Association of educational attainment with incident orientation impairment among Chinese older adults with and without stroke: evidence from a longitudinal study with external validation

Association of educational attainment with incident orientation impairment among Chinese older adults with and without stroke: evidence from a longitudinal study with external validation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Association of educational attainment with incident orientation impairment among Chinese older adults with and without stroke: evidence from a longitudinal study with external validation Ce Shi, Lihua Wu, Qiqi Yang, Fei Wang, Xiang Shang, Tianxin Jiang, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9092847/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background Orientation impairment is an early and clinically meaningful marker of cognitive vulnerability in older adults and is associated with subsequent functional decline, dementia, and mortality. Although stroke is a well-established risk factor for late-life cognitive impairment, its association with incident orientation impairment in community-dwelling populations remains unclear. Educational attainment, a key indicator of cognitive reserve, may modify the cognitive consequences of stroke, but longitudinal evidence addressing this hypothesis is limited. Methods We conducted a population-based longitudinal study using data from the China Health and Retirement Longitudinal Study (CHARLS). Participants aged ≥ 60 years who were free of orientation impairment at baseline (2018) were followed for incident orientation impairment in 2020. Stroke history was defined by self-reported physician diagnosis at baseline. Educational attainment was primarily modeled as a continuous variable (per three-year increase) and additionally evaluated using categorical specifications in sensitivity analyses. Modified Poisson regression with robust variance estimation was used to estimate relative risks (RRs). Effect modification by education was assessed using interaction terms and marginal effect estimation. Sensitivity analyses included inverse probability weighting for loss to follow-up, competing-risk models accounting for death, alternative outcome definitions, and complementary machine-learning analyses. External validation was conducted to examine the transportability of relative risk patterns rather than to formally validate a clinical prediction model. Results Among 9,080 participants free of orientation impairment at baseline, 5,289 participants with observed follow-up orientation assessments were included (complete-case with respect to the outcome), of whom 744 developed incident orientation impairment over two years. Stroke history was not independently associated with incident orientation impairment after multivariable adjustment (RR 0.88, 95% CI 0.63–1.24). In contrast, higher educational attainment showed a strong and consistent protective association: each additional three years of education was associated with a 25.6% lower risk of incident orientation impairment (RR 0.74, 95% CI 0.65–0.86). Formal tests for stroke–education interaction were not statistically significant on multiplicative or additive scales; however, marginal estimates were directionally consistent with attenuation of stroke-associated risk at higher education levels, but estimates were imprecise and confidence intervals crossed the null. Results were robust across sensitivity analyses. Machine-learning models demonstrated modest discrimination and did not materially outperform traditional regression models. External validation showed moderate discrimination but indicated miscalibration of absolute risk estimates, suggesting the need for recalibration in different population settings. Conclusions In this nationally representative cohort of older adults, educational attainment emerged as a consistent and robust protective factor against incident orientation impairment, whereas stroke history showed no independent association after adjustment. Evidence for education modifying the stroke–orientation relationship was suggestive but inconclusive. These findings highlight the central role of educational gradients in late-life cognitive vulnerability and underscore the challenges of detecting domain-specific cognitive consequences of stroke in population-based studies. Orientation impairment Educational attainment Stroke history Cognitive reserve External validation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Background Population ageing has substantially increased the burden of cognitive impairment and related functional decline worldwide, posing major challenges to health and social care systems for older adults[ 1 – 3 ]. Among different cognitive domains, orientation represents a core component of cognitive functioning and is essential for independent living and everyday decision-making in older adults[ 4 , 5 ]. Impairment in orientation has been associated with an increased risk of subsequent dementia, functional dependence, and mortality, making it a clinically relevant and epidemiologically meaningful early marker of cognitive vulnerability in later life[ 6 , 7 ]. Stroke is a leading cause of long-term disability and cognitive impairment in older adults[ 8 ]. Beyond its acute neurological consequences, stroke may initiate or accelerate vascular and neurodegenerative processes that adversely affect cognitive functioning over time[ 9 ]. Previous studies have consistently demonstrated an increased risk of global cognitive decline and dementia among stroke survivors[ 10 ]. However, evidence regarding domain-specific cognitive outcomes—particularly orientation impairment as an early and clinically interpretable marker—remains limited in population-based longitudinal studies. Importantly, the cognitive consequences of stroke are unlikely to be uniform across individuals and may vary according to markers of cognitive reserve[ 11 , 12 ]. Educational attainment, a core indicator of cognitive reserve and a key social determinant of health, has been widely recognised as an important determinant of cognitive ageing trajectories[ 13 ]. Higher educational attainment may buffer the cognitive consequences of brain pathology through enhanced neural efficiency and compensatory mechanisms[ 14 ]. Nevertheless, whether educational attainment modifies the association between stroke history and subsequent orientation impairment remains uncertain, particularly in longitudinal, population-based settings. Most existing studies have treated education primarily as a confounder or have focused on global cognitive outcomes, with limited attention to potential effect modification across specific cognitive domains. From a public health perspective, understanding heterogeneity in stroke-related cognitive risk is essential for identifying vulnerable subgroups and developing targeted prevention strategies. This question is particularly relevant in rapidly ageing populations such as China, where substantial educational disparities persist among older adults as a result of historical and socioeconomic factors[ 15 – 17 ]. Using data from a nationally representative longitudinal cohort of Chinese older adults, this study aimed to examine the association between stroke history and incident orientation impairment and to explore whether this association differs according to educational attainment. We further evaluated the robustness of our findings using multiple analytical approaches, including inverse probability weighting, competing risk models accounting for mortality, and complementary machine learning analyses. Methods Study design and data source This study was a population-based longitudinal analysis using data from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative survey of middle-aged and older adults in China. CHARLS employs a multistage, stratified probability sampling design covering both urban and rural areas and collects comprehensive information on sociodemographic characteristics, health conditions, and cognitive functioning among community-dwelling adults. Baseline data were obtained from the 2018 wave, with follow-up assessments conducted in 2020. The CHARLS study protocol was approved by the Institutional Review Board of Peking University, and all participants provided written informed consent. The present analysis was based on publicly available, de-identified data. Study population The analytic sample was restricted to participants aged 60 years or older at baseline (2018). Among the 5,289 participants included in the analytic sample, the mean age was 61.44 years (SD 10.02). To ensure a clear temporal ordering between exposure and outcome, analyses were further restricted to individuals who were free of orientation impairment at baseline. Participants were included if they had available baseline information on stroke history and educational attainment and completed the orientation assessment at the 2020 follow-up. Individuals without observed orientation assessments at follow-up were excluded from the primary analysis (complete-case with respect to the outcome). Baseline covariates were retained as observed, with missing categories incorporated for categorical variables in regression models. Potential bias arising from loss to follow-up was examined using inverse probability weighting in sensitivity analyses. Exposure assessment: stroke history Baseline stroke history was assessed using self-reported physician diagnosis. Participants were asked whether they had ever been diagnosed with stroke by a physician, and responses were coded as a binary variable (yes or no). Self-reported physician-diagnosed stroke has been widely used in population-based studies and has demonstrated acceptable validity in the CHARLS cohort. Outcome assessment: orientation impairment Cognitive performance in CHARLS is evaluated using a standardized methodological framework consistent with that adopted in the U.S. Health and Retirement Study (HRS)[ 18 ]. As described by Crimmins et al., this framework conceptualizes cognition across four domains: orientation, calculation, memory, and visuospatial ability. These domains are assessed using structured cognitive tests, yielding a total cognitive score ranging from 0 to 31 points. Orientation ability is measured using items derived from the Cognitive Status Telephone Interview, including questions on date and season, with a maximum score of 5 points. Calculation ability is assessed through a serial subtraction task (subtracting 7 from 100 consecutively), with a maximum score of 5 points. Memory is evaluated using immediate and delayed word recall tasks, with a maximum score of 20 points. Visuospatial ability is assessed through a figure-copying task, contributing 1 point to the total score. The present study focuses specifically on the orientation domain as the primary outcome. The decision to concentrate on orientation rather than global cognitive score was based on several considerations. First, orientation reflects a core component of early cognitive decline and demonstrates structural stability across survey waves. Second, orientation items exhibit strong measurement consistency over time, facilitating longitudinal comparisons. Third, impairment in orientation has well-established neurobiological relevance in post-stroke cognitive dysfunction. Incident orientation impairment was defined as the occurrence of impaired orientation at the 2020 follow-up among participants who were free of orientation impairment at baseline in 2018. By restricting the analysis to incident cases, this study minimizes potential reverse causation and strengthens temporal inference between baseline stroke history and subsequent cognitive decline. Educational attainment Educational attainment was measured at baseline. In the primary analysis, education was modeled as years of schooling and scaled per three-year increase to facilitate interpretability. Categorical specifications reflecting the Chinese education system were examined in supplementary analyses: no formal education, primary education, middle school education, and high school or higher education. Educational attainment was conceptualised as a stable indicator of early-life socioeconomic position and cognitive reserve and was prespecified as a potential effect modifier of the association between stroke history and incident orientation impairment. Covariates Baseline covariates were selected a priori based on previous literature and conceptual considerations. These included demographic characteristics (age and sex), socioeconomic factors (marital status and residential area), and major baseline health-related variables. All covariates were measured at baseline to minimise temporal ambiguity and reduce the risk of overadjustment. Statistical analysis The primary objective was to examine the association between baseline stroke history and incident orientation impairment and to explore whether this association differed according to educational attainment. Modified Poisson regression with cluster-robust standard errors was used to estimate relative risks (RRs) and 95% confidence intervals (CIs). Participants with missing orientation assessment at follow-up were excluded from the primary analysis (complete-case with respect to the outcome only). Baseline covariates were retained as observed. For categorical variables, missing values were included as separate indicator categories in regression models rather than excluding participants with incomplete covariate data. This approach is well suited to cohort studies with binary outcomes and provides directly interpretable risk estimates. Educational attainment was first included as an adjustment variable in multivariable models. Potential effect modification by educational attainment was evaluated by introducing an interaction term between stroke history and education. Statistical evidence for interaction was assessed using Wald tests. To facilitate interpretation, marginal and education-specific associations between stroke history and incident orientation impairment were derived from interaction models. Several sensitivity analyses were conducted to assess the robustness of the findings. Inverse probability weighting was applied to address potential selection bias due to loss to follow-up. Competing-risk analyses were performed to account for death during follow-up as a competing event for incident orientation impairment. Additional analyses included alternative definitions of orientation impairment and complementary machine-learning approaches as robustness checks rather than as primary inferential tools. Complementary machine-learning analyses To further assess the robustness of the observed associations across analytical frameworks, complementary machine-learning analyses were conducted. Multiple supervised learning algorithms were trained using baseline variables to predict incident orientation impairment over the follow-up period. Model performance was evaluated using cross-validated out-of-fold procedures, with discrimination and calibration assessed as supportive metrics. These analyses were not intended to develop or optimise a predictive model for clinical use. Rather, they served as robustness checks to examine whether the relative importance and direction of associations for stroke history and educational attainment were consistent with findings from the primary regression-based analyses. External validation External validation was performed using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), an independent population-based cohort of older adults in China[ 19 ]. Inclusion criteria and outcome definitions were aligned as closely as possible with those used in the primary analysis, with an emphasis on conceptual rather than exact measurement equivalence. The purpose of external validation was to assess the transportability of the observed associations and overall risk patterns, rather than to formally validate a prediction model[ 20 ]. Modified Poisson regression models were re-estimated in the CLHLS cohort to assess consistency of relative associations. Cross-validated discrimination and calibration metrics were examined descriptively to evaluate consistency with findings from CHARLS[ 21 ]. Statistical software All statistical analyses were conducted using R software (version 4.5.2). Statistical significance was defined as a two-sided p-value < 0.05. Results Study population and event rates Among CHARLS participants who were free of orientation impairment at baseline in 2018, 9,080 individuals constituted the risk set. Of these, 5,289 participants had observed orientation assessments at the 2020 follow-up and were included in the primary complete-case analysis. The mean age of the analytic sample was 61.44 years (SD 10.02). Over the two-year follow-up, 744 participants developed incident orientation impairment under the primary definition (orientation score ≤ 3). Using a stricter definition (orientation score ≤ 2), 231 incident cases were identified. The cohort derivation and analytic sample selection are shown in Fig. 1 . Participants were drawn from the China Health and Retirement Longitudinal Study (CHARLS). From the original master dataset (N = 19,395), individuals aged < 60 years or without valid baseline orientation assessment were excluded. The final risk set included 9,080 participants free of orientation impairment at baseline (2018). Among them, 5,289 participants with non-missing orientation assessment at follow-up (2020) were included in the primary complete-case analysis, yielding 744 incident cases of orientation impairment. Primary analyses used modified Poisson regression with cluster-robust standard errors; additional spline, interaction, inverse probability weighting, and machine-learning analyses were conducted as sensitivity and supplementary analyses. Baseline characteristics stratified by stroke history among participants included in the analytic sample (complete-case with respect to follow-up outcome only; N = 5,289) are shown in Table 1 . Participants with a history of stroke had a substantially higher prevalence of vascular comorbidities, with very large standardized mean differences observed for hypertension and diabetes, indicating pronounced baseline imbalance between groups, whereas age and educational attainment were broadly comparable between groups. Table 1 Baseline characteristics by stroke history Characteristic No stroke Stroke history SMD N 5024 265 Age, years 61.44 (10.06) 61.53 (9.33) 0.009 Age missing 0 (0.0%) 0 (0.0%) Education, years 3.79 (1.94) 3.97 (1.88) 0.090 Education missing 0 (0.0%) 0 (0.0%) Sex Male 2374 (47.3%) 124 (46.8%) 0.013 Female 2650 (52.7%) 141 (53.2%) Missing 0 (0.0%) 0 (0.0%) Hukou 1 Agricultural Hukou 3560 (70.9%) 187 (70.6%) 0.093 2 Non-agricultural Hukou 1323 (26.3%) 73 (27.5%) 3 Unified Residence Hukou 15 (0.3%) 0 (0.0%) Missing 126 (2.5%) 5 (1.9%) Hypertension No 3214 (64.0%) 81 (30.6%) 0.880 Yes 518 (10.3%) 48 (18.1%) Missing 1292 (25.7%) 136 (51.3%) Diabetes No 3850 (76.6%) 153 (57.7%) 0.541 Yes 514 (10.2%) 48 (18.1%) Missing 660 (13.1%) 64 (24.2%) Myocardial infarction No 4367 (86.9%) 202 (76.2%) 0.383 Yes 262 (5.2%) 29 (10.9%) Missing 395 (7.9%) 34 (12.8%) Notes: Values are mean (SD) for continuous variables and n (%) for categorical variables. SMD = standardized mean difference. Association between stroke history, education, and incident orientation impairment In multivariable modified Poisson regression models with cluster-robust standard errors, baseline stroke history was not independently associated with incident orientation impairment after adjustment for age, sex, educational attainment, hukou status, and vascular comorbidities (RR 0.88, 95% CI 0.63–1.24; Table 2 ). In contrast, educational attainment showed a strong and consistent protective association with incident orientation impairment. When modelled as a continuous variable, each additional three years of education was associated with a 25.6% lower risk of developing orientation impairment over two years (RR 0.74, 95% CI 0.65–0.86). The main multivariable model estimates are shown in Fig. 2 . Table 2 Association of stroke history and educational attainment with incident orientation impairment (CHARLS 2018–2020) Variable RR 95% CI P value Stroke history (yes vs no) 0.888 0.630–1.251 0.498 Education (per 3-year increase) 0.752 0.653–0.867 < 0.001 Age (per 1-year increase) 0.998 0.992–1.005 0.619 Female (vs male) 0.907 0.789–1.041 0.164 Hukou: Non-agricultural (ref: Agricultural) 0.631 0.509–0.783 < 0.001 Hukou: Missing (ref: Agricultural) 0.847 0.529–1.356 0.490 Hukou: Unified residence (ref: Agricultural) 0.504 0.074–3.431 0.484 Hypertension (yes vs no) 0.897 0.705–1.142 0.377 Diabetes (yes vs no) 0.960 0.762–1.209 0.729 Myocardial infarction (yes vs no) 0.895 0.657–1.218 0.480 Notes: Relative risks (RRs) were estimated using modified Poisson regression with a log link and cluster-robust standard errors (clustered by community/household). The model adjusted for stroke history, education (per 3-year increase), age (linear), sex, hukou, hypertension, diabetes, and myocardial infarction. Missing indicator categories were included for categorical covariates in all regression models. Analytic sample: N = 5289, events = 744. Relative risks (RRs) and 95% confidence intervals (CIs) were estimated using modified Poisson regression with cluster-robust standard errors. Models were adjusted for age, sex, hukou status, hypertension, diabetes, and myocardial infarction. Relative risks are shown on a logarithmic scale. To further characterize the education-related gradient, adjusted relative risks across years of education, relative to 9 years of education, are shown in Fig. 3 . Adjusted relative risks of incident orientation impairment across years of education were estimated using a modified Poisson regression model with cluster-robust standard errors. Relative risks are presented relative to 9 years of education (reference). Shaded areas represent 95% confidence intervals. Education non-linearity check To evaluate whether the education–risk relationship was non-linear, we compared the linear education specification against a natural spline specification. The spline model yielded a marginally lower AIC than the linear model (AIC 4361.22 vs 4362.01), but the likelihood ratio test did not provide strong evidence of non-linearity (LRT p = 0.091). Therefore, the linear per-3-year education gradient was retained as the primary, interpretable specification, with spline results considered supportive (Fig. 4 ). Natural spline models were used to assess potential non-linearity in the association between years of education and incident orientation impairment, with 9 years as the reference. Shaded areas represent 95% confidence intervals. The spline analysis was conducted to evaluate the functional form of the association rather than to support causal interpretation at extreme education levels. Effect modification by educational attainment The multiplicative interaction between stroke history and educational attainment (per three-year increase in education) was not statistically significant (interaction term (stroke × education per three-year increase): RR 0.74, 95% CI 0.43–1.26). Although the direction of the interaction suggested a possible attenuation of stroke-associated risk at higher education levels, uncertainty around the estimates was substantial (Fig. 5 ). Predicted risks were derived from the fully adjusted modified Poisson model including a stroke-by-education interaction term. Solid lines represent point estimates and shaded areas indicate 95% confidence intervals. Estimates at extreme education levels should be interpreted cautiously due to limited sample size. To facilitate interpretation, marginal relative risks of stroke history were derived at selected levels of education from the interaction model (Table 3 ). The estimated relative risk of stroke was highest among participants with no formal education and decreased monotonically with increasing education. However, all confidence intervals crossed the null, indicating insufficient statistical evidence to support definitive effect modification. Table 3 Marginal relative risks of stroke history across education levels Education (years) Stroke RR 95% CI 0 1.252 0.656–2.388 6 0.679 0.362–1.271 9 0.500 0.163–1.530 12 0.368 0.071–1.898 Exploratory analyses assessing additive interaction using a binary education threshold (≥ 9 years vs. <9 years) yielded imprecise and unstable estimates, reflecting sparse data within some strata. Accordingly, results from additive interaction analyses were considered exploratory. Sensitivity analyses Inverse probability weighting was applied to address potential bias due to loss to follow-up. The handling of baseline covariate missingness (via missing indicator categories) remained unchanged in weighted models. After weighting and truncation of extreme weights, effect estimates were highly consistent with those from the primary analysis. Stroke history remained non-significantly associated with incident orientation impairment (RR 0.90, 95% CI 0.64–1.27), while educational attainment retained a strong protective association (per + 3 years: RR 0.73, 95% CI 0.64–0.85). When a stricter outcome definition was applied, the association between stroke history and incident orientation impairment remained non-significant, although the point estimate shifted above unity (RR 1.19, 95% CI 0.69–2.03). The protective association of education persisted but was attenuated and less precise, consistent with reduced statistical power due to the smaller number of events. Competing-risk analyses accounting for death as a competing event produced results broadly consistent with the primary findings. Higher educational attainment was associated with a markedly lower subdistribution hazard of incident orientation impairment, whereas stroke history showed no clear association. Complementary machine-learning analyses Across complementary machine-learning models evaluated using cross-validated out-of-fold procedures, discrimination for incident orientation impairment was modest. Elastic Net regression achieved the highest out-of-fold performance after calibration (AUC 0.71), followed by random forest (AUC 0.70) and ridge logistic regression (AUC 0.69). A stacking meta-learner did not materially improve discrimination. Overall, machine-learning models did not materially improve discrimination compared with traditional regression approaches (Fig. 6 ). Importantly, across models, educational attainment consistently ranked among the most influential predictors, whereas stroke history contributed limited incremental predictive information, supporting the primary regression-based findings. Boxplots display the distribution of cross-validated area under the receiver operating characteristic curve (AUC) for logistic regression (LR), Elastic Net, and random forest (RF) models. Each point represents the AUC from one cross-validation repeat with cluster-level grouping to prevent information leakage. Boxes indicate the interquartile range, horizontal lines denote medians, and whiskers represent observed ranges. External validation External validation was conducted using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS). In this independent cohort, discrimination was moderate, with a true out-of-fold AUC of 0.68. Across repeated resampling, discrimination remained stable, indicating reasonable rank-order transportability. Calibration assessment revealed systematic underestimation of absolute risk in the external cohort, as indicated by a positive calibration intercept. A calibration slope greater than 1 suggests that predicted risks were insufficiently dispersed relative to observed risks. These findings suggest that while relative risk patterns were broadly consistent across cohorts, recalibration would be required for accurate absolute risk estimation in different population settings (Fig. 7 ). Observed incidence is plotted against mean predicted risk across deciles of predicted probability. Points represent group-wise observed risks with corresponding 95% confidence intervals. The dashed diagonal line indicates perfect calibration. Model discrimination and calibration metrics are shown within the figure, including the true out-of-fold AUC, calibration intercept, and calibration slope. Discussion In this nationally representative longitudinal cohort of older adults free of orientation impairment at baseline, we examined the association between stroke history and incident orientation impairment over a two-year follow-up, with particular attention to the potential modifying role of educational attainment. Three main findings emerged. First, educational attainment showed a strong, consistent, and robust protective association with incident orientation impairment across all analytical approaches. In contrast, age was not independently associated with incident orientation impairment after multivariable adjustment. This may reflect restriction of the analytic sample to participants free of baseline orientation impairment, the relatively narrow age distribution, and the short follow-up interval. Second, stroke history was not independently associated with incident orientation impairment after multivariable adjustment. Third, although interaction estimates were directionally consistent with attenuation of stroke-associated risk at higher education levels, confidence intervals were wide and crossed the null, and formal tests did not provide statistical evidence for effect modification. Education as a strong and consistent protective factor Across regression-based analyses, sensitivity analyses, competing-risk models, and complementary machine-learning approaches, educational attainment emerged as the most stable determinant of incident orientation impairment. Each additional three years of education was associated with approximately a 25% lower risk of developing orientation impairment over two years, a finding that was remarkably consistent across model specifications and outcome definitions. These results align with the cognitive reserve framework, which posits that education enhances resilience to age-related and pathological brain changes through more efficient neural processing and compensatory mechanisms[ 22 , 23 ]. Importantly, our findings extend prior evidence beyond global cognitive outcomes to orientation impairment, a clinically interpretable and functionally relevant cognitive domain that often represents an early manifestation of cognitive vulnerability in older adults. The absence of strong evidence for non-linearity further supports the interpretation of education as a graded, cumulative protective factor rather than a threshold effect. Non-agricultural hukou status was also associated with a lower risk of incident orientation impairment. This finding may reflect differential access to healthcare resources, social capital, and lifelong cognitive stimulation between urban and rural populations in China. Stroke history and orientation impairment Contrary to expectations based on studies of global cognitive decline and dementia, stroke history was not independently associated with incident orientation impairment in fully adjusted models. The point estimate below unity should not be interpreted as protective; rather, it likely reflects residual confounding, selection of relatively resilient stroke survivors, and limited statistical power. The marked imbalance in vascular comorbidities between stroke and non-stroke groups (e.g., hypertension SMD > 0.8) further underscores the potential for residual confounding. Several factors may explain this finding[ 24 – 26 ]. Stroke history in this study was based on self-reported physician diagnosis and did not capture information on stroke subtype, severity, lesion location, timing, or recurrence. Such heterogeneity may attenuate associations with specific cognitive domains[ 27 – 30 ]. In addition, the analytic cohort was restricted to individuals who were free of orientation impairment at baseline, thereby excluding those with early or persistent post-stroke cognitive deficits and potentially selecting a relatively resilient subgroup of stroke survivors. Adjustment for age, educational attainment, and vascular comorbidities may have further accounted for much of the excess cognitive risk associated with stroke history[ 31 – 33 ]. Notably, when a stricter outcome definition was applied, the point estimate for stroke history shifted above unity, although confidence intervals remained wide, suggesting limited statistical power rather than clear evidence of an effect. Education as a potential modifier of stroke-related risk A central aim of this study was to assess whether educational attainment modifies the association between stroke history and incident orientation impairment. Although the direction of interaction estimates and marginal risk patterns were consistent with a buffering effect of higher education, formal tests for multiplicative and additive interaction did not yield statistically conclusive evidence. These findings should be interpreted cautiously. Detecting interaction effects typically requires substantially larger sample sizes than those needed to detect main effects, particularly when exposures and outcomes are relatively infrequent[ 34 – 36 ]. Moreover, the coarse measurement of stroke history and the short follow-up period may have limited our ability to detect subtle, domain-specific modification effects[ 37 ]. Rather than refuting the cognitive reserve hypothesis, our results suggest that any modifying effect of education on stroke-related orientation impairment is likely modest in magnitude or difficult to detect in population-based settings using available measures[ 38 ]. Insights from machine-learning and external validation analyses Complementary machine-learning analyses provided additional context for interpreting the primary findings. Despite the application of multiple algorithms and calibration procedures, predictive discrimination for incident orientation impairment remained modest, and no model materially outperformed traditional regression approaches. Across models, educational attainment consistently ranked among the most influential predictors, whereas stroke history contributed limited incremental predictive information. These results reinforce the conclusion that education plays a central role in shaping short-term cognitive vulnerability at the population level[ 39 – 41 ]. External validation in an independent cohort demonstrated moderate discrimination and broadly consistent risk patterns, supporting the transportability of relative associations. However, miscalibration of absolute risk estimates highlights the need for recalibration when applying risk models across populations with different characteristics and measurement instruments[ 42 ]. Together, these analyses support the robustness of the main findings while underscoring the challenges of predicting incident cognitive outcomes in older adults. Strengths and limitations Key strengths of this study include its longitudinal design, nationally representative sample, explicit complete-case definition with respect to the outcome, transparent handling of baseline covariate missingness through indicator categories, focus on incident orientation impairment, rigorous handling of loss to follow-up, and the integration of multiple analytical frameworks. The consistent protective association of education across diverse methods strengthens confidence in the central conclusions. Several limitations warrant consideration. Stroke history was measured crudely and lacked clinical detail, limiting etiological inference. The follow-up period was relatively short, potentially underestimating longer-term cognitive consequences of stroke. Orientation impairment represents only one dimension of cognitive function and may not capture subtler deficits in other domains. Finally, limited statistical power constrained definitive conclusions regarding effect modification. Missingness for certain vascular comorbidities, particularly hypertension, was substantial and differential by stroke status, which may have introduced residual confounding despite the use of missing indicator categories. Implications and future directions In this population-based longitudinal study of older adults, educational attainment emerged as a consistent and robust protective factor against incident orientation impairment, whereas stroke history showed no independent association after adjustment for sociodemographic and vascular factors. Evidence for education modifying the stroke–orientation relationship was suggestive but inconclusive. These findings highlight the central role of educational gradients in late-life cognitive vulnerability and underscore the complexity of identifying domain-specific cognitive consequences of stroke in community-dwelling populations. Conclusion In summary, higher education was strongly and consistently associated with a lower risk of incident orientation impairment over two years, whereas stroke history showed no independent association after adjustment. Evidence for education modifying the stroke–orientation relationship was suggestive but not conclusive. These findings highlight the central role of educational gradients in cognitive aging and underscore the complexity of disentangling stroke-related cognitive outcomes in population-based studies. Declarations Ethics approval and consent to participate The China Health and Retirement Longitudinal Study (CHARLS) obtained ethical approval from the Institutional Review Board of Peking University (IRB00001052-11015), and all participants provided written informed consent prior to participation. The Chinese Longitudinal Healthy Longevity Survey (CLHLS) was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-13074), and written informed consent was obtained from all participants or their legal representatives. The present study is a secondary analysis of publicly available, de-identified data from both CHARLS and CLHLS and therefore did not require additional ethical approval from the authors’ institutions. Both studies were conducted in accordance with the Declaration of Helsinki. Consent for publication Not applicable. This study used de-identified secondary data and does not contain any individual person’s identifiable information. Availability of data and materials The datasets analyzed during the current study are publicly available. CHARLS data can be obtained upon registration through the official CHARLS data portal. CLHLS data are available from the CLHLS project upon application and approval via the official CLHLS website. All analytic code used to generate the results is available from the corresponding author upon reasonable request. Funding This study was supported by the Letter of the Department of Personnel and Education of the National Administration of Traditional Chinese Medicine (2022) No. 256; the 9th Anhui Provincial Special Support Program for Talents (2023) No. 35; the Anhui Provincial Natural Science Foundation (Grant No. 2308085MH297); the Key Research Project of Natural Sciences in Anhui Universities (Grant No. 2023AH040099); the Special Project for Clinical Medical Research and Transformation of Anhui Province (Grant No. 202427b10020060); the Scientific Research Project of Anhui Provincial Department of Education (Grant No. 2025AHGXZK40506). Authors’ contributions Ce Shi and Fei Li conceived the research question and designed the study. Ce Shi led data curation, constructed analytic variables, developed and validated the statistical code, conducted the primary analyses and sensitivity analyses, generated the figures and tables, and drafted the initial manuscript. Qiqi Yang and Tianxin Jiang assisted with variable construction, data harmonization, and verification of analytic datasets; they also contributed to interpretation of findings and preparation of supplementary materials. Lihua Wu provided methodological guidance on model specification, interaction assessment, and robustness analyses, and critically revised the manuscript for important intellectual content. Fei Wang and Xiang Shang contributed to the clinical and public health interpretation of results and assisted with manuscript revision. Baoguo Wang supported data checking and interpretation, and provided input on presentation of results. Jianhong Gao, Weiran Li, and Ziyu Ye contributed to methodological input, literature contextualization, and critical revision of the manuscript. Fei Li served as the corresponding author, provided overall supervision, project administration, and oversight of manuscript development. All authors reviewed and approved the final manuscript and agree to be accountable for all aspects of the work. Competing interests The authors declare that they have no competing interests. Acknowledgements We thank the research teams and participants of both the CHARLS and CLHLS studies for providing open-access data that made this work possible. References Tsoy E, Kiekhofer RE, Guterman EL, Tee BL, Windon CC, Dorsman KA, Lanata SC, Rabinovici GD, Miller BL, Kind AJH, et al. Assessment of Racial/Ethnic Disparities in Timeliness and Comprehensiveness of Dementia Diagnosis in California. JAMA Neurol. 2021;78(6):657–65. Kohler IV, Kämpfen F, Bandawe C, Kohler HP. Cognition and Cognitive Changes in a Low-Income Sub-Saharan African Aging Population. J Alzheimer's disease: JAD. 2023;95(1):195–212. Fukuda H, Kanzaki H, Murata F, Maeda M, Ikeda M. Disease Burden and Progression in Patients with New-Onset Mild Cognitive Impairment and Alzheimer's Disease Identified from Japanese Claims Data: Evidence from the LIFE Study. J Alzheimer's disease: JAD. 2023;95(4):1559–72. Guerrero-Berroa E, Luo X, Schmeidler J, Rapp MA, Dahlman K, Grossman HT, Haroutunian V, Beeri MS. The MMSE orientation for time domain is a strong predictor of subsequent cognitive decline in the elderly. Int J Geriatr Psychiatry. 2009;24(12):1429–37. Tian R, Jiang Y, Zhang Y, Yan X, Zhou Y, Chen D. Cognitive training program improves cognitive ability and daily living ability in elderly patients with mild cognitive impairment. Aging Clin Exp Res. 2022;34(5):997–1005. Fauth EB, Schwartz S, Tschanz JT, Østbye T, Corcoran C, Norton MC. Baseline disability in activities of daily living predicts dementia risk even after controlling for baseline global cognitive ability and depressive symptoms. Int J Geriatr Psychiatry. 2013;28(6):597–606. Bangen KJ, Weigand AJ, Thomas KR, Delano-Wood L, Clark LR, Eppig J, Werhane ML, Edmonds EC, Bondi MW. Cognitive dispersion is a sensitive marker for early neurodegenerative changes and functional decline in nondemented older adults. Neuropsychology. 2019;33(5):599–608. Wilcock D, Jicha G, Blacker D, Albert MS, D'Orazio LM, Elahi FM, Fornage M, Hinman JD, Knoefel J, Kramer J, et al. MarkVCID cerebral small vessel consortium: I. Enrollment, clinical, fluid protocols. Alzheimer's Dement J Alzheimer's Assoc. 2021;17(4):704–15. Zhi N, Zhang L, Wang Y, Bai S, Geng J, Yu L, Cao W, Zhuang L, Zhou Y, Guan Y. Modified cerebral small vessel disease score is associated with vascular cognitive impairment after lacunar stroke. Aging. 2021;13(7):9510–21. Zhao J, Jin X, Chen B, Fu C, Ji S, Shen W, Wei J, Zheng H, Zhang Y. Apathy symptoms increase the risk of dementia conversion: a case-matching cohort study on patients with post-stroke mild cognitive impairment in China. Psychogeriatrics: official J Japanese Psychogeriatr Soc. 2021;21(2):149–57. Abdullah AH, Sharip S, Rahman AHA, Bakar L. Cognitive reserve in stroke patients. PsyCh J. 2021;10(3):444–52. Umarova RM, Schumacher LV, Schmidt CSM, Martin M, Egger K, Urbach H, Hennig J, Klöppel S, Kaller CP. Interaction between cognitive reserve and age moderates effect of lesion load on stroke outcome. Sci Rep. 2021;11(1):4478. Mungas D, Fletcher E, Gavett BE, Widaman K, Zahodne LB, Hohman TJ, Mayeda ER, Dowling NM, Johnson DK, Tomaszewski Farias S. Comparison of Education and Episodic Memory as Modifiers of Brain Atrophy Effects on Cognitive Decline: Implications for Measuring Cognitive Reserve. J Int Neuropsychological Society: JINS. 2021;27(5):401–11. Albrecht P, Perisse J, Sauleau EA, Blanc F. Cognitive reserve as one of the determinants of the progression of Alzheimer disease. Geriatrie et psychologie neuropsychiatrie du vieillissement 2022. Choi B, Choi J, Lee J, Choi E, Jang S-N. Socioeconomic disparities in hypertension and diabetes among older adults in South Korea: a trend analysis from a national survey, 2011–2023. Soc Sci Med. 2025;382:118350. Jensen HAR, Davidsen M, Brønnum-Hansen H, Eliasen MH, Christensen AI. Trends in educational inequality in healthy life expectancy in Denmark between 2010 and 2021: a population-based study. BMJ Open. 2025;15(7):e100851. Niu J, Li X, Chen Q, Yang W, Suo L, Chen Z. Association Between Educational Inequality and Income Inequality With Metabolic Diseases and Cause-Specific Mortality. Clin Cardiol. 2025;48(7):e70173. Langa KM, Ryan LH, McCammon RJ, Jones RN, Manly JJ, Levine DA, Sonnega A, Farron M, Weir DR. The Health and Retirement Study Harmonized Cognitive Assessment Protocol Project: Study Design and Methods. Neuroepidemiology. 2020;54(1):64–74. Zhang M, Lv X, Chen Y, Tu L, Fan Z, Yao Y, Yu X, Guan N, Wang H. Excessive sleep increased the risk of incidence of cognitive impairment among older Chinese adults: a cohort study based on the Chinese Longitudinal Healthy Longevity Survey (CLHLS). Int Psychogeriatr. 2022;34(8):725–34. Nieboer D, van der Ploeg T, Steyerberg EW. Assessing Discriminative Performance at External Validation of Clinical Prediction Models. PLoS ONE. 2016;11(2):e0148820. Jonsson H, Nyström L, Blom J. Poisson regression with adjustment for contamination and non-compliance in cohort studies conducted to estimate intervention effectiveness. J Med Screen 2025:9691413251388380. Beker N, Ganz A, Hulsman M, Klausch T, Schmand BA, Scheltens P, Sikkes SAM, Holstege H. Association of Cognitive Function Trajectories in Centenarians With Postmortem Neuropathology, Physical Health, and Other Risk Factors for Cognitive Decline. JAMA Netw open. 2021;4(1):e2031654. Hu D, Liu C, Xia K, Abramowitz A, Wu G. Characterizing the Resilience Effect of Neurodegeneration for the Mechanistic Pathway of Alzheimer's Disease. J Alzheimer's disease: JAD. 2021;84(3):1351–62. Delgado J, Masoli J, Hase Y, Akinyemi R, Ballard C, Kalaria RN, Allan LM. Trajectories of cognitive change following stroke: stepwise decline towards dementia in the elderly. Brain Commun. 2022;4(3):fcac129. Lo JW, Crawford JD, Lipnicki DM, Lipton RB, Katz MJ, Preux PM, Guerchet M, d'Orsi E, Quialheiro A, Rech CR, et al. Trajectory of Cognitive Decline Before and After Stroke in 14 Population Cohorts. JAMA Netw open. 2024;7(10):e2437133. Chen LY, Norby FL, Gottesman RF, Mosley TH, Soliman EZ, Agarwal SK, Loehr LR, Folsom AR, Coresh J, Alonso A. Association of Atrial Fibrillation With Cognitive Decline and Dementia Over 20 Years: The ARIC-NCS (Atherosclerosis Risk in Communities Neurocognitive Study). J Am Heart Association 2018, 7(6). Jamrozik E, Hyde Z, Alfonso H, Flicker L, Almeida O, Yeap B, Norman P, Hankey G, Jamrozik K. Validity of self-reported versus hospital-coded diagnosis of stroke: a cross-sectional and longitudinal study. Cerebrovasc Dis. 2014;37(4):256–62. Yan PJ, Hou LS, Li ME, Lu ZX, Zhan FY, Ran MD, Li JJ, Zhang L, Yang R, Zhou MK, et al. Associations between Lesion Locations and Stroke Recurrence in Survivors of First-ever Ischemic Stroke: A Prospective Cohort Study. Curr Med Sci. 2020;40(4):708–18. Nakling AE, Aarsland D, Næss H, Wollschlaeger D, Fladby T, Hofstad H, Wehling E. Cognitive Deficits in Chronic Stroke Patients: Neuropsychological Assessment, Depression, and Self-Reports. Dement geriatric Cogn disorders extra. 2017;7(2):283–96. Joundi RA, Fang J, Austin PC, Smith EE, Yu AYX, Hachinski V, Sposato LA, Ganesh A, Sharma M, Kapral MK. Magnitude and Time-Course of Dementia Risk in Stroke Survivors: A Population-Wide Matched Cohort Study. Neurology. 2025;104(1):e210131. Weaver NA, Kuijf HJ, Aben HP, Abrigo J, Bae HJ, Barbay M, Best JG, Bordet R, Chappell FM, Chen C, et al. Strategic infarct locations for post-stroke cognitive impairment: a pooled analysis of individual patient data from 12 acute ischaemic stroke cohorts. Lancet Neurol. 2021;20(6):448–59. Ohlmeier L, Nannoni S, Pallucca C, Brown RB, Loubiere L, Markus HS. Prevalence of, and risk factors for, cognitive impairment in lacunar stroke. Int J stroke: official J Int Stroke Soc. 2023;18(1):62–9. Arce Rentería M, Manly JJ, Vonk JMJ, Mejia Arango S, Michaels Obregon A, Samper-Ternent R, Wong R, Barral S, Tosto G. Midlife Vascular Factors and Prevalence of Mild Cognitive Impairment in Late-Life in Mexico. J Int Neuropsychological Society: JINS. 2022;28(4):351–61. Leon AC, Heo M. Sample Sizes Required to Detect Interactions between Two Binary Fixed-Effects in a Mixed-Effects Linear Regression Model. Comput Stat Data Anal. 2009;53(3):603–8. Heo M, Leon AC. Sample sizes required to detect two-way and three-way interactions involving slope differences in mixed-effects linear models. J Biopharm Stat. 2010;20(4):787–802. Yang C, Berkalieva A, Mazumdar M, Kwon D. Power calculation for detecting interaction effect in cross-sectional stepped-wedge cluster randomized trials: an important tool for disparity research. BMC Med Res Methodol. 2024;24(1):57. Oh H, Park J, Seo W. A 2-year prospective follow-up study of temporal changes associated with post-stroke cognitive impairment. Int J Nurs Pract. 2018;24(2):e12618. Kessels RP, Eikelboom WS, Schaapsmeerders P, Maaijwee NA, Arntz RM, van Dijk EJ, de Leeuw FE. Effect of Formal Education on Vascular Cognitive Impairment after Stroke: A Meta-analysis and Study in Young-Stroke Patients. J Int Neuropsychological Society: JINS. 2017;23(3):223–38. Roldán-Tapia MD, Cánovas R, León I, García-Garcia J. Cognitive Vulnerability in Aging May Be Modulated by Education and Reserve in Healthy People. Front Aging Neurosci. 2017;9:340. Mantri S, Nwadiogbu C, Fitts W, Dahodwala N. Quality of education impacts late-life cognition. Int J Geriatr Psychiatry. 2019;34(6):855–62. Ho S, Kozhevnikov M. Cognitive style and creativity: The role of education in shaping cognitive style profiles and creativity of adolescents. Br J Educ Psychol. 2023;93(4):978–96. Pfeiffer RM, Petracci E. Variance computations for functional of absolute risk estimates. Stat Probab Lett. 2011;81(7):807–12. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 07 Apr, 2026 Editor invited by journal 16 Mar, 2026 Editor assigned by journal 13 Mar, 2026 Submission checks completed at journal 13 Mar, 2026 First submitted to journal 11 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9092847","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":620352970,"identity":"b654aa8d-57cb-4f76-8556-a59aa7ffffd8","order_by":0,"name":"Ce Shi","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ce","middleName":"","lastName":"Shi","suffix":""},{"id":620352972,"identity":"fdb0bb52-e0ae-4513-8456-85fa829215f9","order_by":1,"name":"Lihua Wu","email":"","orcid":"","institution":"Henan University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Lihua","middleName":"","lastName":"Wu","suffix":""},{"id":620352974,"identity":"f8f69833-e84e-48fa-8c72-80c14cd799bb","order_by":2,"name":"Qiqi Yang","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Qiqi","middleName":"","lastName":"Yang","suffix":""},{"id":620352975,"identity":"4516f711-8893-4e21-bd8e-fa4d471e3408","order_by":3,"name":"Fei Wang","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Wang","suffix":""},{"id":620352976,"identity":"337ef528-0945-4a85-b40e-d9c89ffaf0cf","order_by":4,"name":"Xiang Shang","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xiang","middleName":"","lastName":"Shang","suffix":""},{"id":620352980,"identity":"ea1ffa03-9a86-4629-aef6-05ac12b50857","order_by":5,"name":"Tianxin Jiang","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Tianxin","middleName":"","lastName":"Jiang","suffix":""},{"id":620352981,"identity":"0c81f167-e1a7-4c1a-a12d-ab7baa359074","order_by":6,"name":"Baoguo Wang","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Baoguo","middleName":"","lastName":"Wang","suffix":""},{"id":620352983,"identity":"2b6bd1e2-8566-4877-8015-28713c35a905","order_by":7,"name":"Jianhong Gao","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jianhong","middleName":"","lastName":"Gao","suffix":""},{"id":620352984,"identity":"704033c4-9d8b-47c5-8e76-c91921211d04","order_by":8,"name":"Weiran Li","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Weiran","middleName":"","lastName":"Li","suffix":""},{"id":620352985,"identity":"d1766449-8714-4c22-a488-bf08fd41cdfd","order_by":9,"name":"Ziyu Ye","email":"","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ziyu","middleName":"","lastName":"Ye","suffix":""},{"id":620352986,"identity":"fea04a1c-d806-41af-ad9e-7ddb159416bf","order_by":10,"name":"Fei Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtElEQVRIiWNgGAWjYBACPigtx8befoA4LWxQ2piP50wCaVoS50k4GBCphb358YsfFXfS2yQYEhh+VGwjQgvPMTPLnjPPctukGw8w9py5TYQWiQQzY8a2w7ltMgcSmBnbiNEi//ybMeO/w+lAvQZEapHgMX7M2HA4gQQtPDlljD3HDhu2AQP5IFF+4Wc/vvnDj5rD8vLt7Qcf/KggQgvYbTDWAaLUAwHzB2JVjoJRMApGwQgFAJREOaFzC61XAAAAAElFTkSuQmCC","orcid":"","institution":"Anhui University of Traditional Chinese Medicine","correspondingAuthor":true,"prefix":"","firstName":"Fei","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2026-03-11 10:08:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9092847/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9092847/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106875005,"identity":"23152253-f113-46e1-b219-df8b47f5361f","added_by":"auto","created_at":"2026-04-14 10:21:59","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":89905,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFlowchart of cohort derivation and analytic sample selection (CHARLS 2018–2020).\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture1.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/00c6303c9a6dd2a170847748.png"},{"id":106875006,"identity":"ca06bc97-2cfe-4bbc-ab4c-af30835853a8","added_by":"auto","created_at":"2026-04-14 10:21:59","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":129045,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAdjusted relative risks for stroke history, educational attainment, and selected covariates in relation to incident orientation impairment.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture2.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/6523df5d6540c30c1d72159d.png"},{"id":106961049,"identity":"b16c567a-fcfe-49e7-84b9-bf7e8092e188","added_by":"auto","created_at":"2026-04-15 09:24:04","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":74780,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEducation gradient and incident orientation impairment.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture3.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/5c3b99392c2e8cdf4eb91cc7.png"},{"id":106960398,"identity":"22212ae8-205e-4638-9c04-4b4d7965e50f","added_by":"auto","created_at":"2026-04-15 09:20:50","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":657779,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNatural spline analysis of education and incident orientation impairment.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture4.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/551e8bc83f3b91cf48695a3e.png"},{"id":106875010,"identity":"dee1c1f2-97e5-4f6c-8e51-8b909dad2ce8","added_by":"auto","created_at":"2026-04-14 10:21:59","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":79394,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredicted risk of incident orientation impairment by stroke status and educational attainment.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture5.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/70ec2649bb9e834238f9cf47.png"},{"id":106875008,"identity":"7e69aeaa-87de-4c65-af6b-301702a56bea","added_by":"auto","created_at":"2026-04-14 10:21:59","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":62824,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCross-validated discrimination of machine-learning models.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture6.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/c0b3d1154cb0aba9438e074f.png"},{"id":106875009,"identity":"8db77221-0f96-45d2-a277-e178faca67d2","added_by":"auto","created_at":"2026-04-14 10:21:59","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":91022,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExternal validation calibration plot (CLHLS cohort).\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture7.png","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/cd463904fb96f030662f00db.png"},{"id":106964768,"identity":"d27c558d-0e31-46bf-a52d-53552eb8652f","added_by":"auto","created_at":"2026-04-15 09:51:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2292391,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9092847/v1/269c4193-a706-450f-bfcf-9beb8225521f.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Association of educational attainment with incident orientation impairment among Chinese older adults with and without stroke: evidence from a longitudinal study with external validation","fulltext":[{"header":"Background","content":"\u003cp\u003ePopulation ageing has substantially increased the burden of cognitive impairment and related functional decline worldwide, posing major challenges to health and social care systems for older adults[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Among different cognitive domains, orientation represents a core component of cognitive functioning and is essential for independent living and everyday decision-making in older adults[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Impairment in orientation has been associated with an increased risk of subsequent dementia, functional dependence, and mortality, making it a clinically relevant and epidemiologically meaningful early marker of cognitive vulnerability in later life[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eStroke is a leading cause of long-term disability and cognitive impairment in older adults[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Beyond its acute neurological consequences, stroke may initiate or accelerate vascular and neurodegenerative processes that adversely affect cognitive functioning over time[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Previous studies have consistently demonstrated an increased risk of global cognitive decline and dementia among stroke survivors[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. However, evidence regarding domain-specific cognitive outcomes\u0026mdash;particularly orientation impairment as an early and clinically interpretable marker\u0026mdash;remains limited in population-based longitudinal studies.\u003c/p\u003e \u003cp\u003eImportantly, the cognitive consequences of stroke are unlikely to be uniform across individuals and may vary according to markers of cognitive reserve[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Educational attainment, a core indicator of cognitive reserve and a key social determinant of health, has been widely recognised as an important determinant of cognitive ageing trajectories[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Higher educational attainment may buffer the cognitive consequences of brain pathology through enhanced neural efficiency and compensatory mechanisms[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Nevertheless, whether educational attainment modifies the association between stroke history and subsequent orientation impairment remains uncertain, particularly in longitudinal, population-based settings.\u003c/p\u003e \u003cp\u003eMost existing studies have treated education primarily as a confounder or have focused on global cognitive outcomes, with limited attention to potential effect modification across specific cognitive domains. From a public health perspective, understanding heterogeneity in stroke-related cognitive risk is essential for identifying vulnerable subgroups and developing targeted prevention strategies. This question is particularly relevant in rapidly ageing populations such as China, where substantial educational disparities persist among older adults as a result of historical and socioeconomic factors[\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eUsing data from a nationally representative longitudinal cohort of Chinese older adults, this study aimed to examine the association between stroke history and incident orientation impairment and to explore whether this association differs according to educational attainment. We further evaluated the robustness of our findings using multiple analytical approaches, including inverse probability weighting, competing risk models accounting for mortality, and complementary machine learning analyses.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design and data source\u003c/h2\u003e \u003cp\u003eThis study was a population-based longitudinal analysis using data from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative survey of middle-aged and older adults in China. CHARLS employs a multistage, stratified probability sampling design covering both urban and rural areas and collects comprehensive information on sociodemographic characteristics, health conditions, and cognitive functioning among community-dwelling adults.\u003c/p\u003e \u003cp\u003eBaseline data were obtained from the 2018 wave, with follow-up assessments conducted in 2020. The CHARLS study protocol was approved by the Institutional Review Board of Peking University, and all participants provided written informed consent. The present analysis was based on publicly available, de-identified data.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy population\u003c/h3\u003e\n\u003cp\u003eThe analytic sample was restricted to participants aged 60 years or older at baseline (2018). Among the 5,289 participants included in the analytic sample, the mean age was 61.44 years (SD 10.02). To ensure a clear temporal ordering between exposure and outcome, analyses were further restricted to individuals who were free of orientation impairment at baseline.\u003c/p\u003e \u003cp\u003eParticipants were included if they had available baseline information on stroke history and educational attainment and completed the orientation assessment at the 2020 follow-up. Individuals without observed orientation assessments at follow-up were excluded from the primary analysis (complete-case with respect to the outcome). Baseline covariates were retained as observed, with missing categories incorporated for categorical variables in regression models. Potential bias arising from loss to follow-up was examined using inverse probability weighting in sensitivity analyses.\u003c/p\u003e\n\u003ch3\u003eExposure assessment: stroke history\u003c/h3\u003e\n\u003cp\u003eBaseline stroke history was assessed using self-reported physician diagnosis. Participants were asked whether they had ever been diagnosed with stroke by a physician, and responses were coded as a binary variable (yes or no). Self-reported physician-diagnosed stroke has been widely used in population-based studies and has demonstrated acceptable validity in the CHARLS cohort.\u003c/p\u003e\n\u003ch3\u003eOutcome assessment: orientation impairment\u003c/h3\u003e\n\u003cp\u003eCognitive performance in CHARLS is evaluated using a standardized methodological framework consistent with that adopted in the U.S. Health and Retirement Study (HRS)[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. As described by Crimmins et al., this framework conceptualizes cognition across four domains: orientation, calculation, memory, and visuospatial ability. These domains are assessed using structured cognitive tests, yielding a total cognitive score ranging from 0 to 31 points.\u003c/p\u003e \u003cp\u003eOrientation ability is measured using items derived from the Cognitive Status Telephone Interview, including questions on date and season, with a maximum score of 5 points. Calculation ability is assessed through a serial subtraction task (subtracting 7 from 100 consecutively), with a maximum score of 5 points. Memory is evaluated using immediate and delayed word recall tasks, with a maximum score of 20 points. Visuospatial ability is assessed through a figure-copying task, contributing 1 point to the total score.\u003c/p\u003e \u003cp\u003eThe present study focuses specifically on the orientation domain as the primary outcome. The decision to concentrate on orientation rather than global cognitive score was based on several considerations. First, orientation reflects a core component of early cognitive decline and demonstrates structural stability across survey waves. Second, orientation items exhibit strong measurement consistency over time, facilitating longitudinal comparisons. Third, impairment in orientation has well-established neurobiological relevance in post-stroke cognitive dysfunction.\u003c/p\u003e \u003cp\u003eIncident orientation impairment was defined as the occurrence of impaired orientation at the 2020 follow-up among participants who were free of orientation impairment at baseline in 2018. By restricting the analysis to incident cases, this study minimizes potential reverse causation and strengthens temporal inference between baseline stroke history and subsequent cognitive decline.\u003c/p\u003e\n\u003ch3\u003eEducational attainment\u003c/h3\u003e\n\u003cp\u003eEducational attainment was measured at baseline. In the primary analysis, education was modeled as years of schooling and scaled per three-year increase to facilitate interpretability. Categorical specifications reflecting the Chinese education system were examined in supplementary analyses: no formal education, primary education, middle school education, and high school or higher education. Educational attainment was conceptualised as a stable indicator of early-life socioeconomic position and cognitive reserve and was prespecified as a potential effect modifier of the association between stroke history and incident orientation impairment.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eCovariates\u003c/h2\u003e \u003cp\u003eBaseline covariates were selected a priori based on previous literature and conceptual considerations. These included demographic characteristics (age and sex), socioeconomic factors (marital status and residential area), and major baseline health-related variables. All covariates were measured at baseline to minimise temporal ambiguity and reduce the risk of overadjustment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eThe primary objective was to examine the association between baseline stroke history and incident orientation impairment and to explore whether this association differed according to educational attainment. Modified Poisson regression with cluster-robust standard errors was used to estimate relative risks (RRs) and 95% confidence intervals (CIs). Participants with missing orientation assessment at follow-up were excluded from the primary analysis (complete-case with respect to the outcome only). Baseline covariates were retained as observed. For categorical variables, missing values were included as separate indicator categories in regression models rather than excluding participants with incomplete covariate data. This approach is well suited to cohort studies with binary outcomes and provides directly interpretable risk estimates.\u003c/p\u003e \u003cp\u003eEducational attainment was first included as an adjustment variable in multivariable models. Potential effect modification by educational attainment was evaluated by introducing an interaction term between stroke history and education. Statistical evidence for interaction was assessed using Wald tests. To facilitate interpretation, marginal and education-specific associations between stroke history and incident orientation impairment were derived from interaction models.\u003c/p\u003e \u003cp\u003eSeveral sensitivity analyses were conducted to assess the robustness of the findings. Inverse probability weighting was applied to address potential selection bias due to loss to follow-up. Competing-risk analyses were performed to account for death during follow-up as a competing event for incident orientation impairment. Additional analyses included alternative definitions of orientation impairment and complementary machine-learning approaches as robustness checks rather than as primary inferential tools.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eComplementary machine-learning analyses\u003c/h3\u003e\n\u003cp\u003eTo further assess the robustness of the observed associations across analytical frameworks, complementary machine-learning analyses were conducted. Multiple supervised learning algorithms were trained using baseline variables to predict incident orientation impairment over the follow-up period. Model performance was evaluated using cross-validated out-of-fold procedures, with discrimination and calibration assessed as supportive metrics.\u003c/p\u003e \u003cp\u003eThese analyses were not intended to develop or optimise a predictive model for clinical use. Rather, they served as robustness checks to examine whether the relative importance and direction of associations for stroke history and educational attainment were consistent with findings from the primary regression-based analyses.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eExternal validation\u003c/h2\u003e \u003cp\u003eExternal validation was performed using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), an independent population-based cohort of older adults in China[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Inclusion criteria and outcome definitions were aligned as closely as possible with those used in the primary analysis, with an emphasis on conceptual rather than exact measurement equivalence.\u003c/p\u003e \u003cp\u003eThe purpose of external validation was to assess the transportability of the observed associations and overall risk patterns, rather than to formally validate a prediction model[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Modified Poisson regression models were re-estimated in the CLHLS cohort to assess consistency of relative associations. Cross-validated discrimination and calibration metrics were examined descriptively to evaluate consistency with findings from CHARLS[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eStatistical software\u003c/h2\u003e \u003cp\u003eAll statistical analyses were conducted using R software (version 4.5.2). Statistical significance was defined as a two-sided p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eStudy population and event rates\u003c/h2\u003e \u003cp\u003eAmong CHARLS participants who were free of orientation impairment at baseline in 2018, 9,080 individuals constituted the risk set. Of these, 5,289 participants had observed orientation assessments at the 2020 follow-up and were included in the primary complete-case analysis. The mean age of the analytic sample was 61.44 years (SD 10.02). Over the two-year follow-up, 744 participants developed incident orientation impairment under the primary definition (orientation score\u0026thinsp;\u0026le;\u0026thinsp;3). Using a stricter definition (orientation score\u0026thinsp;\u0026le;\u0026thinsp;2), 231 incident cases were identified. The cohort derivation and analytic sample selection are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eParticipants were drawn from the China Health and Retirement Longitudinal Study (CHARLS). From the original master dataset (N\u0026thinsp;=\u0026thinsp;19,395), individuals aged\u0026thinsp;\u0026lt;\u0026thinsp;60 years or without valid baseline orientation assessment were excluded. The final risk set included 9,080 participants free of orientation impairment at baseline (2018). Among them, 5,289 participants with non-missing orientation assessment at follow-up (2020) were included in the primary complete-case analysis, yielding 744 incident cases of orientation impairment. Primary analyses used modified Poisson regression with cluster-robust standard errors; additional spline, interaction, inverse probability weighting, and machine-learning analyses were conducted as sensitivity and supplementary analyses.\u003c/p\u003e \u003cp\u003eBaseline characteristics stratified by stroke history among participants included in the analytic sample (complete-case with respect to follow-up outcome only; N\u0026thinsp;=\u0026thinsp;5,289) are shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eParticipants with a history of stroke had a substantially higher prevalence of vascular comorbidities, with very large standardized mean differences observed for hypertension and diabetes, indicating pronounced baseline imbalance between groups, whereas age and educational attainment were broadly comparable between groups.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics by stroke history\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNo stroke\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStroke history\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSMD\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e265\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e61.44 (10.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e61.53 (9.33)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.009\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge missing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation, years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.79 (1.94)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.97 (1.88)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.090\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation missing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2374 (47.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e124 (46.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.013\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2650 (52.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e141 (53.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHukou\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 Agricultural Hukou\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3560 (70.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e187 (70.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.093\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 Non-agricultural Hukou\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1323 (26.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e73 (27.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3 Unified Residence Hukou\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15 (0.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e126 (2.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3214 (64.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e81 (30.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.880\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e518 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e48 (18.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1292 (25.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e136 (51.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3850 (76.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e153 (57.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.541\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e514 (10.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e48 (18.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e660 (13.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e64 (24.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMyocardial infarction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4367 (86.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e202 (76.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.383\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e262 (5.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e29 (10.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e395 (7.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e34 (12.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNotes: Values are mean (SD) for continuous variables and n (%) for categorical variables. SMD\u0026thinsp;=\u0026thinsp;standardized mean difference.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eAssociation between stroke history, education, and incident orientation impairment\u003c/h2\u003e \u003cp\u003eIn multivariable modified Poisson regression models with cluster-robust standard errors, baseline stroke history was not independently associated with incident orientation impairment after adjustment for age, sex, educational attainment, hukou status, and vascular comorbidities (RR 0.88, 95% CI 0.63\u0026ndash;1.24; Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn contrast, educational attainment showed a strong and consistent protective association with incident orientation impairment. When modelled as a continuous variable, each additional three years of education was associated with a 25.6% lower risk of developing orientation impairment over two years (RR 0.74, 95% CI 0.65\u0026ndash;0.86). The main multivariable model estimates are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAssociation of stroke history and educational attainment with incident orientation impairment (CHARLS 2018\u0026ndash;2020)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStroke history (yes vs no)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.888\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.630\u0026ndash;1.251\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.498\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation (per 3-year increase)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.752\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.653\u0026ndash;0.867\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (per 1-year increase)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.998\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.992\u0026ndash;1.005\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.619\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale (vs male)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.789\u0026ndash;1.041\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.164\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHukou: Non-agricultural (ref: Agricultural)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.631\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.509\u0026ndash;0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHukou: Missing (ref: Agricultural)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.847\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.529\u0026ndash;1.356\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.490\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHukou: Unified residence (ref: Agricultural)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.504\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.074\u0026ndash;3.431\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.484\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension (yes vs no)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.897\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.705\u0026ndash;1.142\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.377\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes (yes vs no)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.960\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.762\u0026ndash;1.209\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.729\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMyocardial infarction (yes vs no)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.895\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.657\u0026ndash;1.218\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.480\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNotes: Relative risks (RRs) were estimated using modified Poisson regression with a log link and cluster-robust standard errors (clustered by community/household). The model adjusted for stroke history, education (per 3-year increase), age (linear), sex, hukou, hypertension, diabetes, and myocardial infarction. Missing indicator categories were included for categorical covariates in all regression models. Analytic sample: N\u0026thinsp;=\u0026thinsp;5289, events\u0026thinsp;=\u0026thinsp;744.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRelative risks (RRs) and 95% confidence intervals (CIs) were estimated using modified Poisson regression with cluster-robust standard errors. Models were adjusted for age, sex, hukou status, hypertension, diabetes, and myocardial infarction. Relative risks are shown on a logarithmic scale.\u003c/p\u003e \u003cp\u003eTo further characterize the education-related gradient, adjusted relative risks across years of education, relative to 9 years of education, are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAdjusted relative risks of incident orientation impairment across years of education were estimated using a modified Poisson regression model with cluster-robust standard errors. Relative risks are presented relative to 9 years of education (reference). Shaded areas represent 95% confidence intervals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eEducation non-linearity check\u003c/h2\u003e \u003cp\u003eTo evaluate whether the education\u0026ndash;risk relationship was non-linear, we compared the linear education specification against a natural spline specification. The spline model yielded a marginally lower AIC than the linear model (AIC 4361.22 vs 4362.01), but the likelihood ratio test did not provide strong evidence of non-linearity (LRT p\u0026thinsp;=\u0026thinsp;0.091). Therefore, the linear per-3-year education gradient was retained as the primary, interpretable specification, with spline results considered supportive (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eNatural spline models were used to assess potential non-linearity in the association between years of education and incident orientation impairment, with 9 years as the reference. Shaded areas represent 95% confidence intervals. The spline analysis was conducted to evaluate the functional form of the association rather than to support causal interpretation at extreme education levels.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eEffect modification by educational attainment\u003c/h2\u003e \u003cp\u003eThe multiplicative interaction between stroke history and educational attainment (per three-year increase in education) was not statistically significant (interaction term (stroke \u0026times; education per three-year increase): RR 0.74, 95% CI 0.43\u0026ndash;1.26). Although the direction of the interaction suggested a possible attenuation of stroke-associated risk at higher education levels, uncertainty around the estimates was substantial (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003ePredicted risks were derived from the fully adjusted modified Poisson model including a stroke-by-education interaction term. Solid lines represent point estimates and shaded areas indicate 95% confidence intervals. Estimates at extreme education levels should be interpreted cautiously due to limited sample size.\u003c/p\u003e \u003cp\u003eTo facilitate interpretation, marginal relative risks of stroke history were derived at selected levels of education from the interaction model (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The estimated relative risk of stroke was highest among participants with no formal education and decreased monotonically with increasing education. However, all confidence intervals crossed the null, indicating insufficient statistical evidence to support definitive effect modification.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMarginal relative risks of stroke history across education levels\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation (years)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStroke RR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.252\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.656\u0026ndash;2.388\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.679\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.362\u0026ndash;1.271\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.163\u0026ndash;1.530\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.368\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.071\u0026ndash;1.898\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eExploratory analyses assessing additive interaction using a binary education threshold (\u0026ge;\u0026thinsp;9 years vs. \u0026lt;9 years) yielded imprecise and unstable estimates, reflecting sparse data within some strata. Accordingly, results from additive interaction analyses were considered exploratory.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity analyses\u003c/h2\u003e \u003cp\u003eInverse probability weighting was applied to address potential bias due to loss to follow-up. The handling of baseline covariate missingness (via missing indicator categories) remained unchanged in weighted models. After weighting and truncation of extreme weights, effect estimates were highly consistent with those from the primary analysis. Stroke history remained non-significantly associated with incident orientation impairment (RR 0.90, 95% CI 0.64\u0026ndash;1.27), while educational attainment retained a strong protective association (per +\u0026thinsp;3 years: RR 0.73, 95% CI 0.64\u0026ndash;0.85).\u003c/p\u003e \u003cp\u003eWhen a stricter outcome definition was applied, the association between stroke history and incident orientation impairment remained non-significant, although the point estimate shifted above unity (RR 1.19, 95% CI 0.69\u0026ndash;2.03). The protective association of education persisted but was attenuated and less precise, consistent with reduced statistical power due to the smaller number of events.\u003c/p\u003e \u003cp\u003eCompeting-risk analyses accounting for death as a competing event produced results broadly consistent with the primary findings. Higher educational attainment was associated with a markedly lower subdistribution hazard of incident orientation impairment, whereas stroke history showed no clear association.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eComplementary machine-learning analyses\u003c/h2\u003e \u003cp\u003eAcross complementary machine-learning models evaluated using cross-validated out-of-fold procedures, discrimination for incident orientation impairment was modest. Elastic Net regression achieved the highest out-of-fold performance after calibration (AUC 0.71), followed by random forest (AUC 0.70) and ridge logistic regression (AUC 0.69). A stacking meta-learner did not materially improve discrimination.\u003c/p\u003e \u003cp\u003eOverall, machine-learning models did not materially improve discrimination compared with traditional regression approaches (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Importantly, across models, educational attainment consistently ranked among the most influential predictors, whereas stroke history contributed limited incremental predictive information, supporting the primary regression-based findings.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBoxplots display the distribution of cross-validated area under the receiver operating characteristic curve (AUC) for logistic regression (LR), Elastic Net, and random forest (RF) models. Each point represents the AUC from one cross-validation repeat with cluster-level grouping to prevent information leakage. Boxes indicate the interquartile range, horizontal lines denote medians, and whiskers represent observed ranges.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eExternal validation\u003c/h2\u003e \u003cp\u003eExternal validation was conducted using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS). In this independent cohort, discrimination was moderate, with a true out-of-fold AUC of 0.68. Across repeated resampling, discrimination remained stable, indicating reasonable rank-order transportability.\u003c/p\u003e \u003cp\u003eCalibration assessment revealed systematic underestimation of absolute risk in the external cohort, as indicated by a positive calibration intercept. A calibration slope greater than 1 suggests that predicted risks were insufficiently dispersed relative to observed risks. These findings suggest that while relative risk patterns were broadly consistent across cohorts, recalibration would be required for accurate absolute risk estimation in different population settings (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eObserved incidence is plotted against mean predicted risk across deciles of predicted probability. Points represent group-wise observed risks with corresponding 95% confidence intervals. The dashed diagonal line indicates perfect calibration. Model discrimination and calibration metrics are shown within the figure, including the true out-of-fold AUC, calibration intercept, and calibration slope.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this nationally representative longitudinal cohort of older adults free of orientation impairment at baseline, we examined the association between stroke history and incident orientation impairment over a two-year follow-up, with particular attention to the potential modifying role of educational attainment. Three main findings emerged. First, educational attainment showed a strong, consistent, and robust protective association with incident orientation impairment across all analytical approaches. In contrast, age was not independently associated with incident orientation impairment after multivariable adjustment. This may reflect restriction of the analytic sample to participants free of baseline orientation impairment, the relatively narrow age distribution, and the short follow-up interval. Second, stroke history was not independently associated with incident orientation impairment after multivariable adjustment. Third, although interaction estimates were directionally consistent with attenuation of stroke-associated risk at higher education levels, confidence intervals were wide and crossed the null, and formal tests did not provide statistical evidence for effect modification.\u003c/p\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eEducation as a strong and consistent protective factor\u003c/h2\u003e \u003cp\u003eAcross regression-based analyses, sensitivity analyses, competing-risk models, and complementary machine-learning approaches, educational attainment emerged as the most stable determinant of incident orientation impairment. Each additional three years of education was associated with approximately a 25% lower risk of developing orientation impairment over two years, a finding that was remarkably consistent across model specifications and outcome definitions.\u003c/p\u003e \u003cp\u003eThese results align with the cognitive reserve framework, which posits that education enhances resilience to age-related and pathological brain changes through more efficient neural processing and compensatory mechanisms[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Importantly, our findings extend prior evidence beyond global cognitive outcomes to orientation impairment, a clinically interpretable and functionally relevant cognitive domain that often represents an early manifestation of cognitive vulnerability in older adults. The absence of strong evidence for non-linearity further supports the interpretation of education as a graded, cumulative protective factor rather than a threshold effect.\u003c/p\u003e \u003cp\u003eNon-agricultural hukou status was also associated with a lower risk of incident orientation impairment. This finding may reflect differential access to healthcare resources, social capital, and lifelong cognitive stimulation between urban and rural populations in China.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eStroke history and orientation impairment\u003c/h2\u003e \u003cp\u003eContrary to expectations based on studies of global cognitive decline and dementia, stroke history was not independently associated with incident orientation impairment in fully adjusted models. The point estimate below unity should not be interpreted as protective; rather, it likely reflects residual confounding, selection of relatively resilient stroke survivors, and limited statistical power. The marked imbalance in vascular comorbidities between stroke and non-stroke groups (e.g., hypertension SMD\u0026thinsp;\u0026gt;\u0026thinsp;0.8) further underscores the potential for residual confounding. Several factors may explain this finding[\u003cspan additionalcitationids=\"CR25\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Stroke history in this study was based on self-reported physician diagnosis and did not capture information on stroke subtype, severity, lesion location, timing, or recurrence. Such heterogeneity may attenuate associations with specific cognitive domains[\u003cspan additionalcitationids=\"CR28 CR29\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn addition, the analytic cohort was restricted to individuals who were free of orientation impairment at baseline, thereby excluding those with early or persistent post-stroke cognitive deficits and potentially selecting a relatively resilient subgroup of stroke survivors. Adjustment for age, educational attainment, and vascular comorbidities may have further accounted for much of the excess cognitive risk associated with stroke history[\u003cspan additionalcitationids=\"CR32\" citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. Notably, when a stricter outcome definition was applied, the point estimate for stroke history shifted above unity, although confidence intervals remained wide, suggesting limited statistical power rather than clear evidence of an effect.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eEducation as a potential modifier of stroke-related risk\u003c/h2\u003e \u003cp\u003eA central aim of this study was to assess whether educational attainment modifies the association between stroke history and incident orientation impairment. Although the direction of interaction estimates and marginal risk patterns were consistent with a buffering effect of higher education, formal tests for multiplicative and additive interaction did not yield statistically conclusive evidence.\u003c/p\u003e \u003cp\u003eThese findings should be interpreted cautiously. Detecting interaction effects typically requires substantially larger sample sizes than those needed to detect main effects, particularly when exposures and outcomes are relatively infrequent[\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Moreover, the coarse measurement of stroke history and the short follow-up period may have limited our ability to detect subtle, domain-specific modification effects[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Rather than refuting the cognitive reserve hypothesis, our results suggest that any modifying effect of education on stroke-related orientation impairment is likely modest in magnitude or difficult to detect in population-based settings using available measures[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eInsights from machine-learning and external validation analyses\u003c/h2\u003e \u003cp\u003eComplementary machine-learning analyses provided additional context for interpreting the primary findings. Despite the application of multiple algorithms and calibration procedures, predictive discrimination for incident orientation impairment remained modest, and no model materially outperformed traditional regression approaches. Across models, educational attainment consistently ranked among the most influential predictors, whereas stroke history contributed limited incremental predictive information. These results reinforce the conclusion that education plays a central role in shaping short-term cognitive vulnerability at the population level[\u003cspan additionalcitationids=\"CR40\" citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eExternal validation in an independent cohort demonstrated moderate discrimination and broadly consistent risk patterns, supporting the transportability of relative associations. However, miscalibration of absolute risk estimates highlights the need for recalibration when applying risk models across populations with different characteristics and measurement instruments[\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Together, these analyses support the robustness of the main findings while underscoring the challenges of predicting incident cognitive outcomes in older adults.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eStrengths and limitations\u003c/h2\u003e \u003cp\u003eKey strengths of this study include its longitudinal design, nationally representative sample, explicit complete-case definition with respect to the outcome, transparent handling of baseline covariate missingness through indicator categories, focus on incident orientation impairment, rigorous handling of loss to follow-up, and the integration of multiple analytical frameworks. The consistent protective association of education across diverse methods strengthens confidence in the central conclusions.\u003c/p\u003e \u003cp\u003eSeveral limitations warrant consideration. Stroke history was measured crudely and lacked clinical detail, limiting etiological inference. The follow-up period was relatively short, potentially underestimating longer-term cognitive consequences of stroke. Orientation impairment represents only one dimension of cognitive function and may not capture subtler deficits in other domains. Finally, limited statistical power constrained definitive conclusions regarding effect modification.\u003c/p\u003e \u003cp\u003eMissingness for certain vascular comorbidities, particularly hypertension, was substantial and differential by stroke status, which may have introduced residual confounding despite the use of missing indicator categories.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003eImplications and future directions\u003c/h2\u003e \u003cp\u003eIn this population-based longitudinal study of older adults, educational attainment emerged as a consistent and robust protective factor against incident orientation impairment, whereas stroke history showed no independent association after adjustment for sociodemographic and vascular factors. Evidence for education modifying the stroke\u0026ndash;orientation relationship was suggestive but inconclusive. These findings highlight the central role of educational gradients in late-life cognitive vulnerability and underscore the complexity of identifying domain-specific cognitive consequences of stroke in community-dwelling populations.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn summary, higher education was strongly and consistently associated with a lower risk of incident orientation impairment over two years, whereas stroke history showed no independent association after adjustment. Evidence for education modifying the stroke\u0026ndash;orientation relationship was suggestive but not conclusive. These findings highlight the central role of educational gradients in cognitive aging and underscore the complexity of disentangling stroke-related cognitive outcomes in population-based studies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe China Health and Retirement Longitudinal Study (CHARLS) obtained ethical approval from the Institutional Review Board of Peking University (IRB00001052-11015), and all participants provided written informed consent prior to participation.\u003c/p\u003e\n\u003cp\u003eThe Chinese Longitudinal Healthy Longevity Survey (CLHLS) was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-13074), and written informed consent was obtained from all participants or their legal representatives.\u003c/p\u003e\n\u003cp\u003eThe present study is a secondary analysis of publicly available, de-identified data from both CHARLS and CLHLS and therefore did not require additional ethical approval from the authors\u0026rsquo; institutions. Both studies were conducted in accordance with the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable. This study used de-identified secondary data and does not contain any individual person\u0026rsquo;s identifiable information.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets analyzed during the current study are publicly available.\u003c/p\u003e\n\u003cp\u003eCHARLS data can be obtained upon registration through the official CHARLS data portal.\u003c/p\u003e\n\u003cp\u003eCLHLS data are available from the CLHLS project upon application and approval via the official CLHLS website.\u003c/p\u003e\n\u003cp\u003eAll analytic code used to generate the results is available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by the Letter of the Department of Personnel and Education of the National Administration of Traditional Chinese Medicine (2022) No. 256; the 9th Anhui Provincial Special Support Program for Talents (2023) No. 35; the Anhui Provincial Natural Science Foundation (Grant No. 2308085MH297); the Key Research Project of Natural Sciences in Anhui Universities (Grant No. 2023AH040099); the Special Project for Clinical Medical Research and Transformation of Anhui Province (Grant No. 202427b10020060); the Scientific Research Project of Anhui Provincial Department of Education (Grant No. 2025AHGXZK40506).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCe Shi and Fei Li conceived the research question and designed the study. Ce Shi led data curation, constructed analytic variables, developed and validated the statistical code, conducted the primary analyses and sensitivity analyses, generated the figures and tables, and drafted the initial manuscript. Qiqi Yang and Tianxin Jiang assisted with variable construction, data harmonization, and verification of analytic datasets; they also contributed to interpretation of findings and preparation of supplementary materials. Lihua Wu provided methodological guidance on model specification, interaction assessment, and robustness analyses, and critically revised the manuscript for important intellectual content. Fei Wang and Xiang Shang contributed to the clinical and public health interpretation of results and assisted with manuscript revision. Baoguo Wang supported data checking and interpretation, and provided input on presentation of results. Jianhong Gao, Weiran Li, and Ziyu Ye contributed to methodological input, literature contextualization, and critical revision of the manuscript. Fei Li served as the corresponding author, provided overall supervision, project administration, and oversight of manuscript development. All authors reviewed and approved the final manuscript and agree to be accountable for all aspects of the work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank the research teams and participants of both the CHARLS and CLHLS studies for providing open-access data that made this work possible.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eTsoy E, Kiekhofer RE, Guterman EL, Tee BL, Windon CC, Dorsman KA, Lanata SC, Rabinovici GD, Miller BL, Kind AJH, et al. Assessment of Racial/Ethnic Disparities in Timeliness and Comprehensiveness of Dementia Diagnosis in California. JAMA Neurol. 2021;78(6):657\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKohler IV, K\u0026auml;mpfen F, Bandawe C, Kohler HP. Cognition and Cognitive Changes in a Low-Income Sub-Saharan African Aging Population. J Alzheimer's disease: JAD. 2023;95(1):195\u0026ndash;212.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFukuda H, Kanzaki H, Murata F, Maeda M, Ikeda M. Disease Burden and Progression in Patients with New-Onset Mild Cognitive Impairment and Alzheimer's Disease Identified from Japanese Claims Data: Evidence from the LIFE Study. J Alzheimer's disease: JAD. 2023;95(4):1559\u0026ndash;72.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuerrero-Berroa E, Luo X, Schmeidler J, Rapp MA, Dahlman K, Grossman HT, Haroutunian V, Beeri MS. The MMSE orientation for time domain is a strong predictor of subsequent cognitive decline in the elderly. Int J Geriatr Psychiatry. 2009;24(12):1429\u0026ndash;37.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTian R, Jiang Y, Zhang Y, Yan X, Zhou Y, Chen D. Cognitive training program improves cognitive ability and daily living ability in elderly patients with mild cognitive impairment. Aging Clin Exp Res. 2022;34(5):997\u0026ndash;1005.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFauth EB, Schwartz S, Tschanz JT, \u0026Oslash;stbye T, Corcoran C, Norton MC. Baseline disability in activities of daily living predicts dementia risk even after controlling for baseline global cognitive ability and depressive symptoms. Int J Geriatr Psychiatry. 2013;28(6):597\u0026ndash;606.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBangen KJ, Weigand AJ, Thomas KR, Delano-Wood L, Clark LR, Eppig J, Werhane ML, Edmonds EC, Bondi MW. Cognitive dispersion is a sensitive marker for early neurodegenerative changes and functional decline in nondemented older adults. Neuropsychology. 2019;33(5):599\u0026ndash;608.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilcock D, Jicha G, Blacker D, Albert MS, D'Orazio LM, Elahi FM, Fornage M, Hinman JD, Knoefel J, Kramer J, et al. MarkVCID cerebral small vessel consortium: I. Enrollment, clinical, fluid protocols. Alzheimer's Dement J Alzheimer's Assoc. 2021;17(4):704\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhi N, Zhang L, Wang Y, Bai S, Geng J, Yu L, Cao W, Zhuang L, Zhou Y, Guan Y. Modified cerebral small vessel disease score is associated with vascular cognitive impairment after lacunar stroke. Aging. 2021;13(7):9510\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao J, Jin X, Chen B, Fu C, Ji S, Shen W, Wei J, Zheng H, Zhang Y. Apathy symptoms increase the risk of dementia conversion: a case-matching cohort study on patients with post-stroke mild cognitive impairment in China. Psychogeriatrics: official J Japanese Psychogeriatr Soc. 2021;21(2):149\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdullah AH, Sharip S, Rahman AHA, Bakar L. Cognitive reserve in stroke patients. PsyCh J. 2021;10(3):444\u0026ndash;52.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUmarova RM, Schumacher LV, Schmidt CSM, Martin M, Egger K, Urbach H, Hennig J, Kl\u0026ouml;ppel S, Kaller CP. Interaction between cognitive reserve and age moderates effect of lesion load on stroke outcome. Sci Rep. 2021;11(1):4478.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMungas D, Fletcher E, Gavett BE, Widaman K, Zahodne LB, Hohman TJ, Mayeda ER, Dowling NM, Johnson DK, Tomaszewski Farias S. Comparison of Education and Episodic Memory as Modifiers of Brain Atrophy Effects on Cognitive Decline: Implications for Measuring Cognitive Reserve. J Int Neuropsychological Society: JINS. 2021;27(5):401\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlbrecht P, Perisse J, Sauleau EA, Blanc F. Cognitive reserve as one of the determinants of the progression of Alzheimer disease. \u003cem\u003eGeriatrie et psychologie neuropsychiatrie du vieillissement\u003c/em\u003e 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi B, Choi J, Lee J, Choi E, Jang S-N. Socioeconomic disparities in hypertension and diabetes among older adults in South Korea: a trend analysis from a national survey, 2011\u0026ndash;2023. Soc Sci Med. 2025;382:118350.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJensen HAR, Davidsen M, Br\u0026oslash;nnum-Hansen H, Eliasen MH, Christensen AI. Trends in educational inequality in healthy life expectancy in Denmark between 2010 and 2021: a population-based study. BMJ Open. 2025;15(7):e100851.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiu J, Li X, Chen Q, Yang W, Suo L, Chen Z. Association Between Educational Inequality and Income Inequality With Metabolic Diseases and Cause-Specific Mortality. Clin Cardiol. 2025;48(7):e70173.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLanga KM, Ryan LH, McCammon RJ, Jones RN, Manly JJ, Levine DA, Sonnega A, Farron M, Weir DR. The Health and Retirement Study Harmonized Cognitive Assessment Protocol Project: Study Design and Methods. Neuroepidemiology. 2020;54(1):64\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang M, Lv X, Chen Y, Tu L, Fan Z, Yao Y, Yu X, Guan N, Wang H. Excessive sleep increased the risk of incidence of cognitive impairment among older Chinese adults: a cohort study based on the Chinese Longitudinal Healthy Longevity Survey (CLHLS). Int Psychogeriatr. 2022;34(8):725\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNieboer D, van der Ploeg T, Steyerberg EW. Assessing Discriminative Performance at External Validation of Clinical Prediction Models. PLoS ONE. 2016;11(2):e0148820.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJonsson H, Nystr\u0026ouml;m L, Blom J. Poisson regression with adjustment for contamination and non-compliance in cohort studies conducted to estimate intervention effectiveness. J Med Screen 2025:9691413251388380.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeker N, Ganz A, Hulsman M, Klausch T, Schmand BA, Scheltens P, Sikkes SAM, Holstege H. Association of Cognitive Function Trajectories in Centenarians With Postmortem Neuropathology, Physical Health, and Other Risk Factors for Cognitive Decline. JAMA Netw open. 2021;4(1):e2031654.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu D, Liu C, Xia K, Abramowitz A, Wu G. Characterizing the Resilience Effect of Neurodegeneration for the Mechanistic Pathway of Alzheimer's Disease. J Alzheimer's disease: JAD. 2021;84(3):1351\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDelgado J, Masoli J, Hase Y, Akinyemi R, Ballard C, Kalaria RN, Allan LM. Trajectories of cognitive change following stroke: stepwise decline towards dementia in the elderly. Brain Commun. 2022;4(3):fcac129.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLo JW, Crawford JD, Lipnicki DM, Lipton RB, Katz MJ, Preux PM, Guerchet M, d'Orsi E, Quialheiro A, Rech CR, et al. Trajectory of Cognitive Decline Before and After Stroke in 14 Population Cohorts. JAMA Netw open. 2024;7(10):e2437133.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen LY, Norby FL, Gottesman RF, Mosley TH, Soliman EZ, Agarwal SK, Loehr LR, Folsom AR, Coresh J, Alonso A. Association of Atrial Fibrillation With Cognitive Decline and Dementia Over 20 Years: The ARIC-NCS (Atherosclerosis Risk in Communities Neurocognitive Study). J Am Heart Association 2018, 7(6).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJamrozik E, Hyde Z, Alfonso H, Flicker L, Almeida O, Yeap B, Norman P, Hankey G, Jamrozik K. Validity of self-reported versus hospital-coded diagnosis of stroke: a cross-sectional and longitudinal study. Cerebrovasc Dis. 2014;37(4):256\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan PJ, Hou LS, Li ME, Lu ZX, Zhan FY, Ran MD, Li JJ, Zhang L, Yang R, Zhou MK, et al. Associations between Lesion Locations and Stroke Recurrence in Survivors of First-ever Ischemic Stroke: A Prospective Cohort Study. Curr Med Sci. 2020;40(4):708\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNakling AE, Aarsland D, N\u0026aelig;ss H, Wollschlaeger D, Fladby T, Hofstad H, Wehling E. Cognitive Deficits in Chronic Stroke Patients: Neuropsychological Assessment, Depression, and Self-Reports. Dement geriatric Cogn disorders extra. 2017;7(2):283\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoundi RA, Fang J, Austin PC, Smith EE, Yu AYX, Hachinski V, Sposato LA, Ganesh A, Sharma M, Kapral MK. Magnitude and Time-Course of Dementia Risk in Stroke Survivors: A Population-Wide Matched Cohort Study. Neurology. 2025;104(1):e210131.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeaver NA, Kuijf HJ, Aben HP, Abrigo J, Bae HJ, Barbay M, Best JG, Bordet R, Chappell FM, Chen C, et al. Strategic infarct locations for post-stroke cognitive impairment: a pooled analysis of individual patient data from 12 acute ischaemic stroke cohorts. Lancet Neurol. 2021;20(6):448\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOhlmeier L, Nannoni S, Pallucca C, Brown RB, Loubiere L, Markus HS. Prevalence of, and risk factors for, cognitive impairment in lacunar stroke. Int J stroke: official J Int Stroke Soc. 2023;18(1):62\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArce Renter\u0026iacute;a M, Manly JJ, Vonk JMJ, Mejia Arango S, Michaels Obregon A, Samper-Ternent R, Wong R, Barral S, Tosto G. Midlife Vascular Factors and Prevalence of Mild Cognitive Impairment in Late-Life in Mexico. J Int Neuropsychological Society: JINS. 2022;28(4):351\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeon AC, Heo M. Sample Sizes Required to Detect Interactions between Two Binary Fixed-Effects in a Mixed-Effects Linear Regression Model. Comput Stat Data Anal. 2009;53(3):603\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeo M, Leon AC. Sample sizes required to detect two-way and three-way interactions involving slope differences in mixed-effects linear models. J Biopharm Stat. 2010;20(4):787\u0026ndash;802.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang C, Berkalieva A, Mazumdar M, Kwon D. Power calculation for detecting interaction effect in cross-sectional stepped-wedge cluster randomized trials: an important tool for disparity research. BMC Med Res Methodol. 2024;24(1):57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh H, Park J, Seo W. A 2-year prospective follow-up study of temporal changes associated with post-stroke cognitive impairment. Int J Nurs Pract. 2018;24(2):e12618.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKessels RP, Eikelboom WS, Schaapsmeerders P, Maaijwee NA, Arntz RM, van Dijk EJ, de Leeuw FE. Effect of Formal Education on Vascular Cognitive Impairment after Stroke: A Meta-analysis and Study in Young-Stroke Patients. J Int Neuropsychological Society: JINS. 2017;23(3):223\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRold\u0026aacute;n-Tapia MD, C\u0026aacute;novas R, Le\u0026oacute;n I, Garc\u0026iacute;a-Garcia J. Cognitive Vulnerability in Aging May Be Modulated by Education and Reserve in Healthy People. Front Aging Neurosci. 2017;9:340.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMantri S, Nwadiogbu C, Fitts W, Dahodwala N. Quality of education impacts late-life cognition. Int J Geriatr Psychiatry. 2019;34(6):855\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHo S, Kozhevnikov M. Cognitive style and creativity: The role of education in shaping cognitive style profiles and creativity of adolescents. Br J Educ Psychol. 2023;93(4):978\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePfeiffer RM, Petracci E. Variance computations for functional of absolute risk estimates. Stat Probab Lett. 2011;81(7):807\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Orientation impairment, Educational attainment, Stroke history, Cognitive reserve, External validation","lastPublishedDoi":"10.21203/rs.3.rs-9092847/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9092847/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eOrientation impairment is an early and clinically meaningful marker of cognitive vulnerability in older adults and is associated with subsequent functional decline, dementia, and mortality. Although stroke is a well-established risk factor for late-life cognitive impairment, its association with incident orientation impairment in community-dwelling populations remains unclear. Educational attainment, a key indicator of cognitive reserve, may modify the cognitive consequences of stroke, but longitudinal evidence addressing this hypothesis is limited.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe conducted a population-based longitudinal study using data from the China Health and Retirement Longitudinal Study (CHARLS). Participants aged\u0026thinsp;\u0026ge;\u0026thinsp;60 years who were free of orientation impairment at baseline (2018) were followed for incident orientation impairment in 2020. Stroke history was defined by self-reported physician diagnosis at baseline. Educational attainment was primarily modeled as a continuous variable (per three-year increase) and additionally evaluated using categorical specifications in sensitivity analyses. Modified Poisson regression with robust variance estimation was used to estimate relative risks (RRs). Effect modification by education was assessed using interaction terms and marginal effect estimation. Sensitivity analyses included inverse probability weighting for loss to follow-up, competing-risk models accounting for death, alternative outcome definitions, and complementary machine-learning analyses. External validation was conducted to examine the transportability of relative risk patterns rather than to formally validate a clinical prediction model.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eAmong 9,080 participants free of orientation impairment at baseline, 5,289 participants with observed follow-up orientation assessments were included (complete-case with respect to the outcome), of whom 744 developed incident orientation impairment over two years. Stroke history was not independently associated with incident orientation impairment after multivariable adjustment (RR 0.88, 95% CI 0.63\u0026ndash;1.24). In contrast, higher educational attainment showed a strong and consistent protective association: each additional three years of education was associated with a 25.6% lower risk of incident orientation impairment (RR 0.74, 95% CI 0.65\u0026ndash;0.86). Formal tests for stroke\u0026ndash;education interaction were not statistically significant on multiplicative or additive scales; however, marginal estimates were directionally consistent with attenuation of stroke-associated risk at higher education levels, but estimates were imprecise and confidence intervals crossed the null. Results were robust across sensitivity analyses. Machine-learning models demonstrated modest discrimination and did not materially outperform traditional regression models. External validation showed moderate discrimination but indicated miscalibration of absolute risk estimates, suggesting the need for recalibration in different population settings.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eIn this nationally representative cohort of older adults, educational attainment emerged as a consistent and robust protective factor against incident orientation impairment, whereas stroke history showed no independent association after adjustment. Evidence for education modifying the stroke\u0026ndash;orientation relationship was suggestive but inconclusive. These findings highlight the central role of educational gradients in late-life cognitive vulnerability and underscore the challenges of detecting domain-specific cognitive consequences of stroke in population-based studies.\u003c/p\u003e","manuscriptTitle":"Association of educational attainment with incident orientation impairment among Chinese older adults with and without stroke: evidence from a longitudinal study with external validation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-14 10:21:54","doi":"10.21203/rs.3.rs-9092847/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-04-07T12:32:47+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-03-16T16:19:59+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-13T11:58:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-13T11:58:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Public Health","date":"2026-03-11T09:53:30+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"685a8476-dab0-4916-914d-6d81be72df19","owner":[],"postedDate":"April 14th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-14T10:21:55+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-14 10:21:54","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9092847","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9092847","identity":"rs-9092847","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}