METHODS

We identified 290 eligible TIA patients at the end of planned follow up of the Oxfordshire community stroke project (OCSP),^7,14 and a contemporaneous Oxford cohort study of hospital referred patients with TIA.¹⁵ Both cohorts had been assembled by the same original investigators, using the same diagnostic criteria. The OCSP was a population based study of the incidence of first ever TIA in a population of 105 000 in Oxfordshire, UK, that ran from 1981 to 1986. The hospital referred cohort was recruited during the same period from patients referred from the remaining population of Oxfordshire that was not covered by the OCSP and included both first ever and recurrent TIA. Both studies involved a consecutive series of patients.

Patients were included in our study if they had presented with a TIA in either of the two original studies, and if they were alive and had not had a stroke at the end of 1988. Median time from most recent TIA to study entry was 3.8 years (interquartile range, 2.2 to 5.8 years). After ethical approval and permission from their general practitioner, patients were interviewed by a research nurse at home to ensure that they remained free of stroke, to record their current baseline clinical characteristics, and to take a blood sample for biochemical and haematological tests.

Fatal events during follow up were identified through the Office for National Statistics National Health Service Central Register (ONS NHSCR). The ONS NHSCR coded the underlying cause of death according to ICD-9 and ICD-10. The main outcomes of interest were stroke (ICD-9: 430–438), myocardial infarction/coronary heart disease (ICD-9: 410–414), and cardiovascular death (ICD-9: 390–459). Case notes of all subjects who died were also reviewed centrally. Non-fatal events were identified by direct contact with the family physician of all patients in 1994 and 1998. Non-fatal strokes were defined as events lasting more than 24 hours. Myocardial infarction was defined on the basis of changes in ECG and enzymes. The date of last mortality follow up was August 2000.

Statistical analysis

Kaplan–Meier survival analysis was used to calculate event rates. All analyses were limited to 10 years of follow up—that is, the date of last follow up for non-fatal events in 1998. For fatal events, the expected age and sex standardised death rates were estimated using mortality records for England and Wales. Differences between observed and expected death rates were expressed as standardised mortality ratios (SMR), with confidence intervals and probability (p) values obtained from the Poisson distribution.

RESULTS

Just under half of the patients in the two original cohorts were eligible for inclusion in the current study: 89/184 (48%) in the OCSP and 201/469 (43%) in the hospital referred series. The main reasons for non-inclusion were that the patient had died before 1988, or that they had had a stroke. The characteristics of the 290 study participants are described in table 1. The cohort was elderly (median age 69 years), and predominantly male (62%). The 89 patients derived from the OCSP were older than the 201 patients from the hospital referred cohort (mean age 71.9 v 68.2 years, p = 0.005), but there were no other significant differences in baseline characteristics. Ten year follow up was available in all patients.

Fatal events

During 10 years of follow up (2303 person-years) we observed 147 deaths, of which 20 were attributable to stroke (two with intracerebral haemorrhages) and 53 to myocardial infarction or coronary heart disease. In all, 87 deaths were caused by vascular disease. There were 60 non-vascular deaths. All cause mortality was not significantly greater than expected on the basis of age and sex specific national death rates (table 2). The number of stroke deaths was only slightly greater than expected (SMR = 1.23; 95% confidence interval (CI), 0.75 to 1.91; p = 0.4), but the number of deaths from myocardial infarction or coronary heart disease was significantly higher than expected (SMR = 1.47; 95% CI, 1.10 to 1.93; p = 0.009).

All vascular events

Overall, 45 subjects had a non-fatal or fatal stroke and 67 had a non-fatal or fatal myocardial infarction or died from coronary heart disease. Kaplan–Meier curves for any stroke, any myocardial infarction or coronary heart disease, and anystroke, myocardial infarction, or vascular death are shown in fig 1. There were no significant differences in any of these risks between subjects derived from the OCSP and subjects derived from the concurrent hospital referred cohort. For each outcome, event rates were constant through time. The 10 year risk of any stroke was 18.8% (95% CI, 13.6% to 23.7%; 45 events) and the 10 year risk of myocardial infarction or death from coronary heart disease was 27.8% (95% CI, 21.8% to 33.3%; 67 events). In all, 114 patients had at least one major vascular event (stroke, myocardial infarction, or vascular death), giving a 10 year risk of any first stroke, myocardial infarction, or vascular death of 42.8% (95% CI, 36.4% to 48.5%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Kaplan–Meier event rates for any stroke, any myocardial infarction, or death from coronary heart disease (MI/CHD), and any stroke, myocardial infarction, or vascular death.

DISCUSSION

There have been three previous studies of long term outcome after TIA,^16–18 and one study in patients with minor stroke.¹⁹ However, all of these studies recruited patients within days or weeks of the presenting event and therefore include the early period of follow up during which the stroke risk is high. Two studies were retrospective.^16,17 One of these was confined to patients under 40 years of age,¹⁷ and the other was done in the 1960s and early 1970s before routine use of antiplatelet agents and other preventive treatments.¹⁶ The one prospective study with long term follow up after TIA was based on only 18 patients.¹⁸

Thus our prospective study is the only large one of long term prognosis after a TIA in the modern era. It is also the first published study of the long term risks of stroke, myocardial infarction, and vascular death in patients who have survived free of stroke for several months or years after a TIA. It provides useful data on which clinical decisions about prevention of stroke and other vascular events can be based. We have shown that although the risk of stroke is relatively low in patients who survive the early high risk period, the risk of coronary vascular events remains high, and the overall risk of major vascular events is constant at about 4% per year.

The cost–effectiveness of vascular preventive treatments—such as new antiplatelet drugs, antihypertensive drugs, and statins—depends on the absolute risk of major vascular events.²⁰ The 4% combined annual risk of stroke, myocardial infarction, or vascular death in our cohort did not diminish with time, at least out to 10 years of follow up. Thus even in patients who had remained free of major vascular events for over 10 years after a TIA, the overall vascular risk remains clinically significant. In the absence of evidence of reduced effectiveness of preventive treatments with increasing age or length of treatment, our data suggest that such treatments should be continued for at least 10 years after a TIA, and possibly for life.

Although we have shown that the long term vascular risk is high in patients with a non-recent TIA, these risks are probably underestimates for two reasons. First, our eligibility criteria meant that we had a relatively low risk cohort. We excluded any patient who had had a stroke during follow up in the original studies up to 1988. These patients would have had a higher risk of further strokes, and possibly of other vascular events, during further follow up than those who had remained stroke-free. Second, it is possible that ascertainment of non-fatal events may have been less effective than for fatal events because we relied to a significant extent on ascertainment of events from the medical records of the family physician. However, if our cohort was “low risk” and we did miss some non-fatal vascular events, then this would only further emphasise our overall conclusion that the long term vascular risk remains high in patients with a non-recent TIA.

Our study does, however, have various other shortcomings. First, in order to achieve a sufficient sample size we studied a cohort that was derived from two original studies, one community based and one hospital based. However, although this is not ideal, both cohorts had been assembled by the same investigators from the same geographical region using the same diagnostic criteria. Moreover, we applied the same eligibility criteria to both studies, and both cohorts were at the same time point in the evolution of their disease (mean of four years since the last TIA). Consequently, with the exception of a small difference in age, there were no significant differences between the cohorts in baseline clinical characteristics, and we found no differences in the risks of vascular events on subsequent follow up. Indeed, this lack of heterogeneity, despite the differences in origin of our patients, supports the generalisability of our findings. Second, we did not have sufficient numbers of strokes for a reliable determination of the risk of vascular events separately in the different pathological and aetiological subtypes of TIA, or of risk in relation to particular baseline clinical characteristics.

Limitations aside, this is the first long term follow up study of patients with TIA in the modern era when treatment of hypertension and hyperlipidaemia and the use of antiplatelet agents are likely to have improved prognosis. Despite this, we found that the combined risk of first stroke, myocardial infarction, or vascular death was approximately 4% a year and did not diminish with time. These results suggest that it is important to continue vascular preventive strategies in the long term in patients with a previous TIA, even if they have been free of any major vascular events for several years. In routine clinical practice it is likely that the risks will be higher than those reported here, and it is unlikely that they will be any lower in patients with a previous stroke rather than a TIA. The requirement of the National Service Framework for Older People that all general practices in the United Kingdom set up registers of patients with a previous TIA or stroke⁸ therefore appears to be reasonable, and should facilitate the identification of patients in whom there is scope for more active preventive treatment.