Methods

Our study was based on both the Dijon Stroke Registry, which has been running since 1985,9 10 and the registry of myocardial infarction of Dijon and Côte d’Or (Rico)11 which was launched in 2001. Data from these two registries between 2001 and 2006 have been pooled in a retrospective analysis for a collaborative study called the DIjon VAscular (Diva) project.

Study area and population

The studied population was that of the city of Dijon comprising 151 846 inhabitants (82 144 women and 69 702 men). Fifteen per cent of the population were 65 years of age or older. Less than 5% of migration was observed over the study period with no change in the economic status.

Ascertainment of stroke

Case ascertainment

Since 1985, the stroke registry of Dijon has been using a constant prospective method of ascertainment, for patients of all ages, and without interruption. A detailed description of the Dijon Stroke Registry has been published elsewhere.9 10 To ensure completeness of case ascertainment, information was provided from five sources: (1) the emergency rooms, and all of the clinical and radiological departments of Dijon University Hospital; (2) the emergency rooms and all of the clinical departments of the three private hospitals of the city and its suburbs; (3) the patient’s home or from the nursing homes of the city, with diagnosis assessed by the 250 general practitioners with the help of an outpatients clinic; (4) the three private radiological centres where the medical records were reviewed to identify missed cases that had not been transferred to the registry by the general practitioners; (5) the ultrasound Doppler centres of the University Hospital and private centres where medical records were reviewed. Stroke was defined according to World Health Organization recommendations and the International Classification of Disease.

Diagnosis of stroke subtypes and classification

The diagnosis of stroke subtypes was always performed on clinical criteria, cerebral imaging (CT scan and/or MRI examination) and complementary investigations, including ECG, echocardiography, cervical artery ultrasonography and standard blood and urine tests.9 10 We included: (1) ischaemic strokes from lipohyalinosis of small arteries, so-called lacunar infarct; (2) ischaemic strokes from cardiac embolism originating from atrial fibrillation (AF) or valve disease, patent foramen ovale or spontaneous intracavitar thrombus; (3) other ischaemic strokes defined by focal cortical symptoms, and cortical infarct on a large vascular territory on CT or MRI. This group included ischaemic strokes from atheroma of large arteries and ischaemic stroke from other or undetermined causes (unfortunately, these different types of ischaemic strokes were not separated in our registry); (4) primary cerebral haemorrhagic (PCH); and (5) subarachnoid haemorrhagic (SAH) stroke. When it was difficult to differentiate between ischaemic strokes, lacunar infarct and cardiac embolism strokes, medical staff meetings were held to classify the difficult cases into one of the three categories.

Ascertainment of acute myocardial infarction

The design and methods of the population based Rico survey have been published previously.11 Briefly, since 1 January 2001, the Rico registry has been collecting data from patients hospitalised with AMI in the City of Dijon and in five cities in the Côte d’Or area. To ensure the completeness of case ascertainment, the collaboration of general practitioners and private cardiologists was also required in order to identify patients who had suffered an AMI without hospitalisation in one of the cities. For the Diva study, only AMI patients residing in Dijon were considered. Both ST elevation MI and non-ST-elevation MI were diagnosed according to the European Society of Cardiology and the American College of Cardiology criteria.11 Data were collected at each site by a study coordinator trained in completing the core record form and in extracting data from medical records, using a standardised case report form. Cases were ascertained by prospective collection.

Vascular risk factors and follow-up for stroke and AMI

Vascular risk factors were collected by the two teams for stroke and AMI with a common methodology as previously described.9 10 11 Hypertension was defined by history of known hypertension (systolic blood pressure ⩾140 mm Hg and/or diastolic blood pressure ⩾90 mm Hg) or antihypertensive treatment. Diabetes mellitus was recorded if a glucose level of ⩾7.8 mmol/l had been reported in the medical record or if the patient was receiving insulin or oral hypoglycaemic agents. Hypercholesterolaemia was defined by total cholesterol level ⩾5.7 mmol/l. We also recorded tobacco use (1 cigarette per day, current or former habit), and AF diagnosed from ECG or Holter recordings. For stroke, a history of transient ischaemic attack (TIA) and previous AMI were recorded, and for AMI, a history of TIA and stroke were collected.

After hospital discharge, information at 30 days was acquired by contacting patients individually or their relatives or treating physician, and by reviewing the hospital records if the patient had been rehospitalised. For patients who had died from stroke or AMI in the prehospital setting, the death certificates were obtained from the database of the Regional Health Service in charge of recording deaths in the region of Côte d’Or. Information on case fatalities gathered from these various sources for the whole study population was thus complete.

Statistical analysis

To measure the incidence rates, the National Institute of Statistics provided census data for 1982, 1990 and 1999 for the population in Dijon in l year age groups and by sex. The population was estimated from these censuses by extrapolation. Age specific and sex specific incidence rates per 100 000 of the population per year were calculated for all cases of first ever stroke and AMI, for incident events and for 30 day CFRs. Although all comparisons were made within the same population, rates were standardised to any French, European and World population to allow comparisons with data from the literature. We calculated 95% confidence intervals (CIs) using the Poisson distribution. Differences with regard to age, sex and disease rates were investigated by Poisson regression analyses. The incidence rate ratio by sex, age (<55, 55–75, >75 years) and event was calculated. CFRs were based on survival after 30 days and trends were evaluated with linear regression. Univariate and multivariate analyses were performed using logistic regression to identify the determinants of death before 30 days for stroke and myocardial infarction. The variables included were the time period (2001–2002, 2003–2004, 2005–2006), age, sex, hypertension, hypercholesterolaemia, diabetes, initial glycaemia, tobacco consumption, history of TIA, atrial fibrillation and either history of AMI and stroke subtype for stroke patients or history of stroke for AMI patients. Odds ratios and their 95% CIs were calculated for all the variables in the univariate analysis, and only independent variables were included in the multivariate model. Overall survival was calculated by the Kaplan–Meier method. The survival curves calculated by univariate analysis were compared using the log rank test.

Statistical analysis was performed with STATA 9.0 software and a level of α = 0.05 was used to evaluate the results.

The Diva project was approved by the ethics committee of the University Hospital of Dijon.

Results

Distribution of incidence rates of stroke and AMI

Over the whole study period, 1660 events were recorded. Among these, 1020 (485 men and 535 women) were first ever stroke (61.4%) and 640 (403 men and 237 women) were AMI (38.6%). The distribution of stroke subtypes was 104 PCH (10%), 19 SAH (2%) and 897 ischaemic strokes (88%). Among these, we observed 267 lacunar strokes (29.8%), 181 cardioembolic strokes (20.2%) and 449 (50%) ischaemic strokes from atheroma of large arteries, other and undetermined causes. Identification was based on the first clinical report in 853 cases (83.5%), by first cerebral imaging in 122 cases (12%) and by first ultrasonography in 30 cases (3%). Only 15 cases (1.5%) were ascertained from the death certificate. Cerebral imaging was performed in 999 cases (98%). For AMI, identification was based on the first clinical report in all but two cases which were based on death certificates.

Annual age specific and sex specific incidence rates for stroke and AMI rose steeply with age. In men, mean age at onset was 7 years higher for stroke than for AMI (71.5 (13.1) vs 64.9 (12.7) years; p = 0.01). In women, the age at event onset was similar (75.3 (13.3) vs 75.2 (13.5) years; p = 0.5). The annual incidence of stroke was 98.5/100 000 for ischaemic strokes, 11.4/100 000 for PCH and 2.1/100 000 for SAH. Incidence rates adjusted to French, European and world populations were significantly higher for stroke than for AMI (table 1). The relative incidence of stroke compared with myocardial infarction was 0.88 (95% CI 0.60 to 1.29; p = 0.51) in women <65 years and 2.32 (95% CI 1.95 to 2.75; p<0.001) in those >65 years (fig 1). The incidence of stroke became significantly higher than that of AMI from the age group 65–69 years onwards (95% CI 1.05 to 3.39; p = 0.035). The relative incidence was 0.60 (95% CI 0.42 to 0.86; p<0.001) in men below 55 years, 1.01 (95% CI 0.81 to 1.24; p = 0.957) in those between 55 and 75 years and 2.01 (95% CI 1.48 to 2.71; p<0.001) at 75 years and older (fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Age specific incidence rates of stroke and acute myocardial infarction (MI) in women.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Age specific incidence rates of stroke and acute myocardial infarction (MI) in men.

Distribution of case fatality rates at 30 days for stroke and AMI

CFRs at 30 days were 9.80% for stroke versus 9.84% for AMI without any significant difference (log rank test = 0.946) (fig 3). Nevertheless, differences between stroke subtypes were observed, with lower 30 day CRFs in lacunar stroke (1.5%, 95% CI 0.57 to 3.94) than ischaemic strokes from atheroma of large arteries, other and undetermined causes (10%, 95% CI 13.2 to 17.6), cardioembolic stroke (13.3%, 95% CI 9.1 to 19.1) and haemorrhagic stroke (22%, 95% CI 15.6 to 30.4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Kaplan–Meier estimates of 30 day survival of acute myocardial infarction (MI) and stroke.

Using multivariate regression analyses, the independent determinants for CFR at day 30 for stroke patients were age ⩾85 years, cardioembolic and haemorrhagic strokes, the first study period (2001–2002), history of AMI and glycaemia at onset >7.8 mmol/l. For AMI patients, age ⩾85 years at onset, the first study period 2001–2002, history of hypercholesterolaemia, diabetes, tobacco consumption as well as glycaemia at onset >7.8 mmol/l were associated with increased mortality at 30 days.

Vascular risk factors for stroke and AMI

The distribution of the various vascular risk factors for stroke and myocardial infarction is listed in table 2. The proportion of men (63% vs 47.5%; p<0.001), hypercholesterolaemia (41.5% vs 33.8%; p = 0.002) and diabetes (21.6% vs 17.5%; p = 0.004) was greater in AMI than in stroke while the proportion of blood hypertension was higher in stroke (65.8% vs 51.0%; p<0.001). A history of AMI was observed in 15% (12.8–17.2) of stroke patients while a history of stroke was present in 7.6% (5.6–9.7) of patients with an AMI. A history of TIA and AF was present in 8.8% (7.1–10) and 22.4% (19.9–25) of stroke patients, respectively.

The prevalence of risk factors in both stroke and AMI patients by sex and age groups is shown in table 3. Irrespective of sex and age, glycaemia at onset was higher in AMI patients than in stroke patients. In men, the prevalence of hypertension and tobacco use was higher in stroke patients than in AMI patients from age 55–75 years onwards. In women with AMI, the prevalence of hypercholesterolaemia was higher in the age group <65 years, as was the prevalence of diabetes in the age group >65 years.

Discussion

Diva is a reliable population based registry that has collected data on acute stroke and AMI over a period of 6 years. Data were obtained from numerous sources, there was no age limit, post-mortem and death certificates were checked prospectively, and the cardiology and neurology departments coordinated their efforts. Using data from this registry, we demonstrated that first ever stroke was more frequent than first ever AMI in women aged >65 years and in men aged >75 years. In contrast, in men <55 years, AMI was more frequent than stroke. In addition, for the first time, the distribution of risk factors was compared between stroke and AMI patients in a large comprehensive population based study. Such data are of interest in order to design and implement effective secondary prevention strategies.

Diva and vascular epidemiology

In the literature about vascular epidemiology, many incomplete studies have reported a decreasing incidence of AMI, stroke events and mortality. These studies, however, had some major biases: (a) the majority were either hospital based or had age limitations, including individuals younger than 65 years,12 from 30 to 60 years,13 between 45 and 64 years14 or older than 35 years15; (b) several cardiovascular epidemiology studies included only selected volunteers rather than unselected population cohorts14; (c) some studies were restricted to patients aged 35–74 years admitted to hospitals with AMI,14 or to coronary and peripheral vascular events in patients with diabetes.16 Furthermore, Diva covered the largest population (151 864 inhabitants) versus 5209 in Framingham13 and 17 792 in ARIC.14 In our study, the distribution of vascular risk factors of both stroke and AMI were similar to those in the literature,17 18 19 20 21 as was the distribution of stroke subtypes. Only Oxvasc was able to report all coronary, cerebral and peripheral vascular events.6

Diva and Oxvasc

In Dijon, both the incidence rates and distribution of stroke mechanisms and subtypes, and AMI, were comparable with those described in Oxvasc, despite the large size of the Diva population (150 000) compared with the Oxvasc population (91 106).6 However, differences between Dijon and Oxford have been brought to light.

Firstly, in Dijon, mean age at AMI onset was significantly lower than that for stroke in men whereas there was no difference in Oxford. This correlates with the higher incidence of AMI than of stroke in men <55 years of age in our study.

Secondly, in contrast with Oxvasc, our study did not include TIA and unstable angina. We could argue that TIA and unstable angina should not be included in a comparison of clinical burden. Therefore, we preferred to study first ever stroke and AMI to compare the natural history of these two diseases. In Oxvasc, among the 1657 study patients, the relative rates for acute cerebrovascular events versus coronary events were 1.07 (95% CI 0.98 to 1.18; p = 0.14) overall and 0.97 (0.87 to 1.08; p = 0.57) if TIA and unstable angina were excluded. Nevertheless, considering only first ever events, the incidence of stroke was higher only in women >75 years whereas, in our study, it was higher both in women >65 years and in men >75 years. Moreover, in Dijon, the incidence of AMI was higher than that of stroke in men below 55 years which was not the case in Oxvasc. Hence to understand the different incidence patterns of stroke and AMI in our population, we analysed the distribution of premorbid risk factors by sex and age group. Indeed, it has been suggested that the strength of the association of a given atherothrombotic risk factor varies for different vascular beds (stroke, AMI, peripheral arterial disease). Hence hypertension appeared to be a risk factor strongly associated with stroke whereas hypercholesterolaemia was more strongly related to AMI.22 Nevertheless, in our study, the prevalence of hypertension and tobacco use was higher in male stroke patients compared with male AMI patients, not only in the age group >75 years, but also in the 55–75 year age group, whereas in the latter no difference in the incidence of the two vascular events was noted. In women with AMI, the prevalence of hypercholesterolaemia was higher in the age group <65 years, as was the prevalence of diabetes in the age group >65 years. Therefore, these results suggest that the distribution of classical atherothrombotic risk factor does not totally explain the incidence patterns of stroke and AMI.

The similarity between CFR for stroke and AMI observed in Diva was not observed in Oxvasc where the CFR for AMI was lower than that for stroke. This could suggest that in Dijon, the organisation of health systems and acute medical strategies for stroke and AMI are very similar, using the same care systems. The low CFR of 9% observed in stroke in Diva was similar to that in Oxvasc.6 Finally, we observed the negative impact of hyperglycaemia at onset on prognosis, as previously described for stroke and AMI.23 24

Our findings for AMI in young men have important implications in that they emphasise the need to develop and generalise primary prevention for AMI in young men. Our findings in older age groups suggest that it is necessary to implement an efficient public health policy for primary prevention of stroke in older people. Moreover, there is a need for randomised trials to analyse the effectiveness of prevention or treatment strategies for cerebral and cardiac events in elderly people.

The major advantage of our study is the continuous ascertainment from 2001 to 2006, in a well defined population, and the collaboration of numerous investigators from all fields of patient management, which ensured the exhaustiveness of case ascertainment. In addition, the population of the city was very stable, with less than 5% migration, which avoids bias due to changes in ethnic mix, and there was no change in the economic status of local residents. Diagnosis of stroke subtype was precise as a result of neuroimaging, since CT scan was performed in more than 98% of cases.

However, our study has several possible limitations. Although methods for case ascertainment and criteria were consistent throughout the study period, the incidence of stroke could have been overestimated in the last study periods compared with the first period because of the improvement in diagnostic techniques with the use of MRI. Nevertheless, access to this technique was limited as only 26% of stroke patients had MRI examination over the 6 years and there were no major changes between 2001 and 2006. Furthermore, improvements in the diagnosis of less severe strokes may have been counterbalanced by the fact that the true incidence of non-disabling stroke may have been underestimated because patients with mild symptoms may not have taken medical advice and so may not have been registered. In addition, we were only able to classify ischaemic strokes according to three subtypes instead of using the Toast classification.25 In fact, the Toast classification has been used in the Dijon Stroke Registry since January 2005. Before this period, ischaemic strokes were classified as lacunar infarcts, cardioembolic strokes or other ischaemic strokes, including stroke from atheroma of large arteries. Hence to avoid retrospective diagnostic bias, we chose to limit our study to this classification. Another limitation is that our largely white population with a high socioeconomic level did not allow us to compare the distribution of vascular events according to ethnic group.

The case ascertainment of AMI may not have been exhaustive as the registry of AMI was more recently launched, suggesting underestimation of the crude incidence of AMI in the first years, despite multiple sources in the hospitals and community of Dijon. Although AMI is usually associated with severe symptoms, unrecognised or silent AMI might have been missed by our case ascertainment procedure. As the incidence of unrecognised AMI increases with age, it might have resulted in an underestimation of the crude incidence. Nevertheless, the recent estimates of the rates of unrecognised AMI show that they are in true decline. Awareness of the symptoms and signs of AMI has probably improved over time as a result of a community education campaign. Moreover, for the ascertainment of AMI, the use of biomarkers such as troponin certainly improved the diagnosis of AMI in elderly people, and case fatality is very similar to case fatality reported in current registries of acute AMI, suggesting that this potential bias had little impact on our results.

In conclusion, our population based study demonstrates a higher incidence of stroke in men aged >75 years and in women aged >65 years, contrasting with a higher incidence of AMI in men <55 years. Mean age at onset was 7 years lower for AMI than for stroke in men only whereas no difference was found for CFR at 30 days. Our findings underline the need to put stroke and AMI on an equal footing with regard to health policy.