Cognitive trajectories of older adults before and after the COVID-19 pandemic: socioeconomic and psychological disparities in the Lc65+ cohort Authors: Philippe Voruz, Anthony Nuber-Champier, Yves Henchoz, Gilles Allali, Frederic Assal, Julie A Péron Published: 2026 ABSTRACT AbstractBackground and ObjectivesThe COVID-19 pandemic and related confinement measures have exacerbated health inequalities and posed unique challenges to cognitive aging. This study examined longitudinal changes in cognitive performance and associated factors before and after the COVID-19 pandemic, using data from the Swiss Lc65+ aging study.MethodsAnalyses included participants with complete data across 3 assessment waves (2016-2024): 412 participants from Cohort 1 (born 1934-1938), 702 from Cohort 2 (born 1939-1943), and 826 from Cohort 3 (born 1944-1948), with women representing 58%-63% of each cohort. Cognitive performance was measured using the Mini-Mental State Examination (MMSE). Generalized estimating equations were used to evaluate changes in cognition and associations with socioeconomic, psychological, behavioral, and nutritional indicators.ResultsSignificant declines in MMSE scores were observed over time in Cohorts 1 and 2 (p < .001), with a smaller but significant reduction in Cohort 3. The prevalence of global cognitive efficiency deficits increased notably in Cohort 1 (16.5%-30.6%) and to a lesser extent in Cohort 2, while remaining stable in Cohort 3. Financial assistance for health insurance was associated with cognitive outcomes in several models, although the direction of this association varied across cohorts. Increases in anxiety and apathy were also observed, particularly following the COVID-19 confinement period.Discussion and ImplicationsFindings indicate that the COVID-19 pandemic and confinement period coincided with accelerated cognitive decline and heightened socioeconomic vulnerability among older adults, particularly in the oldest cohort. Results highlight the need for multidimensional public health strategies. This cohort study, with follow-up initiated prior to the COVID-19 pandemic, provides a unique opportunity to evaluate the potential impact of lockdown measures on cognitive outcomes in adults aged 65 years and older. Beyond cognition alone, this study innovatively examines concurrent changes in psychological symptoms, socioeconomic factors, and nutritional status, offering a comprehensive and multidimensional assessment of aging-related vulnerabilities during confinement. Using an advanced statistical framework that integrates epidemiological approaches with fine-grained longitudinal analyses, our findings reveal interactive effects on cognitive performance. Introduction Cognitive health refers to the capacity to think, learn, and remember,1 and it is widely recognized as a core component of healthy aging. According to the World Health Organization2 and several leading studies,3,4 optimal cognitive functioning is underpinned by the structural and functional integrity of the brain and involves the preservation of executive, memory, perceptual, attentional, and language functions. Maintaining cognitive health throughout the lifespan is essential not only for individual well-being and autonomy but also for reducing societal and healthcare burdens associated with neurocognitive disorders. Over recent decades, the scientific community has devoted increasing attention to identifying the biological, psychological, and social determinants that influence the trajectory of cognitive aging. In Switzerland, epidemiological data suggest a progressive increase in global cognitive efficiency deficits across age groups from the age of 30 and a pronounced exponential rise from the age of 60.5 Among individuals aged older than 90, prevalence reaches 44.8%, highlighting the growing public health challenge posed by cognitive decline in an aging population.5 Longitudinal cohort studies have contributed substantially to our understanding of cognitive aging, revealing robust associations between physiological and environmental risk factors and the development of neurocognitive disorders. A well-established example is the apolipoprotein E epsilon 4 (APOE4) allele, which significantly increases the risk of developing Alzheimer’s disease and other major neurocognitive conditions.6 Beyond genetics, accumulating evidence underscores the crucial role of social determinants of health, such as education, income, occupation, and social support networks, as well as psychological factors (e.g., depression, anxiety, sleep) in shaping cognitive trajectories across the lifespan. A recent study by Htun et al. identified 4 multidimensional profiles derived from 72 social indicators, with the most socioeconomically disadvantaged group showing the highest risk of dementia in both men and women, emphasizing the need for targeted interventions.7 More recently, the COVID-19 pandemic, which emerged in 2020, introduced an unprecedented global health and social crisis, whose effects on cognitive health are still being unraveled. Emerging research indicates that SARS-CoV-2 infection may have direct neurobiological impacts,8 including neuroinflammation,9 cerebrovascular alterations,10 and structural brain changes observable through neuroimaging,11 leading to cognitive deficits even in previously healthy individuals.11,12 In parallel, indirect effects of the COVID-19 pandemic, particularly lockdowns, social isolation, and economic disruptions, have exacerbated psychosocial stressors and contributed to declines in mental and cognitive well-being, especially among vulnerable populations such as older adults.13 Moreover, beyond the social stratification and psychological consequences, the public health measures implemented during the COVID-19 pandemic likely had downstream effects on health-related behaviors.13 Increased levels of psychological stress and social isolation may have contributed to higher alcohol and tobacco use in some individuals.14 Nutritional habits may also have been affected, with a possible decline in dietary quality,15 partly due to reduced financial resources, but also because of limited mobility and constrained access to fresh or healthy food during confinement periods.15 In this context, there is a pressing need to disentangle the direct effects of COVID-19 from the broader social consequences of pandemic-related public health measures. This requires high-quality, longitudinal data collected prior to and during the COVID-19 pandemic to capture baseline cognitive function and its evolution over time. The Lc65+ cohort, a population-based longitudinal study launched in Lausanne in 2004, offers a valuable opportunity to examine cognitive aging in relation to socioeconomic, psychological, and health-related factors over time.16 Its repeated assessments make it possible to evaluate changes in cognitive health before and after the onset of the COVID-19 pandemic and to explore whether these changes differed according to pre-existing vulnerabilities. The first aim of the present study, based on data from the Lc65+ cohort, was to examine the evolution of cognitive health and associated factors before and after the onset of the COVID-19 pandemic, using the year 2020 as a temporal benchmark, in 3 birth cohorts of older adults: Cohort 1 (born 1934-1938; pre-World War II [WWII]), Cohort 2 (born 1939-1943; WWII), and Cohort 3 (born 1944-1948; Baby Boomers). We hypothesized an increase in the incidence of global cognitive efficiency deficits across all cohorts, as measured by a standardized cognitive screening tool (Mini-Mental State Examination [MMSE]), reflecting both direct and indirect consequences of the pandemic. Second, through longitudinal analyses, we sought to identify the socioeconomic determinants associated with cognitive decline, with the expectation that older adults experiencing economic and social precarity conditions, likely exacerbated by pandemic-related restrictions, would be disproportionately affected. Together, these analyses aimed at providing a nuanced understanding of how a global health crisis may have influenced cognitive aging, particularly in the context of pre-existing socioeconomic vulnerabilities. Methods Population The Lc65+ study is a population-based longitudinal cohort initiated in Lausanne, Switzerland, to investigate the manifestations, determinants, and consequences of frailty and cognitive aging among community-dwelling older adults.16 For this purpose, 3 representative samples of the non-institutionalized population of Lausanne, aged 65-70 years at baseline, were recruited at 5-year intervals: in 2004 (Cohort 1: birth years 1934-1938, pre-WWII, N = 1564), in 2009 (Cohort 2: birth years 1939-1943, WWII, N = 1489), and in 2014 (Cohort 3: birth years 1944-1948, Baby Boomers, N = 1678). Participants were selected through population registries and gave written informed consent at the first in-person interview. Follow-up combined annual postal questionnaires with in-person interviews and clinical examinations conducted every 3 years at the study center. These assessments included standardized physical and cognitive performance tests, as well as detailed socioeconomic and psychosocial measures. The response rate has remained consistently high over time, with more than 90% of participants retained across waves. Notably, 779 individuals (50% of the initial 2004 sample) were still actively participating in the study 16 years later, attesting to the cohort’s strong longitudinal retention and data quality.16 Importantly, for the present study, cognitive, socioeconomic, and clinical data were analyzed across the 3 cohorts. For the purposes of this study, data collected between 2016 and 2024 were analyzed across all 3 Lc65+ cohorts. This time frame was selected to allow for a comparison of cognitive, socioeconomic, and clinical indicators before and after the onset of the COVID-19 pandemic, using 2020 as a central reference point. The inclusion of this 6-year window enables the identification of temporal trends and potential shifts in cognitive health trajectories associated with the pandemic and its broader societal impacts. Analyses were conducted on participants who had complete data across all 3 assessment waves for their respective cohort. This longitudinal sub-sample included 412 participants in Cohort 1, 702 in Cohort 2, and 826 in Cohort 3. Restricting analyses to individuals with repeated measures ensured consistency in longitudinal comparisons and avoided biases related to differential attrition across time points. Variables of interest In this study, a range of variables was considered to evaluate their association with cognitive performance and impairment in older adults, grouped into 6 conceptual domains reflecting key determinants of cognitive aging. Global cognitive efficiency was assessed using the MMSE,17 with a cut-off ≤24. A cut-off point that is considered conservative and allows us to suggest with some certainty the presence of marked cognitive difficulties. Pandemic exposure was operationalized using a binary temporal indicator, distinguishing assessments conducted before versus after the year 2020, which corresponds to the onset of the COVID-19 pandemic in Switzerland. This variable was used to capture the potential impact of pandemic-related contextual changes on cognitive outcomes. In Switzerland, major confinement measures were implemented from March 2020 onward, including restrictions on social gatherings, reduced mobility, and limited access to healthcare services, particularly affecting older adults. Although the intensity and duration of these measures varied over time, the year 2020 represents a clear structural breakpoint in social and healthcare conditions relevant to cognitive, psychological, and behavioral functioning in this population. Sociodemographic factors included gender, year of birth (used as an indicator of chronological age within each cohort), and household composition, which provide foundational information on biological and social positioning. Although each cohort comprised individuals within a restricted age range, the remaining age variability (approximately 5 years) is sufficient to detect age-related differences in cognitive performance. Therefore, year of birth in the present models reflects chronological age differences rather than cohort effects. Socioeconomic indicators captured aspects of financial vulnerability, specifically through receipt of health insurance premium subsidies and supplementary old-age benefits (AVS/PC), both of which serve as proxies for material precarity in the Swiss social system.18 Psychological factors were assessed through a series of self-reported measures addressing symptoms of depression, anxiety, apathy, and sleep disturbances. These psychological symptoms, which have been widely reported in association with the COVID-19 pandemic, have highlighted an increased vulnerability in mental health within the population. Physical activity was considered low when all 3 of the following conditions were met: (1) engaging in less than 20 min of sports activity once per week, (2) walking for less than 30 min on 3 occasions per week, and (3) avoiding stair climbing or carrying light loads during daily routines.19 Questions were adapted from the Monitoring of Trends and Determinants in Cardiovascular Disease Physical Activity Questionnaire.20 Dietary habits were assessed through regular intake of key food groups known to be related to healthy aging, including fruits or vegetables (≥2 times/day), eggs or legumes (≥2 times/week), dairy products (≥1 time/day), and daily consumption of meat, fish, or poultry. These indicators provide insight into nutritional patterns that may influence cognitive health through inflammation, vascular risk, or micronutrient pathways.21 Alcohol and tobacco consumption were assessed based on frequency of use, with tobacco consumption evaluated according to current smoking status. These variables were included due to their well-documented associations with both vascular and cognitive health, as supported by extensive epidemiological and clinical research. Together, these variables offer a multidimensional view of the biological, behavioral, and social factors that may contribute to late-life cognitive outcomes and allow for exploration of both long-term determinants and pandemic-related disruptions across cohorts. Statistical analysis First, intragroup analyses were conducted within each cohort across time points using Pearson’s Chi-squared tests for ordinal variables and MANOVA for continuous variables, covering all variables of interest. Moreover, based on normative MMSE thresholds (dichotomized as cognitively impaired [≤24 on the MMSE] vs within normal limits),17 annual incidence analyses on the whole cohort and prevalence analysis on participants with responses on 3 time points were conducted for each cohort across the years of interest. Second, to investigate longitudinal associations between cognitive outcomes and a set of behavioral, sociodemographic, and clinical variables potentially influenced by the COVID-19 pandemic, generalized estimating equations (GEEs) were applied separately for each of the 3 cohorts (Cohorts 1, 2, and 3). Two distinct models were estimated: Model 1, using MMSE as a continuous outcome, was fitted using GEE with linear regression; Model 2, based on MMSE dichotomized according to normative cutoffs (impaired vs normal), was analyzed using GEE with a binary logistic function to account for the categorical nature of the outcome. The choice of distribution and link function in each model was adapted to the structure of the dependent variable and the repeated measures design. Results Evolution of cognitive deficits, socioeconomic indicators, physical activity, alcohol use, tobacco use, psychological factors, and dietary habits (Aim 1a) Global cognitive efficiency (MMSE) In Cohort 1, a statistically significant decline in global cognitive performance was observed over time. Mean MMSE scores decreased from 26.80 (SD = 2.40) in 2017 to 26.54 (SD = 2.61) in 2020 and 25.58 (SD = 3.33) in 2023 (χ2 = 65.46, p < .001). The proportion of participants classified as cognitively deficient increased from 16.5% to 19.4% and 30.6% across the 3 time points (χ2 = 26.37, p < .001). In Cohort 2, mean MMSE scores declined significantly from 27.44 (SD = 1.96) in 2016 to 27.03 (SD = 2.38) in 2019 and 26.81 (SD = 2.66) in 2022 (χ2 = 48.11, p < .001). The prevalence of cognitive deficit also increased from 8.4% to 12.5% and 14.8% over time (χ2 = 14.12, p < .001). In Cohort 3, mean MMSE scores showed a smaller but statistically significant decline from 27.47 (SD = 2.10) in 2018 to 27.35 (SD = 2.19) in 2021 and 27.17 (SD = 2.46) in 2024 (χ2 = 7.85, p = .020). The proportion of participants with global cognitive efficiency alterations increased slightly from 9.9% to 9.8% and 12.8%, though this change did not reach statistical significance (χ2 = 5.01, p = .082). Socioeconomic indicators Across all 3 cohorts, there was a significant increase in the proportion of participants receiving financial assistance for health insurance premiums. Assistance uptake rose from 15.5% to 28.4% and 32.8% in Cohort 1 (χ2 = 34.75, p < .001), from 13.2% to 24.5% and 29.1% in Cohort 2 (χ2 = 53.73, p < .001), and from 15.0% to 25.9% and 30.9% in Cohort 3 (χ2 = 59.73, p < .001). Physical activity The proportion of participants reporting low physical activity increased significantly over time in all 3 cohorts: from 9.7% to 11.9% and 18.5% in Cohort 1 (χ2 = 14.91, p < .001); from 8.4% to 8.5% and 13.6% in Cohort 2 (χ2 = 13.28, p = .001); and from 5.4% to 9.1% and 8.6% in Cohort 3 (χ2 = 9.12, p = .010). Psychological factors Anxiety symptoms increased over time in Cohort 2 (from 24.7% to 31.1% and 32.3%; χ2 = 11.19, p = .004) and in Cohort 3 (from 23.1% to 28.3% and 29.9%; χ2 = 10.60, p = .005). In Cohort 1, the increase was not statistically significant (33.4% to 33.3% and 40.1%; χ2 = 8.13, p = .087). Depressive symptoms remained low and stable across time in all cohorts (p > .05). Apathy increased significantly in each cohort: from 17.5% to 23.3% and 26.3% in Cohort 1 (χ2 = 11.50, p = .021); from 14.7% to 16.6% and 19.5% in Cohort 2 (χ2 = 6.01, p = .049); and from 12.1% to 15.9% and 17.2% in Cohort 3 (χ2 = 9.03, p = .011). Sleep quality remained stable across time points in all cohorts (p > .05). Alcohol use Patterns of alcohol consumption shifted modestly. Significant changes in the distribution of consumption frequency were observed in Cohort 2 (χ2 = 23.75, p = .003), while differences in Cohort 1 (χ2 = 9.86, p = .275) and Cohort 3 (χ2 = 11.69, p = .166) did not reach significance. Tobacco use Smoking prevalence did not change significantly over time in any cohort (p > .05). Dietary habits Daily consumption of fruits and vegetables (≥2 times/day) remained stable across all cohorts (p > .60). The proportion of participants consuming eggs or vegetables (≥2 times/week) decreased significantly in Cohort 1 (from 78.4% to 69.7% and 71.8%; χ2 = 8.71, p = .013) and Cohort 2 (from 79.5% to 73.6% and 75.0%; χ2 = 7.20, p = .027), but remained stable in Cohort 3 (p = .310). Daily consumption of dairy products (≥1 time/day) was stable across all cohorts (p > .10). In contrast, daily consumption of meat, fish, or poultry decreased significantly in all 3 cohorts: from 41.5% to 28.4% and 27.9% in Cohort 1 (χ2 = 22.30, p < .001); from 45.4% to 27.9% and 26.4% in Cohort 2 (χ2 = 70.65, p < .001); and from 33.9% to 26.8% and 25.7% in Cohort 3 (χ2 = 16.03, p < .001). All results are displayed in Table 1. Table 1Descriptive data for each cohort and significant differences between time points.VariableResponseCohort 1 (n = 412)Cohort 2 (n = 702)Cohort 3 (n = 826)T1 (2017)T2 (2020)T3 (2023) p T1 (2016)T2 (2019)T3 (2022) p T1 (2018)T2 (2021)T3 (2024) p MMSE raw, mean (SD) 26.80 (2.40)26.54 (2.61)25.58 (3.33) .001 27.44 (1.96)27.03 (2.38)26.81 (2.66) .001 27.47 (2.10)27.35 (2.19)27.17 (2.46) .020 MMSE dichotomized, % Normal83.580.669.4 .001 91.687.585.2 .001 90.190.287.2.082Deficitary16.519.430.68.412.514.89.99.812.8 Gender, % Male37.1 ND 39.0 ND 41.8 ND Female62.961.058.2 Financial assistance, % No84.571.667.2 .001 86.875.570.9 .001 85.074.169.1 .001 Yes15.528.432.813.224.529.115.025.930.9 Physical activity, % No90.388.181.5 .001 91.691.586.4 .001 94.690.991.4 .010 Yes9.711.918.58.48.513.65.49.18.6 Anxiety, % No67.666.759.9.08775.368.967.7 .004 76.971.770.1 .005 Yes33.433.340.124.731.132.323.128.329.9 Depression, % No96.195.194.1.49297.395.995.6.19496.796.694.9.103Yes3.94.95.92.74.14.13.33.45.1 Apathy, % No82.576.773.5 .021 85.383.480.5 .049 87.984.182.8 .011 Yes17.523.326.314.716.619.512.115.917.2 Sleep, % 143.243.640.3.77649.944.445.2.06049.345.543.2.092247.646.248.943.745.445.643.745.348.539.310.210.56.410.19.17.09.28.4 Alcohol, % 110.011.213.1.2759.810.711.0 .003 8.29.38.8.16628.713.311.97.811.715.410.011.514326.522.125.323.922.919.721.923.223.8417.517.218.019.118.719.721.721.521.4537.436.230.039.336.034.138.134.531.8 Tobacco, % 17.56.86.4.85511.410.89.9.66113.312.010.7.486238.238.839.943.345.247.246.148.549.3354.354.413.245.344.042.940.639.540.0 Diet: fruits or vegetables (≥2x day), % No8.07.510.2.8249.510.111.0.66312.112.213.4.666Yes92.092.589.890.589.989.087.987.886.6 Diet: eggs or legumes (≥2x week), % No21.630.328.2 .013 20.526.425.0 .027 23.225.422.3.310Yes78.469.771.879.573.675.076.874.677.7 Diet: dairy products (≥1x day), % No10.211.411.4.81311.314.514.6.11315.516.214.5.633Yes89.888.688.688.785.585.484.583.885.5 Diet: meat, fish, or poultry (daily), % No58.571.672.1 .001 54.672.173.6 .001 66.173.274.3 .001 Yes41.528.427.945.427.926.433.926.825.7Missing data for each variable of interest ranged from 0.0% to 0.6% of the cohort-year. p-values in boldface indicate p < .05.Abbreviation: MMSE, Mini-Mental State Examination. Descriptive data for each cohort and significant differences between time points. Missing data for each variable of interest ranged from 0.0% to 0.6% of the cohort-year. p-values in boldface indicate p < .05. Abbreviation: MMSE, Mini-Mental State Examination. Incidence and prevalence of global cognitive efficiency deficits over time (Aim 1b) The incidence of global cognitive efficiency deficits was examined across the 3 Lc65+ cohorts over the period 2016-2024, with 2020 marking the onset of the COVID-19 pandemic. Distinct temporal patterns were observed across cohorts. In Cohort 1 (1934-1938; pre-WWII), incidence peaked in 2017 (≈0.48), declined in 2020 (≈0.30), and increased again by 2023 (≈0.45). In Cohort 2 (1939-1943; WWII), incidence decreased progressively from 2016 (≈0.32) to 2022 (≈0.21), with no marked change around 2020. Cohort 3 (1944-1948; Baby Boomers) exhibited the lowest and most stable incidence values, ranging from 0.30 in 2018 to 0.19 in 2021, with no subsequent increase (see Figure 1). Cohort-specific trends in (A) incidence and (B) prevalence of cognitive deficits. Cognitive deficits are measured with psychometric thresholds of the Mini-Mental State Examination. Prevalence of global cognitive efficiency deficits was evaluated in the subcohort of participants with complete data across all assessment waves (2016-2024). The overall prevalence varied over time, with values of 8.4% in 2016, 16.5% in 2017, 9.9% in 2018, 12.5% in 2019, 19.4% in 2020, 9.8% in 2021, 14.8% in 2022, 30.6% in 2023, and 12.8% in 2024. When examined by cohort, Cohort 1 presented the highest prevalence with the highest values throughout the observation period, increasing from 2017 to 2023, with a significant increase (p < .001). Cohort 2 showed intermediate and moderately increasing prevalence values from 2016 to 2022, still significant (p < .001), while Cohort 3 consistently displayed the lowest prevalence, with minor fluctuations across years, not significant over time (p = .082) (see Figure 1). Longitudinal relationship between cognitive performances and socioeconomic indicators, physical activity, alcohol use, tobacco use, psychological factors, as well as dietary habits (Aim 2) MMSE total score (as a continuous variable) In Cohort 1, assessment timing was a strong predictor: being evaluated before the COVID-19 pandemic was associated with significantly higher MMSE scores (β = 0.027, p < .001). Insomnia symptoms were negatively related to cognitive performance (β = –0.025, p = .038), consistent with the cognitive burden of sleep difficulties. Two behavioral markers were also significant: lower levels of physical activity (reference vs active) were associated with better cognition (β = 0.026, p = .019), while diet quality was positively associated with MMSE scores (β = 0.014, p = .026). Finally, year of birth showed a positive effect (β = 0.010, p = .001), with later-born individuals exhibiting better cognitive outcomes. Alcohol consumption, depression, anxiety, and apathy were not significant in this cohort (see Figure 2). Heatmap of significant standardized regression coefficients (β) for predictors of cognitive performance (MMSE) across cohorts and model types. MMSE, Mini-Mental State Examination. *p < .05. **p < .01. ***p < .001 In Cohort 2, several predictors emerged in this group. First, financial assistance showed a significant positive effect: not receiving aid was associated with better cognitive performance (β = 0.017, p = .002). Sex was also important: men had lower MMSE scores compared to women (β = –0.016, p = .003). As in Cohort 1, being assessed before the pandemic predicted better cognition (β = 0.010, p < .001). Year of birth again exerted a positive influence (β = 0.007, p < .001. Alcohol consumption and psychological symptoms were not significant (p > .05) (see Figure 2). In Cohort 3, patterns were partly consistent with previous cohorts. Daily alcohol consumption was negatively associated with MMSE scores (β = –0.019, p = .030), whereas moderate intake (level 4) had a positive association (β = 0.010, p = .024), suggesting a potential dose-dependent effect. Anxiety symptoms were also significant: participants without anxiety had higher cognitive scores (β = 0.008, p = .031), highlighting the detrimental impact of psychological distress. Similar to the other cohorts, assessment before the pandemic predicted higher MMSE scores (β = 0.007, p = .003), and year of birth was again positively associated (β = 0.005, p = .003). No strong effects were observed for financial assistance, insomnia, or other behavioral markers in this cohort (see Figure 2). Dichotomized MMSE (impaired vs normal [≤24]) In Cohort 1, receiving financial assistance was positively associated with the likelihood of a normal MMSE score (β = 0.017, p = .002), suggesting that beneficiaries of such aid were more likely to maintain normal cognitive status. Year of birth was also a significant predictor (β = 0.010, p = .001), with more recent birth cohorts showing higher odds of normal cognition. Furthermore, being assessed before the pandemic was positively associated with cognitive normality (β = 0.027, p < .001), implying that participants evaluated prior to the COVID-19 pandemic had better outcomes. No significant associations emerged for depression, anxiety, insomnia, or alcohol use in this cohort (see Figure 2). In Cohort 2, results highlighted several distinct predictors. Sex was significant: men had lower odds of normal MMSE classification compared to women (β = –0.016, p = .003). Year of birth again showed a positive association (β = 0.007, p < .001), meaning that younger individuals were more likely to preserve normal cognitive performance. Being assessed before the pandemic was also positively associated with cognitive normality (β = 0.010, p < .001). No robust effects were observed for financial assistance, depressive symptoms, or alcohol consumption in this group (see Figure 2). In Cohort 3, anxiety symptoms were negatively associated with normal cognition (β = –0.008, p = .031), underlining the cognitive toll of psychological distress. Year of birth again showed a positive effect (β = 0.005, p = .003), with later-born participants faring worse. Being assessed before the pandemic also predicted a higher likelihood of normal MMSE classification (β = 0.007, p = .003). Other health behaviors (e.g., alcohol, insomnia, diet proxies) were not significantly related to cognition in this cohort (see Figure 2). Discussion This study explored cognitive health trajectories in a Swiss aging study, using both continuous and dichotomized MMSE outcomes to examine the impact of sociodemographic, behavioral, psychological, and contextual factors, including those linked to the COVID-19 pandemic. By spanning pre- and post-pandemic periods, this analysis enabled interpretation within a framework of acute societal disruption and chronic population aging. The dual modeling of continuous and dichotomous MMSE outcomes allowed for a comprehensive understanding of cognitive trajectories, capturing both subtle shifts and clinically meaningful impairments. Pandemic timing was associated with cognitive differences. Pre-pandemic assessments were linked to higher MMSE scores in continuous models, supporting the notion that pandemic-related stressors negatively impacted cognition. However, inconsistencies in dichotomized results, particularly in Cohort 1, suggest that binary classification may obscure subtle variations, reinforcing the value of using both types of outcome measures: continuous and dichotomized. Socioeconomic vulnerability, approximated here by receipt of financial assistance for health insurance, emerged as an important factor in several analyses, although the direction and strength of the association were not fully consistent across cohorts and models. Accordingly, this variable should be interpreted with caution, particularly in the dichotomous analyses, where cohort-specific patterns differed. Nevertheless, the results support the broader view that socioeconomic disadvantage may shape cognitive aging through multiple pathways, including healthcare,22 increased chronic stress,23 and diminished cognitive reserve.24 The observed rise in financial assistance uptake across cohorts may also reflect increasing financial fragility in older adults during the study period, potentially reinforced by pandemic-related economic pressures.25 Cohort-specific cognitive trajectories revealed meaningful differences. In Cohort 1, the oldest and most exposed to prolonged pandemic conditions, a significant decline in MMSE scores and in the proportion of participants with normal cognitive status was observed. This pattern likely reflects a compound effect of age-related vulnerability, isolation, and service disruption.13 Cohort 2 showed a more modest decline in MMSE scores without a significant change in categorical impairment, possibly due to shorter exposure or greater resilience, due to a potential cohort effect, with individuals born more recently who may have benefited from societal advancements.26 Cohort 3, the youngest, exhibited cognitive stability, suggesting either resistance to acute disruptions or insufficient follow-up time to detect delayed effects. Given the gradual nature of neurodegenerative processes, late-onset pandemic effects may yet emerge.12,27 Psychological symptoms emerged as critical and distinct contributors to cognitive aging across cohorts. Anxiety prevalence increased significantly in Cohorts 1 and 3 but remained stable in Cohort 2, suggesting differential psychological vulnerability across age groups. The rise in anxiety in Cohort 1 may reflect heightened isolation, health fears, and reduced autonomy, while in Cohort 3, it could signal increased stress linked to digital exclusion, economic concerns, or caregiving responsibilities.28,29 Depression and apathy levels remained low and stable overall, though apathy showed a non-significant upward trend in Cohort 1, possibly indicating early neuropsychiatric manifestations of cognitive decline. Sleep disturbances (insomnia) were also stable, though subtle shifts in sleep patterns were noted and may merit further investigation. Concerning behavioral patterns, alcohol consumption patterns shifted over time, with a general trend toward reduced regular consumption across all cohorts. Continuous models linked daily alcohol use to poorer global cognitive efficiency performance, particularly in older cohorts, while moderate consumption was associated with better cognitive outcomes, notably in Cohorts 2 and 3. These findings echo prior research on the potential benefits of moderate alcohol intake on vascular and cognitive health,30 although the behavioral disruptions of lockdown warrant caution in interpretation.31 Tobacco use declined modestly over time, significantly so only in Cohort 3. Associations with cognition were inconsistent and likely confounded by survivorship bias, lifestyle clustering, or cohort-specific cultural norms. Physical activity patterns were also cohort dependent. In Cohort 2, physical activity was associated with better cognitive scores in continuous models, consistent with literature on the neuroprotective effects of exercise.32 In Cohort 1, however, more frequent outings, possibly a proxy for social engagement, were paradoxically linked to greater impairment, potentially reflecting reverse causality, whereby declining individuals attempt to maintain routines. The association between lower physical activity and better cognitive outcomes observed in the oldest cohort should be interpreted with caution. Reverse causality may be involved, whereby individuals with emerging cognitive difficulties attempt to remain physically active, while cognitively preserved individuals reduce activity due to physical constraints or comorbidities. Moreover, the physical activity indicator used in this study primarily reflects the frequency of daily mobility and routine activities rather than exercise intensity and may therefore capture aspects of social engagement rather than purely physical exertion. Finally, as this measure relies on self-report, it may be subject to recall bias and limited sensitivity, particularly in older adults and during periods of disrupted daily routines such as the COVID-19 pandemic. Nevertheless, the pandemic’s disruption of habitual activity may have also introduced measurement variability and recall bias. Nutritional patterns revealed complex dynamics across cohorts. Although daily fruit and vegetable consumption remained stable, egg/legume intake declined significantly in Cohorts 1 and 2, possibly due to supply issues or financial constraints. Daily dairy consumption remained unchanged. Most notably, meat, fish, or poultry consumption declined significantly across all cohorts, pointing to a possible pandemic-related reduction in protein intake or intentional shifts toward plant-based diets. The observed patterns echo principles of the Mediterranean diet, associated with improved cognitive outcomes and longevity.33 However, disruptions to food access during the pandemic may have led to declines in diet quality, potentially exacerbating vulnerability among older adults. That said, the apparent decrease in the incidence of global cognitive efficiency deficits observed around the onset of the COVID-19 pandemic should not be interpreted as a true improvement in cognitive functioning. Rather, it likely reflects the complex temporal dynamics of cognitive decline and neurocognitive disorders.34 Incidence and prevalence capture distinct processes: while prevalence continues to increase, indicating an accumulation of cognitive vulnerability in the population, incidence may transiently decline when individuals with pre-existing vulnerability progress earlier to more advanced cognitive stages, thereby contributing to prevalence rather than newly incident cases. Moreover, the cognitive consequences of major societal stressors such as the COVID-19 pandemic may not be immediately detectable, as neurocognitive disorders develop gradually and may involve delayed or cumulative effects. It is therefore plausible that the pandemic acted as an accelerator of existing vulnerability or contributed to the emergence of new risk factors (e.g., prolonged social isolation, psychological distress, behavioral changes), whose impact may only become evident at later ages. Importantly, the cohort-sequential design of the Lc65+ study offers a unique opportunity to address this hypothesis in future work by comparing age-specific incidence rates across cohorts assessed before and after the pandemic, thereby allowing the investigation of potential cohort effects related to pandemic exposure. Several limitations must be acknowledged. First, attrition represents a potential source of bias. As described in the published Lc65+ cohort profile,16 drop-out over follow-up was not associated with sex but was more frequent among participants with lower education, financial difficulties, poorer self-rated health, and foreign country of birth. Consequently, the restriction of the present analyses to participants with complete longitudinal data may have resulted in an underrepresentation of socioeconomically and clinically more vulnerable individuals. In addition, selective survival may have occurred, as participants with the most severe health conditions or cognitive impairment are more likely to die during follow-up, potentially further contributing to an underrepresentation of the most vulnerable individuals. Second, the MMSE, while widely used, is a screening tool that primarily captures global cognitive efficiency, lacks sensitivity to subtle or domain-specific impairments, and does not assess functional impact on daily activities, limiting its interpretability in terms of general cognitive health or neurocognitive disorder diagnosis. Third, although pandemic-related patterns are evident, causal interpretations remain speculative due to the observational design. Fourth, infection data were incomplete across cohorts, limiting conclusions on direct viral impacts. Future studies should incorporate richer cognitive testing, mental health measures, biomarkers, and more detailed dietary assessments to fully capture the evolving consequences of the pandemic on cognitive aging. Conclusion These results illustrate how cognitive aging is shaped by a combination of social, behavioral, generational, and contextual factors. The COVID-19 pandemic and its associated public health measures likely amplified existing vulnerabilities, especially among socioeconomically disadvantaged and older individuals. These factors have led to changes in health behaviors and access to care in ways that may influence long-term cognitive trajectories. These findings underscore the importance of designing public health interventions that not only mitigate infectious disease risk but also preserve cognitive and mental health, particularly in times of crisis.