Joint effects of risk factors for stroke and transient ischemic attack in a German population
Single, modifiable risk factors for stroke have extensively been studied. In contrast, differences of their combined effects among stroke and transitoy ischemic attack (TIA) have been rarely investigated. The aim of the present study was to assess single and joint effects of risk factors on the incidence of stroke and TIA and to compare their magnitudes in a large population-based German cohort.
Incident cases of stroke and TIA were identified among 25538 participants (aged 35–65 at baseline) of the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. Relative risks for stroke and TIA related to modifiable risk factors were estimated using Cox proportional hazard models.
During 4.3 years of follow-up 100 stroke cases and 112 TIA cases occurred. Incidences of stroke and TIA were 91.7 and 102.7 per 100 000 person-years, respectively. Relative risks for ischemic stroke (RR 5.12, 95% CI 1.49–17.6, p for trend <0.0001) and for TIA (RR 3.08, 95% CI 1.00–9.44, p for trend <0.024) were highest among participants having 4 or 5 modifiable risk factors. 58.5% of ischemic strokes and 26.2% of TIA cases were attributable to the 5 risk factors hypertension, diabetes mellitus, high alcohol consumption, hyperlipidemia, and smoking.
Our data indicate that classical risk factors may explain almost 60% of ischemic stroke but only one in four TIA cases. Analysing potential differences of known risk factors between ischemic stroke and TIAs and the identification of other determinants of ischemic attacks are important steps to better explain the burden of stroke.
Keywordsstroke TIA incidence relative risk epidemiology
Stroke is among the leading causes of morbidity and mortality in developed countries  and it is the most important cause of impairment and disability in the elderly. Treatment options of the cause are limited to thrombolysis, but only few patients receive this treatment owing to restrictions in application time and indication . Thus, primary prevention remains the most important general strategy for reducing the impact of stroke . Single risk factors have been extensively studied but their joint effects on stroke occurrence have been less frequently investigated . If evaluated, this was mainly done for the purpose of stroke risk score development . As there are several well-established, so called “classical”, modifiable risk factors such as hypertension, smoking, cardiac disease, and diabetes, stroke appears to be a preventable disease to a large extent [18, 25]. In addition, the aim of many primary prevention programs is a change in lifestyle, thus, in various health related behaviors and not just in a single risk factor. Lifestyle changes are likely to influence risk factor prevalences, which in turn may modify the risk of stroke .
The transitory ischemic attack (TIA) constitutes an established risk factor for ischemic stroke . Thus, the primary and secondary prevention of TIA is also an important goal for efficiently decreasing stroke incidence. Diagnosis of TIA is based on clinical symptoms and differs from ischemic stroke in its duration. By definition the symptoms of TIA are lasting less then 24hours . This definition goes back to the era before brain imaging was widely available and is now a matter of controversy [1, 14, 27]. Varying criteria used by clinicians to define and identify TIA cases might be a reason why few prior studies have evaluated risk factors for TIA .
Our study aimed at assessing the incidence of stroke and TIA in the “European Prospective Investigation into Cancer and Nutrition (EPIC) – Potsdam Study” and to compare the single and combined effects of so called classical risk factors on the incidence of stroke and TIA.
The European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study is part of the large-scale European-wide prospective cohort study EPIC and includes 27,548 individuals aged 35 to 65 years who were recruited between 1994 and 1998 from the general population . All participants gave written informed consent and the Ethics Committee of the Federal State Brandenburg gave approval for all study procedures. Information about changes in lifestyle and incident diseases is biennially assessed by a self-administered questionnaire .
In the follow-up questionnaire a validated, brief screening instrument for stroke and TIA was included . It consisted of one general question asking for physician-diagnosed stroke in the past and three questions on common stroke or TIA symptoms. The latter included weakness or paralysis of a limb, slurred speech or problems in speaking and visual symptoms. To improve the understanding of visual symptoms photographs demonstrating hemianopsia, central scotoma, diplopia, monocular loss of vision, and loss of vision at the top of the form were used .
Subsequent validation of stroke and TIA was based on medical records and followed an established protocol [2, 4] that included a standardized form filled in by the treating physician or the study physician. Information was collected about subtype of stroke with ICD-10 code, hospitalization, symptoms, duration, diagnostic procedures and outcome. A stroke was defined as a focal neurological deficit of sudden onset and vascular mechanism lasting longer than 24 hours, and a TIA less than 24 hours, respectively. For definition of TIA we did not consider pathological findings in brain imaging. On the basis of all information strokes were classified into ischemic (ICD-10 I63.0–I63.9 ), intracerebral (ICD-10 I61.0–I61.9 ) and subarachnoidal hemorrhage (ICD-10 I60.0–I60.9 ), undetermined strokes (ICD-10 I64.0–I64.9 ) and TIA (ICD-10 G45.0–G45.9 ) by two physicians in the study center.
Participants with missing follow-up status (n = 723) as well as subjects with self reported prevalent stroke or TIA at baseline (n = 1287) were excluded, leaving 25538 participants for the analyses of incident events. Among these, we identified 100 cases of stroke and 112 cases of TIA occurring between baseline and the second follow-up. For risk estimations we further excluded 2 ischemic stroke cases and 234 non-cases with missing covariate variables as well as those 2 participants with an undefined stroke.
Assessment of risk factors and covariates
Baseline assessment of lifestyle factors, medical history, and socio-demographic characteristics was done by computer-assisted person-to-person interviews or self-administered questionnaires . Anthropometric data and blood pressure were measured by trained and quality-monitored personnel . All participants filled in a food frequency questionnaire that included questions on frequency and portion size of 148 food items eaten during the previous year .
Educational attainment was dichotomized into less than high school education versus high school education or university degree. Blood pressure was measured three times following a standard protocol. Hypertension was defined as a systolic blood pressure (140 mmHg or a diastolic blood pressure of (90 mmHg or taking antihypertensive medication or self report of a diagnosis of hypertension. A history of diabetes was based on self reports of a diagnosis or taking antidiabetic medication. A history of hyperlipidemia was based on self reports of a diagnosis or of taking cholesterol-lowering medication. If participants reported a previous diagnosis of coronary heart disease or myocardial infarction at baseline, a prevalent ischemic heart disease has been assumed. Physical activity was dichotomized into the categories less than two hours sporting activity, and two or more hours of physical activity per week. Smoking status was defined as nonsmoker and former smoker, and current smoker. For the purpose of calculation of the combined risks the three categories of smoking were collapsed into nonsmoker plus former smoker versus current smoker. Body mass index (BMI) was calculated as body weight divided by body height squared. Obesity was defined as a body mass index of ≥30 kg/m2. High alcohol consumption was defined as drinking more than 15 g of alcohol per day in women and drinking more than 30 g alcohol per day in men.
All statistical analyses were performed using SAS software, version 9.13 (SAS Institute, Cary, NC, USA). Age standardized rates of stroke and TIA were calculated for four age categories using the age distribution of all participants. Baseline characteristics of participants are expressed as mean ± standard deviation (SD) or percentages.
We used Cox’s proportional hazard model for analyses of the association between five modifiable classical risk factors (hypertension, current smoking, diabetes, hyperlipidemia and high alcohol consumption) and incident stroke or TIA. Age was used as the underlying time variable in the counting process with entry time defined as the subject’s age at recruitment and exit time defined as the subject’s age at stroke or TIA diagnosis or censoring. Relative risks were estimated sex adjusted and in further analyses additionally adjusted for educational attainment, obesity, physical activity, prevalent coronary heart disease and the respective above-mentioned other modifiable classical risk factors.
Furthermore the association between the number of present risk factors (from hypertension, current smoking, diabetes, hyperlipidemia and high alcohol consumption) and ischemic stroke or TIA was estimated. For this purpose, we defined indicator variables for the presence of one, two, three, and more than three risk factors. Trends in ischemic stroke and TIA risk were assessed in Cox’s proportional hazard models using the number of present risk factors as an independent variable. Population attributable risks were estimated for each of the five selected risk factors separately and combined. Here, we applied the formula of Bruzzi et al.  using the prevalence of risk factors within cases and estimates of adjusted relative risks .
Incidence rates for TIA, ischemic, hemorrhagic and total stroke per 100,000 person years during 4.3 years of follow-up in the EPIC-Potsdam Study
65 + y
Clinical aspects of stroke and TIA in the EPIC-Potsdam Study
TIA (n = 112)
Ischemic stroke (n = 83)
Hemorrhagic stroke (n = 15)
Total Stroke (n = 100)
Baseline characteristicsa for cases and non-cases according to stroke or TIA status in the EPIC-Potsdam Study
TIA (n = 112)
Ischemic strokes (n = 81)
Hemorrhagic stroke (n = 15)
Non-cases (n = 25092)
56.6 ± 7.1
56.4 ± 7.6
53.6 ± 9.2
49.7 ± 9.0
Coronary heart disease
26.8 ± 3.6
27.2 ± 4.3
27.7 ± 5.4
26.2 ± 4.3
Current cigarette smoker
High alcohol consumptione
Physical activity (≥2 h/week)
Relative risks (95% CI) of five selected classical modifiable risk factors for TIA and ischemic stroke in the EPIC-Potsdam Study
TIA (n = 112)
Ischemic stroke (n = 81)
High alcohol consumption
Hyperlipidemia was significantly associated with the risk for TIA in the sex-adjusted model, but no significant association was observed for ischemic or hemorrhagic stroke. Sex- and multivariate-adjusted RR of high alcohol consumption, diabetes mellitus and current smoking for hemorrhagic stroke were not significant.
Prevalence and relative risks a for the increasing number of selected modifiable classical risk factors (hypertension, smoking, diabetes, hyperlipidemia and high alcohol consumption) in the EPIC-Potsdam Study
Number of dichotomised risk factors
Non-cases (n = 25092)
TIA (n = 112)
Ischemic stroke (n = 81)
RR (95% CI)
RR (95% CI)
4 or 5
P for trend:
In this population-based prospective cohort study with more than 25000 participants a lower incidence for stroke than for TIA and a strong age dependency for both outcomes were observed. The stroke incidence observed in this population-based study is comparable with the data reported in other studies of the general population in Germany . Data on the incidence of TIA in population studies are scarce and only reported in one prior study in Germany . Among the few available data two studies evaluating a relationship between TIA and stroke incidence observed a higher incidence for stroke than TIA [9, 20]. However, TIA is still an event difficult to identify in epidemiologial studies, because of varying definitions applied by clinicians . In a retrospective study by Lemesle et al.  a TIA was diagnosed, when the clinical symptoms disappeared within 24 hours and no sequelae were noted on CT. In our study TIA was solely based on clinical symptoms. Brain imaging was not taken into consideration because it was not done in many TIA cases, although it showed pathological findings in about one fourth of our TIA cases. Thus, differences in the definition of TIA may explain, at least in part, differences in incidences.
Hypertension was confirmed as the single most important risk factor for both ischemic and hemorrhagic stroke. The prevalence of two risk factors was associated with a more than 2-fold risk and the prevalence of 4 or 5 factors with a more than 5-fold increased risk for ischemic stroke. The combined risk factors explained as much as 60% of the ischemic stroke incidence in this population. In contrast the combined risk effects were considerably lower in magnitude among TIA cases. A prevalence of 4 or 5 risk factors yielded a 3-fold increased risk and explained only one fourth of incident TIA cases in the EPIC-Potsdam cohort.
Interestingly, associations between most classical modifiable risk factors and incident TIAs were weaker than those observed for stroke. There are only very few studies investigating risk factors for TIA and their results have been inconsistent [26, 28, 29].
In contrast to our data, Whisnant et al.  observed similar odds ratios for stroke and TIA related to the classical risk factors hypertension and diabetes conducted in the United States. Only for smoking did the study detect a higher odds ratio for stroke than for TIA.
Self reported hyperlipidemia is an important risk factor for TIA in our study population. This is biologically plausible, since hyperlipidemia is a key factor in the development of arteriosclerosis. It must be kept in mind that subjects aware of this condition may also be relatively health conscious and more sensitive to mild TIA symptoms, which may in part be responsible for the observed associations.
Our study is one of the very few studies investigating joint effects of classical risk factors on stroke and TIA. In contrast to the few other studies  our aim was not to develop a new stroke risk prediction model but to compare risk factor magnitudes between stroke and TIA. The most commonly used risk score for stroke, developed by the Framingham Study , was based on the inclusion of both TIA and stroke cases. Our results suggest that pooling both types of events would underestimate predicted stroke risks due to the different magnitude of combined risks for both outcomes.
However, it has to be pointed out that previous studies investigating joint effects of classical modifiable risk factors were conducted solely in North America [22, 34], where the impact of modifiable risk factors and their combinations may differ significantly from their impact in a European population. For instance the prevalence of obesity is much higher in the United States than in Germany [6, 12]. Thus our data may be specific for German populations.
The prospective study design, the large sample size and the use of a validated stroke symptom questionnaire that also allowed the classification of TIA  are among the strengths of our investigation. Furthermore, all cases were verified by medical records. Our study also provides information on aspects of diagnostic management and treatment of TIA, which varies among different physicians. Not all patients with a TIA were admitted to a hospital as an emergency, although this is generally recommended.
Owing to difficulties in accurately diagnosing TIA misclassification may be a relevant threat to the validity of studies investigating TIA risk factors, since it may result in substantial attenuation of relative risk estimates and could be responsible for differences in risk factor profiles between TIA and stroke. In particular, this issue may apply to the non-hospitalized TIA cases included in our study. But our findings are in line with findings from a clinical register, which clearly showed differences in the prevalence of classical risk factors between TIA and ischemic stroke cases even though only hospitalized cases were included . As observed in our study, patients with a TIA were less likely to be smokers and to have diabetes mellitus or hypertension, but more likely to have a history of hyperlipidemia.
Our study population is limited to participants aged 35–65 at baseline. It is well-known that stroke is a disease of older ages and the risk of stroke doubles for each successive decade after age 55 [5, 13]. The impact of risk factors may change with subject’s age, as has been shown for hypertension . Thus, the findings of our study may be specific for a middle-aged German population and applicability may be limited for older populations.
The mean follow-up of 4.3 years is rather short, restricting the number of incident TIA and stroke cases. This limits the statistical power to detect significant associations between risk factors and outcomes, especially for hemorrhagic stroke. The assessment of risk factor status was mainly based on self-reports (without measurements of e.g. cholesterol or fasting glucose at baseline), which is a potential source of bias. However, the relatively high socioeconomic status of our study population may be associated with a sufficient quality of self-reports . Another issue may be introduced by dichotomising the risk factor variables. Thereby the information on a specific factor will be simplified and the chosen cut points for selected risk factors such as hypertension and alcohol may not reflect true threshold levels of the dose-response relationship. Finally, we were unable to assess the prevalence of atrial fibrillation in the study population, which is an important risk factor for ischemic stroke in older ages .
In conclusion, almost 60% of incident ischemic stroke in the EPIC-Potsdam cohort could be explained by classical modifiable risk factors, describing the potential of lifestyle changes for stroke risk reduction. Hypertension is confirmed as the most important single risk factor for stroke, but may be less important for TIA. Our findings demonstrate that only 25% of TIA cases can be attributed to the classical risk factors for stroke. TIA is an important condition for the subsequent development of stroke. Thus, a better characterization of the risk factor profile of TIA is of high importance for refining strategies of primary stroke prevention.
The authors thank all study participants for their cooperation. We especially thank our study physician Wolfgang Fleischhauer and our data manager Kay Behling and Ellen Kohlsdorf.
- 13.Goldstein LB, Adams R, Alberts MJ, Appel LJ, Brass LM, Bushnell CD, Culebras A, Degraba TJ, Gorelick PB, Guyton JR, Hart RG, Howard G, Kelly-Hayes M, Nixon JV, Sacco RL (2006) Primary prevention of ischemic stroke: a guideline from the American Heart Association/American Stroke Association Stroke Council: cosponsored by the Atherosclerotic Peripheral Vascular Disease Interdisciplinary Working Group; Cardiovascular Nursing Council; Clinical Cardiology Council; Nutrition, Physical Activity, and Metabolism Council; and the Quality of Care and Outcomes Research Interdisciplinary Working Group: the American Academy of Neurology affirms the value of this guideline. Stroke 37:1583–1633PubMedGoogle Scholar
- 15.Hu G, Sarti C, Jousilahti P, Peltonen M, Qiao Q, Antikainen R, Tuomilehto J (2005) The Impact of History of Hypertension and Type 2 Diabetes at Baseline on the Incidence of Stroke and Stroke Mortality. StrokeGoogle Scholar
- 17.Kroke A, Klipstein-Grobusch K, Voss S, Moseneder J, Thielecke F, Noack R, Boeing H (1999) Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am J Clin Nutr 70:439–447PubMedGoogle Scholar
- 25.Sacco RL, Benjamin EJ, Broderick JP, Dyken M, Easton JD, Feinberg WM, Goldstein LB, Gorelick PB, Howard G, Kittner SJ, Manolio TA, Whisnant JP, Wolf PA (1997) American Heart Association Prevention Conference. IV. Prevention and Rehabilitation of Stroke. Risk factors. Stroke 28:1507–1517PubMedGoogle Scholar
- 30.WHO (1992) International Statistical Classification of Diseases and Related Health Problems. World Health Organisation, GenevaGoogle Scholar