Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2025 Jan 17;80(5):gbaf005. doi: 10.1093/geronb/gbaf005

Modifiable Risk Factors for Cognitive Decline in Community-Dwelling Older Adults Differ by Sex and APOE4

Emilie T Reas 1,✉, Humberto Parada Jr 2,3, Jaclyn Bergstrom 4, Linda K McEvoy 5,6
Editor: Gali Weissberger7
PMCID: PMC11974396  PMID: 39820320

Abstract

Objectives

The extent to which lifestyle shapes trajectories of normal cognitive aging, and the factors with highest potential for mitigating cognitive decline, remain poorly characterized.

Methods

Participants of the Rancho Bernardo Study underwent demographic, health, and behavioral characterization at baseline, along with up to 7 cognitive assessments over a 27-year follow-up period. Factor analysis of 24 baseline risk variables identified 9 composite factors. Mixed effects models on data from 1,489 participants (aged 45–95 years at baseline) assessed prediction of cognitive change by baseline factor scores. Models were repeated stratified by sex and APOE4 status.

Results

Factors of hyperlipidemia and obesity; marriage and depression; occupation and education; and physical activity and subjective health best predicted rates of decline across multiple cognitive domains. Distinct risk profiles were identified for women and men, and for APOE4 carriers and non-carriers. Models of composite risk estimated that potential savings could amount to 7–9.5 years of preserved cognitive health span for low- versus high-risk profiles. Magnitudes of aggregate risk effects were greater among women across cognitive domains, and for APOE4 carriers for memory and verbal fluency.

Discussion

Multifactorial life-course approaches to manage cardiometabolic health and promote physical, cognitive, and social engagement may help to mitigate cognitive decline with age, with composite risk associated with up to a decade of preserved cognitive health span. Differences by sex and APOE4 in risk profiles and their potential for risk reduction, highlight the importance of developing personalized recommendations for multidomain approaches to cognitive health maintenance throughout the life-course.

Keywords: Cognitive aging, Genetics, Lifestyle, Sex differences

Age-related cognitive decline limits independence, compromises quality of life, and introduces significant caregiver burden, ranking among the most pressing public health concerns of our time. Patterns of cognitive aging are highly heterogeneous, shaped by the combined influences of non-modifiable factors such as age, sex, and genetics, together with life-course behaviors and environmental exposures that may support cognitive reserve or resilience to neuropathology. Longitudinal cohort studies have provided invaluable insight into the demographic, lifestyle, health, and psychosocial factors that influence cognitive aging trajectories (Beydoun et al., 2014; Corley et al., 2018; Nyberg et al., 2020).

Mounting evidence suggests that life-course exposures, including vascular health (Bangen et al., 2013; Ekström et al., 2024; Verhaeghen et al., 2003), physical inactivity (Hamer et al., 2018; Josefsson et al., 2012; Lindwall et al., 2012), social and occupational activity (Mousavi-Nasab et al., 2014; Yu et al., 2009), dietary patterns, smoking, and alcohol intake (Dufouil et al., 2000; Ekström et al., 2024; Lourida et al., 2013; Richards et al., 2005), and sensory function (MacDonald et al., 2004), influence rates of cognitive decline. Life-course risk factors have also been estimated to account for 45% of dementia cases, with hearing and visual loss, low education, cigarette smoking, social isolation, depression, traumatic brain injury, hypertension, high low-density lipoprotein (LDL) cholesterol, physical inactivity, air pollution, obesity, diabetes, and excessive alcohol intake, among the most established factors (Livingston et al., 2024). Others with less conclusive evidence that warrant further investigation include dietary patterns (Chen et al., 2019), sleep disturbances (Shi et al., 2018), hyperlipidemia (Anstey et al., 2017), occupational complexity (Andel et al., 2005), and marital status (Håkansson et al., 2009). Yet our understanding of the multifactorial modifiers of cognitive decline remains incomplete, possibly attributable to inadequate follow-up periods or lack of evaluation of common influences among factors or moderating effects of demographics or genetics on life-course exposures, in prior investigations. Multivariate approaches suggest that the cumulative effect of multiple life-course variables better accounts for heterogeneity in cognitive aging than any individual risk factor (Corley et al., 2018; Verhaeghen et al., 2003). The success of multidomain interventions (McMaster et al., 2020; Ngandu et al., 2015; Yaffe et al., 2024), which demonstrated cognitive benefits from programs integrating personalized lifestyle interventions such as physical activity, dietary counseling, cognitive training, and metabolic and vascular risk management, further supports combination strategies for preventing cognitive decline. Well-powered longitudinal studies over adequate follow-up periods with comprehensive health, lifestyle, and psychosocial characterization will be critical to infer associations between exposures and cognitive outcomes while accounting for multicollinearity among health and lifestyle variables. Such investigations are urgently needed to develop precision-prevention strategies integrating personalized therapeutic targets to optimize the cognitive health span.

Despite evidence that genetic, biological, or sociocultural factors may magnify vulnerability to particular risk factors, subgroup differences are not routinely examined in studies of cognitive aging, thus hindering translation of findings to personalized health recommendations. For instance, women may be more susceptible to the adverse effects of hypertension, diabetes, and dyslipidemia on cognitive decline, with fewer protective modifiable factors (Anstey et al., 2021; Gilsanz et al., 2017; Huo et al., 2022), and have higher rates of Alzheimer’s disease (AD) than men even after accounting for greater longevity (Ferretti et al., 2018). Carriage of the apolipoprotein E (APOE) ε4 allele, the strongest single genetic risk factor for AD, independently predicts faster cognitive decline beyond the influence of life-course risk (Corley et al., 2023) and amplifies associations of modifiable factors, including hypertension, diabetes, hypercholesterolemia, education, and alcohol intake, with cognitive decline (Bangen et al., 2013; Blair et al., 2005; Reas et al., 2019). Thus, identifying subpopulation-specific risk factor profiles will be critical to prioritize clinical recommendations to preserve cognitive health that are tailored to the individual, in turn maximizing protocol adherence while minimizing patient burnout.

The present study leveraged the 27-year cognitive follow-up period and extensive clinical, demographic, and genetic characterization of the community-based Rancho Bernardo Study of Healthy Aging (RBS) to conduct multivariate prediction of cognitive aging trajectories. Modifiable risk variables, selected according to established or hypothesized associations with cognitive impairment, were evaluated at baseline and used to derive composite risk factor scores. Associations between baseline composite risk factor scores and longitudinal decline in multiple cognitive domains were assessed while considering differences by sex and APOE4 genotype.

Method

Participants

Eligible participants included community-dwelling participants of the RBS, a population-based study of adults in the Southern California community of Rancho Bernardo. In 1972–1974, 6,726 adults were screened for cardiovascular risk factors (Barrett-Connor, 2013), representing 82% of the target population of 30–79-year-olds, 76% of whom were aged 40 or older. Subsequently, subsets of this initial cohort were invited for periodic research visits until the final (12th) visit in 2014–2016. Cognitive testing was first performed in 1988–1992 (visit 5), considered baseline for the present study, and was subsequently conducted at approximately 4-year intervals until 2014–2016, for a maximum of seven cognitive assessments over a 27-year follow-up. A total of 2,030 participants completed the 1988–1992 visit and were eligible for the present study.

Study procedures were approved by the University of California, San Diego Human Research Protections Program Board and participants provided informed written consent prior to participation. Data access for the RBS, including for all data used in this study, are freely available at https://knit.ucsd.edu/ranchobernardostudy.

Demographic, Lifestyle, and Clinical Assessment

Participant sex, birthdate, occupation, and education level were acquired at enrollment. Education was converted to years of education, and occupational complexity was coded according to the 10-point Hollingshead scale, with decreasing complexity.

Baseline health and lifestyle measures were collected at the first cognitive assessment (visit 5) unless otherwise noted. Standardized questionnaires were administered to assess medical history, psychosocial factors, and health behaviors. Depression was assessed using the Beck Depression Inventory (BDI), with scores from 0 to 63 (severe depression). Individuals were classified as married or not married (separated, divorced, widowed, or single). To evaluate social engagement, the participant was asked how often they saw close friends or relatives, rated from 1 (> once per week) to 5 (< once per month). Participants were asked how many times per week they engaged in light, moderate, and strenuous exercise, and the Godin Leisure-Time Exercise score (Godin & Shephard, 1985) was computed as a composite of the frequency of light, moderate, and strenuous exercise. Participants completed the Willet Food Frequency Questionnaire (Willett et al., 1985) to indicate how frequently during the previous year they consumed a specified portion of foods. A modified Mediterranean diet score was computed based on a Mediterranean scale (Trichopoulou et al., 2003) adapted to the U.S. population (Fung et al., 2018), ranging from 0 to 9, with higher scores representing stricter adherence to the Mediterranean diet. Omega-3 fatty acid intake was computed as the sum of eicosapentaenoic acid and docosahexaenoic acid, estimated in grams using the Harvard nutrient database program, by multiplying the frequency of responses by the nutrient compositions of the corresponding food portion, then divided by daily caloric intake. Alcohol consumption was measured as the average number of alcoholic beverages consumed per week. Smoking was coded as never, former, or current, and sleep was reported as average hours of sleep per night. Self-reported health was coded on a 5-point scale (1 = excellent, 5 = poor).

Blood pressure was measured in seated, resting participants, and systolic and diastolic blood pressure were calculated as the average of two measurements taken 5 min apart. Use of antihypertension medications, including calcium channel blockers, beta blockers, vasopressors, or diuretics, was validated at each visit by a nurse who examined containers and prescriptions brought to the clinic.

Participant height and weight were measured to compute body mass index (BMI, kg/m2), and waist-to-hip ratio was used as an estimate of central obesity.

Ability to detect tones at four frequencies at thresholds of 40, 25, and 20 dB was determined for each ear, with a value of 50 dB assigned when no tone was detected. The pure-tone average (PTA) threshold was calculated as the average across frequencies for each ear, and PTA in the better-hearing ear was used for analysis.

Venous samples were drawn in the morning after fasting. Plasma high-density lipoprotein (HDL) and LDL were measured according to standardized procedures in a Centers for Disease Control and Prevention certified Lipid Research Clinic Laboratory. Triglycerides were measured in the Lipid Research Clinic Laboratory using enzymatic techniques with an ABA-200 biochromatic analyzer (Abbott Laboratories, Abbott Park, Illinois, USA). Plasma glucose was measured using high-performance liquid chromatography with the glucose oxidase method.

Because stroke, diabetes, social engagement, glucose, and blood lipids were not collected for all participants at visit 5 (1988–1992), data collected at visits 4 (1984–1987) and/or 7 (1992–1996) were additionally used for these variables. Participants were considered positive for history of stroke or diabetes if they reported the respective condition at any of visits 4, 5, or 7. Measures for these variables were reliable across visits 4–7 among participants with available data at visit 5, with reports of no stroke or diabetes at visits 5 or 7 consistent with the prior visit (visits 5 vs 4, >99%; visits 7 vs 5, >98%). Available data for social engagement, glucose, and lipids were averaged across visits 4, 5, and 7, with high correlations between visits 5 and 7 (there was inadequate overlap to compare visits 4 and 5); social engagement (r = 0.20, p = .04), glucose (r = 0.57, p < .0001), lipids (r = 0.63–0.80, p < .0001). Because sleep was not assessed at visit 5, values were averaged between visits 4 and 7. Because of incomplete hearing data at visit 5, PTA was used from either visit 5 or 7. Because the BDI was only administered at visit 7 and later, visit 7 was used to assess baseline depression. Sleep, hearing, and BDI scores across visits showed moderate to strong correlations between adjacent visits, with r = 0.62–0.71, p < .0001 for sleep, r = 0.86–0.88, p < .0001 for hearing, and r = 0.59–0.66, p < .0001 for BDI.

Cognitive Assessment

Cognitive testing, initiated in 1988, was conducted by a trained interviewer in a quiet room. The test battery included the Mini-Mental State Examination (MMSE), Trails Making Test, Part B (Trails B) of the Halstead–Reitan Battery, and category fluency test, at all seven visits, and the Buschke–Fuld Selective Reminding test, administered at five visits. The MMSE is a cognitive screening instrument that measures orientation, attention, language, and memory to yield a measure of global cognition. Trails B assesses psychomotor tracking and executive function, requiring participants to connect alternating letters and numbers in sequence, with a maximum completion time of 300 s. Category fluency evaluates verbal semantic fluency by having participants name as many unique animals as possible in 60 s. Buschke–Fuld Selective Reminding assesses verbal episodic memory, requiring participants to recall as many words as possible from a list of 10 words read by the test administrator. Forgotten words are repeated and the recall test is repeated for six trials, with the number of correctly recalled words over all trials analyzed here.

APOE Genotyping

Genetic data were available on 82% (N = 1,221) of participants. APOE genotype was determined as previously described (Siest et al., 1995). DNA was extracted by Sequana Therapeutics (La Jolla, CA, USA) using standard techniques (Puregene; Gentra, Minneapolis, MN, USA). Participants were classified as APOE4 carriers or non-carriers.

Risk Factor Computation

To account for multicollinearity among risk variables, factor analysis was conducted to cluster related risk variables into independent factor scores. Baseline data from all 2,030 participants were used to optimize robustness of risk factor scores. Twenty-four candidate health, social, and lifestyle variables were entered into principal component analysis, using varimax rotation, and a threshold of eigenvalues >1 for factor inclusion. The final model produced nine components representing: lipids/obesity (HDL, BMI, triglycerides, waist-hip-ratio); marriage/low depression; education/occupational complexity; blood pressure; diabetes/glucose; alcohol consumption/smoking; physical activity/subjective health; diet (Mediterranean diet and omega-3 intake); and social isolation/hearing impairment. Stroke, LDL cholesterol, antihypertensive medication, and hours of sleep had communalities <0.4, indicating that they accounted for negligible variability in the respective factor, and were hence excluded from the final model structure. For each participant, component variables were converted to Z-scores (scaled to the mean and standard deviation) before calculating factor scores by combining component variables weighted according to the factor loading (Supplementary Table 1). Missing variables were set to zero, such that missing variables did not contribute to the score, and each factor score was derived from variables with available data. Because hours of sleep demonstrated low correlation with other variables (r = 0.00–0.18) and was not included in any factor, it was considered as a single-variable factor for analysis (Z-scored for comparability).

Statistical Analysis

To determine associations between risk factors and cognitive decline, the sample was restricted to participants with scores for all factors. Participants for whom at least one factor score could not be computed due to missing data for all variables within that factor (N = 541) were excluded, resulting in data from 1,489 participants used for analyses of longitudinal cognitive decline. For each cognitive test, mixed-effects models evaluated associations of baseline modifiable risk factors (from 1988 to 1992 unless otherwise noted) with cognitive change from the 1988–1992 to 2014–2016 assessments. A random intercept and random slope for age were included, allowing subject baseline levels to randomly vary about the mean trajectory. Models included linear and quadratic age, as well as fixed factors of test practice (0 = first assessment, 1 = subsequent assessments), each of 10 risk factors, and interactions between age and risk factor. Preliminary models first considered sex as a covariate, followed by secondary sex-stratified models to explore potential sex differences in associations between risk factors and cognitive trajectories. To interrogate whether APOE4 modified associations, models were repeated stratified by APOE4 on the subset of participants with genetic data. For any significant association of risk factor with cognition in sex- or APOE4-stratified analyses, models were conducted with an additional term for the interaction of sex or APOE4 with risk factor, or the three-way interaction among sex or APOE4, risk factor, and age.

To assess whether differences between the full sample of participants used for factor analysis and the subset used for mixed models influenced results, we conducted sensitivity analyses incorporating only factors available on all participants and compared mixed models between these samples. To evaluate whether differential survival influenced mixed model results, longitudinal joint modeling was conducted for each test, incorporating survival based on death or dropout with covariates of age, sex, and each risk factor using the SAS %SPM macro (Wang et al., 2017).

To evaluate the composite effect of multiple risk factors, cognitive trajectories were compared between high- and low-risk individuals, modeled as hypothetical scores at the 80th percentile (high risk) or 20th percentile (low risk) of the population. For each factor significantly associated with cognitive outcomes in mixed effects models, scores were set to the 80th or 20th percentile, to model cognitive trajectories of individuals with deleterious or beneficial aggregate risk profiles. Regressors that were not significantly associated with cognitive outcomes were modeled as the group mean. To estimate the cognitive health span that may be extended via multifactorial risk modification, the expected age of a low-risk individual with a cognitive score equivalent to a 75-year-old high-risk individual (Age80/20) was computed. To quantify the impact of composite risk on longitudinal cognitive decline, projected cognitive change from age 65 to 90 was computed for high- and low-risk models, and relative difference in change scores for high- versus low-risk trajectories were computed (Decline80/20) = (Score90 − Score65)80th/(Score90 − Score65)20th.

Analyses were conducted in SPSS version 28.0 (IBM Corp, Armonk, NY, USA) and SAS Studio (3.81). Significance was set to p < .05.

Results

Participant Characteristics

Characteristics for participants included in mixed effects models are reported in Table 1, with available baseline data indicated for each variable (N). Participants ranged in age from 45 to 95 (73.0 ± 9.3) years at baseline, participated in 3.0 ± 1.9 cognitive assessments, 59% were female, >99% were White, and 22% carried at least one APOE4 allele. Characteristics for participants stratified by sex and by APOE4 status are reported in Supplementary Tables 2 and 3. Women had longer cognitive follow-up, less education, and lower occupational complexity, better hearing, were less likely to be married, have diabetes, have ever smoked or had a stroke, reported consuming less alcohol, engaging in less physical activity, sleeping less, and had lower WHR, BMI, DBP, and glucose, and higher HDL (p < .001), than men. Women also reported more social engagement (p = .002) and had higher LDL cholesterol (p = .02). APOE4 carriers had shorter follow-up and were more likely to have ever smoked (p = .04), had higher LDL (p < .001) and triglycerides (p = .02), and lower HDL (p = .007), than non-carriers. Although participants excluded from mixed models differed from included participants on several variables (Supplementary Table 4), factor analysis results were essentially identical when limited to the subset of participants used in mixed models (Supplementary Table 5 vs Supplementary Table 1).

Table 1.

Characteristics of Participants Included in Mixed Effect Models

Characteristics Valid N Mean ± SD Range N (%)
Age 1,489 73.0 ± 9.3 45–95
Follow-up (years) 1,489 13.8 ± 7.5 0–27.4
Sex (female) 1,489 875 (59%)
APOE4 carrier 1,221 270 (22%)
Education 1,489 14.3 ± 2.2 10–18
Occupation (higher = low complexity) 1,445 3.0 ± 1.8 1–10
Married 1,489 1112 (75%)
Beck Depression Inventorya 1,155 4.2 ± 3.8 0–22
Social isolation (higher = less social contact)a 1,487 2.0 ± 1.1 1–5
Hearing (higher PTA = worse hearing)a 976 34.0 ± 10.4 20–50
Drinks per week 1,489 5.6 ± 7.4 0–84
Smoking 1,489
 Never 638 (43%)
 Prior 708 (47%)
 Current 143 (10%)
Subjective health (higher = poor) 1,489 1.7 ± 0.9 1–5
Physical activity (Godin score) 1,485 19.4 ± 18.3 0–126
Sleep (hours)a 1,489 7.3 ± 1.1 3–12
Omega-3 (scaled to caloric intake) 1,477 0.00014 ± 0.00013 0–0.002
Mediterranean diet 1,489 4.3 ± 1.8 0–9
Waist-hip ratio 1,480 0.84 ± 0.09 0.42–1.10
Body mass index 1,489 25.2 ± 3.8 15.0–48.2
Systolic blood pressure 1,489 135.5 ± 20.3 82–212
Diastolic blood pressure 1,488 75.6 ± 9.3 48–112
Antihypertensive medication use 1,452 704 (48%)
Diabetesa 1,489 126 (8%)
Glucosea 1,483 99.9 ± 20.5 54–388
HDLa 1,485 60.5 ± 17.7 22–128
LDLa 1,485 132.2 ± 33.0 29–277
Triglyceridesa 1,485 119.4 ± 70.0 22–859
Strokea 1,488 104 (7%)
Open in a new tab

Notes: HDL = high-density lipoprotein; LDL = high-density lipoprotein; PTA = pure tone average; SD = standard deviation.

aAveraged across visits 4, 5, and 7: social isolation, glucose, lipids, stroke, and diabetes. Averaged across visits 4 and 7: sleeping and hearing. Visit 7 only: Beck Depression Inventory.

Risk Factors of Cognitive Decline

Mixed model results for all participants are shown in Supplementary Table 6 and Figure 1. For global cognition, higher lipids/obesity and marriage/lower depression were associated with better performance at younger ages but more rapid decline over time. For executive function, lower education/occupational complexity and lower physical activity/subjective health predicted faster decline. For verbal memory, lower education/occupational complexity and higher lipids/obesity predicted faster decline. For verbal fluency, lower education/occupational complexity was associated with worse performance. Results were comparable when conducted on the full sample used in factor analyses and when using the subset used in primary mixed models, when limiting analyses to include only factor scores available on the full sample (Supplementary Table 7). Results were not materially affected after accounting for survival in joint models (Supplementary Table 8).

Figure 1.

Alt Text: Graphs illustrating cognitive trajectories according to low or high levels of risk for different factors.

Open in a new tab

Modeled cognitive trajectories for all participants are shown for factors significantly associated with cognitive performance or change. Modeled groups represent 20th and 80th percentile scores for each illustrated factor. Other variables are modeled as the population mean.

To estimate the aggregate effect of multiple risk factors, predicted cognitive scores across the age range were computed using factor scores at the 20th (low risk) or 80th (high risk) percentile for each risk factor identified in mixed models described above (lipids/obesity, marriage/lower depression, education/occupation, physical activity/health). Estimated rates of decline from age 65 to 90 (Decline80/20) were 79% (MMSE), 36% (Trails B), 35% (verbal recall), and 7% (verbal fluency) faster for individuals at an 80th versus 20th risk percentile (Figure 2, Supplementary Table 9). This translates to a cognitive health span (Age80/20) 7–9.5 years longer for a low- than high-risk 75-year-old individual.

Figure 2.

Alt Text: Graphs illustrating cognitive trajectories according to low or high aggregate risk from the combined effect of risk factors.

Open in a new tab

Modeled cognitive trajectories for all participants are shown for the combined effect of factors significantly associated with cognition. For each associated factor (lipids/obesity, marriage/depression, occupation/education, physical activity/health), low- and high-risk trajectories are modeled as 20th and 80th percentile scores, respectively. Other variables are modeled as the population mean. Red bars indicate expected age difference between a high-risk 75-year-old and a cognitive performance-equivalent low-risk individual. Percent decline indicates the difference in rate of decline from age 65 to 90 for a high-risk individual compared to a low-risk individual.

Predictors of Cognitive Decline by Sex

Sex-stratified model results are presented in Supplementary Table 10 and Supplementary Figure 1. Education/occupational complexity was associated with better global cognition and verbal fluency at younger ages, but faster verbal fluency decline for women, and slower global cognitive decline for men. For women only, lipids/obesity and marriage/lower depression were associated with better global cognition at younger ages but faster decline. Lipids/obesity also predicted faster recall decline. Higher blood pressure was associated with worse executive function, diabetes was associated with better global cognition, and physical activity/health predicted slower global cognitive and executive function decline. For men only, social isolation/hearing impairment was associated with worse global cognition, and longer sleep duration was associated with better global cognitive and recall performance at younger ages, but faster decline. Sex interactions were significant for the associations of education/occupational complexity with global cognitive decline and sleep duration with recall decline (stronger for men), and for the association of lipids/obesity with memory decline (stronger for women; Supplementary Table 10).

Estimated cognitive trajectories were modeled for women and men according to combined risk factors identified from sex-specific mixed models (lipids/obesity, marriage/depression, education/occupation, blood pressure, diabetes, physical activity/health, social isolation/hearing, and sleep; Figure 3, Supplementary Table 9). Risk-associated 25-year decline (Decline80/20) was faster for women in global cognition, verbal fluency, and memory recall, but faster for men in executive function. Age80/20 scores (low-risk age equivalent for a high-risk 75-year-old) were approximately 1 year greater for women, reflecting accelerated risk-associated cognitive aging, with Age80/20 scores of 7–10 years across domains for women and 6–9 years for men.

Figure 3.

Alt Text: Graphs illustrating cognitive trajectories according to low or high aggregate risk across risk factors, separately for men and women.

Open in a new tab

Modeled cognitive trajectories for women (top) and men (bottom) are shown for the combined effect of factors significantly associated with cognition for either sex. For each associated factor (lipids/obesity, marriage/depression, occupation/education, blood pressure, diabetes, physical activity/health, social isolation/hearing, and sleep), low- and high-risk trajectories are modeled as 20th and 80th percentile scores. Other variables are modeled as the population mean. Red bars indicate expected age difference between a high-risk 75-year-old and a cognitive performance-equivalent low-risk individual. Percent decline indicates the difference in rate of decline from age 65 to 90 for a high-risk individual compared to a low-risk individual.

Predictors of Cognitive Decline by APOE4

Mixed effects model results for APOE4 carriers and non-carriers are presented in Supplementary Table 11 and Supplementary Figure 2. Lipids/obesity were associated with better global cognitive performance at younger ages but faster decline in global cognition and recall for APOE4 non-carriers, and with better performance but faster decline in verbal fluency for carriers. Education/occupational complexity was associated with better verbal fluency and slower executive function decline for APOE4 non-carriers, and slower memory decline for carriers. Social isolation/hearing impairment was associated with better performance at younger ages but faster decline in verbal fluency for APOE4 non-carriers, yet worse performance but slower decline for carriers. For APOE4 non-carriers only, physical activity/health predicted slower global cognitive and executive function decline. For APOE4 carriers only, diabetes was associated with better performance at younger ages but faster decline in memory recall, and alcohol intake/smoking predicted slower decline in recall. APOE4 significantly modified the association of diabetes with memory performance (stronger for APOE4 carriers), as well as the associations of social isolation/hearing impairment with verbal fluency and of alcohol intake/smoking with memory, both reflecting stronger effects on decline for APOE4 carriers (Supplementary Table 11).

Cognitive trajectories for APOE4 carriers and non-carriers were estimated by integrating risk factors from APOE4-stratified mixed effects models (lipids/obesity, education/occupation, physical activity/health, social isolation/hearing, drinking/smoking, and diabetes; Figure 4, Supplementary Table 9). APOE4 carriers demonstrated both more rapid risk-associated 25-year decline (Decline80/20) and older Age80/20 scores for verbal fluency and memory recall, whereas non-carriers showed more rapid Decline80/20 and older Age80/20 scores for global cognition and executive function. The most pronounced difference by APOE4 was observed for memory recall, with high-risk APOE4 carriers predicted to decline 67% faster than low-risk carriers, compared to a risk effect of 16% for non-carriers.

Figure 4.

Alt Text: Graphs illustrating cognitive trajectories according to low or high aggregate risk across risk factors, separately for APOE4 carriers and non-carriers.

Open in a new tab

Modeled cognitive trajectories for APOE4 non-carriers (top) and carriers (bottom) are shown for the combined effect of factors significantly associated with cognition for either group. For each associated factor (lipids/obesity, occupation/education, physical activity/health, social isolation/hearing, drinking/smoking, and diabetes), low- and high-risk trajectories are modeled as 20th and 80th percentile scores. Other variables are modeled as the population mean. Red bars indicate expected age difference between a high-risk 75-year-old and a cognitive performance-equivalent low-risk individual. Percent decline indicates the difference in rate of decline from age 65 to 90 for a high-risk individual compared to a low-risk individual.

Discussion

This study leveraged a large community-dwelling cohort of older adults with extensive baseline characterization and nearly three decades of cognitive follow-up, to identify multivariate predictors of cognitive decline from midlife to older age. In this relatively healthy sample, several modifiable risk factors, including lipids/obesity, education/occupational complexity, physical activity/health, and marriage/lower depression, were associated with domain-specific cognitive trajectories. Models of composite risk underscored the potential impact of multifactorial health and lifestyle approaches to mitigate late-life cognitive decline, with up to a decade of potential savings in cognitive health span. Patterns differed by sex and APOE4 status, with stronger associations between aggregate risk and cognitive decline for women and APOE4 carriers.

Factor analysis identified 20 health and lifestyle variables that clustered into nine factors, providing insight into associated behaviors and health metrics that often covary. Among these candidate factors, low education/occupational complexity reliably predicted decline in multiple cognitive domains, consistent with prior support for protective effects of education on cognitive aging (Clouston et al., 2019) and associations between work control and slower cognitive decline (Yu et al., 2009). Confounding by socioeconomic status or childhood environment has been purported to underlie inconclusive evidence for effects of education on later-life cognition (Corley et al., 2018), which may be reflected in the clustering of education with occupational complexity in our sample. Obesity and hyperlipidemia also predicted faster decline in global cognition and memory. Although both have been linked to poor cognitive outcomes, inverse relationships have been reported, with potentially neuroprotective effects of higher weight and inefficacy of statin treatment for cognitive health, supporting a complex role of lipid metabolism in brain function (Dye et al., 2017; van Vliet, 2012). A wealth of evidence also implicates these factors in dementia incidence (Andel et al., 2005; Anstey et al., 2017; Kivimäki et al., 2018); because the RBS did not conduct clinical evaluation for dementia, further research will be important to disentangle whether these risk factors differentially shape cognitive aging trajectories and development of neurodegenerative disease. Higher physical activity clustered with better perceived health and predicted slower executive function decline, aligned with evidence supporting a role for regular physical activity in preventing age-related diseases (Reiner et al., 2013), and mitigating decline in executive function or other higher level cognitive domains (Hamer et al., 2018; Josefsson et al., 2012; Lindwall et al., 2012). Marital status clustered with lower depression, consistent with prior findings of lower depression rates among married individuals (Yan et al., 2011), and this factor was associated with higher global cognitive function at younger ages but more rapid decline. Considering evidence for higher risk for dementia among those widowed or divorced compared to partnered (Håkansson et al., 2009), we speculate that our finding may be attributed to challenges linked to the loss of a spouse among surviving participants, in contrast to a more stable trajectory of unmarried individuals who may be less likely to experience lifestyle disruption due to partner loss. Indeed, among those who were married at baseline, marital status declined over follow-up, from 91% still married 4 years later, to 65% married at the final assessment.

Composite models demonstrated a profound hypothetical effect of multidomain risk modification on preserving cognitive health in the final decades of life. Contrasting modeled trajectories based on aggregate risk for identified significant factors (lipids/obesity, marriage/depression, education/occupation, physical activity/health), estimated 7%–79% slower 25-year cognitive decline across domains for individuals with low versus high composite risk. Models indicate that this magnitude effect could translate to 7–9.5 years of preserved cognitive health at age 75. Lower composite risk was associated with approximately one-third the rate of decline in memory and executive function, domains particularly vulnerable to aging. These findings extend reports that roughly 45% of dementia risk may be attributable to potentially modifiable factors, to suggest that lifestyle interventions may also help to mitigate cognitive decline in normal aging.

Sex-stratified models revealed distinct risk profiles for women and men. Each of the factors identified in the full cohort (lipids/obesity, marriage/lower depression, education/occupation, physical activity/health) was also associated with cognitive decline among women, and the effect of lipids/obesity on memory decline was significantly stronger for women than for men. Although reports of sex differences in effects of these factors on cognition have been mixed, evidence suggests that low education, physical inactivity, and obesity more profoundly affect cognition for women (Gannon et al., 2019; Sindi et al., 2021). Although men exhibited fewer risk associations, effects of low education and occupational complexity with global cognitive decline, and short sleep duration with worse performance at younger ages but slower decline in global cognition and memory, were significantly stronger for men than women. Unique lifestyle pressures of men and women, particularly during midlife when family or career obligations are prominent may differentially disrupt sleep patterns that in turn affect cognitive abilities. This observation is consistent with findings from an overlapping sample that poor sleep quality in men is associated with abnormal brain microstructure, but sleep duration does not predict future microstructural injury (Tsiknia et al., 2023). It remains to be seen if similar sex-specific associations emerge in younger generations for whom lifestyle behaviors and associated stressors have been equalizing between sexes over recent decades. Composite risk models illustrated stronger difference by risk stratification for women than for men, with greater slowing of late-life decline, and extension of cognitive health span of approximately 1 year longer, with lower risk. These results suggest that women may particularly benefit from multifactorial risk modification, although short-term lifestyle interventions have been found to be similarly effective for women and men (Sindi et al., 2021).

APOE4 carriers showed faster memory decline with diabetes and low levels of alcohol intake and smoking, replicating our prior finding of slower memory decline among APOE4 carriers who consumed alcohol in this population with mostly moderate alcohol intake (Reas et al., 2019). This observation is consistent with prior reports that protective effects of moderate alcohol consumption on cognition are magnified in APOE4 carriers, though this pattern may reverse with excessive intake, and that APOE4 carriers demonstrate weaker negative effects, and potentially positive associations, between smoking and cognition (Carmelli et al., 1999; Dufouil et al., 2000). Although our findings suggest a more deleterious effect of diabetes on cognitive decline among APOE4 carriers, the modifying effects of APOE4 on associations between diabetes and cognition have been inconclusive (Haan et al., 1999; Shinohara et al., 2020), and the low prevalence of diabetes in our cohort may have influenced our results. Non-carriers showed slower decline with physical activity/subjective health, and both genotypes exhibited associations of low education/occupational complexity, lipids/obesity, and social isolation/hearing impairment, with faster decline, though in distinct cognitive domains. APOE4 carriers demonstrated more profound benefit from low composite risk in memory and verbal fluency, domains that are particularly vulnerable to decline early in AD, which aligns with evidence from other cohorts for stronger associations of health and lifestyle risk factors on memory decline for APOE4 carriers (Nyberg et al., 2020). The potential of multidomain lifestyle approaches to disproportionately benefit APOE4 carriers has important implications given the relative inefficacy at slowing clinical progression and higher risk of adverse effects with emerging anti-amyloid immunotherapies among APOE4 carriers (Sims et al., 2023; van Dyck et al., 2022).

Strengths of this study include a large community-dwelling sample with up to 27 years of cognitive follow-up and extensive baseline characterization that permitted accounting for covariance among a variety of lifestyle and health variables. Results of this study, which included predominantly White middle-class participants with reliable access to healthcare, may not extend to other race/ethnic or sociocultural populations. Despite the deep characterization of our sample, not all potential modifiers of cognitive aging were evaluated here, including factors such as exposure to pollutants, socioeconomic status, spiritual or meditative practice, or early-life environment. Because risk variables were not consistently assessed at all visits, only associations of baseline risk factors were examined, and time-varying models could not be evaluated. Finally, while this observational study boosts support for the potential of lifestyle modification to preserve cognitive health, intervention studies are essential to verify the efficacy of multidomain strategies to prevent cognitive decline.

In conclusion, findings from this investigation suggest that multivariate health and lifestyle factors may profoundly shape trajectories of cognitive change from middle to later life, with potential extension of several years of cognitive health span. Whereas certain approaches, such as promoting education and occupational complexity, or controlling obesity and hyperlipidemia, may be effective universal prevention strategies, other lifestyle targets such as physical activity, sleep behaviors, or alcohol intake, could guide precision-medicine recommendations based on sex or APOE status. Further research is necessary to evaluate the efficacy of personally tailored multidomain intervention strategies at preserving cognitive health into the final decades of life.

Supplementary Material

gbaf005_suppl_Supplementary_Tables_S1-S11_Figures_S1-S2
gbaf005_suppl_supplementary_tables_s1-s11_figures_s1-s2.docx (1MB, docx)

Contributor Information

Emilie T Reas, Department of Neurosciences, University of California San Diego, La Jolla, California, USA.

Humberto Parada, Jr., Division of Epidemiology and Biostatistics, School of Public Health, San Diego State University, San Diego, California, USA; UC San Diego Health Moores Cancer Center, La Jolla, California, USA.

Jaclyn Bergstrom, Division of Geriatrics, Gerontology, and Palliative Care, Department of Medicine, University of California San Diego School of Medicine, La Jolla, California, USA.

Linda K McEvoy, Herbert Wertheim School of Public Health and Human Longevity Science, University of California San Diego, La Jolla, California, USA; Kaiser Permanente Washington Health Research Institute, Seattle, Washington, USA.

Gali Weissberger, (Psychological Sciences Section).

Funding

Data collection for the Rancho Bernardo Study of Healthy Aging (RBS) was provided primarily by the National Institutes of Health (HV012160, AA021187, AG028507, AG007181, DK31801, HL034591, HS06726, and HL089622). Archiving and sharing of RBS data was supported by the National Institute on Aging (NIA; AG054067). E T. Reas was supported by NIA (R01 AG077202) and an American Federation for Aging Research/McKnight Brain Research Foundation Innovator Award in Cognitive Aging and Memory. H. Parada was supported by the National Cancer Institute (K01 CA234317), the SDSU/UCSD Cancer Research and Education to Advance HealTh Equity (CREATE) Partnership (U54 CA285117 and U54 CA285115), and the Alzheimer’s Disease Resource Center for Advancing Minority Aging Research at the University of California San Diego (P30 AG059299).

Conflict of Interest

None.

Data Availability

RBS data are available through the RBS website: https://knit.ucsd.edu/ranchobernardostudy/. This study was not preregistered.

References

  1. Andel, R., Crowe, M., Pedersen, N. L., Mortimer, J., Crimmins, E., Johansson, B., & Gatz, M. (2005). Complexity of work and risk of Alzheimer’s disease: A population-based study of Swedish twins. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60(5), P251–P258. https://doi.org/ 10.1093/geronb/60.5.p251 [DOI] [PubMed] [Google Scholar]
  2. Anstey, K. J., Ashby-Mitchell, K., & Peters, R. (2017). Updating the evidence on the association between serum cholesterol and risk of late-life dementia: Review and meta-analysis. Journal of Alzheimer’s Disease, 56, 215–228. https://doi.org/ 10.3233/jad-160826 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Anstey, K. J., Peters, R., Mortby, M. E., Kiely, K. M., Eramudugolla, R., Cherbuin, N., Huque, M. H., & Dixon, R. A. (2021). Association of sex differences in dementia risk factors with sex differences in memory decline in a population-based cohort spanning 20–76 years. Scientific Reports, 11(1), 7710. https://doi.org/ 10.1038/s41598-021-86397-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bangen, K. J., Beiser, A., Delano-Wood, L., Nation, D. A., Lamar, M., Libon, D. J., Bondi, M. W., Seshadri, S., Wolf, P. A., & Au, R. (2013). APOE genotype modifies the relationship between midlife vascular risk factors and later cognitive decline. Journal of Stroke and Cerebrovascular Diseases, 22(8), 1361–1369. https://doi.org/ 10.1016/j.jstrokecerebrovasdis.2013.03.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barrett-Connor, E. (2013). The Rancho Bernardo Study: 40 years studying why women have less heart disease than men and how diabetes modifies women’s usual cardiac protection. Glob Heart, 8(2), 95–104. https://doi.org/ 10.1016/j.gheart.2012.12.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Beydoun, M. A., Beydoun, H. A., Gamaldo, A. A., Teel, A., Zonderman, A. B., & Wang, Y. (2014). Epidemiologic studies of modifiable factors associated with cognition and dementia: Systematic review and meta-analysis. BMC Public Health, 14, 643. https://doi.org/ 10.1186/1471-2458-14-643 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Blair, C., Folsom, A., Knopman, D. S., Bray, M., Mosley, T., & Boerwinkle, E. (2005). APOE genotype and cognitive decline in a middle-aged cohort. Neurology, 64(2), 268–276. https://doi.org/ 10.1212/01.WNL.0000149643.91367.8A [DOI] [PubMed] [Google Scholar]
  8. Carmelli, D., Swan, G. E., Reed, T., Schellenberg, G. D., & Christian, J. C. (1999). The effect of apolipoprotein E epsilon4 in the relationships of smoking and drinking to cognitive function. Neuroepidemiology, 18(3), 125–133. https://doi.org/ 10.1159/000026204 [DOI] [PubMed] [Google Scholar]
  9. Chen, X., Maguire, B., Brodaty, H., & O’Leary, F. (2019). Dietary patterns and cognitive health in older adults: A systematic review. Journal of Alzheimer's Disease, 67, 583–619. https://doi.org/ 10.3233/JAD-180468 [DOI] [PubMed] [Google Scholar]
  10. Clouston, S. A. P., Smith, D. M., Mukherjee, S., Zhang, Y., Hou, W., Link, B. G., & Richards, M. (2019). Education and cognitive decline: An integrative analysis of global longitudinal studies of cognitive aging. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(7), e151–e160. https://doi.org/ 10.1093/geronb/gbz053 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Corley, J., Conte, F., Harris, S. E., Taylor, A. M., Redmond, P., Russ, T. C., Deary, I. J., & Cox, S. R. (2023). Predictors of longitudinal cognitive ageing from age 70 to 82 including APOE e4 status, early-life and lifestyle factors: The Lothian birth cohort 1936. Molecular Psychiatry, 28(3), 1256–1271. https://doi.org/ 10.1038/s41380-022-01900-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Corley, J., Cox, S. R., & Deary, I. J. (2018). Healthy cognitive ageing in the Lothian birth cohort studies: Marginal gains not magic bullet. Psychological Medicine, 48(2), 187–207. https://doi.org/ 10.1017/S0033291717001489 [DOI] [PubMed] [Google Scholar]
  13. Dufouil, C., Tzourio, C., Brayne, C., Berr, C., Amouyel, P., & Alperovitch, A. (2000). Influence of apolipoprotein E genotype on the risk of cognitive deterioration in moderate drinkers and smokers. Epidemiology, 11(3), 280–284. https://doi.org/ 10.1097/00001648-200005000-00009 [DOI] [PubMed] [Google Scholar]
  14. Dye, L., Boyle, N. B., Champ, C., & Lawton, C. (2017). The relationship between obesity and cognitive health and decline. Proceedings of the Nutrition Society, 76(4), 443–454. https://doi.org/ 10.1017/S0029665117002014 [DOI] [PubMed] [Google Scholar]
  15. Ekström, I., Josefsson, M., Bäckman, L., & Laukka, E. J. (2024). Predictors of cognitive aging profiles over 15 years: A longitudinal population-based study. Psychology and Aging, 39(5), 467–483. https://doi.org/ 10.1037/pag0000807 [DOI] [PubMed] [Google Scholar]
  16. Ferretti, M. T., Iulita, M. F., Cavedo, E., Chiesa, P. A., Schumacher Dimech, A., Santuccione Chadha, A., Baracchi, F., Girouard, H., Misoch, S., Giacobini, E., Depypere, H., Hampel, H., Women’s Brain, P., & the Alzheimer Precision Medicine, I. (2018). Sex differences in Alzheimer disease—the gateway to precision medicine. Nature Reviews Neurology, 14, 457–469. https://doi.org/ 10.1038/s41582-018-0032-9 [DOI] [PubMed] [Google Scholar]
  17. Fung, T. T., McCullough, M. L., Newby, P., Manson, J. E., Meigs, J. B., Rifai, N., Willett, W. C., & Hu, F. B. (2018). Diet-quality scores and plasma concentrations of markers of inflammation and endothelial dysfunction. The American Journal of Clinical Nutrition, 82(1), 163–173. https://doi.org/ 10.1093/ajcn.82.1.163 [DOI] [PubMed] [Google Scholar]
  18. Gannon, O. J., Robison, L. S., Custozzo, A. J., & Zuloaga, K. L. (2019). Sex differences in risk factors for vascular contributions to cognitive impairment & dementia. Neurochemistry International, 127, 38–55. https://doi.org/ 10.1016/j.neuint.2018.11.014 [DOI] [PubMed] [Google Scholar]
  19. Gilsanz, P., Mayeda, E. R., Glymour, M. M., Quesenberry, C. P., Mungas, D. M., DeCarli, C., Dean, A., & Whitmer, R. A. (2017). Female sex, early-onset hypertension, and risk of dementia. Neurology, 89(18), 1886–1893. https://doi.org/ 10.1212/WNL.0000000000004602 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the community. Canadian Journal of Applied Sport Sciences, 10(3), 141–146. https://www.ncbi.nlm.nih.gov/pubmed/4053261 [PubMed] [Google Scholar]
  21. Haan, M. N., Shemanski, L., Jagust, W. J., Manolio, T. A., & Kuller, L. (1999). The role of APOE ∊4 in modulating effects of other risk factors for cognitive decline in elderly persons. Journal of the American Medical Association, 282(1), 40–46. https://doi.org/ 10.1001/jama.282.1.40 [DOI] [PubMed] [Google Scholar]
  22. Håkansson, K., Rovio, S., Helkala, E. -L., Vilska, A. -R., Winblad, B., Soininen, H., Nissinen, A., Mohammed, A. H., & Kivipelto, M. (2009). Association between mid-life marital status and cognitive function in later life: Population based cohort study. British Medical Journal, 339, 339. https://doi.org/ 10.1136/bmj.b2462 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hamer, M., Muniz Terrera, G., & Demakakos, P. (2018). Physical activity and trajectories in cognitive function: English longitudinal study of ageing. Journal of Epidemiology and Community Health, 72(6), 477–483. https://doi.org/ 10.1136/jech-2017-210228 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Huo, N., Vemuri, P., Graff-Radford, J., Syrjanen, J., Machulda, M., Knopman, D. S., Jack, C. R., Petersen, R., & Mielke, M. M. (2022). Sex differences in the association between midlife cardiovascular conditions or risk factors with midlife cognitive decline. Neurology, 98(6), e623–e632. https://doi.org/ 10.1212/WNL.0000000000013174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Josefsson, M., de Luna, X., Pudas, S., Nilsson, L. G., & Nyberg, L. (2012). Genetic and lifestyle predictors of 15-year longitudinal change in episodic memory. Journal of the American Geriatrics Society, 60(12), 2308–2312. https://doi.org/ 10.1111/jgs.12000 [DOI] [PubMed] [Google Scholar]
  26. Kivimäki, M., Luukkonen, R., Batty, G. D., Ferrie, J. E., Pentti, J., Nyberg, S. T., Shipley, M. J., Alfredsson, L., Fransson, E. I., Goldberg, M., Knutsson, A., Koskenvuo, M., Kuosma, E., Nordin, M., Suominen, S. B., Theorell, T., Vuoksimaa, E., Westerholm, P., Westerlund, H., & Jokela, M. (2018). Body mass index and risk of dementia: Analysis of individual-level data from 1.3 million individuals. Alzheimer’s & Dementia, 14(5), 601–609. https://doi.org/ 10.1016/j.jalz.2017.09.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lindwall, M., Cimino, C. R., Gibbons, L. E., Mitchell, M. B., Benitez, A., Brown, C. L., Kennison, R. F., Shirk, S. D., Atri, A., Robitaille, A., MacDonald, S. W. S., Zelinski, E. M., Willis, S. L., Schaie, K. W., Johansson, B., Praetorius, M., Dixon, R. A., Mungas, D. M., Hofer, S. M., & Piccinin, A. M. (2012). Dynamic associations of change in physical activity and change in cognitive function: Coordinated analyses of four longitudinal studies. Journal of Aging Research, 2012(1), 493598. https://doi.org/ 10.1155/2012/493598 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Livingston, G., Huntley, J., Liu, K. Y., Costafreda, S. G., Selbæk, G., Alladi, S., Ames, D., Banerjee, S., Burns, A., Brayne, C., Fox, N. C., Ferri, C. P., Gitlin, L. N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Nakasujja, N., Rockwood, K., & Mukadam, N. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet standing commission. Lancet, 404(10452), 572–628. https://doi.org/ 10.1016/S0140-6736(24)01296-0 [DOI] [PubMed] [Google Scholar]
  29. Lourida, I., Soni, M., Thompson-Coon, J., Purandare, N., Lang, I. A., Ukoumunne, O. C., & Llewellyn, D. J. (2013). Mediterranean diet, cognitive function, and dementia: A systematic review. Epidemiology, 24(4), 479–489. https://doi.org/ 10.1097/EDE.0b013e3182944410 [DOI] [PubMed] [Google Scholar]
  30. MacDonald, S. W. S., Dixon, R. A., Cohen, A. -L., & Hazlitt, J. E. (2004). Biological age and 12-year cognitive change in older adults: Findings from the Victoria longitudinal study. Gerontology, 50(2), 64–81. https://doi.org/ 10.1159/000075557 [DOI] [PubMed] [Google Scholar]
  31. McMaster, M., Kim, S., Clare, L., Torres, S. J., Cherbuin, N., DʼEste, C., & Anstey, K. J. (2020). Lifestyle risk factors and cognitive outcomes from the multidomain dementia risk reduction randomized controlled trial, body brain life for cognitive decline (BBL-CD). Journal of the American Geriatrics Society, 68(11), 2629–2637. https://doi.org/ 10.1111/jgs.16762 [DOI] [PubMed] [Google Scholar]
  32. Mousavi-Nasab, S. M., Kormi-Nouri, R., & Nilsson, L. G. (2014). Examination of the bidirectional influences of leisure activity and memory in old people: A dissociative effect on episodic memory. British Journal of Psychology, 105(3), 382–398. https://doi.org/ 10.1111/bjop.12044 [DOI] [PubMed] [Google Scholar]
  33. Ngandu, T., Lehtisalo, J., Solomon, A., Levalahti, E., Ahtiluoto, S., Antikainen, R., Backman, L., Hanninen, T., Jula, A., Laatikainen, T., Lindstrom, J., Mangialasche, F., Paajanen, T., Pajala, S., Peltonen, M., Rauramaa, R., Stigsdotter-Neely, A., Strandberg, T., Tuomilehto, J., & Kivipelto, M. (2015). A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): A randomised controlled trial. Lancet, 385(9984), 2255–2263. https://doi.org/ 10.1016/S0140-6736(15)60461-5 [DOI] [PubMed] [Google Scholar]
  34. Nyberg, L., Boraxbekk, C. -J., Sörman, D. E., Hansson, P., Herlitz, A., Kauppi, K., Ljungberg, J. K., Lövheim, H., Lundquist, A., Adolfsson, A. N., Oudin, A., Pudas, S., Rönnlund, M., Stiernstedt, M., Sundström, A., & Adolfsson, R. (2020). Biological and environmental predictors of heterogeneity in neurocognitive ageing: Evidence from Betula and other longitudinal studies. Ageing Research Reviews, 64, 101184. https://doi.org/ 10.1016/j.arr.2020.101184 [DOI] [PubMed] [Google Scholar]
  35. Reas, E. T., Laughlin, G. A., Bergstrom, J., Kritz-Silverstein, D., Barrett-Connor, E., & McEvoy, L. K. (2019). Effects of APOE on cognitive aging in community-dwelling older adults. Neuropsychology, 33(3), 406–416. https://doi.org/ 10.1037/neu0000501 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Reiner, M., Niermann, C., Jekauc, D., & Woll, A. (2013). Long-term health benefits of physical activity—a systematic review of longitudinal studies. BMC Public Health, 13(1), 813. https://doi.org/ 10.1186/1471-2458-13-813 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Richards, M., Hardy, R., & Wadsworth, M. E. J. (2005). Alcohol consumption and midlife cognitive change in the British 1946 Birth cohort study. Alcohol and Alcoholism, 40(2), 112–117. https://doi.org/ 10.1093/alcalc/agh126 [DOI] [PubMed] [Google Scholar]
  38. Shi, L., Chen, S. J., Ma, M. Y., Bao, Y. P., Han, Y., Wang, Y. M., Shi, J., Vitiello, M. V., & Lu, L. (2018). Sleep disturbances increase the risk of dementia: A systematic review and meta-analysis. Sleep Medicine Reviews, 40, 4–16. https://doi.org/ 10.1016/j.smrv.2017.06.010 [DOI] [PubMed] [Google Scholar]
  39. Shinohara, M., Tashiro, Y., Suzuki, K., Fukumori, A., Bu, G., & Sato, N. (2020). Interaction between APOE genotype and diabetes in cognitive decline. Alzheimer's & Dementia, 12(1), e12006. https://doi.org/ 10.1002/dad2.12006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Siest, G., Pillot, T., Regis-Bailly, A., Leininger-Muller, B., Steinmetz, J., Galteau, M. M., & Visvikis, S. (1995). Apolipoprotein E: An important gene and protein to follow in laboratory medicine. Clinical Chemistry, 41(8 Pt 1), 1068–1086. http://www.ncbi.nlm.nih.gov/pubmed/7628082 [PubMed] [Google Scholar]
  41. Sims, J. R., Zimmer, J. A., Evans, C. D., Lu, M., Ardayfio, P., Sparks, J., Wessels, A. M., Shcherbinin, S., Wang, H., Monkul Nery, E. S., Collins, E. C., Solomon, P., Salloway, S., Apostolova, L. G., Hansson, O., Ritchie, C., Brooks, D. A., Mintun, M., Skovronsky, D. M., & Investigators, T. -A. (2023). Donanemab in early symptomatic Alzheimer disease: The TRAILBLAZER-ALZ 2 randomized clinical trial. Journal of the American Medical Association, 330(6), 512–527. https://doi.org/ 10.1001/jama.2023.13239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sindi, S., Kåreholt, I., Ngandu, T., Rosenberg, A., Kulmala, J., Johansson, L., Wetterberg, H., Skoog, J., Sjöberg, L., Wang, H. -X., Fratiglioni, L., Skoog, I., & Kivipelto, M. (2021). Sex differences in dementia and response to a lifestyle intervention: Evidence from Nordic population-based studies and a prevention trial. Alzheimer's & Dementia, 17(7), 1166–1178. https://doi.org/ 10.1002/alz.12279 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tsiknia, A. A., Parada, H., Banks, S. J., & Reas, E. T. (2023). Sleep quality and sleep duration predict brain microstructure among community-dwelling older adults. Neurobiology of Aging, 125, 90–97. https://doi.org/ 10.1016/j.neurobiolaging.0.02.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Trichopoulou, A., Costacou, T., Bamia, C., & Trichopoulos, D. (2003). Adherence to a Mediterranean Diet and Survival in a Greek Population. New England Journal of Medicine, 348(26), 2599–2608. https://doi.org/ 10.1056/NEJMoa025039 [DOI] [PubMed] [Google Scholar]
  45. van Dyck, C. H., Swanson, C. J., Aisen, P., Bateman, R. J., Chen, C., Gee, M., Kanekiyo, M., Li, D., Reyderman, L., Cohen, S., Froelich, L., Katayama, S., Sabbagh, M., Vellas, B., Watson, D., Dhadda, S., Irizarry, M., Kramer, L. D., & Iwatsubo, T. (2022). Lecanemab in early Alzheimer’s disease. New England Journal of Medicine, 388(1), 9–21. https://doi.org/ 10.1056/nejmoa2212948 [DOI] [PubMed] [Google Scholar]
  46. van Vliet, P. (2012). Cholesterol and late-life cognitive decline. Journal of Alzheimer’s Disease, 30, S147–S162. https://doi.org/ 10.3233/jad-2011-111028 [DOI] [PubMed] [Google Scholar]
  47. Verhaeghen, P., Borchelt, M., & Smith, J. (2003). Relation between cardiovascular and metabolic disease and cognition in very old age: Cross-sectional and longitudinal findings from the Berlin aging study. Health Psychology, 22(6), 559–569. https://doi.org/ 10.1037/0278-6133.22.6.559 [DOI] [PubMed] [Google Scholar]
  48. Wang, W., Wang, W., Mosley, T. H., & Griswold, M. E. (2017). A SAS macro for the joint modeling of longitudinal outcomes and multiple competing risk dropouts. Computer Methods and Programs in Biomedicine, 138, 23–30. https://doi.org/ 10.1016/j.cmpb.2016.10.003 [DOI] [PubMed] [Google Scholar]
  49. Willett, W. C., Sampson, L., Stampfer, M. J., Rosner, B., Bain, C., Witschi, J., Hennekens, C. H., & Speizer, F. E. (1985). Reproducibility and validity of a semiquantitative food frequency questionnaire. American Journal of Epidemiology. 122(1), 51–65. https://doi.org/ 10.1093/oxfordjournals.aje.a114086 [DOI] [PubMed] [Google Scholar]
  50. Yaffe, K., Vittinghoff, E., Dublin, S., Peltz, C. B., Fleckenstein, L. E., Rosenberg, D. E., Barnes, D. E., Balderson, B. H., & Larson, E. B. (2024). Effect of personalized risk-reduction strategies on cognition and dementia risk profile among older adults: The SMARRT randomized clinical trial. JAMA Internal Medicine, 184(1), 54–62. https://doi.org/ 10.1001/jamainternmed.2023.6279 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yan, X. -Y., Huang, S. -M., Huang, C. -Q., Wu, W. -H., & Qin, Y. (2011). Marital status and risk for late life depression: A meta-analysis of the published literature. Journal of International Medical Research, 39(4), 1142–1154. https://doi.org/ 10.1177/147323001103900402 [DOI] [PubMed] [Google Scholar]
  52. Yu, F., Ryan, L. H., Schaie, K. W., Willis, S. L., & Kolanowski, A. (2009). Factors associated with cognition in adults: The Seattle longitudinal study. Research in Nursing and Health, 32(5), 540–550. https://doi.org/ 10.1002/nur.20340 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbaf005_suppl_Supplementary_Tables_S1-S11_Figures_S1-S2
gbaf005_suppl_supplementary_tables_s1-s11_figures_s1-s2.docx (1MB, docx)

Data Availability Statement

RBS data are available through the RBS website: https://knit.ucsd.edu/ranchobernardostudy/. This study was not preregistered.

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES