A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects

Transcription

1 A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects Philippa C Lavallée, Elena Meseguer, Halim Abboud, Lucie Cabrejo, Jean-Marc Olivot, Olivier Simon, Mikael Mazighi, Chantal Nifle, Philippe Niclot, Bertrand Lapergue, Isabelle F Klein, Eric Brochet, Philippe Gabriel Steg, Guy Lesèche, Julien Labreuche, Pierre-Jean Touboul, Pierre Amarenco Summary Background Diagnosis and treatment of cerebral and retinal transient ischaemic attacks (TIAs) are often delayed by the lack of immediate access to a dedicated TIA clinic. We evaluated the effects of rapid assessment of patients with TIA on clinical decision making, length of hospital stay, and subsequent stroke rates. Methods We set up SOS-TIA, a hospital clinic with 24-h access. Patients were admitted if they had sudden retinal or cerebral focal symptoms judged to relate to ischaemia and if they made a total recovery. Assessment, which included neurological, arterial, and cardiac imaging, was within 4 h of admission. A leaflet about TIA with a toll-free telephone number for SOS-TIA was sent to family doctors, cardiologists, neurologists, and ophthalmologists in Paris and its administrative region. Endpoints were stroke within 90 days, and stroke, myocardial infarction, and vascular death within 1 year. Findings Between January, 2003, and December, 2005, we admitted 1085 patients with suspected TIA; 574 (53%) were seen within 24 h of symptom onset. 701 (65%) patients had confirmed TIA or minor stroke, and 144 (13%) had possible TIA. 108 (17%) of the 643 patients with confirmed TIA had brain tissue damage. Median duration of symptoms was 15 min (IQR 5 75 min). Of the patients with confirmed or possible TIA, all started a stroke prevention programme, 43 (5%) had urgent carotid revascularisation, and 44 (5%) were treated for atrial fibrillation with anticoagulants. 808 (74%) of all patients seen were sent home on the same day. The 90-day stroke rate was 1 24% (95% CI ), whereas the rate predicted from ABCD² scores was 5 96%. Interpretation Use of TIA clinics with 24-h access and immediate initiation of preventive treatment might greatly reduce length of hospital stay and risk of stroke compared with expected risk. Introduction Recent advances in stroke management have placed major emphasis on the provision of hospital stroke units and on thrombolysis with alteplase (tissue plasminogen activator) to treat completed or evolving strokes. 1 Transient ischaemic attacks (TIAs) precede up to a quarter of completed strokes, 2 and so can give clinicians an opportunity to avoid a completed stroke and its devastating personal, social, and sometimes fatal consequences. Appropriate detection and treatment of TIAs can also obviate the need for admission to a hospital stroke unit. Diffusion-weighted imaging shows a small amount of tissue damage in most cases of TIA. 3 Thus, TIAs can be thought of as mini strokes, and as medical emergencies. 4 Symptoms of TIA are frequently ignored by patients and their relatives or are unrecognised by doctors, which can delay diagnosis and treatment. Even after a diagnosis of TIA, a patient can be reluctant to be admitted to hospital because the symptoms often last for only a few minutes, and the patient subsequently feels completely recovered. Once a patient agrees to be admitted to hospital, comprehensive testing and diagnosis often cannot be organised quickly or in the same hospital, particularly after 1700 h, which can delay the start of assessment substantially. Patients with a suspected TIA are frequently sent to an emergency department, but emergency medicine physicians in community hospitals might have little or no immediate access to brain and vascular imaging facilities. Frequently, patients are sent back to their family doctor and are examined as outpatients. Many patients give up at this point; others might have their TIA investigated within the next 8 15 days, or later. Because the risk of completed stroke is very high during the first week after a TIA, 5,6 such a care pathway is counterproductive, and greatly reduces the opportunity to prevent stroke. To improve care for patients with symptoms of TIA, we set up a clinic with round-the-clock access for doctors working in the administrative region of Paris (which has 11 million inhabitants). This clinic provided a short and standardised clinical assessment followed by initiation of a comprehensive stroke prevention programme. We also used a leaflet on TIA management and treatment to increase awareness among community doctors of the importance of early diagnosis. We aimed to describe the service provided by the TIA clinic, and to assess the effects of these initiatives on the length of hospital stay and on stroke rates at 90 days as compared with rates predicted by the ABCD² score. 7 We also report the risk of major cardiovascular events within the first year after a TIA. Methods Patients and procedures We mailed a leaflet (webappendix) on TIA to family doctors, cardiologists, neurologists, and ophthalmologists Lancet Neurol 2007; 6: Published Online October 9, 2007 DOI: /S (07)70248-X See Reflection and Reaction page 940 See page 961 See Lancet 2007; DOI: /S (07) Department of Neurology and Stroke Centre (P C Lavallée MD, E Meseguer MD, H Abboud MD, L Cabrejo MD, J-M Olivot MD, O Simon MD, M Mazighi MD, C Nifle MD, P Niclot MD, B Lapergue MD, J Labreuche BS, P-J Touboul MD, P Amarenco MD), Neuroradiology Unit, Department of Radiology (I F Klein MD), Department of Cardiology (E Brochet MD, P G Steg MD), and Department of Vascular Surgery (G Lesèche MD), Bichat Claude Bernard University Hospital, Denis Diderot University and Medical School Paris VII, Assistance Publique-Hôpitaux de Paris, Paris, France Correspondence to: Pierre Amarenco, Department of Neurology and Stroke Centre, Bichat Claude Bernard University Hospital, Denis Diderot University and Medical School Paris VII, 46 Rue Henri Huchard, Paris, France pierre.amarenco@bch.aphp.fr See Online for webappendix Vol 6 November

2 in Paris and its administrative region, and to emergency departments of community and teaching hospitals. This leaflet included information on the definition, key symptoms, risks, main causes, and emergency treatments for TIA, and presented TIA as an opportunity to prevent a completed, potentially devastating stroke. The leaflet also informed doctors of the availability of our TIA clinic, SOS-TIA (called SOS-AIT in French). SOS-TIA was open 24 h a day, 7 days a week, and could be contacted via a tollfree telephone number. The clinic was located in the neurology department of Bichat Claude Bernard University Hospital, which also includes a stroke unit. Between 0900 h and 1700 h each day, a stroke-prevention nurse was in charge of the call centre. Between 1700 h and 0900 h, telephone calls were automatically transferred to the mobile phone of the senior vascular neurologist who was on duty in the stroke unit. The nurse was fully trained on TIA symptoms and stroke, and had completed the Emergencies in neurology university diploma at Denis Diderot Medical School. Patients were admitted to SOS- TIA immediately after the telephone call from their doctor if they had focal symptoms of brain or retinal dysfunction that were of sudden onset and presumed to be related to ischaemia, and if the patients had subsequently made a total recovery. To reduce the number of unnecessary admissions, a vascular neurologist was always available to discuss symptoms with the nurse; the vascular neurologist was responsible for the decision to exclude patients who were judged to have non-ischaemic neurological transient symptoms such as migraine. To avoid overcrowding of the stroke unit by patients with evolving stroke rather than TIA, family doctors of patients with persistent symptoms or signs at the time of the telephone call were directed the French national medical emergency telephone number (15, equivalent to 911 in the USA). The SOS-TIA clinic was organised to provide an initial, standardised assessment within 4 h of admission. Clinical assessment was by the vascular neurologist; if a TIA was suspected, further comprehensive tests were initiated. Brain imaging (either MRI or a default CT scan) was done by staff in the radiological department of Bichat Claude Bernard University Hospital. Duplex ultrasonography and transcranial doppler imaging of brain arteries were done to detect high-grade stenosis, and electrocardiography was done to exclude a high-risk cardiac source of embolism. All vascular neurologists who participated in this programme were fully trained in ultrasound (duplex and transcranial doppler). Urgent transthoracic echo cardiography and transoesophageal echocardiography were done only if the vascular neurologist suspected that there was a high risk of recurrent embolism for cardiac reasons (eg, if endocarditis, aortic dissection, mitral stenosis, acute myocardial infarction, or prosthetic heart valve thrombosis were suspected). Other patients received transoesophageal echocardiography during follow-up testing, unless a definite cause of stroke had been identified. Blood samples were tested for: lipid profile; ionography; red blood cell, white blood cell, and platelet counts; and concentrations of glucose, haemoglobin A 1c, C-reactive protein, and creatinine. Lipid profile and glucose measurements were taken only in fasting patients. Patients who could not be assessed fully during the initial investigation (eg, patients who were not fasting) returned to the neurological department at a later date as outpatients so that testing could be completed. TIA was defined as an acute loss of focal cerebral or ocular function with symptoms that lasted less than 24 h and that were attributed to inadequate blood supply. 8 Symptoms thought to be consistent with TIA were those defined by an ad hoc Committee established by the Advisory Council for the National Institute of Neurological and Communicative Disorders and Stroke. 9 If the brain CT or MRI scans revealed an acute infarction in an area that corresponded to the clinical symptoms, the patient was judged to have a TIA with a new lesion; if not, the event was classified as TIA without a new lesion. Patients who had transient focal neurological symptoms but whose clinical and radiological features were not sufficiently clear to affirm or exclude the diagnosis of definite TIA were judged to have a possible TIA. 10 Patients with incomplete recovery were judged to have a minor stroke. After the end of the evaluation, the referral doctor was contacted by the vascular neurologist to discuss the diagnosis and the most appropriate treatment. Patients were discharged home immediately after the assessment, unless they fulfilled predefined criteria for admission to the stroke unit (panel). The family doctor received a discharge summary that stated the targets of the prevention therapy: blood pressure of less than 140/90 mm Hg in non-diabetic patients and, if that objective was achieved, a further target of less than 130/85 mm Hg (this was the first target in patients with diabetes); low-density lipoprotein (LDL) cholesterol concentrations of less than 2 56 mmol/l; and smoking cessation. In all patients, anti-thrombotic treatment was started immediately (the leaflet also recommended that family doctors give patients a loading dose of mg aspirin, and this recommendation was repeated at the time of initial telephone call). Where possible, therapy to lower blood pressure and blood lipid concentrations was started or modified at the time of discharge from SOS- TIA. For other patients, only broad recommendations were provided to family doctors at discharge, because high blood pressure cannot usually be diagnosed on the basis of recordings taken during the acute phase of stroke, and because statin therapy was not started until after the blood lipid profile had been measured. Whether family doctors followed the recommendations was not recorded. In patients with atrial fibrillation who had not received anticoagulant therapy before symptom onset, subcutaneous low molecular weight heparin was given in Vol 6 November 2007

3 addition to the oral anticoagulant, until the international normalised ratio was more than 2 for immediate full anticoagulation. For patients who had a stroke caused by atherothrombosis, caused by cerebral arteriolopathy, or of unknown cause, antiplatelet drugs were started immediately. Patients with symptomatic high-grade stenosis received carotid revascularisation. We evaluated the risks of stroke at 90 days, and risks of stroke, myocardial infarction, and vascular death at 1 year. Follow-up information on occurrence of these events was obtained by neurologists in the outpatient clinic during face-to-face interviews or by nurses through telephone calls. If the patient could not be contacted, a close relative or their family doctor was interviewed. Patients were informed that they might receive a telephone call but, in line with French regulations, patients did not need to give written informed consent for such a call. Whenever possible, medical records were obtained for patients for whom stroke, myocardial infarction, or vascular death were reported. The decision of whether an endpoint event had occurred was validated by consensus between two neurologists (PCL and PA). A stroke endpoint was defined by focal neurological deficit of sudden onset that lasted for more than 24 h, with or without confirmation from brain imaging results. All such cases, and the TIA qualifying events of these patients, were reviewed by a senior vascular neurologist (PCL) using the GENIC stroke subtyping classification. 11 We compared the risk of stroke at 90 days with that expected using the ABCD² scores of the patients for whom we had information on each component of the score. 7 This seven-point score is calculated on the basis of age (1 point for >60 years), blood pressure (1 point for 140/90 mm Hg), clinical features (2 points for unilateral weakness and 1 point for speech disturbance without weakness), diabetes (1 point), and duration of symptoms (1 point for min and 2 points for >60 min). 7 We also recorded the number of patients discharged with a definite diagnosis of TIA, the number of carotid endarterectomies done, the number of patients discharged with oral anticoagulation treatment for a definite cardiac source of embolism, and the length of hospital stay. We compared the length of stay for patients with TIA in our department with that for the whole of Paris and its administrative region (taken from the Direction Régionale des Affaires Sanitaires et Sociales d Ile de France). Administrative and demographic data, information on previous treatment and TIA characteristics, and results of evaluation, final diagnosis, and follow-up were prospectively collected by the vascular neurologist and were entered into a database by a clinical research assistant. Statistical analysis The risks of stroke at 90 days and 1 year, and the combined risks of stroke, myocardial infarction, and vascular death at 1 year were measured using the Kaplan-Meier method. Patients who died from causes other than stroke or vascular disease were censored at the time of death, depending on the outcome. Because the risk of stroke is high within the first 48 h after TIA, 5 7 we also did a separate analysis of patients seen within 24 h. To investigate whether a recruitment approach that focused only on emergency departments would have selected high-risk patients, we analysed data from patients with an ABCD² score or 4 or more or 5 or more (criteria above which patients are admitted to the stroke unit for at least 48 h in some TIA clinics), and from patients with motor weakness or speech impairment that lasted for more than 10 min within 24 h of symptom onset (FASTER inclusion criteria 12 ). Data were analysed using SAS version 9.1. Role of the funding source SOS-TIA was funded by the SOS-ATTAQUE CEREBRALE Association (a non-profit-making organisation). The corresponding author had full access to all of the data in the study, and had final responsibility for the integrity of the data, for the accuracy of the data analysis, and for the decision to submit for publication. Results Between January, 2003, and December, 2005, 1085 patients with a suspected TIA were entered into the SOS-TIA programme. A mean of 30 patients were seen each month, and the number of patients seen at the TIA clinic increased by 29% between 2003 (n=316) and 2005 (n=407). 946 (87%) patients were seen by a neurologist within 24 h of the telephone call and 574 (53%) were seen by a neurologist within 24 h of symptom onset (figure 1). 665 (61%) were seen within 48 h, and 810 (75%) within 7 days. Figure 2 shows the numbers of patients in each diagnosis group. Baseline characteristics were similar in patients with minor stroke and those with definite TIA (with or without brain tissue damage), and in patients with possible TIA and those with a non-ischaemic diagnosis (table 1). 715 (68%) of 1053 patients had hypertension, as defined Panel: Criteria for admission to the stroke unit after assessment in the SOS-TIA clinic A suspected or identified cause of TIA High-grade stenosis of intracranial or extracranial brain arteries Low blood flow in the middle cerebral artery Potential cardiac sources of high-risk of recurrent embolism: Prosthetic mechanical heart valve Endocarditis Aortic dissection Acute coronary syndrome Overt congestive heart failure TIAs that increase in frequency, duration, or severity (crescendo TIA) 24-h cardiac monitoring warranted (paroxysmal atrial fibrillation strongly suspected) Vol 6 November

4 Median 1 day (IQR 0 5) 631 (58%) within 24 h Median 0 day (IQR 0 1) 946 (87%) within 24 h Symptom onset Telephone call Examination by vascular neurologist Median 1 day (IQR 0 8) 574 (53%) within 24 h Figure 1: Time from symptom onset to telephone call and to examination in TIA clinic We asked the 138 (13%) patients who did not have a precise date of symptom onset to estimate the delay as <1 month, 1 6 months, 6 12 months, >12 months, or unknown. For these patients, we classified the time to telephone call and examination at the TIA clinic as >2 weeks. by medical history of hypertension or blood pressure of more than 140/90 mm Hg; 537 (58%) of 922 patients had hypercholesterolaemia, defined by medical history or a blood cholesterol concentration of 6 21 mmol/l or more. 948 (87%) patients had at least one vascular risk factor (ie, had a body-mass index 30 kg/m², hypertension, hypercholesterolaemia, diabetes, or a history of cardiovascular and cerebrovascular diseases, or were smokers). Median symptom duration was 10 min (IQR 3 30 min) in patients with definite TIA and no brain tissue damage, 15 (5 60) min in patients with definite TIA and brain tissue damage, 15 (3 60) min in patients with possible TIA, and 30 (5 120) min in patients with a non-ischaemic diagnosis (99%) patients had brain imaging; 892 (82%) of these patients had CT scans, 846 (78%) had MRI, 290 (27%) had diffusion-weighted imaging within 15 days of symptom onset, and 760 (71%) had intracranial and extracranial magnetic resonance angiography. 745 (69%) had transthoracic echocardiography and 657 (61%) had both transthoracic and transoesophageal echocardiography (95%) patients had brain imaging, arterial explorations, and cardiac explorations (97%) patients received duplex and transcranial doppler ultrasonography. Causes were identified in 219 (41%) patients with definite TIA and no brain tissue damage, in 37 (64%) patients with minor stroke, and in 80 (74%) patients with definite TIA and brain tissue damage (table 2). Atherothrombosis was the most commonly identified cause. 277 patients (26%) were admitted to the stroke unit, with a median length of stay of 4 days (IQR 2 7). The remaining 808 patients (74%) were judged not to need to be admitted and were discharged home on the day of examination at the SOS-TIA. 478 of these patients had a definite TIA or a minor stroke (figure 3). Antithrombotic medication was given immediately to 824 (98%) of the 845 patients who had minor stroke, definite TIA, or possible TIA. Of the 58 patients with atrial fibrillation, 51 of whom had definite TIA, 44 (76%) were given a prescription for oral anticoagulants; of these patients, 30 were started on oral anticoagulant therapy. Therapy to lower blood pressure was started or modified in 129 (24%) patients who had definite TIA with no brain tissue damage, 46 (43%) patients who had definite TIA with brain tissue damage, 24 (41%) of patients with minor stroke, 14 (10%) patients with possible TIA, and 12 (5%) patients with a non-ischaemic diagnosis. Overall, of 701 patients with definite TIA or minor stroke, therapy to lower blood pressure was started or modified in 199 (28%), and we recommended to family doctors of the remaining 502 (72%) a target blood pressure of 140/90 mm Hg (130/85 mm Hg in diabetics). Therapy to lower blood lipid concentrations was started or modified in 315 (45%) patients, and we recommended to family doctors of the remaining 386 (55%) a target LDL cholesterol concentration of 2 56 mmol/l. Lipid-lowering treatment was started or modified in 232 (44%), 51 (47%), 32 (55%), 49 (34%), and 34 (14%) patients who had, respectively, definite TIA with no brain tissue damage, definite TIA with brain tissue damage, minor stroke, possible TIA, and a non-ischaemic diagnosis. Urgent carotid 1085 patients 643 (59%) definite TIAs 144 (13%) possible TIAs 58 (5%) minor ischaemic strokes 240 (22%) patients with other diagnosis 73 (30%) migraine attacks 70 (29%) bizarre spells 21 (9%) epileptic seizure 10 (4%) anxiety or panic attack 108 with new ischaemic lesions* 535 without new ischaemic lesions 8 (3%) benign positional paroxysmal vertigo Figure 2: Patient numbers relative to diagnosis by vascular neurologists *As diagnosed by MRI (n=105), CT scan (n=3), or both (n=8). Percentages of the 240 patients who had a non-ischaemic diagnosis. Non-focal or not clearly focal transient neurological events for which the mode of onset, topography, and course of symptoms did not fulfil the criteria for definite or possible TIA or another definite or possible neurological syndrome (eg, epilepsy or migraine) Vol 6 November 2007

5 TIA, no new lesion (n=535) TIA, new lesion (n=108) Minor stroke (n=58) Possible TIA (n=144) Other* (n=240) Demographics Mean age (years) 66 1 ( ) 65 8 ( ) 68 3 ( ) 57 3 ( ) 53 0 ( ) Male sex 295 (55%) 66 (61%) 28 (48%) 74 (51%) 90 (38%) Medical history Hypertension 274 (52%) 69 (64%) 31 (54%) 52 (37%) 84 (35%) Hypercholesterolaemia 249 (48%) 52 (49%) 25 (44%) 47 (34%) 64 (27%) Diabetes 63 (12%) 13 (12%) 10 (17%) 7 (5%) 15 (6%) Current smoker 99 (20%) 30 (30%) 15 (28%) 30 (23%) 59 (28%) Cardiovascular disease 130 (24%) 32 (30%) 16 (28%) 21 (15%) 28 (12%) Cerebrovascular disease 84 (16%) 9 (8%) 6 (10%) 11 (8%) 22 (9%) Examination findings BMI (kg/m²) 25 6 ( ) 25 7 ( ) 24 8 ( ) 25 2 ( ) 25 1 ( ) Total cholesterol (mmol/l) 5 51 ( ) 5 30 ( ) 5 51 ( ) 5 64 ( ) 5 37 ( ) LDL cholesterol (mmol/l) 3 25 ( ) 3 23 ( ) 3 26 ( ) 3 36 ( ) 3 13 ( ) HDL cholesterol (mmol/l) 1 63 ( ) 1 47 ( ) 1 78 ( ) 1 71 ( ) 1 71 ( ) Triglycerides (mmol/l) 1 31 ( ) 1 35 ( ) 1 21 ( ) 1 24 ( ) 1 23 ( ) Systolic BP (mm Hg) 144 ( ) 146 ( ) 144 ( ) 140 ( ) 137 ( ) Diastolic BP (mm Hg) 78 (66 90) 79 (68 90) 80 (68 92) 78 (66 90) 78 (66 90) Atherosclerotic plaques in ICA 340 (65%) 70 (67%) 36 (62%) 66 (48%) 75 (34%) Symptoms of TIA Weakness 188 (35%) 61 (57%) 33 (57%) 28 (19%) 36 (15%) Speech impairment 191 (36%) 49 (45%) 30 (52%) 37 (26%) 60 (25%) Previous chronic antithrombotic medication Antiplatelet drugs 120 (23%) 33 (31%) 14 (24%) 22 (15%) 32 (14%) Anticoagulant drugs 24 (5%) 2 (2%) 2 (4%) 7 (5%) 10 (4%) Data are mean (SD) or n (%). In each case, percentages are of the number of patients for whom data were available. BMI=body-mass index. LDL=low-density lipoprotein. HDL=high-density lipoprotein. BP=blood pressure. ICA=internal carotid artery. *Non-ischaemic diagnosis. Table 1: Baseline characteristics of patients grouped according to diagnosis by a vascular neurologist revascularisation was done in 43 patients (24 patients with definite TIA and no brain tissue damage, 14 with definite TIA and brain tissue damage, and five with minor stroke), with a median delay from initial examination of 6 days (IQR 4 10) (97%) patients were followed up for a median of 16 months (IQR 12 19) after presentation at the TIA clinic, and the remaining 33 patients (3%) were lost to follow-up. Information was collected directly from 861 (82%) of these patients, from relatives for 88 (10%) patients, and from doctors for 103 (8%) patients. Two vascular events occurred within 7 days of presentation; both were non-fatal ischaemic stroke. 13 vascular events occurred within 90 days of presentation; all were strokes, including two fatal strokes and one haemorrhagic stroke. During the first year of follow-up, there were 20 strokes (two fatal), two myocardial infarctions (one fatal), and four other vascular deaths, including two sudden deaths. Nine non-vascular deaths and two deaths of unknown cause also occurred within 1 year of presentation. Table 3 shows the resulting 90-day stroke rate and stroke rate expected on the basis of the ABCD² score. All strokes occurred in patients with definite TIA except for one, which occurred in a patient with a possible TIA. Patients with definite TIA and a recent ischaemic brain lesion had the highest risk of stroke When the sample was restricted to 552 patients seen within 24 h of symptom onset, the 90-day stroke rate was 1 63% (95% CI ), whereas the risk expected from the ABCD² scores of these patients was 6 49%. The 90-day stroke rate was 2 08% ( ) for 434 patients with definite or possible TIA or minor stroke who had been seen within 24 h of symptom onset, 2 01% ( ) for the 396 patients who had an ABCD² score of 4 or more, 3 18% ( ) for the 198 patients who had an ABCD² score of 5 or more, and 2 20% ( ) for the 193 patients who had motor weakness or a speech impairment that lasted for more than 10 min and who TIA, no new lesion TIA, new lesion Minor stroke Atherothrombosis 130 (24%) 41 (38%) 18 (31%) Cardioembolism 56 (11%) 15 (149%) 6 (10%) Lacunar stroke 14 (6%) 14 (13%) 8 (14%) Multiple causes 11 (2%) 2 (2%) 3 (5%) Dissection or other rare causes 8 (1%) 8 (7%) 2 (3%) Unknown 316 (59%) 28 (26%) 21 (36%) Data are n (%). In each case, percentages are of the number of patients for whom data were available. Table 2: Causes of confirmed TIAs and minor stroke Vol 6 November

6 Patients (%) Definite TIA with no new lesion Definite TIA with new lesion No admission to stroke unit h Possible TIA Minor stroke 3 7 days More than 1 week Others diagnosis Figure 3: Length of stay in hospital according to diagnosis by vascular neurologists were seen within 24 h of symptom onset. Table 4 shows the 1-year risks of stroke and of the combination of stroke, myocardial infarction, and vascular death for all patients who were followed up. In patients with definite TIA or minor stroke, the 1-year rate of myocardial infarction and vascular death was 1 10%. Discussion SOS-TIA provided physicians with round-the-clock access to assessment, diagnosis, and immediate treatment for patients with suspected TIA. We report a lower rate of stroke at 90 days than was expected on the basis of ABCD² scores. The risk of stroke in the subgroup of patients who were seen at the TIA clinic within 24 h of symptom onset was also lower than that expected from ABCD² scores. The 1-year rate of myocardial infarction and vascular death was lower in our study (1 1%) than were annual risks for myocardial infarction (2 2%) and non-stroke vascular death (2 1%) reported in a meta-analysis. 13 We recorded these low rates even though almost three quarters of patients were discharged home on the same day. The median length of hospital stay for patients with TIA was less than 1 day, whereas the mean stay of such patients in other hospitals in Paris and its area over the same period was 6 8 days. Our data thus suggest that systematic admission to hospital for at least 48 h, as has been recommended previously, 14 is not the only way to reduce risk of stroke within the first 90 days after TIA. Strokes (n) % risk of stroke (95% CI)* % expected risk All patients (n=1052) ( ) 5 96 TIA, no new lesion (n=524) ( ) 6 13 TIA, new lesion (n=105) ( ) 7 76 Possible TIA (n=141) ( ) 4 00 *Estimated by the Kaplan-Meier method. Estimated from the ABCD² score. 7 Table 3: Recorded and expected 90-day risks of stroke All stroke All stroke, myocardial infarction, and vascular death All patients (n=1052) 1 95 ( ) 2 54 ( ) TIA, no new lesion (n=524) 2 17 ( ) 2 78 ( ) TIA, new lesion (n=105) 4 76 ( ) 5 74 ( ) Minor stroke (n=54) 1 96 ( ) 3 81 ( ) Possible TIA (n=141) 2 18 ( ) 2 18 ( ) Other diagnosis (n=228) No events 0 48 ( ) % risk (95% CI) of events as estimated by the Kaplan-Meier method. Table 4: 1-year risk of stroke and combined outcome Rapid evaluation and treatment in a TIA clinic might be at least as effective for acute selection of patients who are at high risk of an ischaemic event and require admission to hospital. This should be tested in randomised controlled trials. Selection of patients by telephone did not lead to a high proportion of non-vascular events, because only 22% of patients did not have a confirmed TIA, possible TIA, or minor stroke. Also, most patients selected by telephone who were misdiagnosed with TIA were patients who needed urgent care (eg, patients with epileptic seizures, intracerebral or subdural haematoma, subarachnoid haemorrhage, or myasthenia gravis). The efficacy of our TIA clinic programme might be explained by the short delay between symptom onset and neurological assessment, by the immediate start of secondary stroke prevention therapy, and by the modification of stroke prevention therapy where necessary once the causes of the TIA or minor stroke had been determined. Nearly all patients who had minor stroke, definite TIA, or suspected TIA received antithrombotic treatment, and more than three quarters of patients with atrial fibrillation received anticoagulants. In cases of symptomatic high-grade stenosis, carotid revascularisation was done in most instances within 8 days. At discharge, therapies to lower blood pressure or blood lipid concentrations were started or titrated wherever possible, and we systematically recommended to family doctors a target blood pressure of 140/90 mm Hg (130/85 mm Hg in diabetics) and a target LDL cholesterol concentration of 2 56 mmol/l. Ongoing trials, such as the FASTER trial, 12 will show whether starting of statin therapy without delay is beneficial. Although we did not record adherence of family doctors to recommendations, development of a comprehensive prevention clinic, in which nurse practitioners ensure that all patients are treated after discharge so as to reach target recommendations, might be beneficial. The strengths of our study include its prospective design, the systematic review of all patients by a vascular neurologist, and the low rate of loss to followup. Another strength is that patient selection was not Vol 6 November 2007

7 based on age, which means that our study represents the general population more closely than might a study with restricted selection. However, our study also has several limitations. First, the cohort was not formally population based because, on the basis of disease-related group records, we saw only about a fifth of all TIAs in the Paris area over the study period. Although our recruitment was oriented towards family doctors and office-based specialists (cardiologists, ophthalmologists, and neurologists), we also recruited patients from emergency departments in community hospitals, when emergency medicine physicians judged that patients could be assessed and treated more quickly and easily at the SOS-TIA clinic than in their own hospital. However, some patients with a high risk of stroke or a previous diagnosis of cerebrovascular or ischaemic heart disease might have gone directly to the emergency department of their closest hospital, and such patients might not all have been referred to our TIA clinic by emergency doctors. Second, we might not have seen patients with TIA who were at the highest risk of stroke within 24 h of symptom onset, because patients with persistent symptoms or signs at the time of the telephone call were directed to the standard national emergency telephone line. Because the risk of stroke is high within the first 48 h, 5 7 we did a separate analysis of patients seen within 24 h, and even then noted a lower than expected 90-day stroke risk. A recruitment approach that would have focused only on emergency departments, such as was used for the California Bay cohort, 5 might have selected patients with a high risk of stroke (those with an ABCD² score of 4 or more or 5 or more, or with motor weakness or speech impairment that lasted for more than 10 min within 24 h of symptom onset). When we used these criteria, we also noted a lower than expected 90-day stroke rate, although, according to the 95% CI, the risk might have been as high as 6 7% for patients with an ABCD² score of 5 or more or with motor weakness or speech impairment. Unfortunately, we had insufficient data for separate analysis of patients who were sent to the clinic by emergency medicine physicians. Third, we had no historical controls and did not randomise patients to admission to the SOS-TIA clinic versus usual care. Only a randomised controlled trial could demonstrate that such a TIA clinic can reduce stroke risk. Although we strongly believe that randomisation would be a more accurate and rigorous way to compare a new strategy with usual care, when we designed SOS-TIA our primary aim was to give rapid access to a TIA clinic to all physicians in Ile de France. To compare our series accurately with other published series of patients with TIA, we used the ABCD² score to compare expected and reported event rates. Although the ABCD² score has been validated only in communities in Oxford, UK, and California, USA, the components of this score include the strongest predictors of stroke in all populations (age, blood pressure, and diabetes) and clinical features well recognised as associated with a high risk of stroke (motor deficit, aphasia, and prolonged duration of symptoms), and therefore are likely to be applicable to other populations, such as those elsewhere in the UK and the USA, and in France. Fourth, although only 3% of patients were lost to follow-up, the median delay for interview was 16 months. We cannot therefore exclude recall bias for minor events. However, hard endpoints were not missed. From January, 2006, onwards, patients were interviewed at 1, 3, and 12 months after the initial assessment, to reduce the time between the interview and the time-points under investigation. Further analysis of this cohort will help to clarify whether recall bias has affected our results. Fifth, our study was limited by patients poor awareness of the warning signs of stroke; many did not classify transient neurological symptoms as an emergency. 42% of the patients did not seek medical attention within the first 24 h. Similar results have been observed in the USA and the UK. 15,16 Public education programmes with simple key messages are needed to improve this lack of awareness. Our results have important clinical implications. A low event rate has been reported in a small series of patients with TIA who were managed in a stroke unit rather than in a dedicated TIA clinic, 17 but we show that prompt evaluation and treatment of patients with TIA in a dedicated outpatient unit is associated with a lower than expected risk of subsequent stroke. Because almost three-quarters of patients were discharged home on the same day as the diagnosis, the TIA clinic is also likely to involve lower costs and greater patient satisfaction about their management than is treatment without such a clinic. Contributors PA and P-JT had the idea for the study. PCL, JL, and PA contributed to the study concept and design. JL analysed the data and PCL, JL, and PA interpreted the data and drafted the manuscript. All authors critically revised the manuscript and have seen and approved the final version. Conflicts of interest We have no conflicts of interest. Acknowledgments We thank Fatma Tounkara and Evelyne Herinomenjanahary (clinical research assistants), Catherine Grognet, Caroline Albany, and Mina Sebane (nurse practitioners), other nurses of the Bichat stroke centre, Genevieve Petre (coordinator), and Christiane Gohin (chief nurse). We also thank Sophie Rushton-Smith, who provided editorial assistance in the preparation of this manuscript and was funded by SOS-ATTAQUE CEREBRALE. References 1 Alberts MJ, Latchaw RE, Selman WR, et al. Recommendations for comprehensive stroke centers: a consensus statement from the Brain Attack Coalition. Stroke 2005; 36: Rothwell PM, Warlow CP. Timing of TIAs preceding stroke: time window for prevention is very short. Neurology 2005; 64: Kidwell CS, Alger JR, Di Salle F, et al. Diffusion MRI in patients with transient ischemic attacks. Stroke 1999; 30: Vol 6 November

8 4 Albers GW, Caplan LR, Easton JD, et al. Transient ischemic attack proposal for a new definition. N Engl J Med 2002; 347: Johnston SC, Gress DR, Browner WS, Sidney S. Short-term prognosis after emergency department diagnosis of TIA. JAMA 2000; 284: Lovett JK, Dennis MS, Sandercock PA, Bamford J, Warlow CP, Rothwell PM. Very early risk of stroke after a first transient ischemic attack. Stroke 2003; 34: e Johnston SC, Rothwell PM, Nguyen-Huynh MN, et al. Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet 2007; 369: Landi G. Clinical diagnosis of transient ischaemic attacks. Lancet 1992; 339: ad hoc Committee established by the Advisory Council for the National Institute of Neurological and Communicative Disorders and Stroke. A classification and outline of cerebrovascular diseases. II. Stroke 1975; 6: Dennis MS, Bamford JM, Sandercock PA, Warlow CP. Incidence of transient ischemic attacks in Oxfordshire, England. Stroke 1989; 20: Touboul PJ, Elbaz A, Koller C, Lucas et al. Common carotid artery intima-media thickness and ischemic stroke: the GÉNIC casecontrol study. Circulation 2000; 102: Kennedy J, Eliasziw M, Hill MD, Buchan AM. The Fast Assessment of Stroke and Transient ischemic attack to prevent Early Recurrence (FASTER) Trial. Semin Cerebrovasc Dis Stroke 2003; 3: Touze E, Varenne O, Chatellier G, Peyrard S, Rothwell PM, Mas JL. Risk of myocardial infarction and vascular death after transient ischemic attack and ischemic stroke: a systematic review and metaanalysis. Stroke 2005; 36: Albucher JF, Martel P, Mas JL. Clinical practice guidelines: diagnosis and immediate management of transient ischemic attacks in adults. Cerebrovasc Dis 2005; 20: Johnston SC, Fayad PB, Gorelick PB, et al. Prevalence and knowledge of transient ischemic attack among US adults. Neurology 2003; 60: Giles MT, Rothwell PM. Determinants of delay in seeking medical attention after a TIA or minor stroke. Cerebrovascular Dis 2004; 17 (suppl 5): Calvet D, Lamy C, Touze E, Oppenheim C, Meder JF, Mas JL. Management and outcome of patients with transient ischemic attack admitted to a stroke unit. Cerebrovasc Dis 2007; 24: Vol 6 November 2007