Cardiovascular health metrics from mid- to late-life and risk of dementia: A population-based cohort study in Finland.

BACKGROUND: Very few studies have explored the patterns of cardiovascular health (CVH) metrics in midlife and late life in relation to risk of dementia. We examined the associations of composite CVH metrics from midlife to late life with risk of incident dementia. METHODS AND
FINDINGS: This cohort study included 1,449 participants from the Finnish Cardiovascular Risk Factors, Aging, and Dementia (CAIDE) study, who were followed from midlife (baseline from1972 to 1987; mean age 50.4 years; 62.1% female) to late life (1998), and then 744 dementia-free survivors were followed further into late life (2005 to 2008). We defined and scored global CVH metrics based on 6 of the 7 components (i.e., smoking, physical activity, and body mass index [BMI] as behavioral CVH metrics; fasting plasma glucose, total cholesterol, and blood pressure as biological CVH metrics) following the modified American Heart Association (AHA)'s recommendations. Then, the composite global, behavioral, and biological CVH metrics were categorized into poor, intermediate, and ideal levels. Dementia was diagnosed following the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria. Data were analyzed with Cox proportional hazards and the Fine and Gray competing risk regression models. During the follow-up examinations, dementia was diagnosed in 61 persons in 1998 and additional 47 persons in 2005 to 2008. The fully adjusted hazard ratio (HR) of dementia was 0.71 (95% confidence interval [CI]: 0.43, 1.16; p = 0.174) and 0.52 (0.29, 0.93; p = 0.027) for midlife intermediate and ideal levels (versus poor level) of global CVH metrics, respectively; the corresponding figures for late-life global CVH metrics were 0.60 (0.22, 1.69; p = 0.338) and 0.91 (0.34, 2.41; p = 0.850). Compared with poor global CVH metrics in both midlife and late life, the fully adjusted HR of dementia was 0.25 (95% CI: 0.08, 0.86; p = 0.028) for people with intermediate global CVH metrics in both midlife and late life and 0.14 (0.02, 0.76; p = 0.024) for those with midlife ideal and late-life intermediate global CVH metrics. Having an intermediate or ideal level of behavioral CVH in both midlife and late life (versus poor level in both midlife and late life) was significantly associated with a lower dementia risk (HR range: 0.03 to 0.26; p < 0.05), whereas people with midlife intermediate and late-life ideal biological CVH metrics had a significantly increased risk of dementia (p = 0.031). Major limitations of this study include the lack of data on diet and midlife plasma glucose, high rate of attrition, as well as the limited power for certain subgroup analyses.
CONCLUSIONS: In this study, we observed that having the ideal CVH metrics, and ideal behavioral CVH metrics in particular, from midlife onwards is associated with a reduced risk of dementia as compared with people having poor CVH metrics. Maintaining life-long health behaviors may be crucial to reduce late-life risk of dementia.

Introduction

Dementia has been recognized as a global public health priority owing to its tremendous economic and societal burden [1]. Despite massive global efforts in the past 4 decades, no cure or even a disease-modifying therapy has been developed for dementia. Encouragingly, epidemiological studies have identified a range of life-long modifiable risk factors for dementia; of them, cardiovascular risk factors from midlife onwards have been shown to play a pivotal role in the development and onset of dementia [2]. Furthermore, simulation research has estimated that up to one-third of global dementia cases might be attributable to these modifiable risk factors [3]. Indeed, numerous epidemiological studies have reported that major cardiovascular risk factors (e.g., smoking and diabetes) and cardiovascular disease (e.g., heart failure and atrial fibrillation) are associated with an elevated risk of dementia [2,4-7]. The associations of cardiometabolic risk factors (e.g., high blood pressure, obesity, and high cholesterol) with the risk of dementia may vary with age from midlife to later in life [8-10]. However, very few studies have explored the potential associations of optimal cardiovascular health (CVH) metrics from mid- to late life with the risk of dementia. This is of high relevance from the public health perspective.

In 2010, the American Heart Association (AHA) defined ideal CVH based on 7 risk factors, also known as Life’s Simple 7, to promote CVH [11]. These CVH metrics include 4 behavioral metrics (e.g., smoking, physical activity, diet, and body weight) and 3 biological metrics (e.g., blood glucose, blood cholesterol, and blood pressure) [11]. The CVH metrics have been shown to be a useful tool for predicting cardiovascular events because numerous studies have linked ideal CVH metrics with a lower risk of coronary heart disease and stroke [12,13].

The AHA’s CVH metrics approach may help achieve healthy brain aging [14]. Having the ideal CVH metrics in midlife has been associated with a lower risk of dementia in late life [15,16]. However, population-based studies investigating the association between late-life composite CVH metrics and risk of dementia have yielded mixed results; some studies suggested an association between a higher CVH metric score in late life and a reduced risk of dementia [17,18], whereas others showed no association [19]. However, most of the previous studies have examined the CVH metrics assessed in either midlife or later in life. The association between changes or patterns of CVH metrics from midlife to late life and risk of dementia have been rarely examined in prospective studies. In addition, very few studies that examine the composite CVH metrics in midlife and late life in association with dementia risk have differentiated behavioral and biological CVH metrics [16]. This is important because the associations between certain biological components of CVH metrics and risk of dementia vary with age from middle age to late life [8-10].

Using data from the Cardiovascular Risk Factors, Aging and Dementia (CAIDE) in Finland, we sought to examine the association between CVH metrics from midlife to late life and risk of dementia. We hypothesized that optimal CVH metrics, especially occurring from midlife onwards, are associated with a reduced risk of incident dementia.

Materials and methods

Study design and participants

The study participants were derived from the Finnish CAIDE study, which is an ongoing population-based cohort study. The study design and purpose of CAIDE study were fully described elsewhere [20-22]. Briefly, the CAIDE study aims to investigate the associations of social, lifestyle, and cardiovascular risk factors from midlife onwards with late-life cognitive phenotypes. The initial participants of the CAIDE study were derived from population-based random samples examined within the framework of the North Karelia project [23] and the FINMONICA studies in 1972 to 1987 [24].

Fig 1 shows the flowchart of participants in the midlife and late-life examinations for this study. Briefly, of the eligible participants (n = 2,293) who were alive by the end of 1997 and were living in 2 geographically defined areas in or close to the towns of Kuopio and Joensuu (target population) [21], a random sample of 2,000 persons was selected for the first late-life evaluation in 1998. Of these, 1,449 (72.5%) undertook the examination in 1998, which consisted of the analytical sample for the association between midlife CVH metrics and late-life dementia detected in the 1998 examinations and the later wave of examination (analytical sample 1 in Fig 1).

Fig 1

Flowchart of the study population.

*The analytical sample 1 was used for analyzing the association between midlife CVH metrics and late-life risk of dementia detected both in the 1998 and 2005 to 2008 examinations. The analytical sample 2 was used for analyzing the association between late-life CVH metrics measured in the 1998 examination as well as the patterns of CVH metrics from midlife to late life (1998) with risk of dementia detected in the 2005 to 2008 examination. CAIDE, Cardiovascular Risk Factors, Aging, and Dementia; CVH, cardiovascular health.

The second late-life examination was conducted in 2005 to 2008. A total of 1,426 persons out of the initial 2,000 persons were still alive in the beginning of 2005 and eligible for the second late-life examination. Of these, 517 persons were not included due to death (n = 405, 78.3%) or poor health or refusal (n = 112, 21.7%), and 909 persons undertook the second late-life examination in 2005 to 2008. Of these, 165 persons either had dementia in 1998 or had missing information on dementia diagnosis, thus, 744 (81.8%) persons, who were free of dementia in 1998 and undertook the 2005 to 2008 examination, were included in the analyses involving late-life CVH metrics and dementia (analytical sample 2 in Fig 1).

The average follow-up time was 21.2 years from baseline to the 1998 examination and 8.3 years from the 1998 examination to the examination in 2005 to 2008.

Ethics statement

The CAIDE study received approval from the local ethics committee at Kuopio University and the Kuopio University Hospital in Kuopio, Finland as well as from the ethics committee at Karolinska Institutet in Stockholm, Sweden. The verbal informed consent (midlife surveys in 1972 to 1987) or the written informed consent (late-life surveys in 1998 and 2005 to 2008) was obtained from all participants prior to the recruitment into each wave of the examination. Research within CAIDE has been carried out in compliance with the ethical principles for medical research involving human subjects expressed in the Declaration of Helsinki. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (S1 Checklist).

Data collection and definitions

Data on demographics (e.g., age, sex, and education), lifestyles (e.g., smoking and physical activity), cardiometabolic factors (e.g., body mass index [BMI], plasma glucose, serum total cholesterol, and blood pressure), and medical history (e.g., diabetes and cerebrovascular and cardiovascular events) were collected through self-administered questionnaire survey, clinical examinations, the patient register, the prescribed drug register, and laboratory tests [21,25]. Cardiovascular diseases included angina pectoris, myocardial infarction, heart failure, and stroke diagnosed by physicians [21,25]. Apolipoprotein E (APOE) genotype was analyzed using polymerase chain reaction and HhaI restriction enzyme digestion [25] and was dichotomized as carriers versus non-carriers of the ε4 allele.

Definition of CVH metrics

We defined and scored the composite global CVH metrics based on 6 of the 7 components following the AHA’s recommendations [11], with some modifications, owing mainly to the lack of data on diet in both midlife and late life as well as the lack of data on fasting plasma glucose in midlife. The details of CVH metric definitions used in our study and the comparisons with the definitions from AHA were described in Table 1. Briefly, we categorized each of the CVH metric components into poor level (score = 0), intermediate level (score = 1), and ideal level (score = 2). The ideal level of each of the 6 CVH metrics was defined as follows. Ideal smoking status was defined as never smoking or having quitted for more than 1 year. Ideal physical activity was defined as having ≥2 times per week of leisure time physical activity that lasted at least 20 to 30 minutes and caused breathlessness and sweating [26]. Ideal BMI was defined as a BMI <25 kg/m2. The AHA recommended cutoffs for blood glucose and total cholesterol were converted to the units that we normally used. Ideal blood glucose in late life was defined as fasting plasma glucose <5.6 mmol/l (corresponding to <100 mg/dl in the AHA’s recommendations) without treatment. However, due to the lack of data on fasting plasma glucose in midlife, information on self-reported history of diabetes, a recorded diagnosis of diabetes from the patient register, and the use of antidiabetic medication from the prescribed drug register were used as a proxy. Thus, ideal level of blood glucose in midlife was defined as having none of the self-reported history of diabetes, registered diagnosis of diabetes, and the use of prescribed antidiabetic medication. Ideal total cholesterol was defined as untreated serum total cholesterol <5.17 mmol/l (corresponding to <200 mg/dl in the AHA’s recommendations). Ideal blood pressure was defined as systolic blood pressure (SBP) <120 mmHg and diastolic blood pressure (DBP) <80 mmHg without antihypertensive treatment.

Table 1

Definitions of CVH metrics in the recommendations of AHA and in the study of CAIDE.

Metrics	Poor level (score = 0)	Intermediate level (score = 1)	Optimal level (score = 2)
Smoking*	Current smoking	Former smoking ≤1 year ago	Never smoking or quit >1 year
Physical activity	No physical activity	AHA: 1–149 min/week moderate intensity or 1–74 min/week vigorous intensity or 1–149 min/week moderate + vigorousCAIDE: ≤1 time/week of leisure time physical activity that lasts at least 20–30 min and causes breathlessness and sweating	AHA: ≥150 min/week moderate intensity or ≥75 min/week vigorous intensity or ≥150 min/week moderate + vigorousCAIDE: ≥2 times/week of leisure time physical activity that lasts at least 20–30 min and causes breathlessness and sweating
Diet	AHA: 0–1 component of a healthy diet scoreCAIDE: No data	AHA: 2–3 components of a healthy diet scoreCAIDE: No data	AHA: 4–5 components of a healthy diet scoreCAIDE: No data
BMI*	≥30 kg/m2	25–29.9 kg/m2	<25 kg/m2
Fasting plasma glucose#	AHA: ≥7.0 mmol/lCAIDE: diabetes treated on drugs (midlife)	AHA: 5.6–6.9 mmol/l or treated to goalCAIDE: diabetes treated on diet (midlife)	AHA: <5.6 mmol/lCAIDE: no diabetes (midlife)
Serum total cholesterol*	≥6.21 mmol/l	5.17–6.20 mmol/l or treated to goal	<5.17 mmol/l untreated
Blood pressure*	SBP ≥140 mmHg or DBP ≥90 mmHg	SBP 120–139 or DBP 80–89 mmHg or treated to goal	SBP <120 mmHg and DBP <80 mmHg untreated

* The definitions for these components in both midlife and late life used in our study were the same as those from AHA’s recommendations (Lloyd-Jones et al., 2010).

# The definition for late-life plasma glucose was the same as that from AHA. In midlife, due to lack of fasting plasma glucose, a proxy was used including the information on self-reported history of diabetes, a recorded diagnosis of diabetes from the inpatient register, and the use of antidiabetic medication from the prescribed drug register.

AHA, American Heart Association; BMI, body mass index; CAIDE, Cardiovascular Risk Factors, Aging, and Dementia; CVH, cardiovascular health; DBP, diastolic blood pressure; SBP, systolic blood pressure.

Then, the global CVH metric score was calculated as the sum of scores of 6 CVH metric components, and the global CVH metrics were categorized into poor (total score ≤5), intermediate (6 to 7), and ideal (≥8) levels. In addition, smoking, physical activity, and BMI were used to define behavioral CVH metrics, and fasting plasma glucose, serum total cholesterol, and blood pressure as biological CVH metrics [11,16]. The levels of composite behavioral and biological CVH metrics were categorized into poor (a composite score ≤2), intermediate (3 to 4), and ideal (≥5) levels, respectively.

Clinical diagnosis of dementia

Comprehensive cognitive testing was performed at the late-life examinations in 1998 and 2005 to 2008. The Mini-Mental State Examination (MMSE) was used to assess global cognitive function. Dementia was ascertained following a 3-step protocol that included a screening phase, a clinical phase, and a differential diagnostic phase [21,22,27]. Briefly, in the 1998 examination, participants with an MMSE score ≤24 were referred to the clinical phase for further evaluations. In the 2005 to 2008 examination, participants with either an MMSE score ≤24, or a decrease in the MMSE score since the 1998 examination ≥3 points, or <70% delayed recall in the Consortium to Establish a Registry for Alzheimer’s Disease word list, or informant’s serious concerns regarding the participant’s cognition, were referred to the clinical phase. In both the 1998 and 2005 to 2008 examinations, the clinical phase included thorough physical examination and neuropsychological testing. If the person was suspected to have dementia, further evaluations were made, including laboratory blood tests, a chest radiograph, an electrocardiogram test, brain MRI/CT scans, and a cerebrospinal fluid analysis as needed. The final diagnoses of dementia were made by a review board after careful evaluation of all available information following the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).

Statistical analysis

The study protocol and analytical plan are reported in Supporting information (S1 Study protocol). The midlife (1972 to 1987) and late-life (1998) characteristics were compared between participants included in the analytical sample and those lost to follow-up by using Student t test for continuous variables and chi-squared test for categorical variables. Incidence of dementia was calculated as the number of dementia cases developed during the follow-up period divided by the total person-years of follow-ups. Cox proportional hazards models were used to estimate the hazard ratio (HR) and 95% confidence interval (CI) of dementia associated with global, behavioral, and biological CVH metrics in midlife and late life as well as the patterns of CVH metrics from midlife to late life, in which the follow-up time was used as the time scale. When examining midlife CVH metrics in relation to risk of dementia detected in 1998 and 2005 to 2008, we estimated the follow-up time from the date of midlife examination to the date of dementia diagnosis or the last contact in the 1998 or 2005 to 2008 examination. When examining the associations of late-life CVH metrics measured in 1998 and CVH metric patterns from midlife to late life (1998) with risk of dementia detected in 2005 to 2008, we estimated the follow-up time from the date of the 1998 examination to the date of the 2005 to 2008 examination. The proportional hazards assumption was verified using time-dependent coefficient in the Cox regression models. To account for possible influence of selective survival, we estimated the cumulative incidence of dementia from Cox proportional hazards models while taking into account death as a competing risk event [28].

We reported the main results from 3 models. Model 1 was adjusted for age, sex, and education, and in model 2, we further adjusted for baseline cardiovascular disease and APOE genotype. We considered APOE genotype as a potential confounder because APOE ɛ4 allele is a genetic risk factor for dementia and also the APOE gene is involved in the lipid transport and lipoprotein metabolism [29,30]. In addition, to assess the potential effect of selective survival, we reported results from the Fine and Gray competing risk regression models (model 3) (S1 Statistical methods) where death was considered as a competing risk event and all covariates in the model 2 were included [31].

IBM SPSS Statistics 26 for Windows (IBM SPSS, Chicago, Illinois) was used for the descriptive analysis and Cox regression analysis. SAS 9.4 Statements (SAS Institute, 2013, Cary, North Carolina, United States of America) was used for estimating the cumulative incidence of dementia and performing the Fine and Gray competing risk regression analysis.

Results

Characteristics of study participants in midlife (1972 to 1987) and late life (1998)

Table 2 shows the midlife characteristics of participants in the analytical sample 1 as well as the comparison of characteristics in both midlife and late life between participants included in the analytical sample 2 (n = 744) and those not included (n = 705). Compared with those not included in the analytical sample 2, the included participants in midlife were younger (p < 0.001), more likely to be female (p = 0.031), more educated (p < 0.001), had higher proportions of ideal smoking (p = 0.002), ideal BMI (p = 0.004), and ideal blood pressure (p < 0.001), a higher score of global, behavioral, and biological CVH metrics (all p < 0.001), and were less likely to carry APOE ε4 allele (p = 0.003) and have cardiovascular diseases (p = 0.024). There was no significant difference in the distributions of physical activity, diabetes, and total cholesterol in midlife between those included and not included. In addition, compared with those not included in the analytical sample 2, the included participants in late life (1998) had higher proportions of ideal smoking (p < 0.001), ideal physical activity (p < 0.001), ideal blood glucose (p < 0.001), higher global and behavioral CVH scores (both p < 0.001), and lower prevalence of cardiovascular disease (p < 0.001).

Table 2

Characteristics of study participants in midlife (1972 to 1987) and late life (1998).

Characteristics*	Analytical sample 1, midlife (n = 1449)	Analytical sample 2 in midlife, 1972–1987			Analytical sample 2 in late life, 1998
		Excluded (n = 705)	Included (n = 744)	p	Excluded (n = 705)	Included (n = 744)	p
Age (years), mean (SD)	50.4 (6.0)	51.4 (5.9)	49.5 (5.9)	<0.001	72.5 (4.2)	70.2 (3.5)	<0.001
Women	900 (62.1)	418 (59.3)	482 (64.8)	0.031	418 (59.3)	482 (64.8)	0.031
Education (years), mean (SD)	8.6 (3.4)	7.9 (3.3)	9.2 (3.5)	<0.001	7.9 (3.3)	9.2 (3.5)	<0.001
Smoking
Poor	61 (4.2)	39 (5.5)	22 (3.0)	0.002	58 (8.5)	34 (4.6)	<0.001
Intermediate	290 (20.0)	158 (22.4)	132 (17.7)		221 (32.3)	198 (26.8)
Ideal	1,098 (75.8)	508 (72.1)	590 (79.3)		406 (59.3)	506 (68.6)
Physical activity
Poor	51 (3.6)	30 (4.4)	21 (2.9)	0.321	73 (10.5)	27 (3.6)	<0.001
Intermediate	772 (54.4)	373 (54.1)	399 (54.6)		146 (21.1)	123 (16.6)
Ideal	597 (42.0)	286 (41.5)	311 (42.5)		473 (68.4)	593 (79.8)
BMI
Poor	241 (16.6)	139 (19.7)	102 (13.7)	0.004	202 (28.7)	192 (25.8)	0.133
Intermediate	718 (49.6)	347 (49.2)	371 (49.9)		321 (45.5)	378 (50.8)
Ideal	490 (33.8)	219 (31.1)	271 (36.4)		182 (25.8)	174 (23.4)
Plasma glucose#
Poor	55 (3.8)	32 (4.5)	23 (3.1)	0.071	65 (9.2)	31 (4.2)	<0.001
Intermediate	3 (0.2)	3 (0.4)	0 (0.0)		97 (13.8)	90 (12.1)
Ideal	1391 (96.0)	670 (95.0)	721 (96.9)		543 (77.0)	623 (83.7)
Total cholesterol
Poor	944 (65.1)	476 (67.5)	468 (62.9)	0.103	229 (32.5)	265 (35.6)	0.095
Intermediate	384 (26.5)	169 (24.0)	215 (28.9)		343 (48.7)	369 (49.6)
Ideal	121 (8.4)	60 (8.5)	61 (8.2)		133 (18.9)	110 (14.8)
Blood pressure
Poor	618 (42.7)	337 (47.8)	281 (37.8)	<0.001	276 (39.1)	276 (37.1)	0.692
Intermediate	760 (52.4)	339 (48.1)	421 (56.6)		415 (58.9)	451(60.6)
Ideal	71 (4.9)	29 (4.1)	42 (5.6)		14 (2.0)	17 (2.3)
Global CVH metrics score, mean (SD)	7.2 (1.5)	7.0 (1.5)	7.4 (1.5)	<0.001	7.1 (1.7)	7.6 (1.5)	<0.001
Behavioral CVH metrics score, mean (SD)	4.2 (1.0)	4.1 (1.1)	4.4 (1.0)	<0.001	4.2 (1.1)	4.4 (1.0)	<0.001
Biological CVH metrics score, mean (SD)	3.0 (1.0)	2.9 (1.0)	3.1 (1.0)	<0.001	3.2 (1.1)	3.2 (1.0)	0.217
APOE ε4 allele carriers	499 (35.1)	265 (39.0)	234 (31.5)	0.003	265 (39.0)	234 (31.5)	0.003
Cardiovascular diseases	114 (7.9)	67 (9.5)	47 (6.3)	0.024	260 (37.6)	190 (25.5)	<0.001

Data were n (%), unless otherwise specified.

* The number of participants with missing values in analytical sample 1 was 21 persons for education, 27 persons for APOE genotype, and 29 persons for midlife physical activity, and in analytical sample 2 was 11 persons for education, 1 person for APOE genotype, 6 persons for late-life smoking, and 1 person for late-life physical activity. In the subsequence analysis, a dummy variable was created to indicate the missing values for education and APOE genotype, whereas missing value on physical activity and smoking was considered as null (i.e., poor level) in the calculation of global CVH metric score.

# In midlife, information on fasting plasma glucose was lacking, instead, the information on self-reported diabetes, a recorded diagnosis of diabetes from the inpatient register, and the use of antidiabetic medication from the prescribed drug register were used as a proxy.

APOE, apolipoprotein E; BMI, body mass index; CVH, cardiovascular health.

Midlife CVH metrics and risk of incident dementia

Compared with people having poor global CVH metrics in midlife, those with an ideal level of global CVH metrics in midlife had a 48% lower risk of dementia after controlling for potential confounding factors and a 54% reduced risk of dementia after further taking into account death as a competing risk event (Table 3, models 2 and 3). When the global CVH metric score in midlife was analyzed as a continuous variable, the fully adjusted HR of dementia was 0.86 (95% CI: 0.76, 0.98; p = 0.019) for per 1-point increment in the CVH metric score, and the HR remained unchanged in model 3 when death as a competing risk event was taken into account. In addition, we analyzed the association of behavioral and biological CVH metrics in midlife with dementia risk in late life. The associations of continuous behavioral CVH metric score and the ideal level of behavioral CVH metrics with a reduced risk of dementia were statistically evident, especially in model 3 when the competing risk due to death was taken into consideration. In contrast, the tendency of an association between ideal biological CVH metrics in midlife and the reduced dementia risk disappeared in the competing risk models (Table 3, models 1 to 3).

Table 3

The association of CVH metrics in midlife (1972 to 1987) and risk of dementia detected in late life (both 1998 and 2005 to 2008) (n = 1,449).

Midlife CVH metrics (score)	No. of participants	No. of dementia cases	Person-years of follow-up	Incidence (per 1,000 person-years)	20-year cumulative incidence (%)*	Model 1#		Model 2#		Model 3#
						HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
Global CVH metrics
Per 1-point increment	1,449	108	30,680	3.52	6.16	0.86 (0.76, 0.97)	0.016	0.86 (0.76, 0.98)	0.019	0.86 (0.76, 0.97)	0.014
Poor (≤5)	183	28	3,619	7.74	9.03	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (6–7)	645	52	13,904	3.74	6.35	0.69 (0.42, 1.13)	0.143	0.71 (0.43, 1.16)	0.174	0.63 (0.38, 1.04)	0.072
Ideal (≥8)	621	28	13,156	2.13	3.82	0.51 (0.29, 0.90)	0.021	0.52 (0.29, 0.93)	0.027	0.46 (0.27, 0.77)	0.003
Behavioral CVH metrics
Per 1-point increment	1,449	108	30,680	3.52	6.16	0.86 (0.72, 1.02)	0.083	0.86 (0.73, 1.02)	0.090	0.81 (0.69, 0.96)	0.013
Poor (≤2)	85	13	1,612	8.06	5.71	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (3–4)	741	65	15,865	4.10	7.24	1.47 (0.73, 2.94)	0.278	1.45 (0.72, 2.91)	0.294	0.67 (0.36, 1.23)	0.195
Ideal (≥5)	623	30	13,204	2.27	3.54	0.73 (0.35, 1.53)	0.409	0.73 (0.35, 1.52)	0.402	0.42 (0.22, 0.83)	0.012
Biological CVH metrics
Per 1-point increment	1,449	108	30,680	3.52	6.16	0.81 (0.65, 1.00)	0.052	0.82 (0.66, 1.01)	0.063	0.90 (0.71, 1.14)	0.374
Poor (≤2)	464	48	9,760	4.92	7.90	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (3–4)	878	55	18,768	2.93	4.73	0.73 (0.49, 1.09)	0.124	0.74 (0.49, 1.10)	0.140	0.83 (0.56, 1.24)	0.363
Ideal (≥5)	107	5	2,152	0.93	2.97	0.62 (0.24, 1.64)	0.338	0.63 (0.24, 1.67)	0.353	1.04 (0.41, 2.62)	0.941

* The cumulative incidence was calculated from the crude Cox models after taking into account the competing risk of death.

# Model 1 was adjusted for age, sex, and education; model 2 was additionally adjusted for APOE ε4 allele and cardiovascular disease in midlife; and model 3 included death as a competing risk event with the adjustment of all covariates in model 2.

APOE, apolipoprotein E; CI, confidence interval; CVH, cardiovascular health; HR, hazard ratio.

Furthermore, of the 6 individual CVH metric components in midlife, intermediate and ideal smoking (versus poor) and ideal BMI (versus poor) were significantly associated with the reduced risk of dementia (S1 Table, model 2). However, the association was no longer significant after taking into account death as a competing risk event (S1 Table, model 3).

Late-life CVH metrics and risk of incident dementia

Compared with poor CVH metrics in late life, having an intermediate or ideal level of global, behavioral, and biological CVH metrics was not significantly associated with the HR of dementia in models 1 and 2, but when the competing risk due to death was taken into account, an ideal level of biological CVH metrics was significantly associated with an increased HR of dementia (Table 4). Similarly, as a continuous variable, none of the global, behavioral, and biological CVH metric scores in late life was significantly associated with dementia risk in models 1 and 2, but after taking into account death as a competing risk event, per 1-point increment in late-life biological CVH metric score was significantly associated with an over 40% increased risk of dementia (HR = 1.41; 95% CI = 1.02, 1.94; p = 0.036) (Table 4). In addition, none of the 6 individual CVH metric components in late life was significantly associated with the risk of dementia in any of the 3 models (S2 Table).

Table 4

The association of CVH metrics in late life (1998) and risk of dementia detected in late life (2005 to 2008) (n = 744).

Late-life CVH metrics (score)	No. of participants	No. of dementia cases	Person-years of follow-up	Incidence (per 1,000 person-years)	8-year cumulative incidence (%)*	Model 1#		Model 2#		Model 3#
						HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
Global CVH metrics
Per 1-point increment	744	47	6,168	7.62	4.24	1.04 (0.86, 1.26)	0.700	1.09 (0.89, 1.33)	0.409	1.10 (0.87, 1.39)	0.433
Poor (≤5)	66	6	537	11.17	6.64	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (6–7)	276	14	2,259	6.20	3.63	0.52 (0.20, 1.38)	0.191	0.60 (0.22, 1.69)	0.338	0.61 (0.23, 1.64)	0.329
Ideal (≥8)	402	27	3,372	8.01	4.33	0.70 (0.29, 1.73)	0.444	0.91 (0.34, 2.41)	0.850	0.96 (0.35, 2.58)	0.929
Behavioral CVH metrics
Per 1-point increment	744	47	6,168	7.62	4.24	0.89 (0.69, 1.15)	0.367	0.93 (0.71, 1.22)	0.608	0.86 (0.62, 1.19)	0.368
Poor (≤2)	36	5	299	16.72	8.88	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (3–4)	353	19	2,909	6.53	3.90	0.53 (0.20, 1.45)	0.218	0.53 (0.19, 1.44)	0.212	0.50 (0.15, 1.66)	0.258
Ideal (≥5)	355	23	2,960	7.77	4.21	0.63 (0.23, 1.72)	0.367	0.70 (0.25, 1.93)	0.491	0.57 (0.17, 1.95)	0.371
Biological CVH metrics
Per 1-point increment	744	47	6,168	7.62	4.24	1.24 (0.93, 1.64)	0.140	1.28 (0.96, 1.70)	0.092	1.41 (1.02, 1.94)	0.036
Poor (≤2)	171	7	1,394	5.02	2.68	1.00 (reference)		1.00 (reference)		1.00 (reference)
Intermediate (3–4)	495	31	4,102	9.02	4.38	1.71 (0.73, 4.00)	0.215	2.06 (0.83, 5.14)	0.120	2.10 (0.84, 5.20)	0.110
Ideal (≥5)	78	9	673	13.37	6.63	2.38 (0.86, 6.64)	0.097	2.85 (0.96, 8.46)	0.059	3.54 (1.16, 10.83)	0.027

* The cumulative incidence was calculated from the crude Cox models after taking into account the competing risk of death.

# Model 1 was adjusted for age, sex, and education; model 2 was additionally adjusted for APOE ε4 allele and cardiovascular disease in late life; and model 3 included death as a competing risk event with the adjustment of all covariates in model 2.

APOE, apolipoprotein E; CI, confidence interval; CVH, cardiovascular health; HR, hazard ratio.

Patterns of CVH metrics from midlife to late life and risk of incident dementia

When midlife and late-life global CVH metric scores were entered simultaneously into the model, the fully adjusted HR of dementia associated with per 1-point increment in CVH score in midlife and late life was 0.82 (95% CI: 0.66, 1.03; p = 0.082) and 1.21 (0.97, 1.51; p = 0.093), respectively. After further taking into account death as a competing risk event, the HR remained unchanged for both midlife and late-life global CVH metric scores. There was a statistical interaction between midlife global CVH metric score and late-life global CVH metric score on the risk of dementia (p for interaction = 0.001).

Fig 2 showed the HRs (95% CI) of dementia associated with the combinations (or patterns) of different levels of global, behavioral, and biological CVH metrics in midlife and late life, controlling for multiple potential confounding factors. (1) Global CVH metrics: Compared with persons having a poor level of global CVH metrics in both midlife and late life, those with an intermediate level of global CVH metrics in both midlife and late life and those with midlife ideal and late-life intermediate global CVH metrics had the fully adjusted HR of 0.25 (95% CI: 0.08, 0.86; p = 0.028) and 0.14 (0.02, 0.76; p = 0.024), respectively, for dementia (Fig 2A). However, the HRs were not significant when taking into account death as a competing risk event. There was no significant association with dementia risk for any other combinations of midlife and late-life CVH metrics. (2) Behavioral CVH metrics: Compared with people with the poor level of behavioral CVH metrics in both midlife and late life, those with an intermediate level of behavioral CVH metrics in both midlife and late life, an intermediate level in midlife but an ideal level in late life, an ideal level in midlife but an intermediate level in late life, and an ideal level in both midlife and late life had a significantly reduced risk of dementia (Fig 2B). The associations remained significant after taking into account death as a competing risk event. We found no significant association between other combining groups of behavioral CVH metrics and risk of dementia. (3) Biological CVH metrics: Of all the 9 combinations of biological CVH metric levels in midlife and late life, only the combination of midlife intermediate and late-life ideal levels of biological CVH metrics was significantly associated with an increased risk of dementia in the fully adjusted Cox regression models (Fig 2C). However, when taking into account death as a competing risk event, compared with people having the poor level of biological CVH metrics in both midlife and late life, the HR of dementia was 7.62 (95% CI: 1.97, 29.48; p = 0.003) for those with midlife intermediate level and late-life ideal level of biological CVH metrics and 7.30 (1.22, 43.78; p = 0.030) for those with midlife ideal level and late-life intermediate level of biological CVH metrics.

Fig 2

The HR (95% CI) of dementia associated with the joint patterns of midlife and late-life CVH metrics: A, global CVH metrics; B, behavioral CVH metrics; C, biological CVH metrics.

HR (95% CI) was derived from the Cox proportional hazards models controlling for age, sex, education, APOE ε4 allele, and cardiovascular disease in midlife. *p < 0.05. APOE, apolipoprotein E; CVH, cardiovascular health; CI, confidence interval; HR, hazard ratio; IM, intermediate; NA, not available; n/N, the number of dementia cases/number of participants.

Discussion

In this population-based cohort study, we found that compared with poor CVH metrics in midlife, an ideal level of midlife global CVH metrics, and behavioral CVH metrics in particular, was associated with a lower risk of dementia in late life. There was no association between CVH metrics and dementia risk when CVH metrics were assessed in late life. However, when the global CVH metric patterns from midlife to late life were examined, compared with the poor global CVH metrics in both midlife and late life, having an intermediate or ideal level of global CVH metrics in midlife or late life was associated with a reduced risk of dementia, although the competing risk due to death could partly account for the associations (Fig 2A). Notably, the association of an intermediate or ideal level of behavioral CVH metrics in either midlife or late life with a reduced risk of dementia was present independent of competing risk and all potential confounders examined (Fig 2B). Finally, having an ideal or intermediate level of late-life biological CVH metrics in combination with either an ideal or intermediate biological CVH metrics in midlife (versus poor level both in midlife and late life) was associated with an increased risk of dementia (Fig 2C).

Population-based studies have rarely investigated the association of composite CVH metrics with the risk of dementia from a life-course perspective. The Whitehall II cohort study, which used the modified AHA’s definitions to define smoking and diet, showed that per 1-point increment in the global CVH metric score at the age of 50 years was associated with an over 10% decreased risk of dementia during an over 25-year follow-up (HR = 0.89; 95% CI = 0.85 to 0.95) [16]. Similarly, the Atherosclerosis Risk in Communities (ARIC) cohort study suggested that a higher global CVH metric score in middle age (mean age, 54 years) was associated with slower cognitive decline over a 20-year follow-up period [32]. In the Finnish CAIDE cohort, we found that having a higher global CVH metric score or having an ideal level of global CVH metrics in midlife (mean age, 50.4 years) was associated with the lower risk of dementia. Taken together, these studies provide consistent evidence supporting the view that achieving optimal CVH in midlife is crucial for reducing late-life dementia risk.

We found no evidence for the association between late-life global CVH metrics and the risk of dementia, which was consistent with the findings from the population-based study of adults (age ≥55 years) from the health insurance settings in Germany [19]. However, the population-based Three-City Study of older adults (age ≥65 years) in France did show that having more optimal CVH metrics or having a higher CVH metric score defined following the AHA’s recommendations was associated with a lower risk of dementia [17]. The inconsistent findings across studies might be partially attributable to differences in methodology (e.g., lack of diet data in our study) and characteristics of study population (e.g., age).

This discrepancies between associations of midlife and late-life CVH metrics with the risk of dementia were in accordance with the fact that the relationships of cardiometabolic components of CVH metrics (e.g., BMI in behavioral CVH metrics and blood pressure and total cholesterol in biological CVH metrics) with the risk of dementia vary with age; the strength of associations between these cardiometabolic risk factors and risk of dementia is attenuated with age from young adulthood and midlife onwards, and may even be reversed among very old people [33-35]. Thus, these individual CVH metrics may render the global CVH metrics less predictive for dementia risk as people age [19]. Indeed, we found that an ideal level of late-life biological CVH metrics was associated with an elevated risk of dementia. However, this association may partly reflect the reverse causality because the ideal (or low) levels of certain biological CVH metric components in late life (e.g., blood pressure <120/80 mmHg and low total cholesterol) might be a marker of preclinical dementia [8,10,33]. This should be kept in mind when interpreting the results from studies that involve late-life composite global and biological CVH metrics in relation to dementia risk.

The association of CVH metric patterns from midlife to late life with the risk of dementia has been rarely evaluated so far in the population-based settings. We found that compared with poor levels of global CVH metrics in both midlife and late life, having the intermediate or ideal level of global CVH metrics, especially behavioral CVH metrics, in either midlife or late life was associated with a substantially reduced risk of dementia. Given the potentially reverse causality of late-life ideal levels of biological components in the global CVH metrics and dementia risk, interpretation of the results involving late-life global CVH metrics needs to be cautious. Our findings implied that maintaining a life-long CVH, and behavioral health in particular, from midlife onwards may help achieve healthy brain aging. This supports the view that intervention strategies to promote CVH, especially behavioral health, from the life-course prospective might help prevent or delay the onset of dementia [2]. Because the CVH metrics are modifiable and manageable, our findings have significant implications for public health. This is particularly relevant given that dementia has become a global public health priority in our aging society.

The mechanisms underlying the association between the CVH metrics over the life course and risk of dementia are not fully understood. The poor levels (as risk factors) of CVH metrics, especially occurring from midlife, are supposed to contribute mainly to macro- and microvascular lesions and neurodegenerative process in the brain, e.g., oxidative stress, inflammation, atherosclerosis, hypoxia, cerebral small vessel lesions, and advanced glycation end products [36]. Optimal or ideal levels of CVH metrics (e.g., no smoking, physical activity, and normal blood glucose) are associated with less burden of cerebrovascular damage [37], which then leads to fewer white matter lesions and brain infarcts as well as less severe neurodegeneration [2,38]. The optimal brain health resulting from optimal CVH profiles may in turn contribute to the lower risk of cognitive impairment [39] and cognitive decline [17,18], and thus lead to a lower risk of dementia.

Strengths of this study include the design of population-based cohort study with long follow-up time from midlife to late life as well as the comprehensive assessments for the diagnosis of dementia. However, our study also has limitations. First, we used the modified AHA’s definitions to assess CVH metrics owing to the lack of data on diet in both midlife and late life as well as midlife plasma glucose. Lack of dietary data might underestimate the potential protective effect of ideal CVH metrics against dementia because data from a small subsample of the CAIDE participants (n = approximately 350) showed that midlife healthy diet and beneficial dietary changes from midlife to late life were associated with a reduced risk of dementia [40,41]. Similarly, the lack of data on midlife fasting plasma glucose might have misclassified some people with abnormal plasma glucose into the ideal category, and thus underestimate the possible beneficial effect of the ideal midlife CVH metrics on dementia risk. Second, the analytic sample was generally healthier compared to dropouts (e.g., death, refusal, or missing information for the diagnosis of dementia). Thus, the selective survival by both CVH metric levels and cognitive conditions (e.g., cognitive impairment and dementia) over the follow-up period from midlife to late life might have led to an underestimation of the true associations between poor CVH metrics and risk of dementia because people with poor CVH metrics or poor cognitive conditions were less likely to survive into old age [21]. However, we have estimated the cumulative incidence of dementia and also used the competing risk models to partially account for the impact of selective survival on the associations between CVH metrics and risk of dementia. Nevertheless, cautiousness is needed when generalizing our study findings to the entire middle-aged and elderly population. Finally, the statistical power was limited for the analysis of some subgroups due to small numbers of participants and dementia cases, especially in the analysis of midlife and late-life CVH metric patterns and risk of dementia. Thus, our findings deserve further confirmation in the large-scale population-based studies.

In conclusion, in this population-based cohort study, we observed that having an ideal or intermediate level of global CVH metrics, especially the behavioral CVH metrics, from midlife to later in life is associated with a lower risk of dementia. Findings from this study reinforce the view that maintaining a life-long optimal CVH profile, and behavioral health profile in particular, may help reduce the late-life risk of dementia.

STROBE checklist.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

(DOCX)

Click here for additional data file.

Fine and Gray competing risk regression model.

(DOCX)

Click here for additional data file.

The association of individual CVH metrics in midlife (1972–1987) with risk of dementia detected in late life (both 1998 and 2005–2008) (n = 1,449).

*The number of participants with missing data was 13 persons for physical activity, and these individuals were included in the analysis by creating a dummy variable to indicate those with missing values. #In midlife, information on fasting plasma glucose was lacking, instead, the information on self-reported history of diabetes, a recorded diagnosis of diabetes from the inpatient register, and the use of antidiabetic medication from the prescribed drug register were used as a proxy. §Model 1 was adjusted for age, sex, education, and other components in the table; model 2 was additionally adjusted for APOE ε4 allele and cardiovascular disease in midlife; and model 3 included death as a competing risk event with the adjustment of all covariates in model 2. APOE, apolipoprotein E; CI, confidence interval; CVH, cardiovascular health; HR, hazard ratio.

(DOCX)

Click here for additional data file.

The association of individual CVH metrics in late life (1998) with risk of dementia detected in late life (2005–2008) (n = 744).

*The number of participants with missing data was 6 persons for smoking and 1 person for physical activity, and these individuals were included in the analysis by creating a dummy variable for each factor to indicate those with missing values. #Model 1 was adjusted for age, sex, education, and other components in the table; model 2 was additionally adjusted for APOE ε4 allele and cardiovascular disease in late life; and model 3 included death as a competing risk event with the adjustment of all covariates in model 2. APOE, apolipoprotein E; CI, confidence interval; CVH, cardiovascular health; HR, hazard ratio.

(DOCX)

Click here for additional data file.

21 Apr 2020

Dear Dr Liang,

Thank you for submitting your manuscript entitled "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based study" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff [as well as by an academic editor with relevant expertise] and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Thomas J McBride, PhD,

PLOS Medicine

7 Jul 2020

Dear Dr. Liang,

Thank you very much for submitting your manuscript "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based study" (PMEDICINE-D-20-01344R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to four independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Jul 28 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Emma Veitch, PhD

PLOS Medicine

On behalf of Clare Stone, PhD, Acting Chief Editor,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

*In the last sentence of the Abstract Methods and Findings section, please include a brief note about any key limitation(s) of the study's methodology.

*At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists (I noted this had been provided as a supplementary file, please include the text immediately following the Abstract in your revised manuscript). This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

*We'd suggest using the STROBE guideline to assist/enhance study reporting - if doing this please include the completed STROBE checklist as Supporting Information. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (SChecklist)." The STROBE guideline can be found here: http://www.equator-network.org/reporting-guidelines/strobe/. When completing the checklist, please use section and paragraph numbers, rather than page numbers.

*In the Methods section of your paper, please clarify if the study had a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

-----------------------------------------------------------

Comments from the reviewers:

Reviewer #1: "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based study" attempts to quantify the association of cardiovascular health metrics (CVH) with the risk of incident dementia, on 1449 participants from the Finnish Cardiovascular Risk Factors, Aging and Dementia (CAIDE) study. Six AHA metrics (excluding diet) were used to represent CVH, with the DSM-IV criteria used to diagnose dementia, with Cox regression models used for the analysis. It was concluded that either maintaning lifelong good CVH, or improving CVH from mid to late lift, is associated with substantially reduced risk of dementia.

The usage of AHA-based CVH metrics follows prior studies ([17]-[19]) with similar structures, while the DSM-IV criteria for diagnosing dementia is also fairly long-established. The study sample size of 1449 participants is comparable to the three works referenced above (1033-6626 participants), as is the regression analysis. A major contribution of this paper over prior works pertains to the examination of mid-to-late-life CVH metrics, over late-life CVH metrics alone, which has yielded mixed results in the previous works.

This manuscript thus represents a new datapoint in CVH to dementia association, that clearly builds upon a coherent body of previous research. However, while the total number of participants is in the ballpark of previous studies, the number of positive dementia diagnoses is relatively low (61 in 1998, 47 in 2005-08). This is further reflected in the especially low number of valid cases for some stratifications (particularly poor/poor & the novel midlife/latelife analysis, Table 4), and correspondingly wide hazard ratio confidence intervals.

There also remain some further concerns:

1. The definition of the two late-life examinations (1998, 2005-08) was not very easy to follow in the Study Design section of the main text, possibly due to the explanation being relegated to the secondary caption of the Figure 1 flowchart. This might be shifted to the main text.

2. The likely relatively large number of participants who died before they could be included in the analytical samples, would appear to have possibly affected the statistics/conclusions. From the main text, it would appear that patients who manifested dementia, but died before a follow-up examination, would be ignored. While dementia incidence might be unknown for these cases, it might be appropriate to consider a survival model for dementia/death, or possibly include fraction/age of death at follow-up in the demographics (e.g. as done in "Midlife cardiovascular fitness and dementia: A 44-year longitudinal population study in women", Horder et al., Neurology, 2018)

3. The adjustment for the APOE genotype in the context of CVH/dementia might be briefly motivated in terms of possible protective effect.

4. The definition for "history of cardiovascular disease", as displayed in Table 1, might be specified in greater detail.

5. The full implementation of intervention assumption for population attributable risk (PAR %) in Page 8 might be more fully elaborated. In particularly, what exactly are these interventions, and what is their expected success rate?

6. For Table 4, "Intermediate or ideal (≥6) in either midlife or late-life" might be clarified as "either midlife or late-life only". Moreover, since improved CVH (i.e. poor in mid-life, intermediate or ideal in late-life) is discussed in the main text, it might also be shown in Table 4 (possibly along with intermediate/ideal in mid-life, poor in late-life)

7. While it is stated that "There was no association between CVH metrics and dementia risk when CVH metrics were assessed in late-life" (Page 10), Table 3 would seem to suggest an overall risk reduction, were the Intermediate & Ideal categories combined and compared against Poor; the authors might also comment on Intermediate & Ideal being analyzed separately in Tables 2 & 3, but combined for the analysis in Table 4.

-----------------------------------------------------------

Reviewer #2: In general this is a well-written manuscript. Weaknesses lie in the fact that not all components of the AHA simple 7 score were available, but that a score was still computed and I don't believe that the score in this format has been validated (if it has been, there should be discussion of this). Because of this, I find less value in use of the AHA Simple 7 metric in this paper and think that it would be more valuable to only report the results of individual components.

* In general the writing is clear, but there are a few sentences that are not grammatically correct and would benefit from some editing. In general, writing is good.

* More detail is needed regarding the definition for physical activity more than twice a week. What was defined as physical activity twice a week? Any physical activity (including walking around the block)? Exercise at a gym? Etc. AHA Simple 7 defines this as >=150 min/week of moderate activity or >=75 min/week of vigorous activity or >=150 min/week of moderate and vigorous activity.

* Since the author's score is missing diet and has a proxy measure for fasting plasma glucose, it isn't really the AHA Simple 7 score anymore. Because of this, it may be more prudent to only evaluate individual components of this score and to not include the computed global CVH metrics score, unless this has been validated previously.

* In Table 1, it looks like participants in analytical sample 2 (late-life) are survivors (slightly more women, much more ideal physical activity, much more ideal total cholesterol), etc. Can authors discuss this further in their manuscript and what this means for their conclusions? There is some discussion of this, but I think that it could be discussed more extensively.

-----------------------------------------------------------

Reviewer #3: In this population based Finnish study of initially n=1400 individuals in their mid-life, authors investigate the association between (a) mid-life and late life 6-item cardiovascular health (CVH) and incident dementia, and (b) change in 6-item CVH between mid-life and late-life with incident dementia. In that study, mid-life but not late life CVH was related to lower risk of dementia. In addition, patterns of change in CVH including improvement in CVH were related to lower risk of dementia.

Primordial prevention is a highly relevant and emergent topic and there is prior evidence that higher CVH is related to lower risk of dementia. Authors add novelty in addressing whether change in CVH is related to dementia risk.

However, the current analysis suffers from major limitations:

1. The study is of relatively small size (at best n=2000 in mid-life) limiting the power of the analysis, and precision of the estimates as shown by the width of the confidence intervals. In particular, this precludes to study with more granularity the patterns of change in CVH between mid-life and late-life: potentially 9 groups of change exist and here only three were investigated. As a corollary, lost to follow-up is substantial: 25% between mid-life and late-life exam1, and an additional lost to follow-up of half the population between late life exam 1 and late life exam 2, so that the analytic sample is likely to be highly selected.

2. As acknowledged by the authors, diet was missing in their CVH evaluation which limits comparison with other studies based on the full (e.g. 7 items) CVH scale. The authors should discuss to which extent this missing metric impacts their results.

3. In that study, there was no significant association between late-life CVH and dementia. The authors concentrate the discussion of this finding by a comparison with a German study that did not find any significant association (ref #19). However, their result is inconsistent with a paper cited in ref #17 that reports significant association with dementia, but the authors did not discuss these inconsistencies.

-----------------------------------------------------------

Reviewer #4: The authors assessed mid-life (ages 45-64) cardiovascular health (CVH) in 2,293 Finnish people in 1972-1987, then ascertained incident dementia among 1,449 survivors in 1998. The authors also assessed late-life (ages 65+) CVH in 1,426 of the survivors in 1998, and again ascertained incident dementia among 744 survivors in 2005-2008 who had been free of dementia in 1998. This study design enabled the authors to relate both mid-life and late-life CVH to incident dementia, which is a strength of the study. The authors correctly stated that few studies have assessed late-life CVH, or a life-course perspective of changes in CVH from mid-life to late-life, in relation to incident dementia, and thus their study would represent an important enhancement to the existing literature.

In my opinion, this study is very good. My comments below are not criticisms of the quality of the study, but rather suggestions for how this already good work might be improved.

MAJOR COMMENTS:

1. Methods/Results: In order to further explore components of mid-life and late-life CVH in relation to incident dementia, I suggest the authors consider adding to their investigation models in which all 6 components are added simultaneously as independent variables, rather than the summary CVH score. These models would show the magnitude and direction of the hazard ratio for each component adjusted for the others, and which of the 6 components matter more than others. This could be especially informative for comparing the results pertaining to mid-life CVH (Table 2) and late-life CVH (Table 3), given that body mass index, blood pressure, and cholesterol may behave differently in mid-life versus late-life in relation to dementia risk.

2. Results, Table 1: I suggest the authors add two columns to Table 1 to enhance readers' understanding of the participants' characteristics. The first new column would be the mid-life characteristics of the subset of analytical sample #1 who were not included in analytical sample #2 (705 people). The second new column would be the mid-life characteristics of analytical sample #2 (744 people). These two columns could be placed in between the existing columns for analytical sample #1 and analytical sample #2. These columns would help us see whether the two halves of analytical sample #1 were different, and would also help us see more directly the change in CVH over time from mid-life to late-life in analytical sample #2. Also, in conjunction with these additions to Table 1, the second paragraph of the Results section could be deleted entirely, because the information in that paragraph partially matches what I am suggesting to add to Table 1.

3. Results, Table 4: I suggest the authors consider adding a model that includes the mid-life CVH score (per 1-point increment) and the late-life CVH score (per 1-point increment) simultaneously as independent variables, and also possibly the interaction of those two CVH scores. This model would complement the model based on combinations of mid-life and late-life CVH categories.

4. Discussion: As the authors note in the paragraph about limitations, this study may suffer from selective survival by levels of CVH. I suggest that the authors also recognize and point out that the study may suffer from selective survival by cognitive decline and dementia. One of the main reasons eligible participants may have not participated in dementia assessment is because they were too cognitively impaired to participate at all, or because they had died as a consequence of dementia, which would be unknown to the authors. I agree with the authors that the magnitude of associations of mid-life CVH and dementia may be under-estimated, and I suggest the authors clarify in their statement that this is likely because of a joint effect of selective survival by CVH level (which can be measured) and selective survival by cognitive decline & dementia (which is unknown for non-survivors and cannot be measured). The first problem, selective survival by CVH level, can be assessed directly in the authors' data, and I suggest the authors conduct this assessment and include the findings in the Results section. For analytical sample #1, compare the mid-life characteristics for those participants who were included in the 1998 dementia assessment versus those who were not. And for analytical sample #2, compare the mid-life and late-life (1998) characteristics for those participants who were included in the 2005-2008 dementia assessment versus those who were not. These analyses will enhance the paper and allow the authors to comment more concretely on the possibility that selective survival influenced the hazard ratios. Fortunately, any bias in the hazard ratios due to selective survival is very likely to be conservative, resulting in under-estimated magnitude of association, rather than anti-conservative.

MINOR COMMENTS:

5. Abstract: The opening statement that associations of CVH with dementia "remain to be explored in cohort studies" seems misleading, given that the authors cited a number of relevant prior publications on CVH and dementia or cognitive impairment or cognitive decline from cohort studies (refs 15-19 cited in the Introduction and refs 28 and 36). I suggest the authors come up with a more appropriate statement for the first sentence of the Abstract.

6. Abstract: The authors stated that the cohort study included 1,449 participants followed from mid-life to late-life. I suggest the authors clarify that 1,449 participants were followed from midlife to late-life (1998), and then a subset of 744 were followed further into late-life (2005-2008). Otherwise, readers of the abstract may wonder why there were relatively few cases of incident dementia (47 cases) identified at the 2005-2008 exam, after 61 incident cases were identified at the 1998 exam.

7. Abstract: I suggest the authors consider including hazard ratios of dementia for intermediate CVH in the abstract, in addition to the hazard ratios for ideal CVH.

8. Abstract: The Conclusions sentence seems slightly mis-matched to the findings. I think the findings supported lower risk of dementia among people who had good CVH in mid-life (Table 2), those who had good CVH in either mid-life or late-life (Table 4), and those who had good CVH in both mid-life and late-life (Table 4). I do not think the authors reported any finding to support the statement that "improving CVH from mid- to late-life" is associated with lower risk of dementia. I suggest the authors re-work this Conclusions statement to accurately reflect their findings.

9. Introduction: After citing refs 15-19, which are cohort studies about cardiovascular health and risk of dementia, the authors assert that "the associations between patterns of CVH metrics from mid- to late-life and risk of dementia remain to be clarified." This sounds vague. What, specifically, remains to be clarified? I suggest the authors revise this sentence to be more specific about the gap in knowledge that remains after refs 15-19, which the present study will address.

10. Methods, Figure 1: The sample assessed for dementia in 1998 was 1,449 people. The sample in whom late-life CVH was measured in 1998 was 1,426 people. Why are these numbers different? What happened to the 23 people who were assessed for dementia in 1998 but who were not assessed for late-life CVH in 1998? Also, I suggest that the authors note more clearly in Figure 1 that mid-life CVH was assessed in 1972-1987 and that late-life CVH was assessed in 1998.

11. Methods: The authors calculated incidence rates of dementia as number of dementia cases divided by number of person-years of follow-up. This method may be OK, but does not account for the competing risk of death, and therefore the comparison of incidence rates across levels of CVH may be biased due to different rates of death across CVH levels. To overcome this limitation, the authors may want to consider estimating cumulative incidence of dementia from Cox proportional hazards models that account for death as a competing risk. See Andersen PK, et al. Int J Epidemiol. 2012;41:861-870. (In contrast, the hazard ratios for dementia are fine without accounting for the competing risk of death.)

12. Methods: The authors state that "the proportional-hazards assumption was verified using time-dependent covariates …" If I understand the authors' intent correctly, they added a timeXexposure interaction to the Cox model to allow for the exposure coefficient (hazard ratio) to vary over follow-up time. If so, the proper terminology is "time-dependent coefficient," not "time-dependent covariate." For an explanation of this distinction and of the appropriate method for adding a timeXexposure interaction to the Cox model, see Therneau T, et al. April 2, 2020. cran.r-project.org/web/packages/survival/vignettes/timedep.pdf.

13. Methods, Table S1: If the journal will allow it, I suggest incorporating Table S1 into the main manuscript instead of putting it in a supplement, as this table is critical for understanding the exposure measures. If the journal limits the main manuscript to a total of 5 tables + figures, then one possibility would be to merge Tables 2 and 3 into a single table to make room for one more table.

14. Results, Table 1: The statistical comparisons of characteristics of analytical sample #1 with characteristics of analytical sample #2, resulting in P values, are probably not strictly valid, because the two samples are not independent, but the methods for calculating the P values (t tests, chi-square tests) probably assume independent samples. If I understand correctly, the participants in analytical sample #2 are a subset of participants in analytical sample #1. The P values are not really meaningful or important, anyway, because the purpose of Table 1 is simply to describe the characteristics of the sample, not to make inference to a population. I suggest the authors omit P values from Table 1 and from the corresponding paragraph in the Results.

15. Results, Table 1: The percentage of participants in the intermediate category of smoking seems quite high, given that this category represents quitting smoking within the past 1 year, which is quite a narrow window given that smokers early in life may quit at any time over the course of decades. How did the authors determine whether former smokers had quit within the past 1 year?

16. Discussion, 3rd paragraph: The authors state that "We found no evidence for the association between late-life CVH metrics and the risk of dementia." While this may be because late-life CVH are less strongly associated with dementia, relative to mid-life (as the authors note), it may also be because of low precision due to small numbers of incident dementia cases in the analysis, relative to the estimated effect sizes (Table 3). The 95% CIs are very wide. I suggest the authors consider whether the precision is adequate in Table 3 for ruling out an effect of late-life CVH, and comment on this in the Discussion.

-----------------------------------------------------------

Any attachments provided with reviews can be seen via the following link:

[LINK]

25 Aug 2020

Submitted filename: Response to reviews_2020Aug25.doc

Click here for additional data file.

28 Sep 2020

Dear Dr. Liang,

Thank you very much for submitting your revised manuscript "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based study" (PMEDICINE-D-20-01344R2) for consideration at PLOS Medicine.

Your revision was evaluated by a senior editor and discussed among all the editors here. It was also discussed with the academic editor, and sent to the original reviewers. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that still we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' remaining comments. Obviously we cannot make any decision about publication until we have seen the newly revised manuscript and your response.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Oct 05 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Thomas McBride, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1- Thank you for agreeing to make your data available. However, PLOS policy does not allow authors to be the primary contact for data access. Please provide a different contact for researchers applying for data access, e.g. a member of the CAIDE data management and maintenance committee who is not a study author.

2- Thank you for providing your pre-specified analysis plan and noting the changes or additions made. In the “Difference in analyses between planned and performed” section, please also note the reasons for the change (e.g. unexpected pattern in the data or in response to a reviewer request).

3- Title: please include the country “Cardiovascular health metrics from mid- to late-life and risk of dementia: a population based study *in Finland*”

4- Abstract Methods and Findings, second sentence: is it more accurate to say “We defined and scored *six of the seven* components of global CVH metrics…”?

5- Please provide p-values alongside 95% CIs throughout the manuscript.

6- Abstract Conclusions: Please address the study implications without overreaching what can be concluded from the data; the phrase "In this study, we observed ..." may be useful. Similarly for the first sentence of the Discussion Conclusions.

7- Was informed consent written or verbal? Please specify when mentioned in the Methods.

Comments from the reviewers:

Reviewer #1: We thank the authors for addressing the points raised in the previous review round, in particular the concern relating to competing risks. The additional Fine-Gray model results appear to generally affirm the previously-observed correlations regarding dementia outcomes between poor & intermediate/ideal CVH, for the main global CVH analysis.

However, the relatively low numbers of dementia cases, especially after stratification into subgroups as presented in Figure 2, has resulted in wide hazard ratio confidence intervals that leave some room for doubt, as well as seemingly counterintuitive findings such as intermediate/ideal biological CVH reported to have a HR of >7 compared to poor biological CVH. While possible explanations about biological CVH components such as blood pressure and total cholesterol have been suggested in the discussion, it would seem that the more fine-grained findings would indeed be well-served through larger-scale studies.

Nonetheless, the manuscript appears improved especially having taken into account suggestions from all reviewers, though there remain some minor issues:

1. Further details about the Fine-Gray competing risk model might be provided, possibly in supplementary material. In particular, how was the risk of death incorporated/quantified in the model?

2. The added statement of APOE being a genetic factor for dementia and being involved in lipid and lipoprotein metabolism (Page 12) might be supported by suitable references.

3. From S1 table, it appears that intermediate smoking (vs. poor) is also significantly associated with reduced risk of dementia (Page 14).

4. "having quitted more than one year" might be "having quitted for more than one year" (Page 9), and "Compared with people having the poor global CVH metrics in midlife" might be "Compared with people having poor global CVH metrics in midlife" (Page 13)

Reviewer #2: I appreciate the authors' thorough and responsive revision, including the authors' separation of global CVH metrics into behavioral and biological metrics and the addition of the S1 and S2 tables. I have only a few remaining minor comments:

1) In the introduction, the AHA's simple 7 are introduced, and, separately, the authors introduce that they will be evaluating CVH metrics. At the point of the introduction, it would be helpful to introduce that this paper will be evaluating the Life's Simple 7 factors, categorized as behavioral (sans diet - important to mention here) and as biological. With the recent revisions, the introduction of the Life's Simple 7 here doesn't quite tie into the rest of the paper because it isn't clear at the stage of the intro whether authors are going to evaluate the Life's Simple 7 verbatim or some other CVH metrics. Fine to refer to the Life's Simple 7 but make it clear that this is the background for the work, and what is being done in this paper, and how it's similar/different. This definition comes in the "Definition of CVH metrics" section later in more detail, but at the stage of the intro, I recommend you add a simple short sentence summarizing this exposure and how it's different/similar to Simple 7.

Reviewer #3: The novelty of the study lies in the investigation of pattern of change in cardiovascular health from midlife to late life with incident dementia. However and despite the answers of the authors, to my view, the paper still suffers from the following major limitations:

- the analysis relies on 6 instead of 7 items of the cardiovascular health; the definition of the diabetic metric is incomplete

- there is too much attrition over time: 50% of the eligible population is lost from 1998 to 2005

- attrition can be handled with IPW technics but this was not done

- the change analysis (which is the novel aspect of the paper) is based on 47 incident dementia cases only

- the results are hard to follow and to interpret, especially those on the change in cardiovascular health.

Reviewer #4: In my opinion, the authors revisions in response to peer review comments—mine and other reviewers—resulted in an improved manuscript. I have no major comments and a few minor comments:

MINOR COMMENTS:

1. Table 2 (formerly Table 1) is now more informative then before about the nature of the sample. I still think the hypothesis test and P values are not very meaningful, given that the purpose of this table is simply to describe the characteristics of the sample, not to make inference to the population. The authors could consider simplifying Table 2 by omitting the P value columns.

2. The addition of cumulative incidence estimates accounting for competing risk of death to Tables 3 and 4 is informative. I suggest clarifying a few details about these estimates: (a) The length of follow-up for cumulative incidence should be reported, such as "20-year cumulative incidence," or whatever number of years of follow-up the estimates correspond to. (b) The word "rate" is not needed for cumulative incidence and could be omitted from the table column heading; the measure is "X-year cumulative incidence (%)," indicating the total incidence over the time span, not a rate. (c) Whether the cumulative incidence was adjusted for Model 1 or Model 2 covariates should be specified.

3. Figure 2 is interesting, but highlights clearly the lack of precision for estimating hazard ratios across so many different combinations of midlife and late life cardiovascular health levels (due to very small sample size for most groups). I know I suggested the authors undertake a more thorough analysis of the combination of midlife and late life cardiovascular health levels, and I am grateful they did so, but now I can see that the sample size is inadequate for such analysis to be very meaningful. The confidence intervals for most bars in Figure 2 are so wide that it is difficult to draw conclusions. Showing the confidence intervals visually instead of with numbers would enhance the figure, but would be difficult to show in the 3-dimensional layout. Thus, I suggest the authors consider revising Figure 2 into a 2-dimensional layout and show the height of the confidence intervals visually on the vertical axis rather than as printed numbers over the bars. Finally, the resolution of Figure 2 was very poor, such that the figure was nearly impossible to read. The authors should provide a high-resolution figure for publication.

4. Tables S1 and S2 are informative because they shed light on the heterogeneous influence of the 6 factors on risk for dementia. Table S2, in particular, helps us see one potential explanation for the late-life cardiovascular health score not being associated with lower dementia risk—"ideal" levels of late-life blood pressure, blood cholesterol, and body mass index are all associated with higher dementia risk, not lower risk. This is in line with prior evidence and especially helps to explain the findings illustrated in the bottom panel of Figure 2 (biological subscore). The authors addressed this briefly in the first paragraph of the Discussion; they could consider expanding this discussion a bit to highlight specific factors that behave differently in late life than in midlife for predicting dementia risk, and implications for using/interpreting the AHA cardiovascular health summary score for late-life data.

5. I agree with the authors' decision to omit the analysis of population attributable risk, for the reasons they gave in their response to Reviewer #1 comment (5).

--Evan L Thacker, PhD, Brigham Young University

Any attachments provided with reviews can be seen via the following link:

[LINK]

5 Oct 2020

Submitted filename: Final_Response to reviews_2020Oct5.doc

Click here for additional data file.

27 Oct 2020

Dear Dr. Liang,

Thank you very much for re-submitting your manuscript "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based study in Finland" (PMEDICINE-D-20-01344R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email.

Our publications team (plosmedicine@plos.org) will be in touch shortly about the production requirements for your paper, and the link and deadline for resubmission. DO NOT RESUBMIT BEFORE YOU'VE RECEIVED THE PRODUCTION REQUIREMENTS.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Nov 03 2020 11:59PM.

Sincerely,

Thomas McBride, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1- Apologies for not asking last round, please add “cohort” to the study design. “… a population-based *cohort* study in Finland”.

2- In the Abstract Methods and Findings, please include the percentage of male participants alongside the mean age.

3- In the first sentence of the Abstract Conclusions, please edit to read: “... from midlife onwards is associated with a reduced risk of dementia *as compared with people with poor CVH metrics*.” or simlar.

4- Please add a similar phrase to point 5 of the Author summary, to make clear what the reduced risk of dementia is in comparison to.

5- Thank you for adding p-values throughout the text. In the tables, rather than a footnote to indicate when p< 0.05 or 0.01, please include the precise p-value in the table, down to p=0.001 (and p<0.001 for any smaller values).

6- Attrition could be added to the study limitations listed in the Abstract.

9 Nov 2020

Dear Dr. Liang,

On behalf of my colleagues and the academic editor, Dr. Raquel C. Gardner, I am delighted to inform you that your manuscript entitled "Cardiovascular health metrics from mid- to late-life and risk of dementia: a population-based cohort study in Finland" (PMEDICINE-D-20-01344R4) has been accepted for publication in PLOS Medicine.

PRODUCTION PROCESS

Before publication you will see the copyedited word document (within 5 business days) and a PDF proof shortly after that. The copyeditor will be in touch shortly before sending you the copyedited Word document. We will make some revisions at copyediting stage to conform to our general style, and for clarification. When you receive this version you should check and revise it very carefully, including figures, tables, references, and supporting information, because corrections at the next stage (proofs) will be strictly limited to (1) errors in author names or affiliations, (2) errors of scientific fact that would cause misunderstandings to readers, and (3) printer's (introduced) errors. Please return the copyedited file within 2 business days in order to ensure timely delivery of the PDF proof.

If you are likely to be away when either this document or the proof is sent, please ensure we have contact information of a second person, as we will need you to respond quickly at each point. Given the disruptions resulting from the ongoing COVID-19 pandemic, there may be delays in the production process. We apologise in advance for any inconvenience caused and will do our best to minimize impact as far as possible.

EARLY VERSION

Please note that an uncorrected proof of your manuscript will be published online ahead of the final version, unless you opted out when submitting your manuscript. If, for any reason, you do not want an earlier version of your manuscript published online, uncheck the box. Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us as soon as possible if you or your institution is planning to press release the article.

PRESS

A selection of our articles each week are press released by the journal. You will be contacted nearer the time if we are press releasing your article in order to approve the content and check the contact information for journalists is correct. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact.

PROFILE INFORMATION

Now that your manuscript has been accepted, please log into EM and update your profile. Go to https://www.editorialmanager.com/pmedicine, log in, and click on the "Update My Information" link at the top of the page. Please update your user information to ensure an efficient production and billing process.

Thank you again for submitting the manuscript to PLOS Medicine. We look forward to publishing it.

Best wishes,

Thomas McBride, PhD

Senior Editor

PLOS Medicine

plosmedicine.org