Dementia is a growing global health challenge. It is currently the sixth leading cause of death, and a major contributor to disability and dependency among older people (World Health Organization [WHO], 2021; GBD 2021 Causes of Death Collaborators, 2024). In 2019, an estimated 55 million people were living with dementia worldwide, a number projected to reach 139 million by 2050 (World Health Organization [WHO], 2021). In Norway, prevalence among those 70 years and older is estimated at 14.6%, with projections suggesting a rise from 101,000 cases in 2020 to nearly 237,000 by 2050 (Gjøra et al., 2021).

Mild cognitive impairment (MCI), affecting around 24% of the global population over 60 years (Salari et al., 2025) and 35.3% of Norwegians 70 years and older (Gjøra et al., 2021), is often considered a prodromal stage of dementia. However, MCI is a heterogeneous condition with multiple potential etiologies, including reversible factors such as depression, excessive alcohol intake, or other medical and psychosocial conditions (Petersen et al., 2018). Despite this variability, MCI remains a valuable early indicator of underlying neurodegenerative disease.

Currently there is no effective treatment for dementia, although research in the field is advancing rapidly. Consequently, early identification of individuals at risk and prevention through modifiable risk factors have become a public health priority (Livingston et al., 2024). The Lancet Commission have identified 14 modifiable risk factors that account for approximately 45% of all cases of dementia worldwide (Livingston et al., 2024). Even small delays in disease onset could substantially reduce the burden of dementia at both individual and societal levels (Lewis et al., 2014).

A growing body of evidence shows that cardiovascular (CVD) risk factors play a central role in the development of cognitive decline and dementia. Conditions such as hypertension, hyperlipidemia, diabetes, obesity, smoking and physical inactivity are among risk factors identified by the Lancet Commission. Importantly, many of these factors exert their greatest impact in midlife, long before clinical symptoms occur (Livingston et al., 2024). Moreover, these risk factors often co-occur and interact with each other, making it difficult to isolate individual effects.

Consequently, composite measures of cardiovascular health have been developed to capture the combined impact of these factors. And several of these have also been associated with cognitive outcomes, such as the Cardiovascular Risk Factors, Aging and Dementia (CAIDE) dementia risk score (Kivipelto et al., 2006; Fayosse et al., 2020), the Lifestyle for Brain Health (LIBRA) index (Schiepers et al., 2018; Sewell et al., 2026) and the Updated Lifestyle for Brain Health (LIBRA2) score (Rosenau et al., 2025), the Systematic Coronary Risk Evaluation 2 (SCORE2) (Conroy et al., 2003; SCORE2 working group and ESC Cardiovascular risk collaboration, 2021; Zheng et al., 2023), the Framingham Risk Score (D’Agostino et al., 2008; Fayosse et al., 2020) and the Framingham Stroke Risk Profile (D’Agostino et al., 1994; Yuan et al., 2024), and composite indices such as Life’s Simple 7 (LS7) (Lloyd-Jones et al., 2010; Huang et al., 2026) and the updated Life’s Essential 8 (LE8) (Lloyd-Jones et al., 2022; Wang et al., 2024). A recent Norwegian study comparing multiple cardiovascular and lifestyle-based risk models (including CAIDE, ANU-ADRI, CogDrisk, LIBRA and LIBRA2) reported that all models were associated with dementia (Stubs et al., 2025). However, the same study concluded that none of the models performed better than a simple model including age and education alone (Stubs et al., 2025). And although many of these tools show statistical associations with cognitive decline or dementia across different populations, no single model has emerged as a gold standard for dementia risk assessment (Mohanannair Geethadevi et al., 2023). Taken together, these findings support the overarching concept that vascular and cardiometabolic risk factors contribute to cognitive decline and dementia, even though the optimal way to operationalize the risk remains unclear.

Building on this evidence, the present study uses a cardiovascular risk score developed and validated for use in the Norwegian population as a proxy for overall cardiovascular health. The NORRISK 2 score is a risk model developed and validated to predict 10-year risk of fatal- and non-fatal incident acute myocardial infarction or cerebral stroke in individuals without known cardiovascular disease (Selmer et al., 2020). The score comprises seven metrics (age, systolic blood pressure, use of antihypertensive medication, total cholesterol, HDL-cholesterol, current smoking status, and family history of early coronary heart disease), and the result is expressed as a percentage representing the estimated risk of experiencing the outcome.

Current prevention guidelines emphasize targeting risk factors during midlife (Livingston et al., 2024). Studies aiming to identify early predictors of cognitive decline must therefore have sufficiently long follow-up periods to include the onset of dementia or MCI, which can occur decades after initial risk exposure (Jia et al., 2024). The HUNT Study is a large population-based cohort with up to 37 years of follow-up, and offers a unique opportunity to explore these associations (Holmen et al., 1990; Holmen et al., 2003; Krokstad et al., 2013; Krokstad et al., 2022; Åsvold et al., 2023; Næss et al., 2024).

In recent years, sex differences in cognitive aging have gained increasing attention, with evidence pointing to distinct patterns in the development and progression of dementia and MCI between men and women. Although females are more likely to develop Alzheimer’s disease (Alzheimer’s Association, 2023), they often perform better on episodic memory tasks and show slower age-related memory decline, particularly in earlier stages such as MCI (Anstey et al., 2021; Bloomberg et al., 2021). This paradox has led researchers to investigate whether modifiable risk factors may affect males and females differently. One such study found that the number of risk factors present moderates sex differences in memory performance (LaPlume et al., 2022). Among individuals with few risk factors, men showed greater age-related memory decline than women. But among those with multiple risk factors, both sexes showed similar decline, and the female memory advantage disappeared (LaPlume et al., 2022). These findings underscore the importance of considering sex-specific pathways in dementia risk, particularly in relation to CVD health.

In this study, we used data from the HUNT Study to examine whether the NORRISK 2 score is associated with dementia and MCI in males and females after a 22-year follow-up period. Using a risk score developed and validated for the Norwegian population within a large Norwegian cohort provides a contextually relevant and population-appropriate measure for examining the longitudinal associations between cardiovascular burden and both dementia and mild cognitive impairment.

In total, 6,971 participants (57.6% females) were included in the study. The mean age at HUNT2 was 55.4 years for males (range 47–78, SD 5.8 years) and 56.5 years for females (range 47–82, SD 6.6 years). The study population was ethnically homogenous; among participants with available data, 98.7% reported Norwegian ethnicity. The males in the study had a mean NORRISK 2 score of 8.4% (range 0.9–25.0, SD 5.0) at HUNT2, while the females had a mean NORRISK 2 score of 4.9% (range 0.5–24.8, SD 4.3). Descriptive statistics of the study population at baseline (HUNT2) are presented in Table 1.

At HUNT4 70+, 12.2% of all males and 15.4% of all females had dementia, and 37.9% of males and 32.2% of females had MCI. Table 2 shows absolute numbers and RRR with 95% CI for MCI and dementia for males and females. Results are presented unadjusted and adjusted for covariates separately.

In unadjusted analyses, there was a clear association between NORRISK 2 score and dementia in both sexes. Although mean NORRISK 2 scores were higher in men, the associations between NORRISK 2 and dementia and MCI were steeper in women, indicating a stronger relative risk per unit increase in the score. For a one-percent increase in NORRISK 2, the relative risk of developing dementia (vs. cognitively unimpaired) increased by 14% for males (RRR = 1.14; 95% CI 1.12–1.17) and 28% for females (RRR = 1.28; 95% CI 1.25–1.31). Adjustments for other covariates showed that age was the only covariate that altered the association between NORRISK 2 and dementia notably. When adjusting for age, the strength of the association was weakened from 14% to 5% for males and from 28% to 12% for females, suggesting that a large part of the association is due to the age component in the NORRISK 2 score, or possibly also that age may act as a confounder or effect modifier in this relationship. Adjustment for education did not meaningfully influence the association. Other covariates similarly had little impact on the association.

The NORRISK 2 score was also associated with MCI, although with a more modest effect size compared to dementia. For each one-percent increase in the NORRISK 2 score, the relative risk of having MCI (vs. cognitively unimpaired) increased by 4% for males (RRR = 1.04; 95% CI 1.02–1.05) and 10% for females (RRR = 1.10; 95% CI 1.08–1.12). Adjustment for covariates did not result in a significant change in the effect estimates for any of the covariates.

The stacked area plots demonstrated marked sex differences in the predicted distribution of outcomes across the NORRISK 2 scale. For females, the curve showed a markedly steeper trajectory, particularly between scores of 10% and 20%, compared to males, indicating a sharper rise in dementia probability with higher cardiovascular risk burden (Figure 2). Investigation of any potential non-linear effects using fractional polynomials did not result in notably different results, data not shown.

Using data from The HUNT2 Survey and The HUNT4 70+ sub-study we investigated whether the CVD risk score NORRISK 2 were associated with dementia and MCI after a 22-year follow-up period. Our results showed that a higher NORRISK 2 score was associated with increased risk of developing dementia. A weaker association was observed for MCI. The associations were evident in both men and women, with the strongest associations being found for women.

To the best of our knowledge, no previous studies have investigated the prospective relationship between NORRISK 2 score and risk of dementia and MCI. However, our findings align with a broader body of evidence showing that several other cardiovascular risk scores and health indices are associated with later-life cognitive decline and dementia (Zheng et al., 2023; Vishwanath et al., 2024; Harrison et al., 2025; Yuan et al., 2024; Song et al., 2020). Together, these studies underscore the importance of cardiovascular health across the life course as a determinant of cognitive aging, supporting the plausibility of the association observed in our analyses.

Previous research have shown that the link between cardiovascular health and cognitive outcomes may differ between populations. In a study comparing associations between cardiometabolic risk and cognitive function in the USA and China, researchers found a significant association in the American population, but these findings were not replicated when studying a Chinese population (Wu et al., 2024). These results imply that there may be regional or genetic differences and highlight the importance of conducting research across diverse populations. A recent comparative study of several CVD-related dementia risk models, also using HUNT data, showed that these scores were associated with dementia, but did not outperform a simpler model including only age and education. Our findings complement this work by evaluating the association between the NORRISK 2 score, and dementia and MCI in Norway. In contrast to the previous study, we draw on a longer follow-up period and additionally assess MCI, allowing us to capture both early and later stages of cognitive decline.

Moreover, our study revealed a stronger association between the CVD risk score and both dementia and MCI in females compared to males. Similar sex-specific patterns have also been reported in other studies (Yuan et al., 2024; Vishwanath et al., 2024). This may suggest that females are more vulnerable to the cognitive consequences of CVD risk factors, potentially due to hormonal influences, differences in vascular physiology, or differences in exposure to risk factors. However, given that females have a higher lifetime risk of dementia, it can also be attributed to survival bias. Further research is needed to clarify the underlying mechanisms and to explore whether sex-specific preventive strategies should be implemented.

In the adjusted models, age was the only covariate that altered the association between the NORRISK 2 score and dementia and MCI significantly, highlighting its potential role as a key confounder. This is a plausible finding, as age is a well-established risk factor for both CVD and cognitive impairment, in addition to influencing both the components in the NORRISK 2 score and the NORRISK 2 score itself. These findings were further supported by age-separated sensitivity analyses. Although the overall analyses showed relatively strong associations between the NORRISK 2 score and the outcomes, age-separated analyses showed more modest effect sizes. Performing separate analyses per age group means that we largely adjust for age. Hence, as expected, the RRRs in these separate analyses are lower than the unadjusted RRRs in Table 2 and closer to the RRRs adjusted for age in Table 2. In both males and females, associations were stronger in the younger groups compared to the older age groups, suggesting that death as a competing risk may be affecting the results. These findings further highlight the importance of considering both sex and age when assessing long-term dementia risk. Although Nord-Trøndelag has a lower number of people with higher education than the national average, this difference did not appear to bias our estimates, as adjustment for education did not materially change the association. The overall minimal impact of other covariates suggests that the observed associations are relatively robust across different subgroups, while the effect of age and sex should be carefully considered in future use or modifications of the risk score.

The inclusion of participants with extreme values of the NORRISK 2 (NORRISK 2 > 25) confirmed the robustness of the primary findings, with only a modest reduction in effect size (0.5–2.2%). This indicates that the exclusion of extreme values in the main analyses did not alter the observed results and supports the validity of the risk score across a broader range of exposure levels. However, the small decrease in effect size may result from the presence of outliers or greater variability at the extremes, which should be considered in future use or validations of the score.

This study has several strengths. It has a large sample size and a population-based inclusion of both home-dwelling and institutionalized participants, a combination that is relatively rare in this field. It has a long follow-up period of 22 years, which is especially important given the long preclinical phase of dementia. Cognitive diagnoses are based on standardized criteria, and cognitive assessments were performed consistently throughout the entire sample, including a consensus process by a clinical expert panel. This is a strength compared to studies using less rigorous registry-based diagnoses. In addition, access to most of the known risk factors for dementia in the dataset reduces the likelihood of residual confounding and strengthens the validity of the observed observation.

However, some limitations must be acknowledged. First, the reliance on self-reported data in some of the components of the NORRISK 2 score may have introduced bias. Second, since cognitive status is only measured in HUNT4 70+, we only know the proportion of participants with dementia and cognitive impairment at that specific time point, but not when they developed the condition, nor whether the NORRISK 2 score at HUNT2 could be affected by a potential preclinical stage of dementia. Consequently, we cannot guarantee that our findings are free from reverse causation. Dementia develops over a long prodromal period during which blood pressure and metabolic parameters may begin to decline (Peters et al., 2020), and such early dementia-related physiological changes could artificially lower NORRISK 2 score and thereby attenuate the observed associations. However, with a follow-up time of 22 years, it is unlikely that this potential bias would be strong enough to impact the results significantly. Third, participation in HUNT4 was lower among individuals with poorer health, raising the possibility of selection bias, which may have resulted in underestimation of the observed associations. Fourth, in a geriatric population with multiple comorbidities, survival bias and the competing risk of death must also be considered. Inclusion of only participants 70 years and older in HUNT4 70+ makes the study prone to survival bias. Further, there may be a missed opportunity to diagnose MCI or dementia before dropout due to death, as people with dementia tend to die earlier than their peers. In addition, participants with a high NORRISK 2 score are more likely to die prematurely from CVD, thus not living long enough to develop cognitive impairment. Fifth, the largely ethnically homogenous composition of the HUNT study may limit the generalizability of the results to people of non-European ancestry. In addition, choosing a risk score intended for a Norwegian population may also limit generalizability to populations that are not closely comparable to the Norwegian population.

The findings of this study contribute to a growing body of evidence indicating that cardiovascular health plays an important role in the development of later life cognitive impairment. By demonstrating that higher CVD risk is associated with both dementia and MCI in a large Norwegian cohort, our results strengthen the evidence that maintaining favorable cardiovascular health throughout adulthood may support healthier cognitive aging. Further, the stronger association observed in females highlights an important sex-specific pattern. Given the high prevalence of dementia among females, these findings underscore the need for further research focusing specifically on dementia in females. Improved understanding of sex-specific pathways may help make more tailored prevention strategies. Overall, the results emphasize the importance of early and sustained efforts to reduce cardiovascular risk as part of a broader dementia-prevention plan and point to the value of incorporating sex-specific perspectives in future research and public health planning.

In conclusion, this study found that the NORRISK 2 score, originally developed to estimate CVD risk, is also associated with dementia and MCI in both males and females, with the strongest association observed in females.