GRAPHICAL ABSTRACT

INTRODUCTION

Ischemic stroke occurs when a cerebral artery is occluded by a thrombus or embolus, resulting in impaired perfusion and neurological deficits [1]. Although less common than other major diseases, acute ischemic stroke carries a 90-day case-fatality rate of 3–7%, compared with up to 17% for intracerebral hemorrhage [2]. Moreover, only 10–15% of patients achieve full recovery, whereas most survivors live with residual disability [3].

Early detection is challenging because prodromal symptoms are often subtle, and fewer than 40% of patients reach the hospital within 3 hours; fewer than 5% receive timely thrombolysis [4]. Therefore, reducing stroke burden requires identifying high-risk individuals before onset and ensuring rapid care once symptoms appear.

Traditional prediction efforts include the Framingham Stroke Risk Score, which identified age, blood pressure, diabetes, smoking, cardiovascular history, atrial fibrillation, and left ventricular hypertrophy as key predictors [5]. Subsequent studies expanded these predictors to include high-density lipoprotein cholesterol and low-density lipoprotein cholesterol [6], physical inactivity, family history, and carotid atherosclerosis [7]. More recently, machine-learning approaches have proposed additional markers such as average glucose level and marital status [8].

Furthermore, repeated health screening information has emerged as an important component in predicting stroke occurrence. Cardiometabolic conditions relevant to stroke—such as hypertension, dysglycemia, obesity, and chronic kidney disease—are typically shaped by sustained lifestyle behaviors, including diet, physical activity, and smoking. Although cross-sectional measurements can estimate relative risk at a single time point, they provide limited insight into trajectories of physiological deterioration, which are essential for personalized prevention. Several multi-year studies have used repeated measures for stroke prediction. Zheng et al. [9] demonstrated that multi-time-point indicators improved stroke prediction compared with Cox models based on single measurements. Other studies have suggested that cumulative exposure scores or trajectory-based classifications show stronger associations than single measurements [10–13].

In Korea, all adults aged ≥20 years who are enrolled in the National Health Insurance Service (NHIS) are eligible for a free national health screening every 2 years. These screenings include anthropometric measurements, blood pressure assessment, and comprehensive blood tests. Because several of these measures are established stroke-related indicators, they provide an opportunity to evaluate how long-term changes influence subsequent stroke risk. Leveraging this framework, the present study aimed to compare 3 consecutive pre-stroke screening cycles between individuals who did and did not develop ischemic stroke and to identify which health indicators were most strongly associated with subsequent stroke occurrence.

METHODS

Data Sources and Study Population

This study used nationwide customized data from the Korean NHIS, including health screening records, socio-demographic variables, and medical claims (2013–2023). The dataset was constructed upon request and consisted of individuals with stroke diagnoses during 2009–2023 and 1:1 non-stroke controls matched on sex and 1-year age strata (n=1 669 108).

To identify incident ischemic stroke cases in claims data, cases were defined as individuals who had an emergency department visit and/or hospitalization, underwent relevant brain imaging (computed tomography/magnetic resonance imaging), and had a primary diagnosis of ischemic stroke.

We then restricted cases to incident stroke events occurring during 2019–2023 and required completion of health screenings in all 3 periods (P1–P3). Participants with missing values in any screening variables were excluded using a complete-case approach. To improve comparability between incident stroke cases and controls, 1:1 propensity score matching was performed, with propensity scores estimated using sex, 1-year age strata, and insurance type. After matching, covariate balance was assessed using standardized mean differences (SMDs), all of which were ≤0.001. The final matched analytic cohort comprised 97 454 incident stroke cases and 97 454 matched controls (n=194 908; Figure 1, Supplemental Material 1).

Time-series Partitioning and Exploratory Analysis

Because the National Health Screening Program in Korea is conducted biennially, screening data were organized into 3 consecutive periods to capture multi-year metabolic patterns:

Period 1 (P1): 2013–2014, Period 2 (P2): 2015–2016, Period 3 (P3): 2017–2018.

To ensure consistent multi-year exposure assessment, only individuals who completed at least 1 health screening examination in all 3 periods were included. The dependent variable was incident ischemic stroke during 2019–2023, ensuring a clear temporal sequence in which metabolic exposures (P1–P3) preceded the outcome.

For descriptive purposes, period-specific summary statistics of metabolic indicators were reported for cases and controls (Table 1). Covariate balance between groups was assessed using SMDs (and variance ratios for continuous variables), rather than hypothesis testing. Urine protein, recorded on a urine dipstick semi-quantitative scale (negative, trace, 1+ to 4+), was treated as an ordinal categorical variable and summarized as n (%) by period in Table 2.

Definition of Derived Variables

Derived variables were constructed to capture multi-year metabolic deterioration rather than single cross-sectional values. Operational definitions were informed by the period-specific mean differences observed in Table 1 and supported by clinical guidelines and prior studies. The detailed rationale for each derived indicator is provided below.

Waist circumference increase

Table 1 showed that waist circumference was consistently higher in the case group across all periods (P1–P3), with statistically significant differences at each time point (all p<0.001). This pattern suggests a persistent central adiposity burden among individuals who later developed stroke. In line with previous studies showing that even modest increases or sustained elevations in waist circumference are associated with higher risks of cardiovascular disease, stroke, and mortality [14,15], waist circumference increase was defined as a ≥5% or ≥10% increase from P1 to P3. Participants with increases above these thresholds were coded as 1, and those below these thresholds were coded as 0.

Sustained high blood pressure

As shown in Table 1, both systolic and diastolic blood pressure values were significantly higher in the case group across all 3 periods (all p<0.001), indicating persistent elevation rather than episodic spikes. Following hypertension guidelines that emphasize repeated measurements over single readings [10] and stroke-risk studies highlighting the importance of sustained blood pressure elevation [13], sustained high blood pressure was defined as meeting the threshold at all 3 periods. Two thresholds were applied: systolic blood pressure ≥120 mmHg or diastolic blood pressure ≥80 mmHg at all 3 periods (120/80 model), or systolic blood pressure ≥130 mmHg or diastolic blood pressure ≥80 mmHg at all 3 periods (130/80 model). These cutoffs were selected because the American Heart Association/American Stroke Association guideline classifies blood pressure <120/80 mmHg as normal and identifies 130/80 mmHg as the overarching blood pressure target for adults [10]. Individuals meeting each criterion across P1–P3 were coded as 1.

Sustained high fasting blood glucose

Sustained high fasting blood glucose was defined on the basis of both the empirical patterns observed in Table 1 and established diagnostic thresholds from the American Diabetes Association (ADA) [11]. Table 1 showed that fasting glucose levels were consistently and significantly higher in the case group across all periods (all p<0.001), indicating a persistent glycemic burden rather than isolated elevations. According to ADA criteria, fasting glucose values of ≥100 mg/dL indicate impaired fasting glucose (prediabetes), whereas values of ≥126 mg/dL represent the diagnostic threshold for diabetes [11]. In alignment with these criteria, sustained hyperglycemia was defined as meeting either threshold at all 3 periods (P1–P3). Individuals who met each cutoff across all 3 periods were coded as 1, and all others were coded as 0.

Creatinine elevation

Creatinine elevation was defined using sex-specific clinical thresholds and the patterns observed across screening periods. As shown in Table 1, creatinine levels were consistently higher in the case group at all 3 periods (P1, p=0.001; P2, p<0.001; P3, p<0.001). Although the chronic kidney disease guideline identifies sex-specific thresholds of ≥0.9 mg/dL for male and ≥0.7 mg/dL for female [16], the distribution of creatinine levels in our study population was notably higher, with mean values ranging from 0.91 mg/dL to 0.94 mg/dL across periods. This pattern is consistent with population-based estimates reported in the Korea National Health and Nutrition Examination Survey, in which mean creatinine levels among Korean adults approximate 0.9 mg/dL [12]. Given these population-specific characteristics, elevated creatinine was defined as ≥1.0 mg/dL for male and ≥0.8 mg/dL for female to better reflect the metabolic profile of this Korean cohort. Individuals whose creatinine exceeded their respective sex-specific thresholds on ≥2 occasions were coded as 1, and all others were coded as 0.

Proteinuria progression

Proteinuria progression was defined using the urine dipstick protein categories (negative, trace, 1+ to 4+) measured across the 3 screening periods (P1–P3). As summarized in Table 2, the distribution of proteinuria categories differed between cases and controls within each period (Mantel–Haenszel chi-square test: P1, p<0.001; P2, p<0.001; P3, p<0.001). Informed by nephrology literature and prior evidence that worsening proteinuria category over time is associated with renal and cardiovascular risk and incident stroke [17], proteinuria progression was operationalized as any upward change in dipstick category from P1 to P2 or from P2 to P3 (coded as 1); no change or any downward change was coded as 0.

Statistical Analysis

Descriptive analyses summarized health screening indicators across the 3 biennial periods (P1: 2013–2014; P2: 2015–2016; P3: 2017–2018). For descriptive comparisons of continuous measures between cases and controls, we reported mean±standard deviation (SD) values and used the independent t-test; equality of variances was assessed using the folded F-test.

To evaluate associations between repeatedly measured metabolic indicators and incident stroke in the 1:1 matched cohort, we fitted conditional logistic regression models with matched pair as the stratum. Eight multi-year models were specified using alternative thresholds for waist circumference increase (≥5 or ≥10%), sustained high blood pressure (≥120/80 or ≥130/80 mmHg across P1–P3), and sustained hyperglycemia (≥100 or ≥126 mg/dL across P1–P3). A cross-sectional model using P3 measurements only (abdominal obesity, high blood pressure, high fasting blood glucose, elevated creatinine, and proteinuria ≥2+) was also examined.

Because matching was performed on sex, 1-year age strata, and insurance type, these variables were not included as covariates in the conditional logistic regression models (i.e., they were accounted for by the matched strata). The models were additionally adjusted for other socio-demographic factors not used in matching, including residential region and income level. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Ethics Statement

The use of NHIS data was approved by the NHIS (approval No. NHIS-2023-1-451). The study protocol was reviewed and approved by the Institutional Review Board of Kangwon National University Hospital (approval No. KNUH-2024-02-006).

RESULTS

In the matched cohort (97 454 cases and 97 454 controls), participant characteristics and derived screening indicators are summarized in Table 3. The sex distribution was 62.3% male among cases and 62.3% among controls (p=0.98). The age-group distribution did not differ between groups (p=0.99), and insurance type also showed no between-group difference (p=0.99). Region and income level differed between cases and controls (both p<0.001); for example, provincial residence was 40.0% among cases versus 34.8% among controls, and the highest income category comprised 29.7% of cases versus 34.2% of controls.

For the derived screening indicators, waist circumference increase ≥5% was observed in 28.2% of cases and 28.0% of controls (p=0.31), and waist circumference increase ≥10% was observed in 10.8% and 10.5%, respectively (p=0.010). Sustained high blood pressure was present in 57.9% of cases versus 50.9% of controls under the ≥120/80 mmHg definition and in 39.2% versus 32.0% under the ≥130/80 mmHg definition (both p<0.001). Sustained hyperglycemia was present in 30.7% of cases versus 25.9% of controls for fasting glucose ≥100 mg/dL and in 7.7% versus 4.5% for fasting glucose ≥126 mg/dL (both p<0.001). Creatinine elevation (sex-specific, ≥2 periods) was observed in 54.8% of cases and 53.0% of controls (p<0.001), and proteinuria progression was observed in 17.4% and 13.1%, respectively (p<0.001).

Results from all 9 conditional logistic regression models are summarized in Supplemental Materials 2–4. For comparability, model 8 and the P3-only model (model 9) are presented in Table 4. In model 8, the cumulative multi-period indicators were associated with incident ischemic stroke: waist circumference increase ≥10% (OR, 1.05; 95% CI, 1.01 to 1.08), sustained blood pressure ≥130/80 mmHg (OR, 1.34; 95% CI, 1.31 to 1.37), sustained fasting glucose ≥126 mg/dL (OR, 1.66; 95% CI, 1.60 to 1.73), creatinine elevation in ≥2 periods (OR, 1.08; 95% CI, 1.06 to 1.10), and proteinuria progression (OR, 1.36; 95% CI, 1.32 to 1.39).

In the P3-only model (model 9), single-time-point indicators were also associated with incident ischemic stroke: abdominal obesity (waist circumference ≥cutoff, P3) (OR, 1.08; 95% CI, 1.06 to 1.10), high blood pressure ≥130/80 mmHg (P3) (OR, 1.36; 95% CI, 1.33 to 1.38), fasting glucose ≥126 mg/dL (P3) (OR, 1.47; 95% CI, 1.43 to 1.51), elevated creatinine (sex-specific, P3) (OR, 1.11; 95% CI, 1.09 to 1.13), and proteinuria ≥2+ (P3) (OR, 1.44; 95% CI, 1.39 to 1.49). No multicollinearity was detected in the final models (all variance inflation factors ≤1.02; Supplemental Material 5).

DISCUSSION

This study evaluated how long-term changes in multiple metabolic indicators—waist circumference, blood pressure, fasting blood glucose, proteinuria, and creatinine—were associated with incident ischemic stroke in a large, nationally representative cohort using 3 biennial health screening periods over 6 years and including nearly 100 000 stroke cases.

The association between abdominal obesity and stroke observed in our study is consistent with prior evidence. Suk et al. [18] found that higher waist-to-hip ratio and waist circumference strongly predicted stroke in Korean adults, outperforming body mass index, and the EPIC-Spanish cohort similarly identified waist circumference and waist-to-height ratio as robust predictors [19]. Consistent with these findings, a ≥10% increase in waist circumference over 6 years in our study was modestly but significantly associated with higher stroke risk, suggesting that progressive central adiposity may contribute to cerebrovascular vulnerability.

Our findings for sustained high blood pressure are also consistent with earlier cohort studies showing that long-term blood pressure burden is more informative than single readings. The CARDIA study demonstrated that repeatedly elevated blood pressure in early adulthood predicted subsequent stroke and cardiovascular disease [20], and the ARIC study reported substantially higher stroke risk among individuals with persistently elevated blood pressure than among those with normal or prehypertensive levels [13]. In our multi-year model, maintaining blood pressure ≥130/80 mmHg across all 3 screenings was associated with 34% higher odds of stroke, supporting the importance of persistent elevation.

We additionally observed a strong association between sustained hyperglycemia and stroke. Participants with fasting glucose ≥126 mg/dL at all periods had markedly higher stroke risk, consistent with ADA thresholds [11] and prior studies linking chronic hyperglycemia to cardiovascular events. Peng et al. [21] reported nearly 1.8-fold higher stroke risk among individuals with 5-year mean glucose levels of 126.0–139.9 mg/dL than among those with levels of 90.0–99.9 mg/dL, underscoring the impact of long-term glycemic burden.

Serum creatinine also contributed to stroke risk. Elevation across at least 2 screenings (≥1.0 mg/dL in male and ≥0.8 mg/dL in female) was significantly associated with stroke, echoing cohort findings that individuals in the upper creatinine decile had substantially higher stroke risk [22].

More recent research has shifted from examining creatinine as an independent predictor to using composite renal-metabolic indicators—such as the blood urea nitrogen-to-creatinine ratio [23], the serum uric acid-to-creatinine ratio [24], and the albumin-to-creatinine ratio [25]—to evaluate future stroke risk. These emerging biomarkers warrant consideration in future analyses.

Finally, proteinuria progression—an indicator of generalized vascular dysfunction and injury—was associated with an increased risk of stroke. A large study of nearly 200 000 Korean health screening participants similarly reported that higher degrees of proteinuria were progressively associated with a greater risk of incident ischemic stroke [26]. Consistent with these findings, a meta-analysis of 10 cohort studies involving approximately 140 000 participants found that proteinuria was an independent risk factor for stroke, conferring nearly 70% higher risk among individuals with proteinuria than among those without [17].

An additional contribution of this study was the direct comparison of multi-year and cross-sectional modeling approaches. The P3 model (model 9), which reflects the most recent health status before stroke, showed larger associations for most single-time-point indicators than the multi-year model, except for sustained hyperglycemia (fasting glucose ≥126 mg/dL). However, health screening values represent an individual’s current status and may also reflect the effects of ongoing health management. Importantly, our results suggest that changes across repeated screenings provide information relevant to incident stroke beyond that captured by a single time-point measurement. Therefore, reviewing prior screening results may help patients and clinicians understand risk over time and underscore the importance of continuous monitoring and sustained risk-factor control, because the multi-year model (model 8) captures whether risk markers persist or worsen across periods rather than relying on a single measurement.

Several limitations should be noted. First, although national health screenings are scheduled at 1–2-year intervals, the actual timing of measurements varied across individuals, which may have introduced imprecision into the estimation of multi-year patterns. Second, important behavioral and treatment-related factors (e.g., smoking, physical activity, diet, and medication use/adherence) could not be incorporated, leaving the possibility of residual confounding. Third, we did not include baseline diagnostic history of chronic conditions (e.g., hypertension or diabetes) in the primary models; diagnostic indicators are most informative when interpreted together with treatment and behavioral information, which could not be reliably defined in the customized dataset used in this study. Future research should develop more refined models that integrate diagnostic history with medication-related and behavior-related measures. Fourth, restricting the sample to individuals who completed all 3 screenings may have introduced selection bias, as regular screening participants may differ systematically from non-participants.

This study found that multi-year patterns in health screening indicators—including increases in waist circumference, sustained elevations in blood pressure and fasting glucose, proteinuria progression, and repeated elevations in sex-specific creatinine —were associated with incident ischemic stroke. These findings underscore the potential value of repeated health screening measurements in supporting continuous monitoring and sustained management of cardiometabolic and renal risk factors. More broadly, our results suggest that longitudinal screening information may complement risk stratification and patient-centered prevention efforts aimed at reducing future cerebrovascular risk.

Supplemental Materials

Notes

Conflict of Interest

The authors have no conflicts of interest associated with the material presented in this paper.

Funding

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024-00461755).

Acknowledgements

None.

Author Contributions

Conceptualization: Hwang YS, Kim SK, Lee GM. Data curation: Hwang YS, Yim I. Formal analysis: Hwang YS, Jo HS. Funding acquisition: Hwang YS, Kim SK, Lee GM. Methodology: Hwang YS, Lee GM, Kim S. Project administration: Hwang YS. Visualization: Hwang YS. Writing – original draft: Hwang YS. Writing – review & editing: Hwang YS, Yim I, Jo HS, Kim S, Kim SK, Lee GM.

Figure 1

Study population flow diagram. Period 1 (P1): 2013–2014; Period 2 (P2): 2015–2016; Period 3 (P3): 2017–2018.

REFERENCES

• 1. Lui F, Hui C, Suheb MZ, Patti L. Ischemic stroke. StatPearls [Internet]. Treasure Island: StatPearls Publishing; 2025
• 2. Kim JY, Kang K, Kang J, Koo J, Kim DH, Kim BJ, et al. Executive summary of stroke statistics in Korea 2018: a report from the Epidemiology Research Council of the Korean Stroke Society. J Stroke 2019;21(1):42-59. https://doi.org/10.5853/jos.2018.03125ArticlePubMedPMC
• 3. Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Blaha MJ, et al. Heart disease and stroke statistics--2014 update: a report from the American Heart Association. Circulation 2014;129(3):e28-e292. https://doi.org/10.1161/01.cir.0000441139.02102.80ArticlePubMedPMC
• 4. Khandelwal A, Sarma K, Hussain M, Dikshit P, Baidya DK. Acute ischemic stroke and the golden hour: critical updates. J Neurosci Rural Pract 2025;16(2):152-159. https://doi.org/10.25259/JNRP_17_2025Article
• 5. Wolf PA, D’Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham Study. Stroke 1991;22(3):312-318. https://doi.org/10.1161/01.str.22.3.312ArticlePubMed
• 6. Ramírez-Moreno JM, Rebollo B, Macías-Sedas P, Valverde N, Parejo A, Felix-Redondo FJ, et al. Strength of association of classical vascular risk factors in young patients with ischaemic stroke: a case-control study. Neurologia (Engl Ed) 2024;39(7):604-613. https://doi.org/10.1016/j.nrleng.2022.07.006ArticlePubMed
• 7. Yi X, Chen H, Wang Y, Yu M, Luo H, Wang C, et al. Prevalence and risk factors of high-risk population for stroke: a population-based cross-sectional survey in southwestern China. Front Neurol 2022;13: 693894. https://doi.org/10.3389/fneur.2022.693894ArticlePubMedPMC
• 8. Hassan A, Gulzar Ahmad S, Ullah Munir E, Ali Khan I, Ramzan N. Predictive modelling and identification of key risk factors for stroke using machine learning. Sci Rep 2024;14(1):11498. https://doi.org/10.1038/s41598-024-61665-4ArticlePubMedPMC
• 9. Zheng W, Chen YH, Sawan M. Longitudinal data to enhance dynamic stroke risk prediction. Healthcare (Basel) 2022;10(11):2134. https://doi.org/10.3390/healthcare10112134ArticlePubMedPMC
• 10. Writing Committee Members, Jones DW, Ferdinand KC, Taler SJ, Johnson HM, Shimbo D, et al. 2025 AHA/ACC/AANP/AAPA/ABC/ACCP/ACPM/AGS/AMA/ASPC/NMA/PCNA/SGIM guideline for the prevention, detection, evaluation and management of high blood pressure in adults: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation 2025;152(11):e114-e218. https://doi.org/10.1161/CIR.0000000000001356ArticlePubMedPMC
• 11. American Diabetes Association. Diabetes diagnosis; [cited 2025 Jul 12]. Available from: https://diabetes.org/about-diabetes/diagnosis
• 12. Choi H, Kim HC, Lee JY, Lee JM, Choi DP, Suh I. Sleep duration and chronic kidney disease: the Korean Genome and Epidemiology Study (KoGES)-Kangwha study. Korean J Intern Med 2017;32(2):323-334. https://doi.org/10.3904/kjim.2015.400ArticlePubMedPMC
• 13. Shimizu Y, Kato H, Lin CH, Kodama K, Peterson AV, Prentice RL. Relationship between longitudinal changes in blood pressure and stroke incidence. Stroke 1984;15(5):839-846. https://doi.org/10.1161/01.str.15.5.839ArticlePubMed
• 14. Kim DH, Nam GE, Han K, Kim YH, Park KY, Hwang HS, et al. Variabilities in weight and waist circumference and risk of myocardial infarction, stroke, and mortality: a nationwide cohort study. Endocrinol Metab (Seoul) 2020;35(4):933-942. https://doi.org/10.3803/EnM.2020.871ArticlePubMedPMC
• 15. Cho JH, Rhee EJ, Park SE, Kwon H, Jung JH, Han KD, et al. The risk of myocardial infarction and ischemic stroke according to waist circumference in 21,749,261 Korean adults: a nationwide population-based study. Diabetes Metab J 2019;43(2):206-221. https://doi.org/10.4093/dmj.2018.0039ArticlePubMedPMC
• 16. Levin A, Stevens PE, Bilous RW, Coresh J, De Francisco AL, De Jong PE, et al.; Kidney disease: improving global outcomes (KDIGO) CKD work group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int Suppl 2013;3(1):1-150. https://doi.org/10.1038/kisup.2012.73Article
• 17. Ninomiya T, Perkovic V, Verdon C, Barzi F, Cass A, Gallagher M, et al. Proteinuria and stroke: a meta-analysis of cohort studies. Am J Kidney Dis 2009;53(3):417-425. https://doi.org/10.1053/j.ajkd.2008.08.032ArticlePubMed
• 18. Suk SH, Sacco RL, Boden-Albala B, Cheun JF, Pittman JG, Elkind MS, et al. Abdominal obesity and risk of ischemic stroke: the Northern Manhattan Stroke Study. Stroke 2003;34(7):1586-1592. https://doi.org/10.1161/01.STR.0000075294.98582.2FArticlePubMed
• 19. Abete I, Arriola L, Etxezarreta N, Mozo I, Moreno-Iribas C, Amiano P, et al. Association between different obesity measures and the risk of stroke in the EPIC Spanish cohort. Eur J Nutr 2015;54(3):365-375. https://doi.org/10.1007/s00394-014-0716-xArticlePubMed
• 20. Nwabuo CC, Appiah D, Moreira HT, Vasconcellos HD, Yano Y, Reis JP, et al. Long-term cumulative blood pressure in young adults and incident heart failure, coronary heart disease, stroke, and cardiovascular disease: the CARDIA study. Eur J Prev Cardiol 2021;28(13):1445-1451. https://doi.org/10.1177/2047487320915342ArticlePubMedPMC
• 21. Peng X, Ge J, Wang C, Sun H, Ma Q, Xu Y, et al. Longitudinal average glucose levels and variance and risk of stroke: a Chinese cohort study. Int J Hypertens 2020: 2020;8953058. https://doi.org/10.1155/2020/8953058ArticlePubMedPMC
• 22. Wannamethee SG, Shaper AG, Perry IJ. Serum creatinine concentration and risk of cardiovascular disease: a possible marker for increased risk of stroke. Stroke 1997;28(3):557-563. https://doi.org/10.1161/01.str.28.3.557ArticlePubMed
• 23. Peng R, Liu K, Li W, Yuan Y, Niu R, Zhou L, et al. Blood urea nitrogen, blood urea nitrogen to creatinine ratio and incident stroke: the Dongfeng-Tongji cohort. Atherosclerosis 2021;333: 1-8. https://doi.org/10.1016/j.atherosclerosis.2021.08.011ArticlePubMed
• 24. Sun X, Lv J, Wu Z, Shi J, Huang H. Serum uric acid to serum creatinine ratio and risk of stroke recurrence in young adults with ischemic stroke. Neuropsychiatr Dis Treat 2022;18: 2031-2039. https://doi.org/10.2147/NDT.S378576ArticlePubMedPMC
• 25. Huang R, Chen X. Increased spot urine albumin-to-creatinine ratio and stroke incidence: a systematic review and meta-analysis. J Stroke Cerebrovasc Dis 2019;28(10):104260. https://doi.org/10.1016/j.jstrokecerebrovasdis.2019.06.018ArticlePubMed
• 26. Lee SM, Ha E, Ryoo JH. Urine protein levels predict future development of cerebral infarction in Koreans. Yonsei Med J 2022;63(1):66-71. https://doi.org/10.3349/ymj.2022.63.1.66ArticlePubMedPMC

Figure & Data

References

Citations

Citations to this article as recorded by