Abstract
Objective
Cerebrovascular disease can be roughly divided into 2 subtypes: Cerebral ischemia (CI) and cerebral hemorrhage (CH). No scale currently exist that can predict the subtypes of cerebrovascular diseases. This study aims to establish a prediction scale for the subtypes of cerebrovascular diseases.
Methods
A total of 1 200 cerebrovascular disease patients were included in this study, data from 1 081 (90%) patients were used to establish the CI-CH risk scale, and data from 119 (10%) patients were used to test it. Risk factors for the CI-CH risk scale were identified by 2 screens, with two-tailed student’s t-test and two-tailed Fisher’s exact test preliminarily and with logistic regression analysis further. The scores of each risk factor for CI-CH risk scale were determined according to the odds rate, and the cut-off point was determined by Youden index.
Results
Nine risk factors were ultimately selected for score system, including age (≥75 years old was -1, <75 years old was 0), BMI (<24 kg/m2 was 0, 24-28 kg/m2 was -1, >28 kg/m2 was -2), hypertension grade (grade 1 was 1, grade 2 was 2, and grade 3 was 3), diabetes status (no was 0, yes was -1), antihypertensive drug use (no was 0, yes was -2), alcohol consumption (<60 g/d was 1, ≥60 g/d was 2), uric acid (less than normal was 0, normal was -1, high than normal was -2), LDL cholesterol (<2 mmol/L was 0, 2-4 mmol/L was -1, and >4 mmol/L was -2), and HDL cholesterol (<1.55 mmol/L was 0, ≥1.55 mmol/L was 2). Patients with a score more than 0 were classified as the CH group, Conversely, they were assigned to the CI group; its sensitivity, specificity, and accuracy were 74.5%, 77.9%, and 76.4%, respectively.
Conclusion
The CI-CH risk scale can help the clinician predict the subtypes of cerebrovascular diseases.
Keywords: cerebral ischemia, cerebral hemorrhage, logistic regression analysis, prediction scale
Abstract
目的
脑血管病大致可分为2个亚型，即脑缺血(cerebral ischemia，CI)和脑出血(cerebral hemorrhage，CH)，目前还没有可以预测脑血管病亚型的模型。本研究旨在建立一个脑血管疾病亚型的预测模型。
方法
本研究共纳入1 200名脑血管病患者，其中1 081名(90%)患者的数据用于建立CI-CH量表，119名(10%)患者的数据对CI-CH量表进行测试。通过t检验和Fisher’s检验对预测因子进行第1次筛选，利用logistic回归对预测因子进行第2次筛选，以确定CI-CH量表的预测因子；参考OR值确定CI-CH量表的各因子的分值，计算约登指数作为CI-CH量表的分界点。
结果
最终选择了9个危险因素作为评分系统，包括年龄(≥75岁为-1；<75岁为0)，BMI(24-28 kg/m2为-1， >28 kg/m2为-2)，高血压等级(1级为1，2级为2，3级为3)，糖尿病(无为0，有为-1)，使用降压药物(否为0，是为 -2)，饮酒量(<60 g/d为1，≥60 g/d为2)，尿酸(低于正常为0，正常为-1，高于正常为-2)，低密度脂蛋白胆固醇 (<2 mmol/L为0，2-4 mmol/L为-1，>4 mmol/L为-2)，高密度脂蛋白胆固醇(<1.55 mmol/L为0，≥1.55 mmol/L为2)。得分大于0的患者被归入CH组，反之，被归入CI组；其敏感性、特异性和准确性分别为74.5%、77.9%和76.4%。
结论
CI-CH量表可以帮助临床医生预测脑血管疾病的亚型。
Keywords: 脑缺血, 脑出血, logistic回归分析, 预测量表
Acute cerebrovascular disease is the main cause of death and long-term disability in adults[1]. From 1994 to 2013, the crude mortality rate of stroke in Chinese male and female showed an increasing trend[2]. The 2 subtypes of cerebrovascular disease are known as cerebral ischemia (CI) and cerebral hemorrhage (CH). Some important medical interventions are contradictory across the 2 subtypes. For example, antiplatelet drugs are widely used to prevent CI in high-risk populations[3], but they increase the risk of CH[4-5]. Therefore, effectively predicting scale of cerebrovascular disease subtypes is important.
At present, the prediction range of the most commonly used tools mostly focuses on the first level, that is, whether cerebrovascular disease will occur; examples of these tools include the Essen[6], stroke prognosis instrument II (SPI-II)[7], and ABCD2 scores[8], which could only be used to predict the incidence of CI. In the 2008 version of the Framingham Risk Score, researchers tried to use the probability of vascular disease to predict a patient’s age at onset, that is, to try to answer the question, when cerebrovascular disease may occur[9]. There is currently no predictive tool for the the subtype of cerebrovascular disease. This study aims to establish a scale suitable for the prediction of cerebrovascular disease subtypes so as to provide a more accurate scheme for their prevention.
1. Patients and methods
1.1. Patients
Sporadic patients with CI or CH admitted to the Department of Neurology of the Third Xiangya Hospital of Central South University from May 2013 to December 2016 were collected for examination.
Inclusion criteria for patients with CI were as follows: 1) New symptoms of neurological impairment within the last 3 days and have been confirmed by a head CT/MRI examination indicating the presence of responsible infarction lesions and 2) at least one of the following risk factors for cerebrovascular diseases, such as hyperlipidemia, history of smoking, diabetes, hypertension, obesity (BMI>28 kg/m2), and advanced age (male ≥45 years old, female ≥55 years old). Exclusion criteria included: 1) History of CH; 2) intracranial tumors; 3) a clear family history of early onset of ischemic cardiovascular disease (first-degree relatives, males <55 years old, females <65 years old); 4) using of heroin or other drugs; 5) pregnant and lactating women.
Inclusion criteria for patients with CH included the following: 1) Clear new symptoms of neurological impairment within the last 3 days and a head CT scan suggesting CH and 2) at least one of the risk factors for cerebrovascular diseases (the same as CH). Exclusion criteria included: 1) CI within 2 weeks, excluding infarct hemorrhage; 2) intracranial tumors; 3) a clear pre-onset coagulation disorder or active liver disease; 4) a clear family history of CH (first-degree relatives); 5) use of heroin or other drugs; 6) pregnant and lactating women; 7) traumatic events. This study was approved by the Ethical Committee of The Third Xiangya Hospital of Central South University (No. I 22111).
1.2. Clinical data
The database of 1 200 cerebrovascular disease patients [including 668 (55.7%) patients with CI] was established by collecting both the personal and disease data of the hospitalized patients, primarily patients’ demographic information (gender, age, height, weight, waist circumference, hip circumference), history of disease (primarily hypertension, hyperlipidemia, diabetes, etc.), personal habits (such as smoking, drinking alcohol, etc), and biochemical indicators within 72 h of onset.
The data of 1 081 (90%) patients (including 600 with CI) were randomly selected from the database as a learning dataset (DB-1), which was build for the establishment of the CI-CH risk scale. The remaining 119 (10%) patients (including 68 with CI) were used as a test dataset (DB-2) to examine the predictive performance of the CI-CH risk scale.
1.3. Statistical analysis
1.3.1. Construction of CI-CH risk scale
DB-1 was used for statistical analysis. Firstly, 2-tailed Student’s t-test was performed for measurement data in the 2 groups (CI group and CH group), and 2-tailed Fisher’s exact test was used for counting data. The difference was considered as statistically significant when P<0.001, and the factors were screened preliminarily for the first time.
Secondly, logistic regression analysis was used for the further analysis. The risk factors for CI-CH risk scale were determined by the second screen and odds ratios (ORs) of each risk factor were calculated.
Thirdly, the scores of each risk factors were determined according to the ORs and the clinical situations.
Fourthly, CI-CH risk scale scores of all patients in DB-1 were calculated. ROC curve of CI-CH risk scale was drawn and area under ROC curve (AUC) was calculated.
Finally, Youden index was calculated as the cut-off point of CI-CH risk scale. Patients with a score more than cut-off point were classified as the CI group. Conversely, they were assigned into the CI group.
1.3.2. Validation
All patients in the DB-2 dataset were evaluated according to CI-CH risk scale. The predicted results were compared with the actual subtypes for each patient to examine the predictive performance of the CI-CH risk scale. A P<0.05 was considered statistically significant.
2. Results
2.1. Construction of CI-CH risk scale
2.1.1. Distribution characteristics of DB-1 data
Demographic data of patients in the DB-1 dataset showed that age, body weight, BMI, waist circumference, and hip circumference of CH group were lower than those of CI group. The difference was statistically significant (P<0.001). The ages of patients in the CH group were (61.83±11.38) years, which was lower than that in the CI group [(64.56±11.80) years]. The body weight of patients in the CH group was lower than that in the CI group [(62.29±9.91) kg vs (64.30±10.84) kg, P<005]. After conversion to BMI, the difference between patients with CH and CI was also statistically significant [(23.00±3.10) kg/m2 vs (24.21±3.78) kg/m2, P<005]. The waist and hip circumferences of patients with CH were (83.15±8.91) cm and (89.95±9.35) cm, respectively, which were lower than those of patients with CI [(85.48±8.51) cm and (91.89±7.12) cm, respectively].
The distribution of disease history was also different between the 2 groups, with more patients with hypertension in the CH group and more patients with diabetes in the CI group. The prevalence of hypertension in the CH group was 84.8% and that in the CI group was 77.0% (P<0.001), and the proportion of patients with grade III hypertension in the CH group was also much higher than that in the CI group (64.2% vs 38.5%, P<0.001). Additionally, the proportion of hypertensive patients taking antihypertensive drugs in the CH group was lower than that in the CI group (49.8% vs 57.1%, P<0.001). The prevalence of diabetes in the CH group was 16.6%, which was lower than that in the CI group (26.5%, P<0.001). The prevalence of hyperlipidemia in the CH group was slightly lower than that in the CI group (33.5% vs 43.7%, P=0.002).
The number of patients with alcoholism in the CH group was higher than that in the CI group (P<0.001). In the CH group, there were 59 patients (12.3%) with general alcohol consumption, and 66 patients (13.7%) with alcoholism (>60 g/d). In the CI group, there were 51 patients (8.5%) with general alcohol consumption and 38 patients (6.1%) with alcoholism.
The biochemical indexes were different in the CH and CI groups (all P<0.001): Prothrombin time, (12.32±1.26) s vs (11.98±1.88) s; activated partial thrombo-plastin time, (29.80±7.68) s vs (28.27±7.28) s; total bilirubin, (17.91±8.73) μmol/L vs (15.31±6.70) μmol/L; direct bilirubin, (6.42±3.93) μmol/L vs (5.27±2.93) μmol/L; high-density lipoprotein cholesterol, (1.52±0.52) mmol/L vs (1.32±0.36) mmol/L. Uric acid was (285.29± 107.81) μmol/L vs (313.87±97.42) μmol/L (P>0.05), and low density lipoprotein cholesterol was (2.63± 0.88) mmol/L vs (2.82±0.85) mmol/L (P>0.05).
2.1.2. Modeling and detection results
We first assigned the 18 previously mentioned indicators, as shown in (Table 1), and then put them into the prediction model as the factors of the preliminary screening.
Table 1.
Assignment of the variables
| Variables | Assignment methods |
|---|---|
| Age | Original value |
| Weight | Original value |
| Waist circumference | Original value |
| Hip circumference | Original value |
| BMI | Original value |
| Hypertension | No=0, good control=1, poor control=2, first found=3 |
| Hypertension classification | No=0, grade 1=1, grade 2=2, grade 3=3 |
| Type 2 diabetes | No=0, Yes=1 |
| Hyperlipemia | No=0, Yes=1 |
| Use of antihypertensive drugs | No=0, Yes=1 |
| Drinking | No alcohol consumption=0, temperance=1, a small amount of alcohol consumption=2, heavy alcohol consumption=3 |
| Prothrombin time | Original value |
| Activated partial thromboplastin time | Original value |
| Total bilirubin | Original value |
| Direct bilirubin | Original value |
| Uric acid | Original value |
| Low-density lipoprotein | Original value |
| How-density lipoprotein | Original value |
After secondary screening and multiple iterations of all the factors, the following 9 factors were finally idenified to be risk factors for CI-CH risk scale: Age, BMI, hypertension grade, diabetes status, antihypertensive drug use, alcohol consumption, uric acid, LDL cholesterol, and HDL cholesterol. Through the prediction model, the CI-CH risk scale was established. The specific assignment method was as follow: For age, 0 points were assigned for <75 years old and 1 point for age ≥75 years old; for BMI, <24 kg/m2 was 0, 24-28 kg/m2 was -1, and >28 kg/m2 was -2; for hypertension, grade 0 was 0, grade 1 was 1, grade 2 was 2, and grade 3 was 3; the score was -1 for patients with diabetes and 0 for patients without diabetes; patients using antihypertensive drugs were scored as -2, while those not using antihypertensive drugs were scored as 0; no alcohol consumption (including abstinence for more than 6 months) was 0, a small amount of alcohol consumption (<60 g/d) was 1, and heavy alcohol consumption (≥60 g/d) was 2; according to the normal reference value range of uric acid (155-428) μmol/L, uric acid less than 155 μmol/L was 0, 155-428 μmol/L was -1, and >428 μmol/L was -2; according to the normal value of HDL cholesterol of 1.16-1.55 mmol/L, HDL cholesterol<1.55 mmol/L was 0, ≥1.55 mmol/L was 2; for LDL cholesterol, <2 mmol/L was 0, 2- 4 mmol/L was -1, and >4 mmol/L was -2. All scores range from -10 to 7.
Different specificity and sensitivity can be obtained by selecting different score values (Table 2). One of the score values, as determined by the largest Youden index, was selected as the cut-off point. As shown in Table 2, when the patient’s score was >0, they were classified into the CH group. Conversely, when the score was ≤0, the patients were classified into the CI group; sensitivity and specificity were 68.0% and 68.0%, respectively. The ROC curve was plotted as shown in Figure 1 and the AUC was 0.74±0.02 (P<0.001).
Table 2.
Cerebral ischemia-cerebral hemorrhage risk scores
| The score value is greater than | Sensitivity/% | Specificity/% | Youden index |
|---|---|---|---|
| -6 | 100.0 | 0 | 0.000 |
| -5 | 99.8 | 0.2 | 0.000 |
| -4 | 99.2 | 2.0 | 0.012 |
| -3 | 97.9 | 9.8 | 0.078 |
| -2 | 94.0 | 26.3 | 0.203 |
| -1 | 82.1 | 48.3 | 0.305 |
| 0 | 68.0 | 68.0 | 0.360 |
| 1 | 50.1 | 83.8 | 0.339 |
| 2 | 27.4 | 92.3 | 0.198 |
| 3 | 13.9 | 97.5 | 0.114 |
| 4 | 6.0 | 99.2 | 0.052 |
| 5 | 2.1 | 99.7 | 0.017 |
| 6 | 0.4 | 100.0 | 0.004 |
| 7 | 0.0 | 100.0 | 0.000 |
Figure 1. ROC curve of cerebral ischemia-cerebral hemorrhage risk scale.
2.2. Validation
DB-2, which included 119 patients (51 were CH and 68 were CI), was then tested; the average age was 63.20 years and 74 of the patients were male. Table 3 lists the indicators of patients in the DB-2. The risk score of CI-CH was used to predict cerebrovascular disease type for the 119 patients in DB-2. The results were as follows: 53 patients with CI were correctly classified into the CI group and 38 patients with CH were correctly classified into the CH group, while 15 patients with CI and 13 patients with CH were incorrectly classified. The sensitivity, specificity, accuracy, and misdiagnosis rates were 74.5%, 77.9%, 76.4%, and 23.5%, respectively. As compared with the actual classification results, the Kappa value was 0.522 (P<0.001).
Table 3.
Data distribution features of the test dataset (n=119)
| Variables | Values | Variables | Values |
|---|---|---|---|
| Age/year | 63.20±11.21 | Drink/[No.(%)] | |
| Height/m | 1.63±0.07 | No alcohol consumption | 90(75.6) |
| Weight/kg | 62.48±9.67 | Temperance | 9(7.6) |
| BMI/(kg·m-2) | 23.34±3.18 | A small amount of alcohol consumption | 10(8.4) |
| Waist circumference/cm | 84.91±8.55 | Heavy alcohol consumption | 10(8.4) |
| Hip circumference/cm | 90.57±6.77 | Drinking consumption/(g·d-1) | 65.21±91.02 |
| Waist-to-hip ratio | 0.94±0.07 | Smoking/[No.(%)] | |
| Hypertension/[No.(%)] | 94(79.0) | Non-smoker | 76(63.9) |
| Hypertension treatment/[No.(%)] | Ex-smokers | 11(9.2) | |
| Good control | 24(25.5) | Smokers | 32(26.9) |
| Poor control | 52(55.3) | Number of cigarettes in smokers | 611.88±396.15 |
| First found | 18(19.1) | Number of cigarettes in ex-smokers | 787.27±777.55 |
| Hypertension classification/[No.(%)] | Body building/[No.(%)] | 31(26.1) | |
| Grade 1 | 19(20.2) | Prothrombin time/s | 12.17±1.66 |
| Grade 2 | 36(38.3) | International normalized ratio (INR) | 0.98±0.16 |
| Grade 3 | 39(41.5) | Activated partial thromboplastin time/s | 30.13±8.10 |
| Type 2 diabetes/[No.(%)] | 24(20.2) | Fibrinogen/(g·L-1) | 3.55±1.17 |
| Diabetes treatment/[No.(%)] | Thrombin time/s | 17.08±2.00 | |
| Good control | 10(41.7) | Diabetic patients with fasting blood | 7.49±1.69 |
| Poor control | 12(50.0) | glucose/(mmol·L-1) | |
| First found | 2(8.3) | Diabetic patients with glycosylated | 8.52±4.21 |
| Hyperlipidemia | 49(41.2) | hemoglobin/% | |
| Hyperlipidemia treatment/[No.(%)] | Fasting blood glucose/(mmol·L-1) | 6.11±2.56 | |
| Good control | 6(12.2) | Alanine aminotransferase/(U·L-1) | 26.96±29.79 |
| Poor control | 10(20.4) | Aspartate aminotransferase/(U·L-1) | 30.13±39.58 |
| First found | 33(67.3) | Total bilirubin/(μmol·L-1) | 17.59±9.77 |
| Coronary heart disease/[No.(%)] | 10(8.4) | Bilirubin direct/(μmol·L-1) | 6.44±4.46 |
| Chronic kidney disease/[No.(%)] | 7(5.9) | Albumin/(g·L-1) | 39.87±4.77 |
| Chronic obstructive pulmonary disease/[No.(%)] | 8(6.7) | Globulin/(g·L-1) | 27.67±4.52 |
| Cancer/[No.(%)] | 3(2.5) | Total bile acid/(μmol·L-1) | 4.49±8.06 |
| Chronic cholecystitis/[No.(%)] | 2(1.7) | Blood urea nitrogen/(mmol·L-1) | 5.61±3.07 |
| Gout/[No.(%)] | 4(3.4) | Creatinine/(μmol·L-1) | 97.09±143.05 |
| Urolithiasis/[No.(%)] | 3(2.5) | Uric acid/(μmol·L-1) | 290.87±103.79 |
| Drug use/[No.(%)] | Low-density lipoprotein/(mmol·L-1) | 2.62±0.80 | |
| Aspirin | 17(14.3) | High-density lipoprotein/(mmol·L-1) | 1.44±0.39 |
| Statatin | 8(6.7) | Total cholesterol/(mmol·L-1) | 4.93±1.09 |
| Hypotensor | 46(48.9) | Triglyceride/(mmol·L-1) | 1.71±1.07 |
| Hypoglycemic drugs | 12(50.0) |
3. Discussion
Controlling the risk factors for cerebrovascular diseases is the main means to prevent the occurrence of cerebrovascular diseases[10-11]. About 90% of all cerebrovascular diseases are linked to these risk factors. Control of these risk factors can maximize the prevention of cerebrovascular diseases[12]. There are many risks of CI and CH, such as hypertension, diabetes, alcoholism, weight, etc.,[13] and many of them are common to both. Hypertension is a risk factor for both lobular and non-lobular hemorrhage[14]. Studies[15] have found an increased incidence of intracerebral hemorrhage in older patients, but the impact of the above factors on CI/CH is not clear. When patients have multiple risk factors at the same time, how to prescribe individualized treatment, such as the need for antiplatelet aggregation drugs, becomes a practical problem for clinicians. Researchers investigated more than 40 risk factors related to CI and CH in this study, and 18 indexes showed differences across CI and CH patients. Based on the analysis of these 18 indicators, it was found that patients with advanced age, obesity, diabetes, and high uric acid were more likely to have CI while patients with hypertension, especially uncontrolled grade 3 hypertension, and alcoholism were more likely to have CH. The results suggested that clinicians should adequately assess the risk of CI and CH in patients with these conditions when developing prevention programs of cerebrovascular diseases.
The establishment of a CI-CH risk scale is one of the key objectives of this study. The AUC is an important index used to evaluate the quality of the CI-CH risk scale; generally, AUC needs to be higher than 0.5 for the CI-CH risk scale to be meaningful. The closer to 1, the better the prediction performance. The AUC values of common cerebrovascular disease prediction models are not high, with Essen’s score being 0.56-0.62[16-18], SPI-II’s score being 0.63-0.65[8, 18], and ABCD2’s score being 0.66[19]. The CI-CH risk scale was successfully established in this study, which included 9 risk factors from 5 dimensions, including previous disease, demographic characteristics, lifestyle, medication, and biochemical indicators. The AUC value of the CI-CH risk scale fluctuated between 0.77 and 0.78. When the CI-CH risk scale was tested using DB-2, its sensitivity, specificity, and accuracy were 74.5%, 77.9%, and 76.4%, respectively, with a misjudgment rate of 23.5%. As compared to the actual classification results, the Kappa value was 0.522 (P<0.001), suggesting a good degree of agreement.
Cerebrovascular disease is a very complex disease system. The known risk factors for cerebrovascular disease can be divided into 3 categories: biological, psychological, and social factors. There are more studies on biological factors, such as previous related diseases, vascular structure, blood composition, etc. Personality traits, social emergencies, and weather also affect the occurrence of cerebrovascular diseases. In future studies, we hope to construct a more perfect database of cerebrovascular disease, collect the risk factors from the 3 aspects more comprehensively, and use more advanced analysis methods to answer the 4 questions: 1) whether cerebrovascular disease will occur; 2) the probability of occurring a cerebrovascular disease; 3) the type of cerebrovascular disease (CH or CI); and 4) specific onset time.
In a word, the CI-CH scale can help clinicians predict the subtypes of cerebrovascular diseases and assist in their prevention.
Funding Statement
This work was supported by the Hunan Provincial Clinical Medical Technology Innovation Guidance Project (2020SK53612) and the Natural Science Foundation of Hunan Province (2021JJ30998), China.
Conflict of Interest
The authors declare that they have no conflicts of interest to disclose.
AUTHORS’CONTRIBUTIONS
Contributions: YAN Wenguang Conceptualized and designed the study, critically reviewed and revised the manuscript; CHEN Ru Collected data, reviewed and revised the manuscript; HU Hao Collected data, drafted, edited, and submitted the manuscript; XU Jiamiao, ZHENG Wen Conducted the statistical analysis, drafted, reviewed and revised the manuscript; SONG Zhi Critically reviewed and revised the manuscript. All authors have approved the final version of this manuscript.
Note
http://xbyxb.csu.edu.cn/xbwk/fileup/PDF/202207928.pdf
References
- 1. Feigin VL, Forouzanfar MH, Krishnamurthi R, et al. Global and regional burden of stroke during 1990―2010: findings from the Global Burden of Disease Study 2010[J]. Lancet, 2014, 383(9913): 245-254. 10.1016/s0140-6736(13)61953-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wang ZK, Hu SB, Sang SP, et al. Age-period-cohort analysis of stroke mortality in China: data from the global burden of disease study 2013[J]. Stroke, 2017, 48(2): 271-275. 10.1161/STROKEAHA.116.015031. [DOI] [PubMed] [Google Scholar]
- 3. Xu J, Wang AX, Wang Q, et al. Efficacy of clopidogrel for stroke depends on CYP2C19 genotype and risk profile[J]. Ann Neurol, 2019, 86(3): 419-426. 10.1002/ana.25535. [DOI] [PubMed] [Google Scholar]
- 4. 刘丽萍. 真实世界研究中抗血小板药物单药治疗对缺血性卒中患者的有效性和安全性[J]. 中华内科杂志, 2019, 58(4): 324-328. 10.3760/cma.j.issn.0578-1426.2019.04.019. [DOI] [Google Scholar]; LIU Liping. The efficacy and safety of antiplatelet monotherapy in patients with ischemic stroke in the real world[J]. Chinese Journal of Internal Medicine, 2019, 58(4): 324-328. 10.3760/cma.j.issn.0578-1426.2019.04.019. [DOI] [PubMed] [Google Scholar]
- 5. Mullen L, Meah MN, Elamin A, et al. Risk of major bleeding with potent antiplatelet agents after an acute coronary event: acomparison of ticagrelor and clopidogrel in 5116 consecutive patients in clinical practice[J/OL]. J Am Heart Assoc, 2021, 10(8): e019467[2021-12-31]. 10.1161/JAHA.120.019467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ling X, Yan SM, Shen B, et al. A modified Essen Stroke Risk Score for predicting recurrent ischemic stroke at one year[J]. Neurol Res, 2018, 40(3): 204-210. 10.1080/01616412.2018.1428389. [DOI] [PubMed] [Google Scholar]
- 7. Kernan WN, Viscoli CM, Brass LM, et al. The stroke prognosis instrument II (SPI-II): a clinical prediction instrument for patients with transient ischemia and nondisabling ischemic stroke[J]. Stroke, 2000, 31(2): 456-462. 10.1161/01.str.31.2.456. [DOI] [PubMed] [Google Scholar]
- 8. MerwickÁ, Albers GW, Amarenco P, et al. Addition of brain and carotid imaging to the ABCD2score to identify patients at early risk of stroke after transient ischaemic attack: amulticentre observational study[J]. Lancet Neurol, 2010, 9(11): 1060-1069. 10.1016/S1474-4422(10)70240-4. [DOI] [PubMed] [Google Scholar]
- 9. Sr D'Agostino RB, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study[J]. Circulation, 2008, 117(6): 743-753. 10.1161/CIRCULATIONAHA.107.699579. [DOI] [PubMed] [Google Scholar]
- 10. Turan TN, Nizam A, Lynn MJ, et al. Relationship between risk factor control and vascular events in the SAMMPRIS trial[J]. Neurology, 2017, 88(4): 379-385. 10.1212/WNL.0000000000003534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Mente A, O’Donnell M, Rangarajan S, et al. Associations of urinary sodium excretion with cardiovascular events in individuals with and without hypertension: a pooled analysis of data from four studies[J]. Lancet, 2016, 388(10043): 465-475. 10.1016/S0140-6736(16)30467-6. [DOI] [PubMed] [Google Scholar]
- 12. O'Donnell MJ, Chin SL, Rangarajan S, et al. Global and regional effects of potentially modifiable risk factors associated with acute stroke in 32 countries (INTERSTROKE): a case-control study[J]. Lancet, 2016, 388(10046): 761-775. 10.1016/S0140-6736(16)30506-2. [DOI] [PubMed] [Google Scholar]
- 13. Aigner A, Grittner U, Rolfs A, et al. Contribution of established stroke risk factors to the burden of stroke in young adults[J]. Stroke, 2017, 48(7): 1744-1751. 10.1161/STROKEAHA.117.016599. [DOI] [PubMed] [Google Scholar]
- 14. Jolink WMT, Wiegertjes K, Rinkel GJE, et al. Location-specific risk factors for intracerebral hemorrhage: systematic review and meta-analysis[J/OL]. Neurology, 2020, 95(13): e1807-e1818 [2021-11-21]. 10.1212/WNL.0000000000010418. [DOI] [PubMed] [Google Scholar]
- 15. Lioutas VA, Beiser AS, Aparicio HJ, et al. Assessment of incidence and risk factors of intracerebral hemorrhage among participants in the Framingham heart study between 1948 and 2016[J]. JAMA Neurol, 2020, 77(10): 1252-1260. 10.1001/jamaneurol.2020.1512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Weimar C, Diener HC, Alberts MJ, et al. The Essen stroke risk score predicts recurrent cardiovascular events: a validation within the reduction of Atherothrombosis for Continued Health (REACH) registry[J]. Stroke, 2009, 40(2): 350-354. 10.1161/STROKEAHA.108.521419. [DOI] [PubMed] [Google Scholar]
- 17. Ohman EM, Bhatt DL, Steg PG, et al. The reduction of Atherothrombosis for Continued Health (REACH) registry: an international, prospective, observational investigation in subjects at risk for atherothrombotic events-study design[J/OL]. Am Heart J, 2006, 151(4): 786.e1-786. 10[2021-11-13]. 10.1016/j.ahj.2005.11.004. [DOI] [PubMed] [Google Scholar]
- 18. Weimar C, Benemann J, Michalski D, et al. Prediction of recurrent stroke and vascular death in patients with transient ischemic attack or nondisabling stroke: a prospective comparison of validated prognostic scores[J]. Stroke, 2010, 41(3): 487-493. 10.1161/STROKEAHA.109.562157. [DOI] [PubMed] [Google Scholar]
- 19. Giles MF, Albers GW, Amarenco P, et al. Addition of brain infarction to the ABCD2 score (ABCD2I): a collaborative analysis of unpublished data on 4574 patients[J]. Stroke, 2010, 41(9): 1907-1913. 10.1161/STROKEAHA.110.578971. [DOI] [PubMed] [Google Scholar]