Frontiers – Association between statin therapy and dementia risk

Volume 17 - 2026 | https://doi.org/10.3389/fphar.2026.1915633

1. College of Pharmacy and Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul, Republic of Korea

2. College of Pharmacy, Chungbuk National University, Cheongju, Republic of Korea

3. Human-Centered Artificial Intelligence Research Institute, Ewha Womans University, Seoul, Republic of Korea

4. Department of Pharmacy, Kangwon National University, Chuncheon, Republic of Korea

5. School of Pharmacy, Sungkyunkwan University, Suwon, Republic of Korea

Statins are widely prescribed for cardiovascular disease prevention in older adults; however, their association with dementia remains controversial. Given the high prevalence of both statin use and cognitive decline in aging populations, clarifying this relationship has critical implications for geriatric clinical practice and shared decision-making.

We conducted a population-based retrospective cohort study using the All of Us Research Program (discovery cohort, n = 83,355; 31,513 statin users) and UK Biobank (replication cohort, n = 173,282; 43,976 statin users). Participants were adults aged ≥40 years who self-identified as White, followed from 2006 through 2023 (discovery) and 2017 (replication). Inverse probability of treatment weighting balanced baseline characteristics. Dementia risk was estimated using Cox proportional hazards models. An exploratory mediation analysis examined low-density lipoprotein cholesterol (LDL-C) and coenzyme Q7, a protein involved in CoQ10 biosynthesis, as potential pathways linking statin use and dementia.

Statin use was associated with significantly increased dementia risk in both cohorts (discovery: adjusted HR 2.22, 95% CI 2.02–2.44; replication: adjusted HR 2.54, 95% CI 1.98–3.33). Clear intensity-dependent relationships were observed, with high-intensity statins conferring 14% greater risk than moderate-to-low intensity therapy. In the exploratory mediation analysis, the estimated indirect effect through coenzyme Q7 was statistically significant and accounted for 8.2% of the total association, whereas the indirect effect through LDL-C was not statistically significant.

Statin use was associated with higher dementia risk in both cohorts, and exploratory mediation findings suggested a modest statistical indirect association through COQ7, which should be considered hypothesis-generating. Given the observational design and conflicting evidence in the literature, these findings warrant further investigation before informing clinical decisions about statin therapy.

Dementia represents one of the most critical public health challenges in geriatric medicine. As of 2019, an estimated 57 million individuals were living with dementia worldwide, projected to rise to 152 million by 2050 (Nichols et al., 2022). As aging populations expand globally, identifying modifiable risk factors for cognitive decline has become a critical priority in the absence of definitive therapeutic interventions.

Statins have become one of the most widely prescribed medication classes globally, with well-established effectiveness in reducing low-density lipoprotein cholesterol (LDL-C) levels and cardiovascular morbidity and mortality (Ridker et al., 2016; Collins et al., 2016). More than 25% of adults aged 40 and over in the United States reported statin use, with particularly high prevalence among the older population (Matyori et al., 2023). Given the substantial overlap between those prescribed statins and those at risk for cognitive decline, there is growing interest in understanding the potential relationship between statin use and dementia.

Current evidence remains inconclusive and contradictory. In 2012, the United States Food and Drug Administration updated statin labeling to include information about potential cognitive side effects, later clarifying in 2016 that such events were generally reversible (U.S. Food and Drug Administration, 2016). Some observational studies have suggested that statin use may increase dementia risk (Roy et al., 2020; Ye et al., 2025), while others have found no association (Ott et al., 2015; Zhou et al., 2021), and recent meta-analysis reported even protective effects (Westphal Filho et al., 2025).

The controversial nature of this relationship has led to diverse mechanistic explanations. Potential neuroprotective pathways include anti-inflammatory effects and cardiovascular benefits that may reduce cerebrovascular disease burden (Liao and Laufs, 2005; Law et al., 2003). Conversely, harmful mechanisms have been suggested, including cerebral cholesterol depletion and mitochondrial dysfunction through depletion of coenzyme Q10 (CoQ10) (Orth and Bellosta, 2012; Fernández-Portero et al., 2023), a compound reduced by statin therapy (Qu et al., 2018). Understanding the role of CoQ10 and LDL-C may offer insights into the biological mechanisms underlying statin-associated cognitive effects.

Because direct measurements of circulating CoQ10 are not currently available in large-scale proteomic resources such as the UK Biobank, alternative biomarkers reflecting CoQ10 biosynthesis are needed. Coenzyme Q7 (COQ7) is a key enzyme involved in the late stages of CoQ10 biosynthesis and has been proposed as a representative marker of CoQ10 biosynthetic capacity. In our recent study investigating statin-associated muscle symptoms (Lee et al., 2026), we successfully utilized plasma COQ7 levels as a proxy for CoQ10 biosynthesis and demonstrated that impaired COQ7-related pathways were associated with statin-induced toxicity. Therefore, we hypothesized that COQ7 may also help generate hypotheses about the biological pathways underlying the relationship between statin therapy and dementia.

To address these knowledge gaps, we conducted a comprehensive population-based study using two large, independent cohorts—the All of Us Research Program and the UK Biobank—focusing on participants aged 40 years or older who self-identified as White. Our study aimed to evaluate the epidemiological association between statin use and incident dementia, and explore potential biological mechanisms.

We conducted a population-based retrospective cohort study utilizing two large-scale biomedical resources: the All of Us Research Program and the UK Biobank. To explore potential biological mechanisms, we incorporated proteomic data from the UK Biobank Pharma Proteomics Project (UKB-PPP). All research procedures adhered to the Declaration of Helsinki. Written or electronic informed consent was obtained from all participants.

The All of Us Research Program is a large-scale, prospective, longitudinal cohort study launched by the U.S. National Institutes of Health in 2018, aiming to enroll at least one million adults aged 18 years and older from across the United States (The All of Us Research Program Investigators, 2019). As of 2025, the program has enrolled approximately 600,000 individuals. Data collected include health questionnaires, electronic health records (EHRs), physical measurements, digital health technology, and genomic data. For this study, we accessed the All of Us Controlled Tier Dataset v8 (C2024Q3R4; released February 2025), with follow-up data available through 1 October 2023. The protocol was reviewed by the Institutional Review Board of the All of Us Research Program, following the regulations and guidance of the National Institutes of Health Office for Human Research Protections.

Statin exposure was defined using prescription records from 1 January 2006 to 1 October 2022, and participants with any statin prescription on or before 31 December 2005 were excluded to restrict the analysis to new users. Statin users were those with a first statin prescription during this period, which served as the index date. Non-users were those with no statin prescription throughout the observation period; this definition therefore relied on information after the index date and may be susceptible to time-related selection bias. Each non-user was assigned a pseudo-index date randomly sampled from the distribution of index dates among users, restricted to the period during which the participant was alive and under observation. In both groups, participants with dementia before the (pseudo-)index date were excluded, and follow-up started at the (pseudo-)index date.

Statin medications were identified using the Anatomical Therapeutic Chemical classification code C10AA (HMG-CoA reductase inhibitors), including atorvastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, and simvastatin. Statins were classified by clinical intensity and lipophilicity as detailed in Supplementary Methods.

Eligible participants were adults aged ≥40 years with available EHR and APOE genotype data who self-identified as White ethnicity. We excluded individuals with a recorded statin prescription before 1 January 2006 (to identify participants whose first recorded statin prescription occurred during the study period); those whose first recorded statin prescription occurred after 1 October 2022 (to ensure sufficient time for outcome ascertainment); participants with prevalent dementia or neurological conditions before the index date (stroke, Parkinson’s disease, or traumatic brain injury); and those with observation duration <1 year (to ensure adequate exposure duration and outcome assessment) (Evans and Golomb, 2009).

The outcome was incident dementia, defined as the first occurrence of any dementia diagnosis during follow-up. Dementia cases were identified using standardized coding systems: International Classification of Diseases, Ninth Revision (ICD-9), International Classification of Diseases, Tenth Revision (ICD-10), and Systematized Nomenclature of Medicine (SNOMED) codes (Supplementary Table S1). For each participant, the observation period began at the index date and ended at the time of censoring, defined as the earliest occurrence of incident dementia, discontinuation of statin therapy among users, loss to follow-up, death, or study completion (1 October 2023).

The UK Biobank represents a large-scale prospective population-based study that enrolled approximately 500,000 participants aged 40–69 years across England, Wales, and Scotland during 2006-2010 (Bycroft et al., 2018). The study collected data from questionnaires, interviews, EHRs, genotype data, and protein level data. Follow-up data are available through 31 May 2022 for inpatient records, and 31 May 2017 for primary records. Ethical approval was obtained from the National Research Ethics Committee (16/NW/0274) under application ID 126305.

Eligible participants met the same inclusion and exclusion criteria as the discovery cohort, except only those with first recorded statin prescription between 1 January 2006 and 31 May 2016 were included. White ethnicity was defined as British, Irish, or other White background. Statin exposure and index date definitions were identical to the discovery cohort. Statin medications were identified using the read code and drug name (Supplementary Table S2). Dementia cases were identified using ICD-9, ICD-10, and Read codes (Supplementary Table S1). The observation period extended from index to dementia diagnosis, statin discontinuation among users, death, loss to follow-up, or 31 May 2017.

To explore potential biological pathways linking statin use to dementia, we conducted a mediation analysis using proteomic data from the UKB-PPP. Because direct CoQ10 measurements were unavailable, we evaluated COQ7, a protein involved in CoQ10 biosynthesis, as a potential mediator (Acosta et al., 2016). We constructed a cohort from UKB-PPP participants based on self-reported medication use at enrollment. Participants were classified as statin users or non-users based on reported regular statin use. Statin medications were coded using standardized classification systems (Supplementary Table S3). Dementia case definitions were identical to those used in the replicati