aParticipants could have had more than 1 reason for exclusion. bIncludes 12 nonspecific vertebrobasilar attacks, 129 possible transient neurological attacks (TNAs) that could not be classified due to insufficient data, and 82 events that did not fulfill the diagnostic criteria for TNA.
Bos MJ, van Rijn MJE, Witteman JCM, Hofman A, Koudstaal PJ, Breteler MMB. Incidence and Prognosis of Transient Neurological Attacks. JAMA. 2007;298(24):2877-2885. doi:10.1001/jama.298.24.2877
Author Affiliations: Department of Epidemiology and Biostatistics (Drs Bos, van Rijn, Witteman, Hofman,
and Breteler), and Department of Neurology (Drs Bos and Koudstaal),
Erasmus Medical Center, Rotterdam, the Netherlands.
Context Transient neurological attacks (TNAs) are attacks with temporary (<24 hours) neurological symptoms. These symptoms can be focal,
nonfocal, or a mixture of both. The prognostic significance of TNAs with focal symptoms (better known as transient ischemic attacks [TIAs])
is well understood. Conversely, hardly anything is known about the prognostic significance of TNAs with nonfocal or mixed symptoms.
Objective To study the incidence and prognosis of focal TNAs (or TIAs),
nonfocal TNAs, and mixed TNAs.
Design, Setting, and Participants The study population comprised 6062 community-dwelling Rotterdam Study participants who were aged 55 years or older and free from stroke,
myocardial infarction, and dementia at baseline (1990-1993). They were followed up for events until January 1, 2005. We analyzed the associations between incident TNAs and subsequent adverse events with age- and sex-adjusted Cox regression models.
Main Outcome Measures Stroke, ischemic heart disease, or dementia.
Results During 60 535 person-years, 548 participants developed TNA (282 focal, 228 nonfocal, and 38 mixed). The incidence rate per 1000 person-years was 4.7 (95% confidence interval [CI], 4.1-5.2)
for focal TNA, 3.8 (95% CI, 3.3-4.3) for nonfocal TNA, and 0.6 (95%
CI, 0.4-0.9) for mixed TNA. Participants with focal TNA were at higher risk of subsequent stroke than participants without TNA (n = 46
vs 540; hazard ratio [HR], 2.14; 95% confidence interval [CI]; 1.57-2.91)
but had an equal risk of ischemic heart disease and dementia. Nonfocal TNA patients were at higher risk of stroke (27 vs 540; HR, 1.56; 95%
CI, 1.08-2.28) and dementia (30 vs 552; HR, 1.59; 95% CI, 1.11-2.26)
than participants without TNA. Mixed TNA patients were at higher risk of stroke (6 vs 540; HR, 2.48; 95% CI, 1.11-5.56), ischemic heart disease (8 vs 779; HR, 2.26; 95% CI, 1.07-4.78), vascular death (8
vs 594; HR, 2.54; 95% CI, 1.31-4.91), and dementia (7 vs 552; HR,
3.46; 95% CI, 1.72-6.98) than participants without TNA.
Conclusion Patients who experience nonfocal TNAs, and especially those with mixed TNAs, have a higher risk of major vascular diseases and dementia than persons without TNA.
The first internationally accepted clinical classification for cerebrovascular disorders was formulated in 1975.1 It included diagnostic criteria for transient attacks of neurological dysfunction, which in the present article we will call transient neurological attacks (TNAs).
The classification attempted to create a formal distinction between TNAs with an unfavorable clinical course (which were presumed to be of vaso-occlusive origin and were therefore called transient ischemic attacks [TIAs]) and more benign attacks. Transient ischemic attacks were defined as temporary attacks (commonly 2-15 minutes, maximum 24 hours) with focal symptoms, which are attributable to dysfunction of one arterial territory of the brain. The remaining TNAs, with diffuse, nonlocalizing cerebral symptoms, were considered more benign. We will call these TNAs either nonfocal TNAs, if they present with only nonfocal symptoms, or mixed TNAs, if they present with a mix of focal and nonfocal symptoms.
Although the conventional diagnostic criteria for TIA are clear,
it is uncertain how mixed TNAs should be classified and treated. In our experience, mixed TNAs have mostly been classified as TIAs until now, both by neurologists and by general practitioners,2,3 but it seems reasonable to assume that some physicians may take a wait-and-see attitude in patients with mixed attacks. For nonfocal TNAs, a wide variety of diagnoses is commonly applied and none fulfills criteria for TIA. The relations between symptoms, the conventional clinical classification, and the classification used in this article follows: conventional classifies focal symptoms as TIAs and nonfocal symptoms as nonspecific, which could include syncope, confusion or transient global amnesia. The conventional classification is ambiguous on mixed symptoms. The proposed TNA classification categorizes the symptoms as TNA focal, TNA mixed,
and TNA nonfocal.
Although the assumption that patients with TIA are at high risk of major vascular disease has repeatedly been confirmed since 1975,4- 8 hardly any studies have challenged and verified the assumption that nonfocal TNAs have a benign clinical course. Those that did, focused only on small subgroups of patients with TNAs, included a limited number of adverse events, and included mostly a selected subgroup of patients referred to a neurologist.9- 13 Only one previous study, with participants of the Dutch TIA trial, assessed the prognosis of mixed TNAs, and suggested a higher risk of cardiac events in patients with mixed attacks.14
We studied incidence and prognosis of focal TNAs, nonfocal TNAs,
and mixed TNAs in a population-based cohort.
The present study is part of the Rotterdam Study, a prospective population-based cohort study of chronic and disabling diseases.15 Invitations were sent to all inhabitants of Ommoord, a district in the city of Rotterdam in the Netherlands,
aged 55 years and older. People living in homes for the elderly were included. The participation rate of those invited for the study was 78%; in total, 7983 individuals participated (Figure 1). The study was approved by the Medical Ethics Committee of Erasmus University Rotterdam. Written informed consent to retrieve information from treating physicians was obtained from all included participants. Baseline assessments were obtained from 1990 through 1993 and consisted of a home interview and 2 visits to the research center for physical examination.
History of stroke at study baseline was positive if a stroke was self-reported during the baseline interview and confirmed by medical records.16,17 History of myocardial infarction was positive if it was reported during the baseline interview and confirmed by baseline electrocardiograph or medical records. History of coronary artery bypass graft (CABG) surgery or percutaneous transluminal coronary angioplasty (PTCA) was positive if it was reported during the baseline interview and confirmed by medical records. Angina pectoris was assessed using the World Health Organization (WHO) angina pectoris questionnaire.18 History of dementia was assessed using a 3-step protocol19: 2 brief tests of cognition, the Mini-Mental State Examination (MMSE) and the Geriatric Mental State (GMS) schedule organic level, were used to screen all participants. Patients with positive screens (MMSE score <26 or GMS organic level >0) underwent the Cambridge Examination for Mental Disorders of the Elderly (Camdex). Those who were suspected of having dementia were examined by a neuropsychologist if additional neuropsychological testing was required for diagnosis. The diagnosis of dementia and subtype of dementia was made in accordance with internationally accepted criteria for dementia (Diagnostic and Statistical Manual of Mental Disorders [Third Edition Revised]), Alzheimer disease (National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer's Disease and Related Disorders Association),20 and vascular dementia (NINDS–Association Internationale pour la Recherché et l'Enseignement en Neurosciences)21 by a panel consisting of a neurologist, a neuropsychologist, and a research physician. We had no adequate information on TNAs at baseline.
Blood pressure was measured twice in the right arm using a random-zero sphygmomanometer, with the participant in a sitting position. We used the average of these 2 measurements. Hypertension was defined as stage 2 hypertension according to the seventh Joint National Committee on Detection, Evaluation, and Diagnosis of High Blood Pressure criteria (diastolic blood pressure ≥100 mm Hg and/or a systolic blood pressure ≥160 mm Hg), use of antihypertensive medication indicated to treat high blood pressure, or both. C-reactive protein, total cholesterol,
high-density lipoprotein cholesterol (HDL-C), and uric acid were measured in nonfasting baseline blood with automated enzymatic procedures.
We considered diabetes mellitus to be present if a random or postload glucose level was 200 mg/dL (to convert to mmol/L, multiply by 0.0555)
or higher, or if a person used antidiabetic medication. Atrial fibrillation at baseline was considered to be present if it was seen on electrocardiogram during the center visit or if reported in medical records. The intima-media thickness was measured by longitudinal 2-dimensional ultrasound of the carotid artery.22 During the home interview, smoking status and medication use were assessed. Genotyping for apolipoprotein E (APOE) was performed on coded DNA specimens. Two groups were formed on the basis of presence or absence of an APOE ε4 allele. Education was dichotomized into high (higher vocational or university education)
We excluded 1858 participants because they had myocardial infarction (n = 871), stroke (n = 261), or dementia (n = 482)
at baseline, or because they did not undergo cognitive screening at baseline (n = 455). Some participants had more than one of these conditions. Participants who refused informed consent to the retrieval of information from general practitioners were also excluded (n = 63). This left 6062 participants eligible for the analyses.
After enrollment into the Rotterdam Study, participants were continuously monitored for strokes, TIA, ischemic heart disease, dementia,
and death through automated linkage of the study database with files held by general practitioners, the Regional Institute for Outpatient Mental Health Care, and the municipality. If available, additional information (including brain imaging) was obtained from hospitals.
Only neurologists could request neuroimaging. In addition, during 3 follow-up surveys (1993-1995, 1997-1999, and 2002-2004), a research physician screened all eligible participants in person for occurrence of TIA by asking for transient neurological symptoms and carefully registered all mentioned symptoms and attack characteristics. During these surveys, we also assessed presence of dementia as described below. All information thus collected was verified as described below.
Follow-up for all events was completed until January 1, 2005,
for 96.2% of potential person-years.23
The information collected on potential strokes and TNAs during follow-up was reviewed and structured by a research physician. An experienced stroke neurologist (P.J.K.) verified all diagnoses. A stroke was diagnosed if a patient had typical symptoms that lasted longer than 24 hours. Strokes were subclassified as ischemic if a computed tomographic or magnetic resonance imaging scan made within 4 weeks after onset of symptoms ruled out other diagnoses or when indirect evidence (deficit limited to 1 limb or completely resolved within 72 hours or atrial fibrillation in absence of anticoagulant therapy) indicated the ischemic nature of the stroke. Transient neurological attacks were defined as attacks of sudden neurological symptoms that completely resolved within 24 hours, with no clear evidence for the diagnosis of migraine, epilepsy, Ménière disease,
hyperventilation, cardiac syncope, hypoglycaemia, or orthostatic hypotension.
If only focal brain symptoms were reported, the event was classified as a focal TNA . If only nonfocal brain symptoms were reported, the event was classified as nonfocal TNA . If focal and nonfocal symptoms were reported for one and the same attack,
a mixed TNA was diagnosed. Focal brain symptoms included one or more of the following: hemiparesis, hemihypesthesia, dysphasia,
dysarthria, amaurosis fugax, hemianopia, hemiataxia, diplopia, or vertigo. If the vertebrobasilar symptoms diplopia and vertigo were present in isolation, this was not considered sufficient evidence for the diagnosis of focal TNA and the event was classified as aspecific vertebrobasilar attack. Nonfocal symptoms were defined as broadly as possible because our aim was to have as few prejudgements as possible.
These attacks had to include 1 or more of the following symptoms:
decreased consciousness, unconsciousness, confusion, amnesia, unsteadiness,
nonrotatory dizziness, positive visual phenomena, cardiac or vegetative signs, paresthesias, bilateral weakness, and unwell feelings. Symptoms had to set in suddenly and to clear up within seconds to a maximum of 24 hours. Cardiac and vegetative signs were not indicative of the diagnosis of TNA by themselves but could shift the diagnosis from focal TNA to mixed TNA if they occurred in combination with focal neurological symptoms. In case of an unwell feeling attack, suspicion of underlying neurological disease should include a sudden onset and rapid clearance within seconds to hours. Although the patients could not further specify what had been wrong, the treating physician included TIA in the differential diagnosis.
The information collected on potential ischemic heart disease during follow-up was independently coded by 2 research physicians.
Diagnoses on which the research physicians disagreed were discussed to reach consensus. Finally, a medical expert in cardiovascular disease,
whose judgment was considered final, reviewed all events. Ischemic heart disease was coded according to the International Statistical Classification of Diseases, 10th Revision (ICD-10).
Incident ischemic heart disease during follow-up was defined as the occurrence of a fatal or nonfatal myocardial infarction (ICD-10 code I21), a revascularization procedure (PTCA or CABG), other forms of acute (ICD-10 code I24) or chronic ischemic (ICD-10 code I25)
heart disease, sudden (cardiac) death (ICD-10 codes I46 and R96), and death due to ventricular fibrillation (ICD-10 code I49) and death due to congestive heart failure (ICD-10 code I50).24
Vascular death was diagnosed when a participant died and the most probable underlying cause of death was stroke or ischemic heart disease.
Presence of dementia was assessed during the 3 follow-up surveys with the same protocol that was used at baseline (see above), in combination with information from general practitioners, the Regional Institute for Outpatient Mental Health Care, the municipality, and the hospitals.19
First, we assessed the age- and sex-specific incidence rates of all types of first TNAs. Subsequently, we assessed the associations between baseline characteristics and risk of TNA with adjusted Cox proportional hazards models. Then we assessed the association between occurrence of TNA and risk of adverse events with time-dependent Cox proportional hazards models, comparing the risk of an adverse event developing after TNA to the risk of an adverse event developing in TNA-free participants. We included occurrence of TNA as a time-dependent covariate in the model, which means that the TNA status changed from absent to present at the exact moment during follow-up that it had occurred. We adjusted for baseline covariates age and sex (model 1)
and additionally for a propensity score25 that was based on C-reactive protein, systolic blood pressure, carotid intima-media thickness, total cholesterol, HDL-C, uric acid, waist-hip ratio, MMSE, current smoking, ever smoking, atrial fibrillation, diabetes mellitus, hypertension, PTCA or CABG, angina pectoris, APOE ε4 carriership, and education (model 2), and present results as hazard ratios (HRs) with 95% confidence intervals (CIs).
Only first TNAs during follow-up were used in the analyses. Follow-up ended at time of the outcome event, end of study, loss to follow-up,
or death, whichever occurred first. Participants were censored at the time of stroke, ischemic heart disease, or dementia if it occurred before TNA. All analyses were performed with SPSS for Windows, release 11.0.1 (SPSS Inc, Chicago, Illinois). Two-sided P values <.05 were considered statistically significant.
The median age at baseline was 67.7 years, and 3758 participants (62%) were women (Table 1).
During 60 535 person-years of follow-up, a TNA occurred in 548
participants; 282 of these were classified as focal, 228 as nonfocal,
and 38 as mixed, on the basis of the symptoms with which they presented (Table 2). Twelve participants experienced an attack that did not fit into either category because it consisted of a single isolated vertebrobasilar symptom (isolated diplopia, vertigo, or dysphagia).
Of the patients with focal TNA, 100 (35%) consulted a neurologist,
141 (50%) consulted a general practitioner only, and 41 (15%) did not consult any physician; 93 (33%) reported the event at the subsequent visit to the research center. Of the patients with mixed TNA, 23 (61%)
consulted a neurologist, 14 (37%) consulted another physician only,
and 1 (3%) did not consult any physician; 7 (18%) reported the event at the subsequent visit to the research center. Of the patients with nonfocal TNA, 68 (30%) consulted a neurologist 106 (46%) consulted another physician, and 54 (24%) did not consult any physician but reported the event at the subsequent visit to the research center;
60 (26%) reported the event at the subsequent visit to the research center (Figure 1). Eighty-two (13%) reported possible TNAs were left out of the analyses; this was mainly because in these cases the symptoms lasted longer than 24 hours or because a clear alternative diagnosis could be made (most notably migraine, epilepsy, Ménière disease, hyperventilation,
cardiac syncope, hypoglycemia, or orthostatic hypotension). We had to leave 129 possible TNAs out of the analyses because the symptoms had been described inadequately. The median age at the time of possible TNA was 79 years, and 69% of patients were women. These cases were all retrieved from the general practitioners' medical records. The documentation was considered inadequate if general practitioners just mentioned the diagnosis of TIA without describing symptoms or when their handwriting was illegible. The prognosis of a possible TNA was very similar to that of a TIA, with the risk of stroke being higher following possible TNA than in participants without TNA (HR, 2.86;
95% CI, 1.42-5.75), but with no clear effect other than chance of possible TNA on risk of coronary heart disease (HR, 0.24: 95% CI,
0.03-1.71), or dementia (HR, 1.87; 95% CI, 0.83-4.19). Censoring the analyses for possible TNAs did not change the associations we found.
Other events that were observed during follow-up were stroke (n = 619), ischemic heart disease (n = 848), vascular death (n = 662; these participants were also classified as having stroke [n = 192] or ischemic heart disease [n = 430]),
and dementia (n=609).
Both in men and in women, nonfocal TNAs were almost as frequent as focal TNAs (the incidence rate per 1000 person-years, 4.7 for focal TNA; nonfocal TNA, 3.8), and for both types of events the incidence rates strongly increased with increasing age (Table 3). Mixed TNAs were less frequent (incidence rate,
0.6 per 1000 person-years) and their relation with age was less clear.
Age, systolic blood pressure, total cholesterol, and angina pectoris were independently and statistically significantly associated with the risk of focal TNA at an α level of .05 (Table 4). Age and systolic blood pressure were statistically significantly associated with the risk of nonfocal TNA, and age, systolic blood pressure, serum HDL-C, current smoking, APOE ε4 carriership, and high vocational or university education with mixed TNA.
The Kaplan-Meier curves showing the event-free survival time after TNA are drawn in Figure 2.
The Cox regression analysis showed that, compared with participants without TNA, participants with focal TNA had a higher risk of stroke (HR, 2.14; 95% CI. 1.57-2.91) and ischemic stroke (HR, 2.61; 95% CI,
1.78-3.84; Table 5); the risk of stroke within 90 days after focal TNA (TIA) was 3.5% (10 of 282
patients). Patients with nonfocal TNA had a higher risk of stroke (HR, 1.56; 95% CI, 1.08-2.28) and dementia (HR, 1.59; 95% CI, 1.11-2.26; Table 6), especially vascular dementia (HR, 5.05; 95% CI, 2.21-11.6), than participants without TNA. Patients with mixed TNA were at increased risk of stroke, and especially ischemic stroke (HR, 2.99; 95% CI, 1.11-8.03); ischemic heart disease, and especially myocardial infarction (HR, 3.34; 95% CI, 1.24-8.99); vascular death (HR, 2.54; 95% CI, 1.31-4.91); and dementia, especially vascular dementia (HR, 21.5; 95% CI 6.48-71.3) compared with participants without TNA. Adjustment for confounding did not materially change these associations.
Of the 12 patients with an aspecific vertebrobasilar TNA, during follow-up,
1 experienced an unspecified stroke and 2 were diagnosed with Alzheimer disease.
In this large, prospective population-based study, TNAs with nonfocal symptoms were almost as frequent as focal TNAs, and had an equally unfavorable overall subsequent clinical course with a slightly higher risk of stroke and a higher risk of vascular dementia than persons without TNA. Transient neurological attacks with combined focal and nonfocal symptoms had a particularly bad prognosis, indicating that patients are at high risk of stroke, ischemic heart disease,
vascular dementia, and vascular death.
Before these results can be interpreted, some methodological issues need to be discussed. The strengths of our study are the large study population (N = 6062), the intense monitoring of occurrence of diseases, the nearly complete follow-up (loss of potential person-years,
3.8%), and the close collaboration with general practitioners, who are the gatekeepers of the Dutch medical system. We had full access to the general practitioners' medical records and verified each event for which the treating physician considered TIA, stroke, ischemic heart disease, or dementia as a differential diagnosis. In addition,
we interviewed the participants in person at regular intervals. These procedures enabled us to ascertain cases that had not been referred to a neurologist but that had only been seen by a general practitioner,
nursing home physician, Rotterdam Study physician, or other physician,
which was the case for 201 strokes (32%), 15 mixed TNAs (39%), 182
focal TNAs (65%), and 160 nonfocal TNAs (70%). We think that this makes our database unique. However, due to lack of neuroimaging and thorough examinations, 194 strokes (31%) could not be subclassified as ischemic or hemorrhagic. In addition, 129 possible TNAs had to be left out of the analyses because the symptoms had been described inadequately. It is likely that the adequacy of the description of symptoms in medical files was not related to patient-related characteristics but rather to physician-related factors, although younger persons were more often referred to a neurologist than older persons, and the documentation of neurologists was generally more extensive than that of general practitioners.
Regarding the allocation of TNAs to the various subgroups, a possible source of error is that, because most physicians (including those of the Rotterdam Study research center) are conditioned to look for focal symptoms, nonfocal symptoms may have been underreported,
which may have led to an underreporting of mixed TNAs as well as to the inclusion of mixed TNAs in the focal TNA group.
Because patients do not always consult a physician after a TNA,
we tried to make the case finding more complete by screening the study participants in person at regular intervals. However, since patients who became diseased or died shortly after the TNA may have been preferentially missed by this additional screening step, it may have caused our estimates to be underestimations of the true associations.
The workup was done as part of routing clinical practice and not standardized. It is possible that upon a more thorough workup for some patients, the diagnosis would have changed their diagnosis from nonfocal TNA to an alternative diagnosis, possibly with a different and more favorable prognosis. This means that the estimates we found may be underestimations of the true relations between TNAs and risk of subsequent diseases due to the nonrandom allocation of non-TNA patients (with a favorable prognosis) into the TNA groups.
Physicians may have started preventive treatment in TNA patients,
which may have caused the associations between TNAs and outcome events that we found to be weaker than they would have been if TNA patients had not received preventive treatment.
Participants with TNA before baseline were not excluded from our present study: no medical records on TNA before baseline were available and we feared that participants would not be able to reliably recall TNA symptoms years after the event.
In our study, the risk of stroke within 90 days after focal TNA (TIA) was lower than in previous reports: 3.5% in our study compared with 9.2%26 and 9.5%27 reported previously. We think this may be because most other studies recruited patients from hospitals or asked the general practitioners to call a stroke consultant, whereas we had direct access to general practitioners' files and could also question the participants in person. Our study may therefore have included more focal TNAs with mild symptoms of short duration, which would have remained unnoticed by other studies, and which have a better prognosis.28 Because our study was population-based and not emergency department–based, some TIAs immediately prior to a stroke may not have been captured, which may also result in a lower 90-day event rate.
Because nonfocal TNAs present with a wide variety of symptoms,
which are often ascribed to an equally wide variety of relatively harmless nonvascular conditions (although this is not supported by scientific evidence), they have not been studied as a group before.
It is likely that our nonfocal TNAs are a heterogeneous group with respect to etiology and prognosis. Although further subdivision of nonfocal TNAs may be tempting, this should be done with great caution,
not only because the subdivision has to be based on unreliable anamnestic data (with the patient often being confused, anxious, or even unconscious),
but also because there is no empirical basis for the subdivision of nonfocal TNAs.
Our findings challenge the strong but unfounded conviction that nonfocal TNAs are harmless. On the contrary, our findings suggest that nonfocal TNAs are not only a risk factor for stroke, but also for dementia. For 2 small subgroups of nonfocal TNAs—transient global amnesia11,12 and dizziness10,29—it has been reported that the clinical course is usually more benign.
Unfortunately, our data do not allow us to determine the prognosis of various subgroups of nonfocal TNAs because of small sample size.
Only 1 previous study reported on the prognosis of mixed TNAs.14 Although the definition of mixed TNAs differed slightly, in this study those with mixed TNA had a significantly higher risk of ischemic heart disease than those with focal TNA, which matches our findings very well. In addition, we found that patients who experienced mixed TNA are at increased risk of dementia, especially of vascular dementia. Of course our findings regarding mixed TNAs have to be interpreted cautiously because the number of cases and events are relatively small.
Compared with persons without TNA, patients with focal TNA (ie,
TIA) had a higher risk of stroke, and patients with nonfocal TNA a higher risk of stroke and dementia. Although the number was quite small, patients with mixed TNA had a higher risk of stroke, dementia,
ischemic heart disease, and vascular death than persons without TNA.
Corresponding Author: Monique M.
B. Breteler, MD, PhD, Department of Epidemiology and Biostatistics,
Erasmus Medical Center, Dr Molewaterplein 50, PO Box 2040, 3000 CA Rotterdam, the Netherlands (firstname.lastname@example.org).
Author Contributions: Dr Breteler had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bos, Witteman,
Hofman, Koudstaal, Breteler
Acquisition of data: Bos, van Rijn,
Analysis and interpretation of data:
Bos, Koudstaal, Breteler
Drafting of the manuscript: Bos, Koudstaal,
Critical revision of the manuscript for important intellectual content: Bos, van Rijn, Witteman, Hofman, Koudstaal,
Statistical analysis: Bos
Obtained funding: Hofman, Koudstaal,
Study supervision: Koudstaal, Breteler
Financial Disclosures: None reported.
Funding/Support: The Rotterdam Study is supported by Erasmus Medical Center Rotterdam, the Erasmus University Rotterdam, the Netherlands Organization for Scientific Research, the Netherlands Organization for Health Research and Development, the Research Institute for Diseases in the Elderly, the Ministry of Education,
Culture and Science, and the Ministry of Health, Welfare, and Sports.
This study was supported by the Netherlands Organization for Scientific Research grants 904-61-093 and 918-46-615.
Role of the Sponsors: The sponsors had no role in the design and conduct of the study, in the collection,
management, analysis, and interpretation of the data, or in the preparation,
review, or approval of the manuscript.
Additional Contributions: We thank the general practitioners and pharmacists of the Ommoord district to the Rotterdam Study for their participation.