This association remained significant in multivariable Cox models (intermediate vs optimal cardiovascular health, HR 1.35; 95% CI 1.21–1.51, and poor vs optimal cardiovascular health, HR 1.94; 95% CI 1.72–2.18; p trend <0.001).

Discussion

Among middle-aged adults enrolled in 2 large population studies, poor cardiovascular health profiles were associated with two-fold higher risk of developing a composite outcome that captures the most important diseases related to poor brain health. Because the evaluated risk factors are all modifiable, our findings highlight the potential brain health benefits of using the Life's Essential 8 to guide cardiovascular health optimization.

Introduction

Brain health is defined as the absence of brain disease or the presence of a healthy state.¹ It is most important that brain health is paramount for the optimal well-being of every person, enabling us to function adaptively in the world. Unfortunately, clinical manifestations of poor brain health are on the rise.² Stroke stands out as the most burdensome neurologic disorder in the United States, accounting for 3.6 million disability-adjusted life years (DALYs), and is the second leading cause of death worldwide, with nearly 800,000 people in the United States experiencing a stroke annually.^3,4 Dementia, encompassing Alzheimer disease, vascular, and other dementias, is the second most burdensome neurologic condition in the United States, with 2.6 million DALYs,⁵ ranking as the fourth most common neurologic disorder in the nation, affecting more than 10% of adults aged older than 65 years. Late-life depression, a form of depression prevalent in older adults and strongly associated with cardiovascular risk factors,^6,7 not only leads to greater functional and cognitive impairments compared with depression in younger individuals but also imposes a considerable emotional and economic toll on patients and their families.^8,9

Mounting evidence indicates that cardiovascular health optimization during middle-age leads to brain health benefits in later life.^1,10 The American Heart Association has developed the Life's Essential 8 (LE8), a research and public health construct that captures modifiable contributors to cardiovascular health (blood pressure, non–high-density lipoprotein (HDL) blood cholesterol, blood glucose, smoking, physical activity, diet, body mass index (BMI), and sleep duration).¹¹ Although this framework has been instrumental in guiding cardiovascular research and health, a critical knowledge gap persists related to the overall effect of the LE8 on global metrics of brain health. There is an increasing need for a comprehensive, composite outcome that captures the most important brain diseases that affect a person's quality of life and survival. Our study aims to fill this gap by examining the association between overall cardiovascular health, as denoted by the LE8, and a novel composite outcome of poor brain health-related conditions.

We leveraged data from 2 large population studies to examine the association between the LE8 and a comprehensive composite outcome of clinically meaningful poor brain health-related diseases, including stroke, dementia, and late-life depression. We included late-life depression because there is ample evidence linking this debilitating disorder to cerebrovascular disease, cardiovascular risk factors (i.e., hypertension), and known markers of cerebral small vessel disease (i.e., white matter hyperintensities).^6,7 We hypothesize that poor cardiovascular health, as denoted by the LE8, during middle age, is strongly associated with a higher incidence of a composite outcome that comprehensively encompasses the end-stage clinical manifestations of long-term poor brain health.

Methods

Study Design

We conducted a 2-stage (discovery and replication) observational study using individual-level data from the UK Biobank (UKB) (discovery) and All of Us (AoU) (replication). The UK Biobank is an ongoing, large population-based cohort study based in the United Kingdom that enrolled more than 500,000 Britons from 2006 to 2010. The NIH All of Us Research Program is also an ongoing prospective cohort study. It is being conducted in the United States, and it aims to enroll 1 million Americans, with approximately 400,000 participants enrolled since 2018. For those who consent, both the UKB and the AoU collect electronic health records, genomic, physical measurements (i.e., blood pressure and body mass index), blood draws (i.e., blood glucose and cholesterol), and survey questionnaire data (i.e., smoking). Further details and description about the UKB's and AoU's study protocols are available elsewhere.^12,13 To maintain our focus in primary prevention, we excluded all participants with a previous diagnosis of stroke, dementia, or depression. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Standard Protocol Approvals, Registrations, and Patient Consents

The institutional review boards for both the UKB and the AoU approved their study protocols, respectively. All participants or their legally designated surrogates provided written informed consent.

Exposure Ascertainment and Modeling

Our primary exposure was the LE8, a well-established research and public health construct that captures 8 modifiable contributors to cardiovascular health: blood pressure, non-HDL blood cholesterol, blood glucose, smoking, physical activity, diet, BMI, and sleep duration. The specific definition for each component as per the American Heart Association's guidelines is summarized in eTable 1.¹¹ We modeled the LE8 using an approach previously validated in population cohorts different from the ones used in this study that entails creating a cardiovascular health score that ranges from 0 (worst) to 100 (best).^1,2 We calculated the score for each component of the LE8 and then calculated the overall total LE8 score for each participant as the unweighted average of the 8 components. We then followed a similar approach to previous research^1,2 and organized the overall LE8 score into categories of <20, 20–80, and >80, where <20 represents the fraction of study population with worst (poor) cardiovascular health and >80 the study population with the best (optimal) cardiovascular health. Details about the protocols followed in the UKB and AoU to obtain self-reported information, electronic health records, blood pressure measurements, and laboratory data are available elsewhere.^12,13 Briefly, clinical and biometric data were obtained during the baseline interview, while laboratory data were obtained from blood draws performed during the baseline visit. In the UKB, we ascertained the LE8 scores using values for blood pressure and blood pressure medication, non-HDL blood cholesterol and lipid-lowering medication, glycemic status (fasting blood glucose, hemoglobin A1c, and diabetes mellitus diagnosis), smoking and nicotine exposure (smoking status, years of abstinence, and passive household smoking exposure), exercise (minutes of moderate and vigorous activity in a week), diet (using the Dietary Approaches to Stop Hypertension tool),^14,15 BMI, and sleep duration (self-reported hours of sleep per day, including day-time napping), all measured at the time of enrollment.¹¹ In AoU, we calculated the scores for the LE8 using the same protocol for blood pressure, glucose, and cholesterol; BMI; and smoking and nicotine exposure. Because of data constraints within the AoU replication cohort, we were not able to calculate the diet, physical activity, and sleep duration scores. Thus, the LE8 score was calculated as the average sum of the blood pressure, blood cholesterol, blood glucose, BMI, and smoking scores for the replication.

Outcome Ascertainment and Modeling

Our primary outcome was the composite risk of stroke (defined as any ischemic or hemorrhagic stroke), dementia, or late-life depression, all first-ever events. Late-life depression was included because it is increasingly recognized that cerebrovascular disease may predispose, precipitate, or perpetuate this disease.^6,7 Our secondary outcomes were incident stroke, dementia, and late-life depression considered separately. Stroke and dementia are part of the UKB's algorithmically defined variables, which were ascertained through a combination of self-reported data obtained during the baseline visit, death registry, and previously validated International Statistical Classification of Diseases and Related Health Problems, Ninth and Tenth Revision (ICD-9 and 10)¹⁶ codes abstracted from electronic health records of hospital admissions and primary care visits that took place between enrollment and the last participant follow-up (eTable 2). Incident late-life depression was ascertained using validated first occurrence F-32 (single episode of major depressive disorder) and F-33 (major depressive disorder, recurrent) ICD-10 codes after 60 years of age (eTable 2). In AoU, stroke, dementia, and late-life depression were diagnosed using a similar combination of diagnostic codes as for the UKB (eTable 3) obtained through chart abstraction of hospital admissions and outpatient visits. To ensure direct comparability between our discovery and replication analyses, follow-up for both cohorts was restricted to 5 years. This is because 5 years is the maximum available amount of follow-up for AoU participants, in the current AoU release version 7.

Covariates

Demographic information was acquired during baseline interviews, for both the UKB and the AoU.^12,13 Race and ethnicity were defined as Asian, Black, and White in the UKB, and as Asian, Black, Hispanic, and White in AoU, based on participants' self-identification at baseline questionnaires. Comorbidities (myocardial infarction, atrial fibrillation, heart failure, and aortic valve disease) were ascertained using validated ICD-9 and 10 in UKB, and using validated ICD-9 and 10 and Systematized Nomenclature of Medicine¹⁷ codes in AoU (eTable 1 and 2). Myocardial infarction is part of the UKB's algorithmically defined variables (see section ‘Outcome ascertainment and modelling’). Socioeconomic deprivation, a comprehensive social determinant of health known to associate with worse cardiovascular disease incidence,¹⁸ was ascertained as the Townsend deprivation index in the UKB,¹⁹ which was calculated centrally by the UKB research team. The Townsend deprivation index uses 4 variables: percentage of persons over 16 not employed, percentage of all households not owning a car, percentage of all households not owning a house, and household overcrowding. In AoU, deprivation was ascertained using a previously validated deprivation index that uses 5 census tract level variables derived from the 2015 American Community Survey²⁰: fraction of the population below the poverty level, median household income, fraction of the population 25 years and older with high-school education, fraction of the population without health insurance, fraction of households receiving public assistance income, and fraction of houses that are vacant in the individuals' neighborhood.

Statistical Analyses

Discrete variables are presented as counts (percentages [%]) and continuous variables as means (SD). We used Pearson Chi-square tests to evaluate unadjusted associations between categories of the LE8 and the risk of the primary (composite risk of stroke, dementia, and late-life depression) and secondary outcomes of interest (each disease considered separately), and used pairwise proportion tests to calculate 95% confidence intervals for them. Subsequently, we conducted survival analyses using Kaplan-Meier survival curves with associated log-rank tests followed by multivariable Cox proportional hazard models (reporting hazard ratios [HR] and their 95% CIs) to test for association between categories of the LE8 and the primary and secondary outcomes, after adjusting for potential confounders. We fit 3 different models: for model 1, we included demographic covariates (age, sex, race, and ethnicity); for model 2, we included demographic covariates plus cardiovascular (history of myocardial infarction, as a proxy for poor cardiovascular health; atrial fibrillation because it accounts for almost 50% of cardioembolic strokes; and heart failure and aortic valve disease because they have been shown to associate with cognitive impairment)^21-23 and psychiatric comorbidities (history of anxiety, bipolar, substance abuse disorders, and schizophrenia because they are known to be associated with both dementia and depression^24-28); for model 3, we included demographic covariates, comorbidities, and socioeconomic deprivation. The proportional hazards assumption was checked using Schoenfeld residuals. All analyses used a complete case approach, where study participants without data for the exposure, outcome, or covariates of interest were excluded. We then conducted 2 sensitivity analyses: first, both in the UKB and in the AoU, we tested for interaction between race and ethnicity and the LE8 score for the primary composite outcome and each secondary outcome, using product terms; second, in the UKB, we repeated the primary analysis using the LE8 score as exposure with only the 5 LE8 components available in AoU (blood pressure, glucose, non-HDL cholesterol, body mass index, and smoking) to verify how the lack of these other 3 components (physical activity, sleep, and diet) was affecting our results. Finally, in secondary analysis, we evaluated the independent contribution of each LE8 factor separately for the primary and secondary outcomes in the UKB, fitting a multivariable Cox regression model adjusting for age, sex, and race and ethnicity, including each separate component of the LE8. For this analysis, and the sensitivity analysis that tested for interaction between race and ethnicity and the LE8 score, we modelled the score of each LE8 component (exposure) as a continuous variable, standardized and inverted to model the increase in cardiovascular risk per 1 SD increase in the score. We set statistical significance at p < 0.05 (2-tailed) for the single association test required to test our hypothesis that poorer LE8 scores are associated with a higher composite risk of stroke, dementia, or late-life depression. All analyses were conducted using R statistical software version 4.1.3 (R Project for Statistical Computing) between January 2023 and February 2024.

Data Availability

The data from the UKB analyzed in our study are available to researchers on completion of the registration process and on successful application.²⁹ The UKB access committee approved this analysis as part of project 58743. Data from the All of Us Research Program are publicly available.³⁰ We used release version 7, which comprises data from all participants who enrolled from the beginning of the program on May 30, 2017, to June 23, 2022.

Results

The UKB enrolled 502,389 participants. As shown in Figure 1A, 154,163 were excluded because of missing values among the components of the LE8 score and 32,099 were excluded for having prevalent outcome events. Our final analytical sample included 316,127 middle-aged adults. This group had a mean (SD) age of 56 (SD 8) years, and 52% were female. As shown in eFigure 1A, LE8 scores were normally distributed before standardization, with a mean (SD) of 70.2 (11). Following previous research, participants were divided into 3 LE8 categories: 60,734 with poor (LE8 <20), 190,919 with intermediate (20> LE8 <80), and 64,474 with optimal (LE8 >80) cardiovascular health (Table 1).

Figure 1. Flowchart Describing the Steps Taken to Obtain the Final Analytical Samples.

Table 1.

Baseline Characteristics of the UK Biobank and All of Us Study Populations

Variable	UK Biobank (n = 316,127)			All of Us (n = 68,407)
	Optimal LE8	Intermediate LE8	Poor LE8	Optimal LE8	Intermediate LE8	Poor LE8
Number of participants	64,474	190,919	60,734	13,681	41,044	13,682
Demographics
Age (mean [SD])	54 (8)	57 (8.)	57 (8)	49 (17)	58 (16)	60 (13)
Male sex, n (%)	20,511 (32)	95,476 (50)	36,767 (61)	3,498 (26)	16,618 (40)	6,034 (44)
Race/ethnicity, n (%)
White	61,250 (95)	181,017 (95)	57,444 (95)	8,881 (65)	24,090 (59)	6,638 (48)
Non-White
Black	655 (1.0)	2,874 (1.5)	1,133 (1.9)	1,174 (8.6)	7,201 (18)	4,168 (31)
Asian	1,531 (2.4)	3,842 (2.0)	1,067 (1.8)	965 (7.1)	1,096 (2.7)	142 (1.0)
Hispanic	—	—	—	2,043 (15)	6,660 (16)	2,073 (15)
Vascular risk factors, n (%)
Hypertension	7,315 (11.4)	49,382 (26)	22,936 (38)	2,586 (19)	23,226 (57)	11,249 (82)
Diabetes	765 (1.2)	7,270 (4)	6,268 (10.3)	488 (3.6)	8,527 (21)	8,057 (59)
Hyperlipidemia	5,646 (8.8)	26,723 (14.0)	11,148 (18.4)	4,376 (32)	25,198 (61)	10,919 (80)
Smoking
Current	468 (0.7)	12,742 (6.7)	17,125 (28.2)	0	3,212 (7.8)	4,067 (30)
Former	15,039 (23.3)	70,944 (37.2)	24,226 (39.9)	1,004 (7.3)	11,640 (28)	5,839 (43)
Comorbidities, n (%)
Atrial fibrillation	749 (1.2)	2,989 (1.6)	1,083 (1.8)	562 (4.1)	3,649 (8.9)	1,693 (12)
Myocardial infarction	856 (1.3)	3,956 (2.1)	1,816 (3)	250 (1.8)	2,598 (6.3)	1,669 (12)
Heart failure	153 (0.2)	787 (0.4)	446 (0.7)	261 (1.9)	2,506 (6.1)	1,708 (12.5)
Aortic valve disease	118 (0.2)	484 (0.3)	199 (0.3)	102 (0.8)	600 (1.5)	174 (1.3)
Schizophrenia	40 (0.1)	170 (0.1)	152 (0.3)	<20a	87 (0.2)	47 (0.3)
Bipolar disorder	96 (0.2)	390 (0.2)	208 (0.3)	72 (0.5)	466 (1.1)	290 (2.1)
Anxiety disorders	1,405 (2.2)	4,426 (2.3)	1,507 (2.5)	1,575 (11.5)	4,757 (11.6)	1,682 (12.3)
Substance abuse disorder	89 (0.1)	632 (0.3)	504 (0.8)	122 (0.9)	775 (1.9)	451 (3.3)

^aPer the All of Us Research Program's policy, all numbers <20 are to be published as “<20” without an accompanying % to reduce the risk of re-identification.

Abbreviation: LE8 = Life's Essential 8.

Association Between LE8 and Composite Outcome

Over a mean (SD) follow-up time of 5 (0.4) years, the unadjusted cumulative incidence of the poor brain health composite outcome was 1.2% (95% CI 1.15–1.23 [3,753 events]). This unadjusted cumulative incidence across the optimal, intermediate, and poor LE8 categories was 0.7% (95% CI 0.61–0.74 [434 events]), 1.2% (95% CI 1.11–1.22 [2,227 events]), and 1.8% (95% CI 1.70–1.91 [1,092 events]), respectively (unadjusted p value <0.001, Table 2, Figure 2A). These differences remained significant in multivariable Cox proportional hazard regression analyses adjusting for age, sex, race, ethnicity, cardiovascular and psychiatric comorbidities, and social deprivation (model 3, Table 3): compared with optimal cardiovascular health, intermediate and poor cardiovascular health was associated with a 37% (HR 1.37; 95% CI 1.24–1.52) and more than two-fold (HR 2.11; 95% CI 1.88–2.36) higher risk of presenting an event within the poor brain health composite, respectively (p for trend across both comparisons <0.001). Other models (Models 1 and 2) were equally significant (Table 3). Schoenfeld residual results confirmed the Cox proportional hazard assumptions and are presented in eFigure 2. Sensitivity analyses showed no significant interactions between the LE8 and racial and ethnic categories (all p > 0.05) for the composite outcome, stroke, dementia, and late-life depression (eFigures 3 and 4A). Finally, modelling the LE8 score using only the 5 components available in AoU (blood pressure, glucose, non-HDL cholesterol, BMI, and smoking) showed similar results to the primary analysis (eTable 4).

Table 2.

Unadjusted Associations for Both Primary and Secondary Outcomes in the Discovery Stage (UK Biobank), and the Replication Stage (All of Us)

Outcome	Life Essential 8 Optimal	Life Essential 8 Intermediate	Life Essential 8 Poor	p Valuea
Discovery—all participants, % (95% CI)
Composite	0.7 (0.61–0.74)	1.2 (1.11–1.22)	1.8 (1.70–1.91)	<0.001
Stroke	0.3 (0.22–0.31)	0.5 (0.48–0.54)	0.8 (0.74–0.88)	<0.001
Dementia	0.06 (0.046–0.086)	0.10 (0.091–0.12)	0.14 (0.11–0.18)	<0.001
Late-life depression	0.4 (0.33–0.43)	0.6 (0.55–0.62)	0.9 (0.83–0.99)	<0.001
Replication—all participants, % (95% CI)
Composite	2.8 (2.49–3.05)	6 (5.76–6.22)	9.7 (9.24–10.24)	<0.001
Stroke	0.7 (0.58–0.86)	1.7 (1.61–1.19)	3.4 (3.10–3.70)	<0.001
Dementia	0.7 (0.61–0.90)	1.3 (1.12–1.41)	1.6 (1.36–1.79)	<0.001
Late-life Depression	1.5 (1.29–1.71)	3.6 (3.42–3.79)	5.9 (5.51–6.31)	<0.001

^aPearson χ² test. 95% CI were calculated using pairwise proportion tests.

Figure 2. Kaplan-Meier Curves Showing Composite Outcome Events for (A) the Discovery Stage in the UK Biobank and (B) the Replication Stage in All of Us.

LE8 = Life's Essential 8.

Table 3.

Cox Proportional Hazard Analysis Results for the Composite Outcome in Both the Discovery (UK Biobank) and Replication (All of Us) Stages

Study stage	Covariate model	Life's essential 8 category	Composite outcome
			HR (95% CI)	p Trend
Discovery stage UK Biobank	Model 1: age, sex, race, and ethnicity	Optimal	Reference	<0.001
		Intermediate	1.39 (1.26–1.55)
		Poor	2.26 (2.02–2.52)
	Model 2: model 1 + comorbidities	Optimal	Reference	<0.001
		Intermediate	1.38 (1.25–1.54)
		Poor	2.19 (1.96–2.45)
	Model 3: model 2 + social deprivation	Optimal	Reference	<0.001
		Intermediate	1.37 (1.24–1.52)
		Poor	2.11 (1.88–2.36)
Replication stage All of Us	Model 1: age, sex, race, and ethnicity	Optimal	Reference	<0.001
		Intermediate	1.42 (1.27–1.58)
		Poor	2.20 (1.96–2.48)
	Model 2: model 1 + comorbidities	Optimal	Reference	<0.001
		Intermediate	1.35 (1.21–1.51)
		Poor	1.95 (1.73–2.19)
	Model 3: model 2 + social deprivation	Optimal	Reference	<0.001
		Intermediate	1.35 (1.21–1.51)
		Poor	1.94 (1.72–2.18)

Abbreviation: HR = hazard ratio.

Association Between LE8 and Incident Stroke, Dementia, and Late-Life Depression

The unadjusted incidence for stroke, dementia, and late-life depression also differed significantly across cardiovascular health categories (Table 2). The cumulative number of separate outcomes of stroke, dementia, and late-life depression during the total 5 years of follow-up is shown in eFigure 5A. Multivariable Cox proportional hazard models adjusting for age, sex, race, ethnicity, myocardial infarction, atrial fibrillation, and social deprivation showed that poorer cardiovascular health was associated with increased risk of experiencing each of these outcomes (p for trend across all comparisons <0.001 for stroke and late-life depression and 0.02 for dementia, eTable 5). Other covariate models were also significant (eTable 5).

Replication in All of Us

In All of Us, among the 413,477 participants, 68,407 had available data for analysis after relevant exclusions (Figure 1). As for the UKB and shown in eFigure 1B, LE8 scores were normally distributed before standardization, with a mean (SD) of 73 (16). Over a mean (SD) follow-up time of 2.9 (1.41) years, the unadjusted cumulative incidence of the poor brain health composite outcome was 2.8% (95% CI 2.50–3.10 [378 events]), 6% (95% CI 5.76–6.22 [2,456 events]), and 9.7% (95% CI 9.24–10.24 [1,331 events]) for participants with optimal, intermediate, and poor cardiovascular health, respectively (unadjusted p value <0.001). In the fully adjusted multivariable Cox proportional hazard regression analysis (model 3), participants with intermediate cardiovascular health had a 35% increased risk of the poor brain health composite outcome (HR 1.35; 95% CI 1.21–1.51), while those with poor cardiovascular health had almost twice the risk of experiencing the poor brain health composite outcome (HR 1.94; 95% CI 1.72–2.18), relative to participants with optimal cardiovascular health (p for trend across both comparisons <0.001). Results from the other models were comparable (Table 3). Figure 2B shows the cumulative number of poor brain health composite outcomes during the 5-year follow-up period. The unadjusted incidence for stroke, dementia, and late-life depression differed significantly across cardiovascular health categories, as summarized in Table 2 (unadjusted p value <0.001). The fully adjusted multivariable Cox proportional hazard model showed that poorer cardiovascular health was significantly associated with increased risk of experiencing stroke and late-life depression when compared with optimal cardiovascular health (eTable 4) but was not significantly associated with incident dementia (all models p > 0.4). eFigure 5B shows the cumulative number of stroke, dementia, and late-life depression events during the 5-year follow-up period in the replication cohort. Finally, in sensitivity analysis, there was no evidence of significant interactions between the LE8 and racial and ethnic categories (all p > 0.05) for the composite outcome, stroke, dementia, and late-life depression (eFigure 4B).

Secondary Analysis

Figure 3 presents the association results between the individual components of the LE8 and both the primary and secondary outcomes of interest. Except for non-HDL cholesterol and blood pressure, all LE8 components were linked to the combined risk of stroke, dementia, and disability. The strongest associations with the composite outcome were observed for smoking, sleep duration, and BMI. Notably, evaluating each component of the composite outcome separately, revealed blood pressure as a major determinant of stroke risk, underscoring the well-established link between high blood pressure and cerebrovascular disease. For dementia, the most significant factors were sleep duration and physical activity, while smoking and sleep duration were paramount for late-life depression. Of interest, higher blood pressure levels were associated with a lower risk of late-life depression. This counterintuitive finding suggests cautious interpretation, proposing that individuals with higher blood pressures might receive better medical attention, including more frequent clinical visits for blood pressure management and antihypertensive medication monitoring.

Figure 3. Forest Plot Comparing the Risk of Sustaining a Primary and Secondary Outcome Event, Among Each Separate Component of the Life's Essential 8 (LE8), in the UK Biobank (Discovery).

LE8 was modelled as a continuous, standardized variable. HDL = high-density lipoprotein; OR = odds ratio. Adjusted for age, sex, and ethnicity.

Discussion

We conducted a large observational study to evaluate the extent to which cardiovascular health optimization during middle age influences brain health later in life. We ascertained cardiovascular health using the American Heart Association LE8, a well-validated and widely used research and public health tool intended to capture cardiovascular health (not disease) through 8 known modifiable risk factors for cardiovascular disease. Following recent statements from the American Heart Association and the American Academy of Neurology,^31-33 we used a novel and clinically relevant composite poor brain health outcome that captures a collection of highly prevalent neurologic conditions that significantly modify quality of life and overall survival. We found that, in ∼400,000 middle-aged adults without a history of cerebrovascular disease and depression, poorer LE8 profiles were significantly associated with an increased risk of a composite outcome of incident stroke, dementia, or late-life depression. It is important that among the different components of the LE8, smoking, sleep, and body mass index were the most strongly associated with the composite of poor brain health. Finally, this increased risk was observed equally across all racial and ethnic groups.

Our study provides important new evidence related to the role of cardiovascular health optimization in improving long-term outcomes that capture the full spectrum of brain functions and associated wellbeing. Along these lines, recent statements by the American Heart Association and the American Academy of Neurology emphasize the need to advance research focused on brain health considered broadly. Approximately 1 in 5 Americans aged 65 years or older live with mild cognitive impairment and 1 in 7 have dementia.³⁴ It is important that it is estimated that by 2050, the number of persons with dementia will triple.³⁵ In line with this new research framework and public health necessity, we evaluated a composite outcome that in addition to stroke, which is well-known to relate to cardiovascular risk factors, also includes dementia and late-life depression. The connection between cardiovascular health and dementia is well illustrated by SPRINT-MIND,³⁶ a large clinical trial that evaluated intensive blood pressure reduction vs standard treatment among almost 10,000 adults 50 years or older with hypertension. They found that aggressive blood pressure reduction significantly reduced the risk of a composite of mild cognitive impairment and probable dementia. In line with these data, our study found that poor cardiovascular health was associated with an increased risk of developing dementia. However, these results were not replicated in AoU, possibly due to power limitations and to the absence of the sleep and physical activity components in the AoU LE8, which our analysis indicates are key drivers of the association between LE8 and dementia risk in the UKB.

One important novel component of the new paradigm of poor brain health is the inclusion of late-life depression as one of its elements. Of note, this disease may not be immediately perceived as being caused by cardiovascular mechanisms. However, there now exists ample evidence for the ‘vascular depression’ hypothesis,⁷ which states that cerebrovascular disease may predispose, precipitate, or perpetuate late-life depression: first, late-life depression has been associated with greater white matter hyperintensity severity and volumes,^37,38 a well-known neuroimaging marker of cerebral small vessel disease; second, hypertension and blood pressure variability strongly associate with late-life depression^39,40; third, renin-angiotensin system allelic variants have been shown to increase depression risk⁴¹ and modulate monoamine metabolite concentrations in CSF⁴¹ besides modulating vascular function, and finally, cognitive deficits, a main finding of late-life depression, are not only strongly associated with vascular risk factors and white matter hyperintensity severity but also predict poor antidepressant response.⁷ Our findings support the hypothesis that poor cardiovascular health is linked to late-life depression and that late-life depression alongside with stroke, drive the association between poor cardiovascular health and the composite outcome of poor brain health. This might be due to the higher incidence of late-life depression and stroke compared with that of dementia, highlighting the importance of a comprehensive assessment of poor brain health.

It is critical to emphasize that the LE8 was designed to, by definition, capture modifiable risk factors. Our work provides important new results on overall brain health that bring a new perspective for possible users of programs aimed to optimize cardiovascular health using the LE8. Specifically, most of the evidence collected thus far for the LE8 has focused on isolated or combined cardiovascular endpoints like myocardial infarction and stroke.^42,43 Given the rising interest of health care providers and the general population in overall well-being as we age, our study highlights an area where LE8 optimization can result in significant brain health benefits. Along these lines, all 3 components of our composite outcome of interest carry significant morbidity and mortality in older adults. Concretely, almost 20% of all stroke patients are older adults ≥85 years,⁴⁴ almost 11% of older adults ≥65 years have dementia,³¹ and 10–50% of older adults ≥65 years will manifest clinically significant depressive symptoms.⁶ Although our findings recapitulate this, they also illustrate the complexity of the associations between cardiovascular health and poor brain health. In our secondary analysis, we found no associations between blood pressure, as well as non-HDL cholesterol and the composite outcome. This might be due to the inclusion of 3 heterogenous and pathophysiologic distinct disorders, as a single-composite measure. Furthermore, we found no association between non-HDL cholesterol and stroke. This might be related to the fact that in some previous studies, higher levels of non-HDL cholesterol have been associated with a lower risk intracerebral hemorrhage.^45,46 However, future research is needed to understand the complex relationship between cholesterol levels and brain health.

Our results point to important questions that should be tackled by follow-up research. First, the observed associations between adverse LE8 profiles and poor brain health outcomes point to an opportunity for deploying early and comprehensive brain health optimization programs that aim to tackle several of the LE8 components, with a potential strong emphasis on smoking cessation, sleep optimization, and body mass index reduction. Along these lines, the SPRINT-MIND study³⁶ showed important benefits when focusing on only 1 LE8 component (blood pressure) and 1 poor brain health outcome (cognition and dementia). In addition, it is imperative to further understand how social determinants of health, like race, ethnicity, and social deprivation influence and modify the link between cardiovascular risk factors and poor brain health. In our study, we found no evidence of statistical interaction between the LE8 score and each primary and secondary outcome. However, because the interplay between social determinants of health and brain health is complex,⁴⁷ future research that takes into consideration these additional variables is needed. These future investigations are important because they could potentially identify high risk groups that will benefit from comprehensive interventions tackling the most important components of the LE8.

Our study has some limitations. First, secondary to data constraints within AoU, we were unable to use sleep, physical activity, and diet as part of the LE8 score within the replication stage. Nevertheless, sensitivity analysis results showed that modelling the LE8 score without these 3 components in the UKB led to similar results, with slightly smaller estimates, indicating that the estimates from the replication might be somewhat underestimated. Second, because of the known healthy volunteer bias inherent of the UKB, results may not be directly generalizable to the broad population.⁴⁸ Third, socioeconomic deprivation was assessed using different methods in the UKB and AoU, which may have influenced our findings. Although the deprivation index used in AoU includes more domains than the Townsend deprivation index (UKB), the Townsend index has been shown to correlate well with other modern deprivation indexes that have a similar number and type of domains as the one used in AoU.⁴⁹ Despite this methodological difference, we could replicate our results successfully.⁴⁹ Fourth, the LE8 score was calculated based on single measures at baseline; thus, we did not account for potential lifestyle modifications during the 5-year follow-up. However, previous literature has shown that during a 4-year follow-up period, cardiovascular health profiles ascertained using the Life's Simple 7 metric (the previous version of the LE8) had little variation.⁵⁰ Finally, the strengths of our study include its large sample size and prospective design, the use of a replication stage to validate results, and the use of a comprehensive, clinically relevant, poor brain health composite outcome. It is important to note that a considerable strength of this study lies in the consistency of results despite differences in study location, prevalence of comorbidities, baseline cardiovascular profiles, and outcome incidence across both discovery and replication stages (i.e., participants with poor cardiovascular health in the UKB had a 39% prevalence of hypertension in comparison with 82% in AoU participants of the same group), results were strikingly similar.

In conclusion, this large prospective study found that poorer American Heart Association's LE8 profiles significantly increase a composite risk of stroke, dementia, and late-life depression. These findings underscore the strong association between cardiovascular health during middle age, ascertained by the LE8, and a composite outcome of the most important diseases related to poor brain health. These findings support the utilization of this composite outcome in clinical trials focused on brain health and the utilization of the LE8 as a tool to potentially improve overall brain health.

Glossary

AoU

All of Us

BMI

body mass index

DALY

disability-adjusted life year

HDL

high-density lipoprotein

HR

hazard ratio

LE8

Life's Essential 8

UKB

UK Biobank

Appendix. Authors

Name	Location	Contribution
Santiago Clocchiatti-Tuozzo, MD, MHS	Department of Neurology; Yale Center for Brain and Mind Health; Department of Internal Medicine, Geriatrics, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data
Cyprien A. Rivier, MD, MSc	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Daniela Renedo, MD	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Shufan Huo, MD, PhD	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Maximiliano A. Hawkes, MD	Department of Neurology, Mayo Clinic, Rochester, MN	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Adam de Havenon, MD, MSc	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Lee H. Schwamm, MD	Department of Neurology, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Kevin N. Sheth, MD	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Thomas M. Gill, MD	Department of Internal Medicine, Geriatrics, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data
Guido J. Falcone, MD, ScD, MPH	Department of Neurology; Yale Center for Brain and Mind Health, Yale School of Medicine, New Haven, CT	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data

Study Funding

S. Clocchiatti-Tuozzo is funded by NIH T32 AG019134 and together with T.M. Gill are funded by P30 AG021342. C.A. Rivier is supported by the American Heart Association (817874) and the AAN/AHA Ralph L. Sacco Scholars Fellowship (https://doi.org/10.58275/AHA.24RSSPOST1328228.pc.gr.197089). A. de Havenon reports NIH/NINDS funding (K23NS105924, UG3NS130228, and R01NS130189). A. de Havenon has received consultant fees from Integra and Novo Nordisk, royalty fees from UpToDate, and has equity in TitinKM and Certus.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.