Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study

Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study | 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 Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study Zejing Lin, Rulan Ma, Xing Chen, Yinzhou Wang, Xingyong Chen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6950624/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: This study aimed to evaluate the utility of machine learning (ML) algorithms in predicting cognitive impairment among elderly individuals with hypertension in the United States and to identify key associated risk factors. Methods: Data were obtained from 19,931 participants enrolled in the 2011–2012 and 2013–2014 cycles of the National Health and Nutrition Examination Survey (NHANES). The dataset was randomly split into training and test sets (70:30). Seven ML algorithms—logistic regression (LR), extreme gradient boosting (XGB), decision tree (DT), categorical boosting (CatBoost), random forest (RF), light gradient boosting machine (LGBM), and support vector machine (SVM)—were trained to predict cognitive impairment. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). The SHapley Additive exPlanation (SHAP) method was applied for feature importance interpretation. Results: Among all models, LR exhibited the best overall performance, achieving an AUC of 0.791 on the test set, with superior F1-score and accuracy. Calibration plots demonstrated good agreement between predicted and observed outcomes. DCA confirmed the clinical utility of the LR model. SHAP analysis identified the key variables contributing to model predictions. A web-based calculator based on the final LR model, incorporating 12 predictors, is available at: https://cognitiveimpairment.shinyapps.io/cognitiveimpairment. Conclusion: An interpretable ML model was developed and validated to predict the risk of cognitive impairment in elderly hypertensive patients in the United States. This model enables clinicians to quickly identify high-risk patients, which in turn supports more effective prevention and intervention strategies. hypertension cognitive impairment machine learning predictive model NHANES Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction The 2021 Global Status Report on Dementia by the World Health Organization estimated that fifty-five million individuals worldwide were affected by dementia in 2019, with this number projected to increase to seventy-eight million by 2030 [ 1 ] . Despite this alarming trend, a recent report in The Lancet suggested that approximately forty percent of dementia cases could theoretically be prevented or postponed through effective control of risk factors [ 2 ] . The etiology of cognitive impairment is complex; however, autopsy analyses have shown that in more than half of clinically diagnosed dementia cases, vascular factors, particularly hypertension, are the primary or exclusive cause. [ 3 ] . Chronic hypertension can gradually damage the structure and function of cerebral blood vessels, resulting in the breakdown of the blood-brain barrier due to inflammatory pathological mechanisms. [ 4 ] . This, in turn, may result in amyloid deposition and the development of Alzheimer’s disease (AD) pathology [ 4 ] . Additionally, hypertension-induced damage to small cerebral vessels, ranging from capillaries to arterioles, referred to as cerebral small vessel disease (CSVD), accounts for forty-five percent of clinically diagnosed cases of vascular cognitive impairment [ 5 ] . Furthermore, it has been established that elevated blood pressure during midlife increases the risk of developing dementia in older age [ 6 ] , independent of genetic risk factors for dementia [ 7 ] . Accordingly, it is crucial to examine the contributing factors to cognitive decline in individuals with hypertension, develop clinical prediction models, and further explore the underlying mechanisms. Such efforts could provide more effective decision-making support for the prevention and intervention of early cognitive impairment. To date, studies on predictive models for evaluating the likelihood of cognitive decline in individuals with hypertension have been published [ 8 ]–[ 13 ] . Unfortunately, the target populations in these studies have predominantly been from China, and most models were developed using a single method, such as logistic regression, which limits their practical utility. As clinical data have become increasingly large and complex, artificial intelligence (AI) has been increasingly adopted to address these challenges. Among the subfields of AI, machine learning (ML) has achieved remarkable success in constructing predictive models and is now widely applied to the diagnosis, prognosis, and treatment selection of various diseases [ 14 ],[ 15 ] . Therefore, by comparing multiple ML algorithms and selecting the optimal model for clinical prediction, clinicians can be better supported in identifying patients at risk of cognitive impairment, thereby facilitating improved prevention and intervention strategies. In this study, predictive models for identifying cognitive impairment in older adults with hypertension were developed and validated using seven ML algorithms, utilizing data derived from the National Health and Nutrition Examination Survey (NHANES) database. These models address a significant gap in research related to cognitive impairment prediction in hypertensive patients within the U.S. population. Furthermore, a web-based calculator was constructed on an online platform to facilitate the clinical application of the model. This research aims to enhance the timely and rapid diagnosis of cognitive impairment in high-risk populations, thereby supporting the formulation of individualized prevention approaches and therapeutic interventions to reduce the socioeconomic impact on families and the broader community. 2. MATERIALS AND METHODS 2.1 Data Collection and Study Population All data was obtained from NHANES, a nationally representative database made available by the U.S. Centers for Disease Control and Prevention. The NHANES protocol underwent review by the ethics committee of the National Center for Health Statistics (https://www.cdc.gov/nchs/nhanes/about/erb.html), and documented informed consent was secured from all subjects. Because each survey cycle of NHANES is independent and the participants in each cycle are different, it creates good conditions for the development and verification of clinical prediction models. The study population comprised 19,931 participants were recruited from two NHANES survey cycles (2011-2012 and 2013-2014). Participants were excluded according to the subsequent criteria: (a) participants younger than 60 years; (b) participants missing cognitive function test questionnaires; (c) participants who were determined not to have hypertension based on the test questionnaire; (d) participants with missing data for other candidate predictor variables. After the screening process, a final cohort of 1,373 older adults with hypertension was enrolled. The screening procedure is depicted in Fig. 1. 2.2 Definition of hypertension Diagnosis of hypertension was assigned based on either: self-reported use of antihypertensive medications, or confirmation by a healthcare provider of the condition. 2.3 Cognitive assessment Among the participants we included from the two survey cycles, cognition was evaluated employing the subsequent three assessment tools: (1) the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) [16] , primarily used to evaluate immediate and delayed learning ability of new verbal information (memory domain). It comprises three consecutive learning trials and a delayed recall trial, with cumulative scores ranging from 0 to 40. (2) the Animal Fluency Test (AFT) [17] , which probes categorical verbal fluency, a component of executive function, with a maximum score of 10. (3) the Digit Symbol Substitution Test (DSST) [18] , a performance module from the Wechsler Adult Intelligence Scale (WAIS III), chiefly employed to gauge processing speed, sustained attention, and working memory, with a maximum score of 10. Currently, there are no established cutoff values for the tests mentioned above to classify participants as cognitively impaired or normal. Therefore, we referred to previous studies [19] and used the average z-scores from the three assessment tools to obtain a composite z-score. This approach was used to further analyze the overall cognitive performance across the three tests. The lowest quartile of the composite z-score served as the demarcation point. Individuals scoring below this value were stratified into the cognitive impairment group, while others comprised the non-cognitive impairment group. The z-score was derived as [19] : z-score , where denotes an individual’s test result, and and represent the mean and standard deviation of scores across the entire sample, respectively. 2.4 Determination of Candidate Predictors We collected a range of widely recognized and commonly used indicators from the NHANES database as candidate predictors to include in the final predictive model. The candidate predictors consist of three components: demographic data, laboratory test results, and medical history. Demographic data include age, gender, poverty income ratio (PIR), race, education status, and marital status. Laboratory test data include standard blood biochemical tests, red blood cell folate (RBC folate), and serum vitamin B12 levels. Standard blood biochemical tests include alanine aminotransferase (ALT), glycohemoglobin, albumin, blood urea nitrogen (BUN), blood glucose, serum creatinine, serum uric acid, triglycerides (TG), serum sodium, and high-density lipoprotein cholesterol (HDL-C). Medical history data include smoking, body mass index (BMI), alcohol consumption, age at hypertension onset, heart failure, coronary heart disease, stroke, diabetes, chronic bronchitis, sleep disorders, and cancer. Smoking status was categorized: Never-smokers (100 cigarettes, no current use), and Current smokers (all others). Participants reporting ≥12 alcoholic drinks/year were designated as alcohol consumers. 2.5. Statistical analysis 2.5.1 Data Cleaning and Variable Selection After excluding patients with missing data according to the inclusion and exclusion standards detailed in Fig. 1, hypertensive patients aged ≥60 years with complete cognitive assessments and candidate predictor data were included. Continuous variables that did not follow a normal distribution were summarized as medians with interquartile ranges and analyzed using non-parametric methods. Categorical data were reported as counts and percentages and compared using chi-square tests. The dataset was randomly split into a training subset (70%) and a validation subset (30%) based on the commonly adopted 70/30 rule. Within the training cohort, candidate predictors for the final model were identified using the Least Absolute Shrinkage and Selection Operator (LASSO) regression technique. 2.5.2 Model Construction and Evaluation In ML, four primary paradigms exist: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Given that the objective of this study was to categorize elderly individuals with hypertension into groups with either intact cognitive function or cognitive impairment, supervised learning techniques were deemed most suitable for this classification problem [20] . To develop predictive models for cognitive impairment in this population, seven ML algorithms were implemented: logistic regression (LR), categorical boosting (CatBoost), extreme gradient boosting (XGB), decision tree (DT), random forest (RF), light gradient boosting machine (LGBM), and support vector machine (SVM). The models were evaluated using several commonly used metrics in the test set, including the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, accuracy, recall, F1-score, precision, Matthews correlation coefficient (MCC), and specificity. Model calibration was examined through calibration plots, while decision curve analysis (DCA) was utilized to determine the clinical utility of the models across various decision thresholds. The optimal model was identified based on comprehensive evaluation of these metrics in the test set. 2.5.3 Model Interpretation and Web Calculator Construction SHAP (SHapley Additive exPlanation) is a post-hoc model interpretation technique for ML, grounded in game theory [21] . It assesses the significance of individual features by determining their contribution to the model's output, thus enhancing the transparency and interpretability of the model. Finally, based on the best-performing prediction model selected, a web-based calculator was constructed on the Shiny application platform. By inputting the patient's clinical data, this tool can assist healthcare providers and individuals in assessing the risk of cognitive impairment. All statistical procedures were performed using R software (version 4.3.1) and DecisionLinnc1.0 software [22] , with p < 0.05 considered statistically significant. 3. Results 3.1 Baseline Characteristics of Study Participants This study included 1,373 older adults with hypertension who had complete cognitive evaluations and relevant predictor data. An overview of baseline characteristics is presented in Table 1. Most variables displayed similar distributions between the training and testing subsets, as indicated by p -values predominantly exceeding 0.05. Participants were categorized into two cohorts based on cognitive status: the cognitive impairment group and the non-cognitive impairment group. Notable differences were observed between the two groups in demographic data, including age, gender, race, education status, and PIR (all p < 0.05). For laboratory indicators, the cognitive impairment group had higher levels of ALT, BUN, blood creatinine, and blood glucose compared to their counterparts without impairment (all p < 0.05). Additionally, significant differences were found in medical history, including BMI, diabetes, heart failure, and stroke (all p < 0.05). 3.2 Variable Selection Lasso regression was utilized to select independent variables, and ten-fold cross-validation was conducted to determine the optimal regularization parameter λ (Fig. 2). With cognitive impairment as the dependent variable, a total of twenty-nine predictor variables from three domains—demographic data, laboratory tests data, and medical history data—were included in the Lasso regression. The λ values corresponding to lambda.min and lambda.1se, as well as the number of selected predictor variables, were obtained. To improve the predictive accuracy of the model, λ value at lambda.min was selected, resulting in a λ of 0.0150 (log(lambda.min) = −4.202). This approach yielded 12 variables with non-zero coefficients, including age, gender, BMI, race, PIR, education status, serum creatinine, serum sodium, diabetes, stroke, RBC folate, and age at hypertension onset. 3.3 Construction and Performance Evaluation of Predictive Models 12 variables selected by Lasso regression were used to construct predictive models with seven ML algorithms. Fig. 3 illustrates the discriminatory performance of the seven models on the ROC curves for the training and test datasets. As shown in Fig. 3B, on the test set, LR achieved the highest AUC (AUC = 0.791), indicating the best discriminatory ability, followed by CatBoost (AUC = 0.785), SVM (AUC = 0.773), RF (AUC = 0.757), LGBM (AUC = 0.720), XGB (AUC = 0.720), and DT (AUC = 0.713). Moreover, LR also had higher accuracy (0.777), F1-Score (0.452) and MCC (0.359) compared to the other six models (Table 2). Additionally, as shown in Table 2, XGB had the highest recall (0.369) and false positive rate (FPR) (0.140), SVM had the highest precision (0.737), and LGBM exhibited the highest specificity (0.977) and false negative rate (FNR) (0.856). Fig. 4A and 4B display the calibration curves for the training and test sets, respectively. Based on the calibration curve for the test set (Fig. 3B), LR shows impressive calibration performance compared to the other six models. Fig. 5A and 4B present the DCA curves for the training and test sets, respectively. As shown in Fig. 5B, under a defined probability threshold, LR achieved the highest net benefit among the seven machine learning models, suggesting its greater clinical applicability. Considering its overall superiority in predictive accuracy, calibration reliability, and decision utility, LR was determined to be the most effective model for forecasting cognitive impairment in older adults with hypertension. 3.4 Model Interpretation As shown in Fig. 6, the SHAP algorithm was applied to interpret the output of the final LR model by quantifying each feature’s influence on the prediction results. In Fig. 6A, the SHAP summary bar plot was employed to evaluate feature importance based on the mean absolute SHAP values. Predictive variables were ranked in descending order according to their contribution to cognitive impairment risk, with the top five features being education status, age, PIR, BMI, and diabetes. Additionally, the SHAP summary dot plot in Fig. 6B visually represented both the direction and magnitude of each variable’s effect on the model’s predictions. The vertical axis represents SHAP values of zero, with variables to the right of the line, colored in purple, indicating a positive contribution to the prediction, while those colored in yellow indicate a negative contribution. The variables were also ranked in descending order of their contribution. As a result, age and diabetes were found to be positively correlated with the occurrence of cognitive impairment, meaning that older age and diabetes in hypertensive patients increase the probability of cognitive impairment. Conversely, education status, PIR, and BMI were found to be negatively correlated with cognitive impairment, indicating that higher education status, PIR, and BMI levels in hypertensive patients reduce the risk of cognitive impairment. Furthermore, as illustrated in Fig. 6C, the SHAP waterfall plot is used to illustrate how specific features affect the prediction output at the individual level. The exact values of each feature, along with their associated SHAP values, indicate whether the feature contributes positively or negatively to the final prediction. Education status (college or above) and age (60–69 years) made significant positive contributions to the prediction outcome, with SHAP values of -0.104 and -0.0593, respectively, while PIR (≤1.3) and race (Other Race) contributed negatively with SHAP values of +0.0382 and +0.029, respectively. Other features, such as RBC folate, gender, creatinine, and diabetes, also contributed to the prediction result to varying degrees. The process by which the machine learning model generates results for an individual patient is clearly demonstrated in the SHAP waterfall plot through the cumulative SHAP values. Therefore, the use of the SHAP method provides a more transparent and interpretable approach, significantly improving insight into the model’s underlying decision-making process. 3.5 Development of a Web-Based Calculator As shown in Fig. 7, a web-based calculator was developed based on the LR prediction model to facilitate its clinical application. By entering the actual values of the 12 required variables, the calculator allows for the estimation of cognitive impairment risk in individual older adults with hypertension. The web-based calculator is available via the following link: https://cognitiveimpairment.shinyapps.io/cognitiveimpairment/. 4. Discussion As previously mentioned, studies on predictive models for evaluating the likelihood of cognitive decline in individuals with hypertension has been documented [8]–[13] . However, the target populations and modeling approaches in these studies have been relatively limited, resulting in constrained clinical applicability. To the best of our understanding, this represents the first investigation to construct a cognitive impairment prediction model using multiple ML algorithms based on data from elderly hypertensive individuals in the United States. This study analyzed information from the 2011-2012 and 2013-2014 cycles of the NHANES, which incorporated cognitive function evaluations. 12 predictive factors were identified from three domains: demographic data, laboratory tests data, and medical history data. Based on these 12 predictive factors, we developed and validated prediction models using seven ML algorithms. Ultimately, the LR model was identified as the optimal clinical prediction model, demonstrating the highest discrimination, accuracy, F1-score, and clinical net benefit in the validation set. Finally, the SHAP method was employed to elucidate the results of the final predictive model, and a web-based calculator was developed to facilitate its implementation in clinical practice. In our study, we employed the SHAP framework to evaluate the contribution of the 12 variables included in our model. The top five variables ranked by importance were education status, age, PIR, BMI, and diabetes. Our findings indicate that low education level, advanced age, low PIR, low BMI, and diabetes are significant risk factors contributing to cognitive decline in elderly individuals with hypertension. Numerous studies have demonstrated that higher educational attainment is linked to a decreased likelihood of developing cognitive impairment [23],[24] . Bielak et al. proposed that education benefits individuals throughout their lifetime through a mechanism known as "preserved differentiation" [25] . Simply put, if the rate of cognitive decline is uniform across individuals, differences in peak cognitive function in early adulthood would lead to variations in when cognitive function falls below a critical threshold. In other words, higher levels of formal education may reduce the risk of cognitive impairment by enhancing an individual's cognitive reserve. This concept might help explain why some individuals with severe neurodegenerative diseases, such as AD, still exhibit only mild cognitive decline [26] . Aging is a well-established risk factor for cognitive decline and a major contributor to the progression of neurological disorders. Research shows that cognitive decline accelerates with age, with detectable signs appearing as early as middle age (45–55 years). [27] . Furthermore, approximately 40% of adults aged 65 years or above exhibit some degree of memory impairment [28] . In hypertensive patients, prolonged exposure to high blood pressure with advancing age can result in chronic inflammation, oxidative stress, and cerebrovascular damage [29] . This not only increases the risk of stroke-related cognitive impairment but also leads to subclinical vascular brain damage and white matter integrity deterioration, collectively contribute to progressive cognitive decline [30] . Moreover, our study indicates that individuals with lower socioeconomic status are at greater risk for cognitive impairment. Similarly, Zeki et al. found that prolonged exposure to low income over two decades was significantly correlated with diminished cognitive performance, indicating that socioeconomic conditions during both early and later stages of life play a crucial role in shaping cognitive health in old age [31] . Additionally, individuals experiencing long-term low-income exposure often have poorer socioeconomic status and engage in more unhealthy behaviors, which further exacerbates cognitive impairment [32] . A meta-analysis incorporating 39 prospective studies revealed a heightened overall risk of cognitive decline and dementia among individuals with lower socioeconomic status. Subgroup analyses further showed that low-income and low-education groups faced a significantly greater risk of developing cognitive deficits and dementia [33] . Interestingly, our study suggests that higher BMI in later life serves as a protective factor for cognitive function in patients with hypertension. The relationship between BMI and cognitive performance is complex. While extensive research supports a link between higher midlife BMI and worsened cognitive outcomes in old age [34],[35] , the impact of BMI in old age remains inconclusive. In our study, as well as in other investigations, lower BMI in old age was linked to worse cognitive function [34],[36] , though some studies have reported no association or even the opposite relationship [36],[37] .Lastly, our study suggests that elderly hypertensive patients with diabetes face an elevated likelihood of experiencing cognitive decline. Previous studies indicate that approximately 20% of individuals with diabetes develop mild cognitive impairment (MCI). Furthermore, those diagnosed with both MCI and diabetes have a 1.53 times greater risk of advancing to dementia compared to individuals without diabetes [38] . A meta-analysis also identified diabetes as the only independent predictor of MCI conversion to dementia [39] . Apart from stroke and age at hypertension onset, which have been briefly analyzed earlier, three additional modifiable risk factors among the remaining five predictors include serum creatinine, RBC folate, and serum sodium levels. Our study indicates high serum creatinine, low RBC folate, and high serum sodium as contributing factors to cognitive decline in older hypertensive individuals. Serum creatinine serves as a standard indicator of renal function, and it is well known that individuals receiving hemodialysis for end-stage renal disease (ESRD) frequently suffer from cognitive impairment. Consistent with our findings, a 10-year prospective study in the United Kingdom demonstrated an association between poor kidney function and dementia, independent of major cardiometabolic diseases and stroke [40] . However, the relationship between serum creatinine and cognitive impairment may be nonlinear, as low creatinine levels have also been linked to poorer cognitive outcomes. A recent Finnish study, which followed young adults with normal serum creatinine levels for 10 years, found that in men, persistently high creatinine levels were associated with better memory and learning function in midlife compared to lower creatinine levels [41] . Thus, the precise relationship between serum creatinine and cognitive function still requires further investigation. Mounting evidence suggests that folate, an essential vitamin, plays a pivotal role in the pathogenesis of AD. A prospective cohort study conducted in China among elderly individuals indicated that lower folate concentrations were linked to a heightened risk of mild MCI [42] , aligning with our findings and other previous studies [43],[44] . As the primary extracellular cation, sodium ions regulate fluid-electrolyte balance, circulating blood volume, and osmotic stability [45] . Abnormal serum sodium levels can disrupt intracellular and extracellular osmotic equilibrium, leading to cerebral blood volume disturbances and neuronal dysfunction. Theoretically, both low and high serum sodium levels may increase the risk of cognitive dysfunction. Research indicates that reduced serum sodium may cause astrocyte swelling and trigger the release of excitatory neurotransmitters (e.g., glutamate), which are closely linked to neuronal damage [46] . Conversely, higher serum sodium levels have been correlated with increased AD pathology, reduced hippocampal volume, and cognitive decline [47] , which aligns with our observations. However, the association between serum sodium levels and cognitive performance warrants further investigation. Despite this study utilized a large, nationally representative sample of older adults in the United States and benefited from the methodological strengths and rigorous quality assurance of NHANES, several limitations should be acknowledged. First, the NHANES dataset is cross-sectional in nature, which precludes the ability to evaluate the longitudinal risk of developing cognitive impairment. Second, our predictive model was developed based on data from United States patients and validated using internal validation only, which may limit its generalizability and external applicability to other populations. Lastly, due to the extensive number of variables in the NHANES database and our exclusion of variables with missing data to enhance model reliability, several variables potentially associated with cognitive impairment were not included. This omission may have influenced the final results. Future research should consider incorporating longitudinal data, including a broader population, and conducting external validation of the model to enhance its generalizability and robustness. Conclusion In summary, we effectively established an interpretable ML model that can assist clinicians in the early and rapid detection of cognitive impairment risk among elderly hypertensive patients, which may ultimately help improve patient outcomes. Declarations Acknowledgements The authors would like to acknowledge the invaluable contributions of both the participants and staff of the NHANES, whose efforts were essential to the success of this study. Author contributions Zejing Lin: Conceptualization, Methodology, Formal analysis, Writing – original draft. Rulan Ma: Writing – review & editing, Data curation, Software, Visualization. Xing Chen: Investigation, Validation, Visualization. Yinzhou Wang: Project administration, supervision, writing- review and editing. Xingyong Chen: Conceptualization, Writing – review & editing, Supervision, Project administration. F unding This work was supported by National Natural Science Foundation of China (81771250); Joint Funds for the innovation of science and Technology, Fujian province (2023Y9286); Fujian Provincial Natural Science Foundation (2020R1011004,2021J01374,2024J011004); Fujian Research and Training Grants for Young and Middle-aged Leaders in Healthcare (2022). Data availability The data utilized in this study are available in the NHANES database (https://wwwn.cdc.gov/nchs/nhanes/default.aspx). Additional datasets analyzed or referenced in the present research can be obtained from the corresponding authors upon reasonable request. Ethical approval The NHANES dataset is freely available for public access through its official website. The NHANES study protocol received approval from the ethics review committee of the U.S. Centers for Disease Control and Prevention, and all subjects provided written informed consent prior to participation. Consent for publication Not applicable. Consent to participate Not applicable. Competing interests The authors declare that there are no conflicts of interest. Clinical trial number Not applicable. References World Health Organization. Global status report on the public health response to dementia. 2021. Available from: https://digitalcommons.fiu.edu/srhreports/health/health/65. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet . 2020;396(10248):413–446. Azarpazhooh MR, Avan A, Cipriano LE, et al. Concomitant vascular and neurodegenerative pathologies double the risk of dementia. Alzheimers Dement . 2018;14(2):148–156. Santisteban MM, Iadecola C, Carnevale D. Hypertension, neurovascular dysfunction, and cognitive impairment. Hypertension . 2023;80(1):22–34. Pacella J, Lembo G, Carnevale L. A translational perspective on the interplay between hypertension, inflammation and cognitive impairment. Can J Cardiol . 2024;40(12):2368–2377. McGrath ER, Beiser AS, DeCarli C, et al. Blood pressure from mid- to late life and risk of incident dementia. Neurology . 2017;89(24):2447–2454. Littlejohns TJ, Collister JA, Liu X, et al. Hypertension, a dementia polygenic risk score, APOE genotype, and incident dementia. Alzheimers Dement . 2023;19(2):467–476. Zhong X, Yu J, Jiang F, et al. A risk prediction model based on machine learning for early cognitive impairment in hypertension: development and validation study. Front Public Health . 2023;11:1143019. Cerezo GH, Fernández RA, Enders JE, et al. Predicting cognitive function and dementia risk in patients with hypertension. Hypertens Res . 2024;47(6):1728–1734. Jingyu L, Wen D, Liping Z, et al. A nomogram for predicting mild cognitive impairment in older adults with hypertension. BMC Neurol . 2023;23(1):363. Qu L, Dong Z, Ma S, et al. Gut microbiome signatures are predictive of cognitive impairment in hypertension patients—a cohort study. Front Microbiol . 2022;13:841614. Li Y, Xin J, Fang S, et al. Development and validation of a predictive model for early identification of cognitive impairment risk in community-based hypertensive patients. J Appl Gerontol . 2024;43(12):1867–1877. Jiang X, Huo M, Yu Q, et al. Analysis of factors affecting subjective cognitive decline in elderly hypertensive patients and the construction of a risk prediction model. Geriatr Nurs . 2025;61:302–310. Greener JG, Kandathil SM, Moffat L, et al. A guide to machine learning for biologists. Nat Rev Mol Cell Biol . 2022;23(1):40–55. Tran KA, Kondrashova O, Bradley A, et al. Deep learning in cancer diagnosis, prognosis and treatment selection. Genome Med . 2021;13(1):152. Morris JC, Heyman A, Mohs RC, et al. The consortium to establish a registry for Alzheimer’s disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer’s disease. Neurology . 1989;39(9):1159–1165. Carone DA. Strauss E, Sherman EMS, Spreen O. A compendium of neuropsychological tests: administration, norms, and commentary (3rd ed.). Appl Neuropsychol . 2007;14(1):62–63. Ryan JJ, Lopez SJ. Wechsler adult intelligence scale-III. In: Dorfman WI, Hersen M, eds. Understanding Psychological Assessment. Springer US; 2001:19–42. Shi Y, Wang H, Zhu Z, et al. Association between exposure to phenols and parabens and cognitive function in older adults in the United States: a cross-sectional study. Sci Total Environ . 2023;858(Pt 3):160129. Choi RY, Coyner AS, Kalpathy-Cramer J, et al. Introduction to machine learning, neural networks, and deep learning. Transl Vis Sci Technol . 2020;9(2):14. Hou F, Zhu Y, Zhao H, et al. Development and validation of an interpretable machine learning model for predicting the risk of distant metastasis in papillary thyroid cancer: a multicenter study. EClinicalMedicine . 2024;77:102913. DecisionLinnc Core Team. DecisionLinnc software, version 1.0. 2023. Available from: https://www.statsape.com/. Stern Y, Arenaza-Urquijo EM, Bartrés-Faz D, et al. Whitepaper: defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement . 2020;16(9):1305–1311. Lövdén M, Fratiglioni L, Glymour MM, et al. Education and cognitive functioning across the life span. Psychol Sci Public Interest . 2020;21(1):6–41. Bielak AAM, Cherbuin N, Bunce D, et al. Preserved differentiation between physical activity and cognitive performance across young, middle, and older adulthood over 8 years. J Gerontol B Psychol Sci Soc Sci . 2014;69(4):523–532. Zhong T, Li S, Liu P, et al. The impact of education and occupation on cognitive impairment: a cross-sectional study in China. Front Aging Neurosci . 2024;16:1435626. Kohler IV, Kämpfen F, Bandawe C, et al. Cognition and cognitive changes in a low-income sub-Saharan African aging population. Zuelsdorff M, ed. J Alzheimers Dis . 2023;95(1):195–212. Brito DVC, Esteves F, Rajado AT, et al. Assessing cognitive decline in the aging brain: lessons from rodent and human studies. NPJ Aging . 2023;9(1):23. Coleman CG, Wang G, Faraco G, et al. Membrane trafficking of NADPH oxidase p47(phox) in paraventricular hypothalamic neurons parallels local free radical production in angiotensin II slow-pressor hypertension. J Neurosci . 2013;33(10):4308–4316. Levine DA, Galecki AT, Langa KM, et al. Blood pressure and cognitive decline over 8 years in middle-aged and older Black and White Americans. Hypertension . 2019;73(2):310–318. Zeki Al Hazzouri A, Elfassy T, Sidney S, et al. Sustained economic hardship and cognitive function. Am J Prev Med . 2017;52(1):1–9. Lynch JW, Kaplan GA, Salonen JT. Why do poor people behave poorly? Variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse. Soc Sci Med . 1997;44(6):809–819. Wang AY, Hu HY, Ou YN, et al. Socioeconomic status and risks of cognitive impairment and dementia: a systematic review and meta-analysis of 39 prospective studies. J Prev Alzheimers Dis . 2023;10(1):83–94. Pedditzi E, Peters R, Beckett N. The risk of overweight/obesity in mid-life and late life for the development of dementia: a systematic review and meta-analysis of longitudinal studies. Age Ageing . 2016;45(1):14–21. Ravona-Springer R, Schnaider-Beeri M, Goldbourt U. Body weight variability in midlife and risk for dementia in old age. Neurology . 2013;80(18):1677–1683. Wotton CJ, Goldacre MJ. Age at obesity and association with subsequent dementia: record linkage study. Postgrad Med J . 2014;90(1068):547–551. Singh-Manoux A, Dugravot A, Shipley M, et al. Obesity trajectories and risk of dementia: 28 years of follow-up in the Whitehall II Study. Alzheimers Dement . 2018;14(2):178–186. Ding X, Yin L, Zhang L, et al. Diabetes accelerates Alzheimer’s disease progression in the first year post mild cognitive impairment diagnosis. Alzheimers Dement . 2024;20(7):4583–4593. Li JQ, Tan L, Wang HF, et al. Risk factors for predicting progression from mild cognitive impairment to Alzheimer’s disease: a systematic review and meta-analysis of cohort studies. J Neurol Neurosurg Psychiatry . 2016;87(5):476–484. Singh-Manoux A, Oumarou-Ibrahim A, Machado-Fragua MD, et al. Association between kidney function and incidence of dementia: 10-year follow-up of the Whitehall II cohort study. Age Ageing . 2022;51(1):afab259. Hakala JO, Pahkala K, Juonala M, et al. Repeatedly measured serum creatinine and cognitive performance in midlife: the Cardiovascular Risk in Young Finns Study. Neurology . 2022;98(22):e2268–e2281. Fu J, Liu Q, Zhu Y, et al. Circulating folate concentrations and the risk of mild cognitive impairment: a prospective study on the older Chinese population without folic acid fortification. Eur J Neurol . 2022;29(10):2913–2924. Wang Q, Zhao J, Chang H, et al. Homocysteine and folic acid: risk factors for Alzheimer’s disease—an updated meta-analysis. Front Aging Neurosci . 2021;13:665114. Hama Y, Hamano T, Shirafuji N, et al. Influences of folate supplementation on homocysteine and cognition in patients with folate deficiency and cognitive impairment. Nutrients . 2020;12(10):3138. Cook NR, He FJ, MacGregor GA, et al. Sodium and health—concordance and controversy. BMJ . 2020;369:m2440. Hui Z, Wang L, Deng J, et al. Joint association of serum sodium and frailty with mild cognitive impairment among hospitalized older adults with chronic diseases: a cross-sectional study. Front Nutr . 2024;11:1467751. Chen Y, Wang Z, Liu X, et al. Elevated serum sodium is linked to increased amyloid-dependent tau pathology, neurodegeneration, and cognitive impairment in Alzheimer’s disease. J Neurochem . 2024;:jnc.16257. Tables Table 1. Characteristics of datasets. Characteristics Overall Non-cognitive impairment Cognitive impairment p -value Training set Test set p -value N = 1,373 N = 1,030 N = 343 N = 961 N = 412 Cognitive assessment 0.272 Non-cognitive impairment 1,030 (75.0%) 729 (75.9%) 301 (73.1%) Cognitive impairment 343 (25.0%) 232 (24.1%) 111 (26.9%) Age, years <0.001 0.646 60-69 705 (51.3%) 577 (56.0%) 128 (37.3%) 501 (52.1%) 204 (49.5%) 70-79 426 (31.0%) 308 (29.9%) 118 (34.4%) 295 (30.7%) 131 (31.8%) ≥80 242 (17.6%) 145 (14.1%) 97 (28.3%) 165 (17.2%) 77 (18.7%) Gender 0.005 0.855 Male 648 (47.2%) 463 (45.0%) 185 (53.9%) 452 (47.0%) 196 (47.6%) Female 725 (52.8%) 567 (55.0%) 158 (46.1%) 509 (53.0%) 216 (52.4%) Race <0.001 0.406 Mexican American 107 (7.8%) 81 (7.9%) 26 (7.6%) 67 (7.0%) 40 (9.7%) Other Hispanic 120 (8.7%) 71 (6.9%) 49 (14.3%) 84 (8.7%) 36 (8.7%) Non-Hispanic White 677 (49.3%) 538 (52.2%) 139 (40.5%) 485 (50.5%) 192 (46.6%) Non-Hispanic Black 358 (26.1%) 253 (24.6%) 105 (30.6%) 246 (25.6%) 112 (27.2%) Other Race 111 (8.1%) 87 (8.4%) 24 (7.0%) 79 (8.2%) 32 (7.8%) Educational status <0.001 0.476 Less than 9th grade 149 (10.9%) 56 (5.4%) 93 (27.1%) 98 (10.2%) 51 (12.4%) High school education 539 (39.3%) 382 (37.1%) 157 (45.8%) 378 (39.3%) 161 (39.1%) College or above 685 (49.9%) 592 (57.5%) 93 (27.1%) 485 (50.5%) 200 (48.5%) Marital status 0.105 0.505 Married, living with partner 784 (57.1%) 604 (58.6%) 180 (52.5%) 553 (57.5%) 231 (56.1%) Widowed, divorced, or separated 511 (37.2%) 367 (35.6%) 144 (42.0%) 350 (36.4%) 161 (39.1%) Unmarried 78 (5.7%) 59 (5.7%) 19 (5.5%) 58 (6.0%) 20 (4.9%) PIR <0.001 0.171 ≤ 1.3 421 (30.7%) 265 (25.7%) 156 (45.5%) 286 (29.8%) 135 (32.8%) 1.3-3.5 535 (39.0%) 408 (39.6%) 127 (37.0%) 390 (40.6%) 145 (35.2%) ≥3.5 417 (30.4%) 357 (34.7%) 60 (17.5%) 285 (29.7%) 132 (32.0%) Glycohemoglobin, % 5.80 (5.50, 6.40) 5.80 (5.50, 6.30) 5.90 (5.50, 6.60) 0.056 5.80 (5.50, 6.40) 5.85 (5.50, 6.30) 0.582 Albumin, g/dL 4.20 (4.00, 4.40) 4.20 (4.00, 4.40) 4.20 (4.00, 4.40) 0.051 4.20 (4.00, 4.40) 4.20 (4.00, 4.40) 0.782 ALT, U/L 20.00 (16.00, 26.00) 20.00 (16.00, 26.00) 19.00 (14.00, 24.00) 0.002 20.00 (16.00, 26.00) 19.00 (15.00, 25.50) 0.184 Blood urea nitrogen, mmol/L 5.36 (4.28, 7.14) 5.36 (4.28, 6.78) 5.71 (4.64, 7.14) 0.008 5.36 (4.28, 7.14) 5.36 (4.28, 7.14) 0.885 Serum creatinine, umol/L 84.86 (70.72, 103.43) 83.98 (68.95, 100.78) 90.17 (74.26, 111.38) <0.001 84.86 (70.72, 102.54) 84.86 (69.40, 106.08) 0.914 Serum glucose, mmol/L 5.66 (5.11, 6.61) 5.66 (5.05, 6.61) 5.72 (5.22, 6.77) 0.031 5.66 (5.11, 6.66) 5.66 (5.11, 6.55) 0.916 Serum uric acid, umol/L 345.00 (285.50, 404.50) 345.00 (285.50, 404.50) 345.00 (285.50, 416.40) 0.777 345.00 (291.50, 410.40) 339.00 (285.50, 401.50) 0.284 Serum sodium, mmol/L 140.00 (138.00, 141.00) 140.00 (138.00, 141.00) 140.00 (138.00, 141.00) 0.085 140.00 (138.00, 141.00) 140.00 (138.00, 142.00) 0.092 TG, mmol/L 1.49 (1.02, 2.19) 1.50 (1.02, 2.20) 1.42 (1.01, 2.09) 0.193 1.46 (0.99, 2.19) 1.54 (1.10, 2.18) 0.282 Direct HDL-C, mmol/L 1.32 (1.09, 1.60) 1.32 (1.11, 1.60) 1.29 (1.06, 1.55) 0.063 1.32 (1.09, 1.60) 1.32 (1.11, 1.58) 0.670 Serum vitamin B12, pmol/L 411.10 (288.60, 594.80) 414.00 (291.50, 587.40) 403.70 (265.70, 631.00) 0.571 416.20 (288.60, 586.70) 398.15 (288.20, 635.40) 0.936 RBC folate, nmol/L 1,270.00 (906.00, 1,740.00) 1,290.00 (926.00, 1,750.00) 1,180.00 (859.00, 1,730.00) 0.055 1,280.00 (928.00, 1,740.00) 1,230.00 (878.50, 1,740.00) 0.185 BMI, kg/m 2 0.008 0.048 <25 284 (20.7%) 201 (19.5%) 83 (24.2%) 184 (19.1%) 100 (24.3%) 25-29.9 496 (36.1%) 360 (35.0%) 136 (39.7%) 363 (37.8%) 133 (32.3%) ≥30 593 (43.2%) 469 (45.5%) 124 (36.2%) 414 (43.1%) 179 (43.4%) Smoking 0.084 0.124 Never 657 (47.9%) 483 (46.9%) 174 (50.7%) 472 (49.1%) 185 (44.9%) Former 547 (39.8%) 427 (41.5%) 120 (35.0%) 366 (38.1%) 181 (43.9%) Current 169 (12.3%) 120 (11.7%) 49 (14.3%) 123 (12.8%) 46 (11.2%) Alcohol consumption 0.128 0.926 No 429 (31.2%) 310 (30.1%) 119 (34.7%) 301 (31.3%) 128 (31.1%) Yes 944 (68.8%) 720 (69.9%) 224 (65.3%) 660 (68.7%) 284 (68.9%) Age at hypertension onset 0.074 0.613 <35 118 (8.6%) 80 (7.8%) 38 (11.1%) 85 (8.8%) 33 (8.0%) ≥35 1255 (91.4%) 950 (92.2%) 305 (88.9%) 876 (91.2%) 379 (92.0%) History of diabetes 0.028 0.351 No 986 (71.8%) 756 (73.4%) 230 (67.1%) 683 (71.1%) 303 (73.5%) Yes 387 (28.2%) 274 (26.6%) 113 (32.9%) 278 (28.9%) 109 (26.5%) History of heart failure 0.013 0.886 No 1,252 (91.2%) 951 (92.3%) 301 (87.8%) 877 (91.3%) 375 (91.0%) Yes 121 (8.8%) 79 (7.7%) 42 (12.2%) 84 (8.7%) 37 (9.0%) History of coronary heart disease 0.374 0.737 No 1,217 (88.6%) 918 (89.1%) 299 (87.2%) 850 (88.4%) 367 (89.1%) Yes 156 (11.4%) 112 (10.9%) 44 (12.8%) 111 (11.6%) 45 (10.9%) History of stroke <0.001 0.678 No 1,253 (91.3%) 956 (92.8%) 297 (86.6%) 879 (91.5%) 374 (90.8%) Yes 120 (8.7%) 74 (7.2%) 46 (13.4%) 82 (8.5%) 38 (9.2%) History of chronic bronchitis 0.815 0.194 No 1,263 (92.0%) 949 (92.1%) 314 (91.5%) 890 (92.6%) 373 (90.5%) Yes 110 (8.0%) 81 (7.9%) 29 (8.5%) 71 (7.4%) 39 (9.5%) History of Sleep disorder 0.592 0.395 No 1,183 (86.2%) 884 (85.8%) 299 (87.2%) 833 (86.7%) 350 (85.0%) Yes 190 (13.8%) 146 (14.2%) 44 (12.8%) 128 (13.3%) 62 (15.0%) History of Cancer 0.053 0.309 No 1,080 (78.7%) 797 (77.4%) 283 (82.5%) 763 (79.4%) 317 (76.9%) Yes 293 (21.3%) 233 (22.6%) 60 (17.5%) 198 (20.6%) 95 (23.1%) ALT: Alanine aminotransferase; BMI: body mass index; HDL-C: high-density lipoprotein cholesterol; PIR: poverty income ratio; RBC: folate red blood cell folate; TG: Triglycerides. Table 2. Performance comparison of models. Model Accuracy Prevalence Recall F1-Score MCC AUC of ROC Presicion Specificity FNR FPR LR 0.777 0.269 0.342 0.452 0.359 0.791 0.667 0.937 0.658 0.063 CatBoost 0.760 0.269 0.252 0.361 0.286 0.785 0.636 0.947 0.748 0.053 SVM 0.774 0.269 0.252 0.376 0.336 0.773 0.737 0.967 0.748 0.033 RF 0.762 0.269 0.198 0.310 0.283 0.757 0.710 0.970 0.802 0.030 LGBM 0.752 0.269 0.144 0.239 0.234 0.720 0.696 0.977 0.856 0.023 XGB 0.728 0.269 0.369 0.423 0.254 0.720 0.494 0.860 0.631 0.140 DT 0.767 0.269 0.225 0.342 0.306 0.713 0.714 0.967 0.775 0.033 AUC: area under the curve; CatBoost: categorical boosting; LR: logistic regression; DT: decision tree; LGBM: light gradient boosting machine; MCC: Matthews correlation coefficient; RF: random forest; ROC: receiver operating characteristic; SVM: support vector machine; XGB: extreme gradient boosting. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-6950624","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":492947699,"identity":"af771639-abe1-43a4-bc0a-359e75d2abf6","order_by":0,"name":"Zejing Lin","email":"","orcid":"","institution":"Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zejing","middleName":"","lastName":"Lin","suffix":""},{"id":492947700,"identity":"18a1dc75-486d-44cf-91ac-742f96e48b5c","order_by":1,"name":"Rulan Ma","email":"","orcid":"","institution":"Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Rulan","middleName":"","lastName":"Ma","suffix":""},{"id":492947702,"identity":"489049b5-d060-45e1-bf27-033168d15fa6","order_by":2,"name":"Xing Chen","email":"","orcid":"","institution":"Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xing","middleName":"","lastName":"Chen","suffix":""},{"id":492947703,"identity":"3d6ba66b-d1c1-49c3-a045-522468dc13b3","order_by":3,"name":"Yinzhou Wang","email":"","orcid":"","institution":"Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yinzhou","middleName":"","lastName":"Wang","suffix":""},{"id":492947704,"identity":"97f9428e-edf9-4023-a0de-0072f3820b18","order_by":4,"name":"Xingyong Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxElEQVRIiWNgGAWjYPCC//X9zMyHH5CihZlxZjtbmgFpWjac51GQIEot/4wcw8cFv9iYjQ/zMBgw1NhEE9QiceaMsfHMPh42s8O8Bx4wHEvLbSCkxYC9x0yat0eCx+wwX4IBY8NhIrQw84C0GEgYN/MYSBCnBWQLz48EA6BeIrVInDlWbMzbcCBB4jAwkBOI8Qv/jOSNj3n+HEjg7z98+MGHGhvCWhgYOAwYGNug7ATCykGA/QEDwx/ilI6CUTAKRsEIBQCp5zoeDvhd3QAAAABJRU5ErkJggg==","orcid":"","institution":"Fujian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xingyong","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2025-06-22 16:23:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6950624/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6950624/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88097433,"identity":"4a0aa26e-79d5-476f-aba3-70fcf9ecc90f","added_by":"auto","created_at":"2025-08-01 11:04:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":122117,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart of the study design. NHANES: National Health and Nutrition Examination Survey.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/f02ca4844885fc86135e0e8b.png"},{"id":88098990,"identity":"ae49c3ca-2f39-476c-8971-8bb67cd53dd6","added_by":"auto","created_at":"2025-08-01 11:12:28","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":173278,"visible":true,"origin":"","legend":"\u003cp\u003eLeast Absolute Shrinkage and Selection Operator (LASSO) regression was applied to the training set. \u003cstrong\u003e(A)\u003c/strong\u003e The Lasso coefficient path plot for the 29 variables is shown. \u003cstrong\u003e(B)\u003c/strong\u003e The 10-fold cross-validation curve is presented, with the left dashed line marking lambda.min and the right dashed line marking lambda.1se.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/7cca1cafb9f5e2c83884c871.jpeg"},{"id":88098991,"identity":"d807061e-1d8c-42b3-b9c9-ed01594f757b","added_by":"auto","created_at":"2025-08-01 11:12:28","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":179148,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of seven machine learning (ML) prediction models in the training set \u003cstrong\u003e(A) \u003c/strong\u003eand the test set\u003cstrong\u003e (B)\u003c/strong\u003e. AUC: area under the curve; CatBoost: categorical boosting; LR: logistic regression; DT: decision tree; LGBM: light gradient boosting machine; RF: random forest; SVM: support vector machine; XGB: extreme gradient boosting.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/a9ae46972ae9ea69979d8ad2.png"},{"id":88097439,"identity":"4cdc5ca2-fa9b-4efb-a3ad-852bfc040546","added_by":"auto","created_at":"2025-08-01 11:04:28","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":184859,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curves of seven machine learning (ML) prediction models in the training set\u003cstrong\u003e(A) \u003c/strong\u003eand the test set \u003cstrong\u003e(B)\u003c/strong\u003e. CatBoost: categorical boosting; LR: logistic regression; DT: decision tree; LGBM: light gradient boosting machine; RF: random forest; SVM: support vector machine; XGB: extreme gradient boosting.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/9ed023a807ac6789877c739b.jpeg"},{"id":88097437,"identity":"199c3c1a-7d4a-4385-9082-e620d7c57d4c","added_by":"auto","created_at":"2025-08-01 11:04:28","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":115730,"visible":true,"origin":"","legend":"\u003cp\u003eThe decision curve analysis curves of seven machine learning (ML) prediction models in the training set\u003cstrong\u003e (A)\u003c/strong\u003e and the test set \u003cstrong\u003e(B)\u003c/strong\u003e. CatBoost: categorical boosting; LR: logistic regression; DT: decision tree; LGBM: light gradient boosting machine; RF: random forest; SVM: support vector machine; XGB: extreme gradient boosting.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/85b71c4742de2cfdd24fd318.png"},{"id":88100136,"identity":"7e04ea64-fcfc-4d9a-9d08-377f02f09e25","added_by":"auto","created_at":"2025-08-01 11:20:28","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":509653,"visible":true,"origin":"","legend":"\u003cp\u003eThe final prediction model was interpreted using the SHapley Additive exPlanation (SHAP) algorithm.\u003cstrong\u003e (A) \u003c/strong\u003eSHAP summary bar plot. \u003cstrong\u003e(B)\u003c/strong\u003e SHAP summary dot plot. \u003cstrong\u003e(C)\u003c/strong\u003e SHAP waterfall plot. SHAP: SHapley Additive explanation.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/4028467d3f405db720d732bb.png"},{"id":88097444,"identity":"95295c48-d2aa-456f-a069-f301cf39293a","added_by":"auto","created_at":"2025-08-01 11:04:28","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":163876,"visible":true,"origin":"","legend":"\u003cp\u003eThe web-based calculator predicts cognitive impairment in elderly hypertensive patients using this model. By entering demographic data, laboratory test results, and medical history—such as age, gender, race, education status, poverty income ratio (PIR), body mass index (BMI), serum creatinine, serum sodium, diabetes, stroke, folate red blood cell folate (RBC) folate, and age at hypertension onset—the risk of cognitive impairment can be estimated.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/53ff8bba83ae9e56e65f6eb5.jpeg"},{"id":89840494,"identity":"385bbbba-242b-4d3a-a403-2e8890652dde","added_by":"auto","created_at":"2025-08-25 15:17:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2338510,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6950624/v1/e5fed89c-23a5-4ac4-af83-3d42e90be2be.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe 2021 Global Status Report on Dementia by the World Health Organization estimated that fifty-five million individuals worldwide were affected by dementia in 2019, with this number projected to increase to seventy-eight million by 2030\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. Despite this alarming trend, a recent report in The Lancet suggested that approximately forty percent of dementia cases could theoretically be prevented or postponed through effective control of risk factors\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. The etiology of cognitive impairment is complex; however, autopsy analyses have shown that in more than half of clinically diagnosed dementia cases, vascular factors, particularly hypertension, are the primary or exclusive cause.\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. Chronic hypertension can gradually damage the structure and function of cerebral blood vessels, resulting in the breakdown of the blood-brain barrier due to inflammatory pathological mechanisms.\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. This, in turn, may result in amyloid deposition and the development of Alzheimer\u0026rsquo;s disease (AD) pathology\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. Additionally, hypertension-induced damage to small cerebral vessels, ranging from capillaries to arterioles, referred to as cerebral small vessel disease (CSVD), accounts for forty-five percent of clinically diagnosed cases of vascular cognitive impairment\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. Furthermore, it has been established that elevated blood pressure during midlife increases the risk of developing dementia in older age \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e, independent of genetic risk factors for dementia \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. Accordingly, it is crucial to examine the contributing factors to cognitive decline in individuals with hypertension, develop clinical prediction models, and further explore the underlying mechanisms. Such efforts could provide more effective decision-making support for the prevention and intervention of early cognitive impairment.\u003c/p\u003e\u003cp\u003eTo date, studies on predictive models for evaluating the likelihood of cognitive decline in individuals with hypertension have been published \u003csup\u003e[\u003cspan additionalcitationids=\"CR9 CR10 CR11 CR12\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u0026ndash;[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e. Unfortunately, the target populations in these studies have predominantly been from China, and most models were developed using a single method, such as logistic regression, which limits their practical utility. As clinical data have become increasingly large and complex, artificial intelligence (AI) has been increasingly adopted to address these challenges. Among the subfields of AI, machine learning (ML) has achieved remarkable success in constructing predictive models and is now widely applied to the diagnosis, prognosis, and treatment selection of various diseases\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e],[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Therefore, by comparing multiple ML algorithms and selecting the optimal model for clinical prediction, clinicians can be better supported in identifying patients at risk of cognitive impairment, thereby facilitating improved prevention and intervention strategies.\u003c/p\u003e\u003cp\u003eIn this study, predictive models for identifying cognitive impairment in older adults with hypertension were developed and validated using seven ML algorithms, utilizing data derived from the National Health and Nutrition Examination Survey (NHANES) database. These models address a significant gap in research related to cognitive impairment prediction in hypertensive patients within the U.S. population. Furthermore, a web-based calculator was constructed on an online platform to facilitate the clinical application of the model. This research aims to enhance the timely and rapid diagnosis of cognitive impairment in high-risk populations, thereby supporting the formulation of individualized prevention approaches and therapeutic interventions to reduce the socioeconomic impact on families and the broader community.\u003c/p\u003e"},{"header":"2. MATERIALS AND METHODS","content":"\u003cp\u003e\u003cstrong\u003e2.1 Data Collection and Study Population\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data was obtained from NHANES, a nationally representative database made available by the U.S. Centers for Disease Control and Prevention. The NHANES protocol underwent review by the ethics committee of the National Center for Health Statistics (https://www.cdc.gov/nchs/nhanes/about/erb.html), and documented informed consent was secured from all subjects. Because each survey cycle of NHANES is independent and the participants in each cycle are different, it creates good conditions for the development and verification of clinical prediction models.\u003c/p\u003e\n\u003cp\u003eThe study population comprised 19,931 participants were recruited from two NHANES survey cycles (2011-2012 and 2013-2014). Participants were excluded according to the subsequent criteria: (a) participants younger than 60 years; (b) participants missing cognitive function test questionnaires; (c) participants who were determined not to have hypertension based on the test questionnaire; (d) participants with missing data for other candidate predictor variables. After the screening process, a final cohort of 1,373 older adults with hypertension was enrolled. The screening procedure is depicted in Fig. 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Definition of hypertension\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDiagnosis of hypertension was assigned based on either: self-reported use of antihypertensive medications, or confirmation by a healthcare provider of the condition.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Cognitive assessment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAmong the participants we included from the two survey cycles, cognition was evaluated employing the subsequent three assessment tools: (1) the Consortium to Establish a Registry for Alzheimer\u0026rsquo;s Disease (CERAD)\u003csup\u003e[16]\u003c/sup\u003e, primarily used to evaluate immediate and delayed learning ability of new verbal information (memory domain). It comprises three consecutive learning trials and a delayed recall trial, with cumulative scores ranging from 0 to 40. (2) the Animal Fluency Test (AFT)\u003csup\u003e[17]\u003c/sup\u003e, which probes categorical verbal fluency, a component of executive function, with a maximum score of 10. (3) the Digit Symbol Substitution Test (DSST)\u003csup\u003e[18]\u003c/sup\u003e, a performance module from the Wechsler Adult Intelligence Scale (WAIS III), chiefly employed to gauge processing speed, sustained attention, and working memory, with a maximum score of 10.\u003c/p\u003e\n\u003cp\u003eCurrently, there are no established cutoff values for the tests mentioned above to classify participants as cognitively impaired or normal. Therefore, we referred to previous studies\u003csup\u003e[19]\u003c/sup\u003e and used the average z-scores from the three assessment tools to obtain a composite z-score. This approach was used to further analyze the overall cognitive performance across the three tests. The lowest quartile of the composite z-score served as the demarcation point. Individuals scoring below this value were stratified into the cognitive impairment group, while others comprised the non-cognitive impairment group. The z-score was derived as\u003csup\u003e[19]\u003c/sup\u003e: z-score \u0026nbsp;, where \u0026nbsp; denotes an individual\u0026rsquo;s test result, and \u0026nbsp; and \u0026nbsp; represent the mean and standard deviation of scores across the entire sample, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4 Determination of Candidate Predictors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe collected a range of widely recognized and commonly used indicators from the NHANES database as candidate predictors to include in the final predictive model. The candidate predictors consist of three components: demographic data, laboratory test results, and medical history. Demographic data include age, gender, poverty income ratio (PIR), race, education status, and marital status. Laboratory test data include standard blood biochemical tests, red blood cell folate (RBC folate), and serum vitamin B12 levels. Standard blood biochemical tests include alanine aminotransferase (ALT), glycohemoglobin, albumin, blood urea nitrogen (BUN), blood glucose, serum creatinine, serum uric acid, triglycerides (TG), serum sodium, and high-density lipoprotein cholesterol (HDL-C). Medical history data include smoking, body mass index (BMI), alcohol consumption, age at hypertension onset, heart failure, coronary heart disease, stroke, diabetes, chronic bronchitis, sleep disorders, and cancer. Smoking status was categorized: Never-smokers (\u0026lt;100 lifetime cigarettes), Former smokers (\u0026gt;100 cigarettes, no current use), and Current smokers (all others). Participants reporting \u0026ge;12 alcoholic drinks/year were designated as alcohol consumers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5. Statistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5.1 Data Cleaning and Variable Selection\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;After excluding patients with missing data according to the inclusion and exclusion standards detailed in Fig. 1, hypertensive patients aged \u0026ge;60 years with complete cognitive assessments and candidate predictor data were included. Continuous variables that did not follow a normal distribution were summarized as medians with interquartile ranges and analyzed using non-parametric methods. Categorical data were reported as counts and percentages and compared using chi-square tests. The dataset was randomly split into a training subset (70%) and a validation subset (30%) based on the commonly adopted 70/30 rule. Within the training cohort, candidate predictors for the final model were identified using the Least Absolute Shrinkage and Selection Operator (LASSO) regression technique.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5.2 Model Construction and Evaluation\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;In ML, four primary paradigms exist: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Given that the objective of this study was to categorize elderly individuals with hypertension into groups with either intact cognitive function or cognitive impairment, supervised learning techniques were deemed most suitable for this classification problem\u003csup\u003e[20]\u003c/sup\u003e. To develop predictive models for cognitive impairment in this population, seven ML algorithms were implemented: logistic regression (LR), categorical boosting (CatBoost), extreme gradient boosting (XGB), decision tree (DT), random forest (RF), light gradient boosting machine (LGBM), and support vector machine (SVM). The models were evaluated using several commonly used metrics in the test set, including the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, accuracy, recall, F1-score, precision, Matthews correlation coefficient (MCC), and specificity. Model calibration was examined through calibration plots, while decision curve analysis (DCA) was utilized to determine the clinical utility of the models across various decision thresholds. The optimal model was identified based on comprehensive evaluation of these metrics in the test set.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5.3 Model Interpretation and Web Calculator Construction\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;SHAP (SHapley Additive exPlanation) is a post-hoc model interpretation technique for ML, grounded in game theory\u003csup\u003e[21]\u003c/sup\u003e. It assesses the significance of individual features by determining their contribution to the model\u0026apos;s output, thus enhancing the transparency and interpretability of the model. Finally, based on the best-performing prediction model selected, a web-based calculator was constructed on the Shiny application platform. By inputting the patient\u0026apos;s clinical data, this tool can assist healthcare providers and individuals in assessing the risk of cognitive impairment. All statistical procedures were performed using R software (version 4.3.1) and DecisionLinnc1.0 software\u003csup\u003e[22]\u003c/sup\u003e, with \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05 considered statistically significant.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cp\u003e\u003cstrong\u003e3.1 Baseline Characteristics of Study Participants\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study included 1,373 older adults with hypertension who had complete cognitive evaluations and relevant predictor data. An overview of baseline characteristics is presented in Table 1. Most variables displayed similar distributions between the training and testing subsets, as indicated by \u003cem\u003ep\u003c/em\u003e -values predominantly exceeding 0.05. Participants were categorized into two cohorts based on cognitive status: the cognitive impairment group and the non-cognitive impairment group. Notable differences were observed between the two groups in demographic data, including age, gender, race, education status, and PIR (all \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). For laboratory indicators, the cognitive impairment group had higher levels of ALT, BUN, blood creatinine, and blood glucose compared to their counterparts without impairment (all \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). Additionally, significant differences were found in medical history, including BMI, diabetes, heart failure, and stroke (all \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Variable Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLasso regression was utilized to select independent variables, and ten-fold cross-validation was conducted to determine the optimal regularization parameter \u0026lambda; (Fig. 2). With cognitive impairment as the dependent variable, a total of twenty-nine predictor variables from three domains\u0026mdash;demographic data, laboratory tests data, and medical history data\u0026mdash;were included in the Lasso regression. The \u0026lambda; values corresponding to lambda.min and lambda.1se, as well as the number of selected predictor variables, were obtained. To improve the predictive accuracy of the model, \u0026lambda; value at lambda.min was selected, resulting in a \u0026lambda; of 0.0150 (log(lambda.min) = \u0026minus;4.202). This approach yielded 12 variables with non-zero coefficients, including age, gender, BMI, race, PIR, education status, serum creatinine, serum sodium, diabetes, stroke, RBC folate, and age at hypertension onset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Construction and Performance Evaluation of Predictive Models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e12 variables selected by Lasso regression were used to construct predictive models with seven ML algorithms. Fig. 3 illustrates the discriminatory performance of the seven models on the ROC curves for the training and test datasets. As shown in Fig. 3B, on the test set, LR achieved the highest AUC (AUC = 0.791), indicating the best discriminatory ability, followed by CatBoost (AUC = 0.785), SVM (AUC = 0.773), RF (AUC = 0.757), LGBM (AUC = 0.720), XGB (AUC = 0.720), and DT (AUC = 0.713). Moreover, LR also had higher accuracy (0.777), F1-Score (0.452) and MCC (0.359) compared to the other six models (Table 2). Additionally, as shown in Table 2, XGB had the highest recall (0.369) and false positive rate (FPR) (0.140), SVM had the highest precision (0.737), and LGBM exhibited the highest specificity (0.977) and false negative rate (FNR) (0.856). Fig. 4A and 4B display the calibration curves for the training and test sets, respectively. Based on the calibration curve for the test set (Fig. 3B), LR shows impressive calibration performance compared to the other six models. Fig. 5A and 4B present the DCA curves for the training and test sets, respectively. As shown in Fig. 5B, under a defined probability threshold, LR achieved the highest net benefit among the seven machine learning models, suggesting its greater clinical applicability. Considering its overall superiority in predictive accuracy, calibration reliability, and decision utility, LR was determined to be the most effective model for forecasting cognitive impairment in older adults with hypertension.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 Model Interpretation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Fig. 6, the SHAP algorithm was applied to interpret the output of the final LR model by quantifying each feature\u0026rsquo;s influence on the prediction results. In Fig. 6A, the SHAP summary bar plot was employed to evaluate feature importance based on the mean absolute SHAP values. Predictive variables were ranked in descending order according to their contribution to cognitive impairment risk, with the top five features being education status, age, PIR, BMI, and diabetes. Additionally, the SHAP summary dot plot in Fig. 6B visually represented both the direction and magnitude of each variable\u0026rsquo;s effect on the model\u0026rsquo;s predictions. The vertical axis represents SHAP values of zero, with variables to the right of the line, colored in purple, indicating a positive contribution to the prediction, while those colored in yellow indicate a negative contribution. The variables were also ranked in descending order of their contribution. As a result, age and diabetes were found to be positively correlated with the occurrence of cognitive impairment, meaning that older age and diabetes in hypertensive patients increase the probability of cognitive impairment. Conversely, education status, PIR, and BMI were found to be negatively correlated with cognitive impairment, indicating that higher education status, PIR, and BMI levels in hypertensive patients reduce the risk of cognitive impairment.\u003c/p\u003e\n\u003cp\u003eFurthermore, as illustrated in Fig. 6C, the SHAP waterfall plot is used to illustrate how specific features affect the prediction output at the individual level. The exact values of each feature, along with their associated SHAP values, indicate whether the feature contributes positively or negatively to the final prediction. Education status (college or above) and age (60\u0026ndash;69 years) made significant positive contributions to the prediction outcome, with SHAP values of -0.104 and -0.0593, respectively, while PIR (\u0026le;1.3) and race (Other Race) contributed negatively with SHAP values of +0.0382 and +0.029, respectively. Other features, such as RBC folate, gender, creatinine, and diabetes, also contributed to the prediction result to varying degrees. The process by which the machine learning model generates results for an individual patient is clearly demonstrated in the SHAP waterfall plot through the cumulative SHAP values. Therefore, the use of the SHAP method provides a more transparent and interpretable approach, significantly improving insight into the model\u0026rsquo;s underlying decision-making process.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.5 Development of a Web-Based Calculator\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Fig. 7, a web-based calculator was developed based on the LR prediction model to facilitate its clinical application. By entering the actual values of the 12 required variables, the calculator allows for the estimation of cognitive impairment risk in individual older adults with hypertension. The web-based calculator is available via the following link: https://cognitiveimpairment.shinyapps.io/cognitiveimpairment/.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eAs previously mentioned, studies on predictive models for evaluating the likelihood of cognitive decline in individuals with hypertension has been documented\u003csup\u003e[8]\u0026ndash;[13]\u003c/sup\u003e. However, the target populations and modeling approaches in these studies have been relatively limited, resulting in constrained clinical applicability. To the best of our understanding, this represents the first investigation to construct a cognitive impairment prediction model using multiple ML algorithms based on data from elderly hypertensive individuals in the United States. This study analyzed information from the 2011-2012 and 2013-2014 cycles of the NHANES, which incorporated cognitive function evaluations. 12 predictive factors were identified from three domains: demographic data, laboratory tests data, and medical history data. Based on these 12 predictive factors, we developed and validated prediction models using seven ML algorithms. Ultimately, the LR model was identified as the optimal clinical prediction model, demonstrating the highest discrimination, accuracy, F1-score, and clinical net benefit in the validation set. Finally, the SHAP method was employed to elucidate the results of the final predictive model, and a web-based calculator was developed to facilitate its implementation in clinical practice.\u003c/p\u003e\n\u003cp\u003eIn our study, we employed the SHAP framework to evaluate the contribution of the 12 variables included in our model. The top five variables ranked by importance were education status, age, PIR, BMI, and diabetes. Our findings indicate that low education level, advanced age, low PIR, low BMI, and diabetes are significant risk factors contributing to cognitive decline in elderly individuals with hypertension. Numerous studies have demonstrated that higher educational attainment is linked to a decreased likelihood of developing cognitive impairment\u003csup\u003e[23],[24]\u003c/sup\u003e. Bielak et al. proposed that education benefits individuals throughout their lifetime through a mechanism known as \u0026quot;preserved differentiation\u0026quot;\u003csup\u003e[25]\u003c/sup\u003e. Simply put, if the rate of cognitive decline is uniform across individuals, differences in peak cognitive function in early adulthood would lead to variations in when cognitive function falls below a critical threshold. In other words, higher levels of formal education may reduce the risk of cognitive impairment by enhancing an individual\u0026apos;s cognitive reserve. This concept might help explain why some individuals with severe neurodegenerative diseases, such as AD, still exhibit only mild cognitive decline\u003csup\u003e[26]\u003c/sup\u003e. Aging is a well-established risk factor for cognitive decline and a major contributor to the progression of neurological disorders. Research shows that cognitive decline accelerates with age, with detectable signs appearing as early as middle age (45\u0026ndash;55 years).\u003csup\u003e[27]\u003c/sup\u003e. Furthermore, approximately 40% of adults aged 65 years or above exhibit some degree of memory impairment\u003csup\u003e[28]\u003c/sup\u003e. In hypertensive patients, prolonged exposure to high blood pressure with advancing age can result in chronic inflammation, oxidative stress, and cerebrovascular damage\u003csup\u003e[29]\u003c/sup\u003e. This not only increases the risk of stroke-related cognitive impairment but also leads to subclinical vascular brain damage and white matter integrity deterioration,\u0026nbsp;collectively contribute to progressive cognitive decline\u003csup\u003e[30]\u003c/sup\u003e. Moreover, our study indicates that individuals with lower socioeconomic status are at greater risk for cognitive impairment. Similarly, Zeki et al. found that prolonged exposure to low income over two decades was significantly correlated with diminished cognitive performance, indicating that socioeconomic conditions during both early and later stages of life play a crucial role in shaping cognitive health in old age\u003csup\u003e[31]\u003c/sup\u003e. Additionally, individuals experiencing long-term low-income exposure often have poorer socioeconomic status and engage in more unhealthy behaviors, which further exacerbates cognitive impairment\u003csup\u003e[32]\u003c/sup\u003e. A meta-analysis incorporating 39 prospective studies revealed a heightened overall risk of cognitive decline and dementia among individuals with lower socioeconomic status. Subgroup analyses further showed that low-income and low-education groups faced a significantly greater risk of developing cognitive deficits and dementia\u003csup\u003e[33]\u003c/sup\u003e. Interestingly, our study suggests that higher BMI in later life serves as a protective factor for cognitive function in patients with hypertension. The relationship between BMI and cognitive performance is complex. While extensive research supports a link between higher midlife BMI and worsened cognitive outcomes in old age\u003csup\u003e[34],[35]\u003c/sup\u003e, the impact of BMI in old age remains inconclusive. In our study, as well as in other investigations, lower BMI in old age was linked to worse cognitive function\u003csup\u003e[34],[36]\u003c/sup\u003e, though some studies have reported no association or even the opposite relationship\u003csup\u003e[36],[37]\u003c/sup\u003e.Lastly, our study suggests that elderly hypertensive patients with diabetes face an elevated likelihood of experiencing cognitive decline. Previous studies indicate that approximately 20% of individuals with diabetes develop mild cognitive impairment (MCI). Furthermore, those diagnosed with both MCI and diabetes have a 1.53 times greater risk of advancing to dementia compared to individuals without diabetes\u003csup\u003e[38]\u003c/sup\u003e. A meta-analysis also identified diabetes as the only independent predictor of MCI conversion to dementia\u003csup\u003e[39]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eApart from stroke and age at hypertension onset, which have been briefly analyzed earlier, three additional modifiable risk factors among the remaining five predictors include serum creatinine, RBC folate, and serum sodium levels. Our study indicates high serum creatinine, low RBC folate, and high serum sodium as contributing factors to cognitive decline in older hypertensive individuals. Serum creatinine serves as a standard indicator of renal function, and it is well known that individuals receiving hemodialysis for end-stage renal disease (ESRD) frequently suffer from cognitive impairment. Consistent with our findings, a 10-year prospective study in the United Kingdom demonstrated an association between poor kidney function and dementia, independent of major cardiometabolic diseases and stroke\u003csup\u003e[40]\u003c/sup\u003e. However, the relationship between serum creatinine and cognitive impairment may be nonlinear, as low creatinine levels have also been linked to poorer cognitive outcomes. A recent Finnish study, which followed young adults with normal serum creatinine levels for 10 years, found that in men, persistently high creatinine levels were associated with better memory and learning function in midlife compared to lower creatinine levels\u003csup\u003e[41]\u003c/sup\u003e. Thus, the precise relationship between serum creatinine and cognitive function still requires further investigation. Mounting evidence suggests that folate, an essential vitamin, plays a pivotal role in the pathogenesis of AD. A prospective cohort study conducted in China among elderly individuals indicated that lower folate concentrations were linked to a heightened risk of mild MCI\u003csup\u003e[42]\u003c/sup\u003e, aligning with our findings and other previous studies\u003csup\u003e[43],[44]\u003c/sup\u003e. As the primary extracellular cation, sodium ions regulate fluid-electrolyte balance, circulating blood volume, and osmotic stability\u003csup\u003e[45]\u003c/sup\u003e. Abnormal serum sodium levels can disrupt intracellular and extracellular osmotic equilibrium, leading to cerebral blood volume disturbances and neuronal dysfunction. Theoretically, both low and high serum sodium levels may increase the risk of cognitive dysfunction. Research indicates that reduced serum sodium may cause astrocyte swelling and trigger the release of excitatory neurotransmitters (e.g., glutamate), which are closely linked to neuronal damage\u003csup\u003e[46]\u003c/sup\u003e. Conversely, higher serum sodium levels have been correlated with increased AD pathology, reduced hippocampal volume, and cognitive decline\u003csup\u003e[47]\u003c/sup\u003e, which aligns with our observations. However, the association between serum sodium levels and cognitive performance warrants further investigation.\u003c/p\u003e\n\u003cp\u003eDespite this study utilized a large, nationally representative sample of older adults in the United States and benefited from the methodological strengths and rigorous quality assurance of NHANES, several limitations should be acknowledged. First, the NHANES dataset is cross-sectional in nature, which precludes the ability to evaluate the longitudinal risk of developing cognitive impairment. Second, our predictive model was developed based on data from United States patients and validated using internal validation only, which may limit its generalizability and external applicability to other populations. Lastly, due to the extensive number of variables in the NHANES database and our exclusion of variables with missing data to enhance model reliability, several variables potentially associated with cognitive impairment were not included. This omission may have influenced the final results. Future research should consider incorporating longitudinal data, including a broader population, and conducting external validation of the model to enhance its generalizability and robustness.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn summary, we effectively established an interpretable ML model that can assist clinicians in the early and rapid detection of cognitive impairment risk among elderly hypertensive patients, which may ultimately help improve patient outcomes.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to acknowledge the invaluable contributions of both the participants and staff of the NHANES, whose efforts were essential to the success of this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eZejing Lin: Conceptualization, Methodology, Formal analysis, Writing \u0026ndash; original draft. Rulan Ma: Writing \u0026ndash; review \u0026amp; editing, Data curation, Software, Visualization. Xing Chen: Investigation, Validation, Visualization. Yinzhou Wang: Project administration, supervision, writing- review and editing. Xingyong Chen: Conceptualization, Writing \u0026ndash; review \u0026amp; editing, Supervision, Project administration.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eF\u003c/strong\u003e\u003cstrong\u003eunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by National Natural Science Foundation of China (81771250); Joint Funds for the innovation of science and Technology, Fujian province (2023Y9286); Fujian Provincial Natural Science Foundation (2020R1011004,2021J01374,2024J011004); Fujian Research and Training Grants for Young and Middle-aged Leaders in Healthcare (2022).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data utilized in this study are available in the NHANES database (https://wwwn.cdc.gov/nchs/nhanes/default.aspx). Additional datasets analyzed or referenced in the present research can be obtained from the corresponding authors upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe NHANES dataset is freely available for public access through its official website.\u0026nbsp;The NHANES study protocol received approval from the ethics review committee of the U.S. Centers for Disease Control and Prevention, and all subjects provided written informed consent prior to participation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that there are no conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWorld Health Organization. Global status report on the public health response to dementia. 2021. Available from: https://digitalcommons.fiu.edu/srhreports/health/health/65.\u003c/li\u003e\n\u003cli\u003eLivingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. \u003cem\u003eLancet\u003c/em\u003e. 2020;396(10248):413\u0026ndash;446.\u003c/li\u003e\n\u003cli\u003eAzarpazhooh MR, Avan A, Cipriano LE, et al. Concomitant vascular and neurodegenerative pathologies double the risk of dementia. \u003cem\u003eAlzheimers Dement\u003c/em\u003e. 2018;14(2):148\u0026ndash;156. \u003c/li\u003e\n\u003cli\u003eSantisteban MM, Iadecola C, Carnevale D. Hypertension, neurovascular dysfunction, and cognitive impairment. \u003cem\u003eHypertension\u003c/em\u003e. 2023;80(1):22\u0026ndash;34. \u003c/li\u003e\n\u003cli\u003ePacella J, Lembo G, Carnevale L. A translational perspective on the interplay between hypertension, inflammation and cognitive impairment. \u003cem\u003eCan J Cardiol\u003c/em\u003e. 2024;40(12):2368\u0026ndash;2377. \u003c/li\u003e\n\u003cli\u003eMcGrath ER, Beiser AS, DeCarli C, et al. Blood pressure from mid- to late life and risk of incident dementia. \u003cem\u003eNeurology\u003c/em\u003e. 2017;89(24):2447\u0026ndash;2454.\u003c/li\u003e\n\u003cli\u003eLittlejohns TJ, Collister JA, Liu X, et al. Hypertension, a dementia polygenic risk score, APOE genotype, and incident dementia. \u003cem\u003eAlzheimers Dement\u003c/em\u003e. 2023;19(2):467\u0026ndash;476. \u003c/li\u003e\n\u003cli\u003eZhong X, Yu J, Jiang F, et al. A risk prediction model based on machine learning for early cognitive impairment in hypertension: development and validation study. \u003cem\u003eFront Public Health\u003c/em\u003e. 2023;11:1143019. \u003c/li\u003e\n\u003cli\u003eCerezo GH, Fern\u0026aacute;ndez RA, Enders JE, et al. Predicting cognitive function and dementia risk in patients with hypertension. \u003cem\u003eHypertens Res\u003c/em\u003e. 2024;47(6):1728\u0026ndash;1734.\u003c/li\u003e\n\u003cli\u003eJingyu L, Wen D, Liping Z, et al. A nomogram for predicting mild cognitive impairment in older adults with hypertension. \u003cem\u003eBMC Neurol\u003c/em\u003e. 2023;23(1):363. \u003c/li\u003e\n\u003cli\u003eQu L, Dong Z, Ma S, et al. Gut microbiome signatures are predictive of cognitive impairment in hypertension patients\u0026mdash;a cohort study. \u003cem\u003eFront Microbiol\u003c/em\u003e. 2022;13:841614. \u003c/li\u003e\n\u003cli\u003eLi Y, Xin J, Fang S, et al. Development and validation of a predictive model for early identification of cognitive impairment risk in community-based hypertensive patients. \u003cem\u003eJ Appl Gerontol\u003c/em\u003e. 2024;43(12):1867\u0026ndash;1877. \u003c/li\u003e\n\u003cli\u003eJiang X, Huo M, Yu Q, et al. Analysis of factors affecting subjective cognitive decline in elderly hypertensive patients and the construction of a risk prediction model. \u003cem\u003eGeriatr Nurs\u003c/em\u003e. 2025;61:302\u0026ndash;310. \u003c/li\u003e\n\u003cli\u003eGreener JG, Kandathil SM, Moffat L, et al. A guide to machine learning for biologists. \u003cem\u003eNat Rev Mol Cell Biol\u003c/em\u003e. 2022;23(1):40\u0026ndash;55. \u003c/li\u003e\n\u003cli\u003eTran KA, Kondrashova O, Bradley A, et al. Deep learning in cancer diagnosis, prognosis and treatment selection. \u003cem\u003eGenome Med\u003c/em\u003e. 2021;13(1):152. \u003c/li\u003e\n\u003cli\u003eMorris JC, Heyman A, Mohs RC, et al. The consortium to establish a registry for Alzheimer\u0026rsquo;s disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer\u0026rsquo;s disease. \u003cem\u003eNeurology\u003c/em\u003e. 1989;39(9):1159\u0026ndash;1165. \u003c/li\u003e\n\u003cli\u003eCarone DA. Strauss E, Sherman EMS, Spreen O. A compendium of neuropsychological tests: administration, norms, and commentary (3rd ed.). \u003cem\u003eAppl Neuropsychol\u003c/em\u003e. 2007;14(1):62\u0026ndash;63. \u003c/li\u003e\n\u003cli\u003eRyan JJ, Lopez SJ. Wechsler adult intelligence scale-III. In: Dorfman WI, Hersen M, eds. \u003cem\u003eUnderstanding Psychological\u003c/em\u003e Assessment. Springer US; 2001:19\u0026ndash;42.\u003c/li\u003e\n\u003cli\u003eShi Y, Wang H, Zhu Z, et al. Association between exposure to phenols and parabens and cognitive function in older adults in the United States: a cross-sectional study. \u003cem\u003eSci Total Environ\u003c/em\u003e. 2023;858(Pt 3):160129. \u003c/li\u003e\n\u003cli\u003eChoi RY, Coyner AS, Kalpathy-Cramer J, et al. Introduction to machine learning, neural networks, and deep learning. \u003cem\u003eTransl Vis Sci Technol\u003c/em\u003e. 2020;9(2):14.\u003c/li\u003e\n\u003cli\u003eHou F, Zhu Y, Zhao H, et al. Development and validation of an interpretable machine learning model for predicting the risk of distant metastasis in papillary thyroid cancer: a multicenter study. \u003cem\u003eEClinicalMedicine\u003c/em\u003e. 2024;77:102913. \u003c/li\u003e\n\u003cli\u003eDecisionLinnc Core Team. DecisionLinnc software, version 1.0. 2023. Available from: https://www.statsape.com/.\u003c/li\u003e\n\u003cli\u003eStern Y, Arenaza-Urquijo EM, Bartr\u0026eacute;s-Faz D, et al. Whitepaper: defining and investigating cognitive reserve, brain reserve, and brain maintenance. \u003cem\u003eAlzheimers Dement\u003c/em\u003e. 2020;16(9):1305\u0026ndash;1311. \u003c/li\u003e\n\u003cli\u003eL\u0026ouml;vd\u0026eacute;n M, Fratiglioni L, Glymour MM, et al. Education and cognitive functioning across the life span. \u003cem\u003ePsychol Sci Public Interest\u003c/em\u003e. 2020;21(1):6\u0026ndash;41.\u003c/li\u003e\n\u003cli\u003eBielak AAM, Cherbuin N, Bunce D, et al. Preserved differentiation between physical activity and cognitive performance across young, middle, and older adulthood over 8 years. \u003cem\u003eJ Gerontol B Psychol Sci Soc Sci\u003c/em\u003e. 2014;69(4):523\u0026ndash;532. \u003c/li\u003e\n\u003cli\u003eZhong T, Li S, Liu P, et al. The impact of education and occupation on cognitive impairment: a cross-sectional study in China. \u003cem\u003eFront Aging Neurosci\u003c/em\u003e. 2024;16:1435626. \u003c/li\u003e\n\u003cli\u003eKohler IV, K\u0026auml;mpfen F, Bandawe C, et al. Cognition and cognitive changes in a low-income sub-Saharan African aging population. Zuelsdorff M, ed. \u003cem\u003eJ Alzheimers Dis\u003c/em\u003e. 2023;95(1):195\u0026ndash;212. \u003c/li\u003e\n\u003cli\u003eBrito DVC, Esteves F, Rajado AT, et al. Assessing cognitive decline in the aging brain: lessons from rodent and human studies. \u003cem\u003eNPJ Aging\u003c/em\u003e. 2023;9(1):23. \u003c/li\u003e\n\u003cli\u003eColeman CG, Wang G, Faraco G, et al. Membrane trafficking of NADPH oxidase p47(phox) in paraventricular hypothalamic neurons parallels local free radical production in angiotensin II slow-pressor hypertension. \u003cem\u003eJ Neurosci\u003c/em\u003e. 2013;33(10):4308\u0026ndash;4316.\u003c/li\u003e\n\u003cli\u003eLevine DA, Galecki AT, Langa KM, et al. Blood pressure and cognitive decline over 8 years in middle-aged and older Black and White Americans. \u003cem\u003eHypertension\u003c/em\u003e. 2019;73(2):310\u0026ndash;318. \u003c/li\u003e\n\u003cli\u003eZeki Al Hazzouri A, Elfassy T, Sidney S, et al. Sustained economic hardship and cognitive function. \u003cem\u003eAm J Prev Med\u003c/em\u003e. 2017;52(1):1\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003eLynch JW, Kaplan GA, Salonen JT. Why do poor people behave poorly? Variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse. \u003cem\u003eSoc Sci Med\u003c/em\u003e. 1997;44(6):809\u0026ndash;819. \u003c/li\u003e\n\u003cli\u003eWang AY, Hu HY, Ou YN, et al. Socioeconomic status and risks of cognitive impairment and dementia: a systematic review and meta-analysis of 39 prospective studies. \u003cem\u003eJ Prev Alzheimers Dis\u003c/em\u003e. 2023;10(1):83\u0026ndash;94. \u003c/li\u003e\n\u003cli\u003ePedditzi E, Peters R, Beckett N. The risk of overweight/obesity in mid-life and late life for the development of dementia: a systematic review and meta-analysis of longitudinal studies. \u003cem\u003eAge Ageing\u003c/em\u003e. 2016;45(1):14\u0026ndash;21. \u003c/li\u003e\n\u003cli\u003eRavona-Springer R, Schnaider-Beeri M, Goldbourt U. Body weight variability in midlife and risk for dementia in old age. \u003cem\u003eNeurology\u003c/em\u003e. 2013;80(18):1677\u0026ndash;1683. \u003c/li\u003e\n\u003cli\u003eWotton CJ, Goldacre MJ. Age at obesity and association with subsequent dementia: record linkage study. \u003cem\u003ePostgrad Med J\u003c/em\u003e. 2014;90(1068):547\u0026ndash;551.\u003c/li\u003e\n\u003cli\u003eSingh-Manoux A, Dugravot A, Shipley M, et al. Obesity trajectories and risk of dementia: 28 years of follow-up in the Whitehall II Study. \u003cem\u003eAlzheimers Dement\u003c/em\u003e. 2018;14(2):178\u0026ndash;186. \u003c/li\u003e\n\u003cli\u003eDing X, Yin L, Zhang L, et al. Diabetes accelerates Alzheimer\u0026rsquo;s disease progression in the first year post mild cognitive impairment diagnosis. \u003cem\u003eAlzheimers Dement\u003c/em\u003e. 2024;20(7):4583\u0026ndash;4593. \u003c/li\u003e\n\u003cli\u003eLi JQ, Tan L, Wang HF, et al. Risk factors for predicting progression from mild cognitive impairment to Alzheimer\u0026rsquo;s disease: a systematic review and meta-analysis of cohort studies. \u003cem\u003eJ Neurol Neurosurg Psychiatry\u003c/em\u003e. 2016;87(5):476\u0026ndash;484. \u003c/li\u003e\n\u003cli\u003eSingh-Manoux A, Oumarou-Ibrahim A, Machado-Fragua MD, et al. Association between kidney function and incidence of dementia: 10-year follow-up of the Whitehall II cohort study. \u003cem\u003eAge Ageing\u003c/em\u003e. 2022;51(1):afab259. \u003c/li\u003e\n\u003cli\u003eHakala JO, Pahkala K, Juonala M, et al. Repeatedly measured serum creatinine and cognitive performance in midlife: the Cardiovascular Risk in Young Finns Study. \u003cem\u003eNeurology\u003c/em\u003e. 2022;98(22):e2268\u0026ndash;e2281.\u003c/li\u003e\n\u003cli\u003eFu J, Liu Q, Zhu Y, et al. Circulating folate concentrations and the risk of mild cognitive impairment: a prospective study on the older Chinese population without folic acid fortification. \u003cem\u003eEur J Neurol\u003c/em\u003e. 2022;29(10):2913\u0026ndash;2924. \u003c/li\u003e\n\u003cli\u003eWang Q, Zhao J, Chang H, et al. Homocysteine and folic acid: risk factors for Alzheimer\u0026rsquo;s disease\u0026mdash;an updated meta-analysis. \u003cem\u003eFront Aging Neurosci\u003c/em\u003e. 2021;13:665114. \u003c/li\u003e\n\u003cli\u003eHama Y, Hamano T, Shirafuji N, et al. Influences of folate supplementation on homocysteine and cognition in patients with folate deficiency and cognitive impairment. \u003cem\u003eNutrients\u003c/em\u003e. 2020;12(10):3138. \u003c/li\u003e\n\u003cli\u003eCook NR, He FJ, MacGregor GA, et al. Sodium and health\u0026mdash;concordance and controversy. \u003cem\u003eBMJ\u003c/em\u003e. 2020;369:m2440. \u003c/li\u003e\n\u003cli\u003eHui Z, Wang L, Deng J, et al. Joint association of serum sodium and frailty with mild cognitive impairment among hospitalized older adults with chronic diseases: a cross-sectional study. \u003cem\u003eFront Nutr\u003c/em\u003e. 2024;11:1467751. \u003c/li\u003e\n\u003cli\u003eChen Y, Wang Z, Liu X, et al. Elevated serum sodium is linked to increased amyloid-dependent tau pathology, neurodegeneration, and cognitive impairment in Alzheimer\u0026rsquo;s disease. \u003cem\u003eJ Neurochem\u003c/em\u003e. 2024;:jnc.16257. \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. Characteristics of datasets.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 132px;\"\u003e\n \u003cp\u003eCharacteristics\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003eOverall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003eNon-cognitive impairment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003eCognitive impairment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 46px;\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003eTraining set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003eTest set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 38px;\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u003cem\u003eN\u003c/em\u003e = 1,373\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u003cem\u003eN\u003c/em\u003e = 1,030\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cem\u003eN\u003c/em\u003e = 343\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cem\u003eN\u003c/em\u003e = 961\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u003cem\u003eN\u003c/em\u003e = 412\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eCognitive assessment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.272\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNon-cognitive impairment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,030 (75.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e729 (75.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e301 (73.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eCognitive impairment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e343 (25.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e232 (24.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e111 (26.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.646\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e60-69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e705 (51.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e577 (56.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e128 (37.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e501 (52.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e204 (49.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e70-79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e426 (31.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e308 (29.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e118 (34.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e295 (30.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e131 (31.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026ge;80\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e242 (17.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e145 (14.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e97 (28.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e165 (17.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e77 (18.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.855\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e648 (47.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e463 (45.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e185 (53.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e452 (47.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e196 (47.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e725 (52.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e567 (55.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e158 (46.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e509 (53.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e216 (52.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eRace\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.406\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eMexican American\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e107 (7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e81 (7.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e26 (7.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e67 (7.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e40 (9.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eOther Hispanic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e120 (8.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e71 (6.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e49 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e84 (8.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e36 (8.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNon-Hispanic White\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e677 (49.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e538 (52.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e139 (40.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e485 (50.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e192 (46.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNon-Hispanic Black\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e358 (26.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e253 (24.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e105 (30.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e246 (25.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e112 (27.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eOther Race\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e111 (8.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e87 (8.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e24 (7.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e79 (8.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e32 (7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eEducational status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.476\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eLess than 9th grade\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e149 (10.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e56 (5.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e93 (27.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e98 (10.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e51 (12.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHigh school education\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e539 (39.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e382 (37.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e157 (45.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e378 (39.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e161 (39.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eCollege or above\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e685 (49.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e592 (57.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e93 (27.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e485 (50.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e200 (48.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eMarital status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.105\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.505\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eMarried, living with partner\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e784 (57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e604 (58.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e180 (52.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e553 (57.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e231 (56.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eWidowed, divorced, or separated\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e511 (37.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e367 (35.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e144 (42.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e350 (36.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e161 (39.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eUnmarried\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e78 (5.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e59 (5.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e19 (5.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e58 (6.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e20 (4.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003ePIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.171\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026le; 1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e421 (30.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e265 (25.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e156 (45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e286 (29.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e135 (32.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e1.3-3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e535 (39.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e408 (39.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e127 (37.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e390 (40.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e145 (35.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026ge;3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e417 (30.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e357 (34.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e60 (17.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e285 (29.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e132 (32.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eGlycohemoglobin,\u0026nbsp;%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e5.80 (5.50, 6.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e5.80 (5.50, 6.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.90 (5.50, 6.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.056\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.80 (5.50, 6.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e5.85 (5.50, 6.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.582\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eAlbumin, g/dL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e4.20 (4.00, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e4.20 (4.00, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e4.20 (4.00, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e4.20 (4.00, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e4.20 (4.00, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.782\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eALT, U/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e20.00 (16.00, 26.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e20.00 (16.00, 26.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e19.00 (14.00, 24.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e20.00 (16.00, 26.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e19.00 (15.00, 25.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.184\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eBlood urea nitrogen, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e5.36 (4.28, 7.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e5.36 (4.28, 6.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.71 (4.64, 7.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.36 (4.28, 7.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e5.36 (4.28, 7.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.885\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSerum creatinine, umol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e84.86 (70.72, 103.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e83.98 (68.95, 100.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e90.17 (74.26, 111.38)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e84.86 (70.72, 102.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e84.86 (69.40, 106.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.914\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSerum glucose, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e5.66 (5.11, 6.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e5.66 (5.05, 6.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.72 (5.22, 6.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e5.66 (5.11, 6.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e5.66 (5.11, 6.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.916\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSerum uric acid, umol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e345.00 (285.50, 404.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e345.00 (285.50, 404.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e345.00 (285.50, 416.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.777\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e345.00 (291.50, 410.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e339.00 (285.50, 401.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.284\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSerum sodium, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e140.00 (138.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e140.00 (138.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e140.00 (138.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.085\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e140.00 (138.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e140.00 (138.00, 142.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eTG, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1.49 (1.02, 2.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e1.50 (1.02, 2.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1.42 (1.01, 2.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.193\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1.46 (0.99, 2.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.54 (1.10, 2.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.282\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eDirect HDL-C, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1.32 (1.09, 1.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e1.32 (1.11, 1.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1.29 (1.06, 1.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.063\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1.32 (1.09, 1.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.32 (1.11, 1.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.670\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSerum vitamin B12, pmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e411.10 (288.60, 594.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e414.00 (291.50, 587.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e403.70 (265.70, 631.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e416.20 (288.60, 586.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e398.15 (288.20, 635.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.936\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eRBC folate, nmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,270.00 (906.00, 1,740.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e1,290.00 (926.00, 1,750.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1,180.00 (859.00, 1,730.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1,280.00 (928.00, 1,740.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1,230.00 (878.50, 1,740.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.185\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eBMI, kg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.048\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026lt;25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e284 (20.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e201 (19.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e83 (24.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e184 (19.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e100 (24.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e25-29.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e496 (36.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e360 (35.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e136 (39.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e363 (37.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e133 (32.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026ge;30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e593 (43.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e469 (45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e124 (36.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e414 (43.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e179 (43.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eSmoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.124\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNever\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e657 (47.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e483 (46.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e174 (50.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e472 (49.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e185 (44.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eFormer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e547 (39.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e427 (41.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e120 (35.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e366 (38.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e181 (43.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eCurrent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e169 (12.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e120 (11.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e49 (14.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e123 (12.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e46 (11.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eAlcohol consumption\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.128\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.926\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e429 (31.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e310 (30.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e119 (34.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e301 (31.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e128 (31.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e944 (68.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e720 (69.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e224 (65.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e660 (68.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e284 (68.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eAge at hypertension onset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.074\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.613\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026lt;35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e118 (8.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e80 (7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e38 (11.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e85 (8.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e33 (8.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003e\u0026ge;35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1255 (91.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e950 (92.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e305 (88.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e876 (91.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e379 (92.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of diabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.351\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e986 (71.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e756 (73.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e230 (67.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e683 (71.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e303 (73.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e387 (28.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e274 (26.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e113 (32.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e278 (28.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e109 (26.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of heart failure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.886\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,252 (91.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e951 (92.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e301 (87.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e877 (91.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e375 (91.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e121 (8.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e79 (7.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e42 (12.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e84 (8.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e37 (9.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of coronary heart disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,217 (88.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e918 (89.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e299 (87.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e850 (88.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e367 (89.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e156 (11.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e112 (10.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e44 (12.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e111 (11.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e45 (10.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of stroke\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.678\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,253 (91.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e956 (92.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e297 (86.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e879 (91.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e374 (90.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e120 (8.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e74 (7.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e46 (13.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e82 (8.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e38 (9.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of chronic bronchitis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.815\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.194\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,263 (92.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e949 (92.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e314 (91.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e890 (92.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e373 (90.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e110 (8.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e81 (7.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e29 (8.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e71 (7.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e39 (9.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of Sleep disorder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.592\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.395\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,183 (86.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e884 (85.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e299 (87.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e833 (86.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e350 (85.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e190 (13.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e146 (14.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e44 (12.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e128 (13.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e62 (15.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eHistory of Cancer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e0.309\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1,080 (78.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e797 (77.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e283 (82.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e763 (79.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e317 (76.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 132px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e293 (21.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e233 (22.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e60 (17.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e198 (20.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e95 (23.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eALT: Alanine aminotransferase; BMI: body mass index; HDL-C: high-density lipoprotein cholesterol; PIR: poverty income ratio; RBC: folate red blood cell folate;\u0026nbsp;TG: Triglycerides.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Performance comparison of models.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eModel\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003ePrevalence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eRecall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003eF1-Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eMCC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eAUC of ROC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003ePresicion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003eFNR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003eFPR\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.777\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.342\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.452\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.359\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.791\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.937\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.658\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.063\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eCatBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.252\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.361\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.286\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.785\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.636\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.947\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.748\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.053\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.774\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.252\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.376\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.336\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.967\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.748\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.033\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.762\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.198\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.310\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.283\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.757\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.710\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.970\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.802\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.030\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eLGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.752\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.144\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.239\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.234\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.720\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.696\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.977\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.856\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eXGB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.423\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.254\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.720\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.494\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.860\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.631\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.140\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003eDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e0.767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e0.269\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.225\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.342\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.306\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.713\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.967\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.775\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 46px;\"\u003e\n \u003cp\u003e0.033\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAUC: area under the curve; CatBoost: categorical boosting; LR: logistic regression; DT: decision tree; LGBM: light gradient boosting machine; MCC: Matthews correlation coefficient; RF: random forest; ROC: receiver operating characteristic; SVM: support vector machine; XGB: extreme gradient boosting.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"hypertension, cognitive impairment, machine learning, predictive model, NHANES","lastPublishedDoi":"10.21203/rs.3.rs-6950624/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6950624/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eThis study aimed to evaluate the utility of machine learning (ML) algorithms in predicting cognitive impairment among elderly individuals with hypertension in the United States and to identify key associated risk factors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eData were obtained from 19,931 participants enrolled in the 2011–2012 and 2013–2014 cycles of the National Health and Nutrition Examination Survey (NHANES). The dataset was randomly split into training and test sets (70:30). Seven ML algorithms—logistic regression (LR), extreme gradient boosting (XGB), decision tree (DT), categorical boosting (CatBoost), random forest (RF), light gradient boosting machine (LGBM), and support vector machine (SVM)—were trained to predict cognitive impairment. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). The SHapley Additive exPlanation (SHAP) method was applied for feature importance interpretation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eAmong all models, LR exhibited the best overall performance, achieving an AUC of 0.791 on the test set, with superior F1-score and accuracy. Calibration plots demonstrated good agreement between predicted and observed outcomes. DCA confirmed the clinical utility of the LR model. SHAP analysis identified the key variables contributing to model predictions. A web-based calculator based on the final LR model, incorporating 12 predictors, is available at: https://cognitiveimpairment.shinyapps.io/cognitiveimpairment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eAn interpretable ML model was developed and validated to predict the risk of cognitive impairment in elderly hypertensive patients in the United States. This model enables clinicians to quickly identify high-risk patients, which in turn supports more effective prevention and intervention strategies.\u003c/p\u003e","manuscriptTitle":"Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-01 11:04:23","doi":"10.21203/rs.3.rs-6950624/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d1d923e7-007c-4cc6-880e-ee1ff79ad351","owner":[],"postedDate":"August 1st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-08-25T15:09:02+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-01 11:04:23","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6950624","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6950624","identity":"rs-6950624","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}