Sex- and age-specific impacts of smoking, overweight/obesity, hypertension, and diabetes mellitus in the development of disabling dementia in a Japanese population

Background Sex- and age-specific impacts of cardiovascular risk factors on the development of dementia have not been well evaluated. We investigated these impacts of smoking, overweight/obesity, hypertension, and diabetes mellitus on the risk of disabling dementia. Methods The study participants were 25,029 (10,134 men and 14,895 women) Japanese aged 40–74 years without disabling dementia at baseline (2008–2013). They were assessed on smoking status (non-current or current), overweight/obesity (body mass index ≥25 kg/m2 and ≥30 kg/m2, respectively), hypertension (systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg or any antihypertensive medication use), and diabetes mellitus (a fasting serum glucose ≥126 mg/dL, non-fasting glucose ≥200 mg/dL, hemoglobin A1c ≥6.5% by the National Glycohemoglobin Standardization Program or glucose-lowering medication use) at baseline. Disabling dementia was identified as the level of care required ≥1 and cognitive disability grade ≥IIa according to the National Long-term Care Insurance Database. We used a Cox proportional regression model to estimate hazard ratios and 95% confidence intervals (95% CIs) of disabling dementia according to the cardiovascular risk factors and calculated the population attributable fractions (PAFs). Results During a median follow-up of 9.1 years, 1,322 (606 men and 716 women) developed disabling dementia. Current smoking and hypertension were associated with a higher risk of disabling dementia in both sexes, whereas overweight or obesity was not associated with the risk in either sex. Diabetes mellitus was associated with a higher risk only in women (p for sex interaction = 0.04). The significant PAFs were 13% for smoking and 14% for hypertension in men and 3% for smoking, 12% for hypertension, and 5% for diabetes mellitus in women. The total PAFs of the significant risk factors were 28% in men and 20% in women. When stratified by age, hypertension in midlife (40–64 years) was associated with the increased risk in men, while diabetes mellitus in later-life (65–74 years) was so in women. Conclusions A substantial burden of disabling dementia was attributable to smoking, and hypertension in both sexes and diabetes mellitus in women, which may require the management of these cardiovascular risk factors to prevent dementia.


Introduction
Dementia has emerged as a major public health issue worldwide [1]. In Japan, a super-aging society, 15% of older people aged 65 and over had dementia in 2012, and the prevalence is estimated to be more than double by 2060 [2]; thus, the prevention of dementia is urgently needed.
The 2020 Lancet Commission reported that approxi-mately 40% of worldwide dementia could be prevented by 12 risk factors identified as having consistent evidence for the association from meta-analytic reviews: less education for early life (<45 years), hearing loss, traumatic brain injury, hypertension, alcohol consumption (²21 drinks/week), obesity (body mass index [BMI] ²30 kg/ m 2 ) for middle life (45-65 years), smoking, depression, social isolation, air pollution, physical inactivity, and diabetes mellitus for late life (²65 years) [3].
Of the above 12 risk factors, smoking, obesity, hypertension, and diabetes mellitus are mandated for annual screening in the Japanese health system [4], and are well-known risk factors for cardiovascular disease [5]. Cardiovascular disease often occurs before dementia, including vascular dementia as well as Alzheimer's disease [6,7]. Vascular dementia accounts for a higher percentage in Japan than in Western countries [8], provably due to the high burden of stroke in Japan. Thus, preventing and controlling for cardiovascular risk factors could reduce risk of dementia. However, only one study of 8,563 older Japanese aged ²65 years estimated the burden of dementia attributable to these risk factors via the population attributable fraction (PAFs) [9]; approximately one-third of disabling dementia cases were attributed to seven major risk factors (smoking, physical inactivity, hypertension, diabetes mellitus, obesity with BMI of ²30 kg/m 2 , severe psychological distress, and low educational attainment).
That study, as well as the 2020 Lancet Commission, did not provide the sex-specific impacts of cardiovascular risk factors on dementia. The prevalence and incidence of vascular dementia was lower in women than in men [10,11], while that of Alzheimer's disease was reported to be higher in women than in men probably because women live longer than men [12]. Women with a history of cardiovascular diseases are more likely to develop Alzheimer's disease, but not vascular dementia compared with men [13]. The sex difference in the prevalence and incidence of dementia subtypes suggests that some risk factors associated with risk of dementia may differ between men and women. Raised blood pressures and diabetes mellitus were associated stronger with risk of dementia in women than in men [14,15]. On the other hand, smoking rate, higher blood pressure levels, and higher prevalence of diabetes mellitus were observed in men than in women [16,17]. Especially, smoking rate was much higher in Japanese men than in Japanese women [17]. These findings warrant further studies to evaluate the sex-specific burden of dementia attributable to cardiovascular risk factors.
In addition, the incidence of dementia increases with age while levels of cardiovascular risk factors vary with age. The Framingham Heart Study reported that systolic blood pressure level and diabetes mellitus at age 55, diabetes mellitus at ages 70, 75, and 80 were regarded as vascular risk factors for dementia [18]. Therefore, it is necessary to evaluate the sex-and age-specific risk factors and their impacts and attributable to dementia.
Our prior hypothesis is that cardiovascular risk factors and their impacts on the development of dementia vary by sex and age.
In the present prospective cohort study, we examined the sex-and age-specific impacts of smoking, overweight/ obesity, hypertension, and diabetes mellitus on the risk of dementia in a Japanese population.

Design, setting, and participants
We used the data from the collaborative research between Yao City, Osaka Prefecture, and Osaka University, which aimed to promote health among Yao City residents. The population size of Yao City, Osaka Prefecture (southwestern Japan), was 266,143 citizens (128,974 men and 137,169 women) on 31 March 2008. The percentage for ages 40-74 years accounted for 46% of the total census population, which was similar to the percentage in the 2010 national census population. Approximately 10% of people aged 40-74 years (n = 12,122; 4,824 men and 7,298 women) received health check-ups under the national health insurance in fiscal 2008.
We used the data from the annual national health checkup, the national long-term care insurance, and notification of moving-out, moving-in, and death in this collaborative research. The present study was a prospective cohort study of individuals aged 40-74 years who underwent health check-ups under the national health insurance between April 2008 and March 2013. There were initially 28,352 participants at baseline. Individuals were excluded if they had disabling dementia (n = 97) or had a history of cardiovascular disease, including stroke and heart disease from a self-reported questionnaire (n = 2,628), or had outliers in the anthropometric measurements or laboratory tests or missing data (n = 598). Finally, 25,029 participants (10,134 men and 14,895 women) were enrolled in the present study (Fig. 1). The protocol was approved by the ethics committees of Osaka University, Yao City, and Osaka Center for Cancer and Cardiovascular Disease Prevention.

Definition of disabling dementia and follow-up
We used the data from the screening judgment results for long-term care requirement certification under the national long-term care insurance system established in 2000 [19,20].
The certification of the long-term care requirement was provided based on the evaluation results of the applicants' activities of daily living, psychophysical condition, and cognitive function from a national standardized in-home assessment by trained investigators in the municipality and a medical assessment by primary care physicians. In the certification, a person's level of required care was classified as preventive support level of 1 or 2, long-term care levels of 1 to 5, and a person's cognitive disability grade was classified as non-cognitive symptoms, I, IIa, IIb, IIIa, IIIb, IV, and M (M = needs treatments in a specialized medical facility because of serious cognitive symptoms) [19,20]. Disabling dementia was defined as the level of long-term care required ²1 and cognitive disability grade ²IIa in the present study as defined in the previous study [21]. The validation of the criteria previously confirmed that the sensitivity and specificity values were 73% and 96%, respectively, compared with the diagnosis by neuropsychiatrists [22]. The follow-up lasted until the end of December 2018 and was terminated at the first incident of disabling dementia, exit from the community, or death.

Risk factors and baseline examination
In Japan, screening for smoking, being overweight, hypertension, diabetes mellitus, and dyslipidemia have been mandated in annual health check-ups for persons aged 40-74 years since 2008 to prevent cardiovascular disease [4]. Of these risk factors, we focused on smoking, obesity (BMI ²30 kg/m 2 ), hypertension, and diabetes mellitus, which were reported as the risk factors with the most consistent evidence on the association of dementia in the 2020 Lancet Commission [3]. The present study also examined the association between overweight (BMI ²25 kg/ m 2 ) and the risk of disabling dementia because the prevalence of obesity in Japan is very low compared to that in Western countries [23].
The dataset in the present study includes information on general characteristics (sex, date of birth, and date of health check-up), a self-administered questionnaire on lifestyle (smoking status and alcohol consumption), anthropometric measurements (height, weight, and BMI), blood pressure (systolic blood pressure [SBP], diastolic blood pressure [DBP]), laboratory values (fasting serum glucose, non-fasting serum glucose, hemoglobin A1c [HbA1c], low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], and protein in the urine), medication use (antihypertensive medication, glucose-lowering medication, and lipid-lowering medication), and medical history (stroke, heart disease, and chronic kidney disease including artificial dialysis). The baseline HbA1c value was determined using the previous Japanese stand-ard substance and measurement (Japan Diabetes Society [JDS]). Therefore, the HbA1c value was estimated as the National Glycohemoglobin Standardization Program (NGSP) equivalent value calculated using the following formula: HbA1c (NGSP) (%) = 1.02 © HbA1c (JDS) (%) + 0.25% [24].

Statistical analysis
The values of baseline characteristics according to disabling dementia during follow-up in men and women were reported as means (standard deviations) for continuous variables and percentages for categorical variables. The analysis of covariance after adjustment for age (continuous) was used to compare the presence or absence of disabling dementia. Person-years were calculated as the sum of individual follow-up time until the occurrence of disabling dementia, death, exit from the community, or the end of follow-up, whichever occurred first.
Cox proportional hazards regression models were used to estimate the sex-and age-specific hazard ratios (HRs) and 95% confidence intervals (CIs) for the risk of incident disabling dementia according to risk factors (smoking, overweight/obesity, hypertension, and diabetes mellitus). We divided the participants into two groups (40-64 and 65-74 years) based on the median age (65 years) of the Environmental Health and Preventive Medicine (2023) 28:11 participants. To estimate the sex-specific HRs, the initial model was adjusted only for age, while the multivariable model was adjusted for age, smoking status (non-current or current), drinking status (non-current or current), obesity (yes or no), hypertension (yes or no), diabetes mellitus (yes or no), LDL-C, HDL-C, and lipid-lowering medication use (yes or no). The test for effect modification by sex in the analysis of each risk factor was conducted with interaction terms generated by multiplying the categorical variable of each risk factor (0 or 1) by sex (0 or 1). In the multivariable analysis, the test for effect modification by age were performed with interaction terms generated by multiplying the categorical variable of each risk factor (0 or 1) by age group (0 or 1).
We estimated the sex-and age-specific multivariableadjusted PAFs (95% CIs) of incident disabling dementia according to each risk factor, which is the proportion of disabling dementia events in the population that would be attributable to a particular risk factor. We used a categoryspecific attributable fraction with the formula pdi (1 ¹ 1/ RRi), which produces internally valid estimates when confounding exists [28], where pdi represents the proportion of total events in the population arising from the ith exposure category and RRi is the multivariable-adjusted HR for the ith exposure category relative to the unexposed group. We also calculated approximate estimates of the 95% CI for the PAF [29]. The sum of the category-specific attributable fractions was calculated using the following formula [28]: X i pdi 1 À 1 RRi All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC, USA). Statistical significance was set at P < 0.05. Table 1 shows the baseline characteristics of 25,029 participants aged 40-74 years (10,134 men and 14,895 women) according to the absence or presence of incident disabling dementia during follow-up. In both men and women, compared with the participants without disabling dementia, those with disabling dementia were more likely to be older and high-risk individuals who suffered from hypertension. In men, those with disabling dementia had a higher rate of smoking and lower means of LDL-cholesterol level and BMI, whereas women with disabling dementia had a lower rate of smoking and a higher mean LDL-cholesterol level. In women, those with disabling dementia had more diabetes mellitus and a lower mean HDL-cholesterol level.

Results
During 211,804 person-years of follow-up (median follow-up of 9.1 years), 1,322 (606 men and 716 women) disabling dementia events occurred. The incidence of disabling dementia tended to be lower in women than in men, but the multivariable HR (95% CI) for women versus men was 0.89 (0.78-1.02). Table 2 shows the sex-specific HRs and PAFs of incident disabling dementia according to the absence or presence of cardiovascular risk factors. Current smoking and hypertension were positively and similarly associated with a higher risk of disabling dementia in both sexes (p for sex interaction = 0.94 and 0.88, respectively), while over-   Table 3 shows the sex-and age-specific HRs and PAFs of incident disabling dementia according to the absence or

Discussion
This community-based prospective study of over 25,000 dementia-free Japanese individuals showed that smoking and hypertension were associated with an increased risk of disabling dementia in both sexes and diabetes mellitus in women. Moreover, we found that hypertension was associated with the risk in midlife than in late-life among men, and diabetes mellitus was stronger associated with the risk in late-life than in midlife among women. Our estimate of PAF indicated that 28% of male disabling dementia cases and 20% of female disabling dementia cases could be prevented by modifying these risk factors. To the best of our knowledge, our study is the first to evaluate the sex-and age-specific impacts of these risk factors on the risk of disabling dementia. The smoking rate has been high in Japanese men, although the rate declined over the years from 24.2% in 2009 to 17.7% in 2019, while in contrast, the rate has been low in women from 11.7% in 2009 to 8.1% in 2019 [30]. The larger burden of dementia attributable to smoking in men than in women in the present study was due to men's higher smoking rate despite the similar association between smoking and dementia in both sexes.
In the present study, overweight or obesity was not associated with the risk of disabling dementia, regardless of sex or age, probably because of the low prevalence of overweight (24%) and obesity (3%) compared with non-Asian populations. Our result was consistent with the finding from a previous study in older Japanese [9].
Hypertension has been acknowledged to be a strong risk factor for mortality and morbidity from stroke among Japanese [17], and midlife high blood pressure has been re-ported as a significant predictor of reduced cognitive function and increased dementia incidence in later life in the Honolulu-Asia Aging Study [31,32], the Hisayama study [33], and a Chinese study [34]. In the present study, midlife hypertension in men and late-life hypertension in women associated with the risk of disabling dementia.
We found that diabetes mellitus was a strong predictor for disabling dementia in women but not in men, which was consistent with the finding from a pooled analysis of 2.3 million people comprising more than 100,000 cases of dementia in 11 Western and 3 Asian studies [15]; women with diabetes had a 19% greater risk of vascular dementia than men (the women-to-men ratio of the pooled multipleadjusted relative risks 1.19 [95% CI 1.08-1.30]; p < 0.001), but not for non-vascular dementia. Diabetic women had a larger number of other cardiovascular risk factors than diabetic men, which may be due to the greater increase in adiposity and insulin resistance associated with diabetes mellitus in women than in men [35].
Potential mechanisms for the development of dementia by smoking, hypertension, and diabetes mellitus are as follows. Smoking and diabetes mellitus increase oxidative stress which may enhance the development of arteriolosclerosis in cerebral arteries (small vessel disorders), leading to vascular dementia [36,37], and also may initiate and progress the accumulation of amyloid and tau proteins, leading to an increased risk of Alzheimer's disease [38]. Hypertension is a strong determinant of arteriolosclerosis in cerebral arteries, leading to vascular dementia [39]. There is no good explanation on the more evident contribution of diabetes mellitus to dementia in women than in men. A recent cross-sectional imaging study of older persons showed that diabetes mellitus was associated with cerebral lacunes, brain atrophy, and impaired cognitive function in women but not in men, possibly due to more oxidative stress and pro-thrombotic state in women [40].
The strengths of our study include a large populationbased prospective study of over 25,000 dementia-free individuals and the first study to systematically evaluate the sex-specific impacts of cardiovascular risk factors on the risk of disabling dementia. Our finding from the large population-based sample could be extrapolated to general Japanese populations. Moreover, we evaluated the impact of midlife risk factors for dementia, although a previous study of 8,563 older Japanese individuals aged 65 years or older reported those of late-life risk factors [9].
Our study has several limitations. First, because a detailed clinical diagnosis of dementia was not obtained, we did not evaluate the types of dementia, such as Alzheimer's disease or vascular dementia. Second, some misclassification of diagnosis for incident dementia could occur in the present study. Because our dementia criteria had sensitivity (73%) and specificity (96%) against clinical diagnosis by neuropsychiatrists [22], which unlikely biased the estimation of the association mathematically. Third, we did not examine the potential risk factors such as less education, hearing loss, traumatic brain injury, de-Environmental Health and Preventive Medicine (2023) 28:11 pression, social isolation, physical inactivity, and air pollution due to no data availability. Further studies are needed to build evidence on the associations between modifiable risk factors and the risk of dementia, especially in Asian populations.

Conclusion
In conclusion, a substantial burden of disabling dementia was attributable to smoking and hypertension in both sexes and diabetes mellitus in women, which may require the management of smoking, hypertension, and diabetes mellitus to prevent dementia.