First, there are no formal studies of the cost-effectiveness of risk stratification after infarction, and in their absence it is not warranted to extrapolate from the effectiveness and cost of individual tests to a statement on cost-effectiveness. To do so ignores induced costs (for instance when an abnormal exercise ECG leads to angiography without subsequent intervention), it ignores the relatively high failure rate of the exercise ECG after infarction (submaximal exercise, abnormal resting ECG, etc), and it ignores the need for further investigation in the chronic phase (such as myocardial perfusion imaging if this has not already been performed). For the diagnosis of coronary artery disease, there is now powerful evidence to indicate that strategies that include myocardial perfusion imaging are more cost effective than those that do not. Because diagnosis involves the detection of ischaemia and because persistent ischaemia after infarction is a key factor in prognosis, it is likely that formal studies of cost effectiveness of risk stratification would provide similar results.
Second, although the positive predictive value of any form of pre-discharge non-invasive testing is low, Dr Timmis ignores the superior positive predictive value of perfusion imaging compared with the exercise ECG in patients who have not received thrombolysis (13% versus 9% for death and 24% versus 18% for death or reinfarction). Since the distinction between patients who have and who have not received thrombolysis is mainly the size of their infarcts, it seems to us that a test that is capable of assessing infarct size and prognosis in those with large infarcts should not be cast aside so easily. Shaw and colleagues also emphasised the relatively poor quality of the studies available. For instance, three quarters of the studies were retrospective and one third reported fewer than 5 deaths, and they recognised the need for further studies to provide more reliable information. Further high quality studies are now available and we believe that the weight of evidence is that perfusion imaging is more effective than the exercise ECG for risk stratification after myocardial infarction.
Numerous studies have shown that inducible perfusion abnormalities are more common than ST segment changes on the exercise ECG. This extends observations from the era before thrombolysis when an inducible thallium defect was shown to be better than the exercise ECG for detecting and localising ischaemia and for identifying multi-vessel disease. However, because thrombolysis has reduced the event rate after infarction, there are those who argue that the event rate is now so low that myocardial perfusion imaging is less valuable after infarction than it was.  However these studies included a relatively low risk group with younger patients, preserved left ventricular function, and a low prevalence of multi-vessel disease. The higher risk patients were excluded because they mainly went straight to angiography and revascularisation and it is precisely these who would have been best identified by myocardial perfusion imaging if doubt remained after clinical assessment. In addition, selection bias was operating against perfusion imaging because when it was performed the results were used to guide intervention and hence to exclude the patient from follow-up. It is not surprising that non-invasive testing was not found to be valuable in the few remaining low risk patients.
In better designed studies there is now ample evidence that perfusion imaging is a valuable tool for risk stratification after infarction. Dakik and colleagues reported that a predischarge exercise myocardial perfusion scan in patients who had received thombolysis provided incremental prognostic information over and above clinical and ejection fraction data. In contrast, none of the variables from the exercise ECG contributed to the assessment of prognosis. Similarly, Travin and colleagues found that myocardial perfusion imaging within 14 days of infarction frequently revealed residual ischaemia and was better than clinical and exercise ECG variables in identifying patients at high risk. More recently, Brown and colleagues reported a multi-centre trial in patients with first acute infarction. Three hundred and thirty nine patients were randomised to early (2 to 4 days) dipyridamole myocardial perfusion imaging followed by predischarge (6 to 12 days) sub-maximal exercise imaging, and 112 patients were randomised to submaximal predischarge imaging alone. The findings of the early perfusion study were not available to the responsible physicians and hence patient management was not affected. Early dipyridamole imaging was safe and was predictive of both in-hospital and late cardiac events, and it was a stronger predictor than predischarge submaximal exercise ECG or submaximal exercise perfusion imaging. This prognostic value was independent of thrombolytic status.
Third, with regard to the influence of left ventricular function on prognostic power, Mahmarian and colleagues compared adenosine perfusion imaging and ejection fraction after infarction and found them to have complementary roles. They found that even patients with an ejection fraction greater than 40% were further stratified into low and high risk groups by the extent of the inducible perfusion defect. This study and the guidelines of the American College of Physicians contradict Dr Timmis's recommendation that further risk stratification is not required in asymptomatic patients with an ejection fraction greater than 40%. Indeed, the ACP guidelines suggest that these are the patients in whom non-invasive risk stratification is most successful.
With the growing pressure to reduce costs but maintain quality of care, accurate risk stratification at an early stage using myocardial perfusion imaging with vasodilator stress in all but the highest risk patients could have important benefits. Low risk patients could be discharged earlier than those at a higher risk and in-hospital cardiac events might be prevented. Because of the complementary role of perfusion and functional information, it is conceivable that combined assessment of both using gated SPECT could be highly effective in routine use, particularly since left ventricular end systolic volume appears to be more powerful in prognostic terms than ejection fraction alone.
In summary, we believe that there is sufficient evidence to modify Dr Timmis's algorithm to require early myocardial perfusion imaging rather than exercise ECG in all but the highest risk patients assessed by clinical criteria, and to include patients with relatively preserved LVEF in this strategy. We predict that such an approach will be cost-effective.
S Richard Underwood, MD, FRCP, FRCR, FESC
Imperial College School of Medicine
Royal Brompton Hospital
Sydney St, London SW3 6NP
Constantinos Anagnostopoulos MD, PhD
Royal Brompton Hospital
Sydney St, London SW3 6NP
Leslee J Shaw PhD
Emory University School of Medicine
1518 Clifton Road NE, Rm638
Atlanta, Georgia 30322, USA
1 Timmis A. Acute coronary syndromes: risk stratification. Heart 2000;83:241-6.
2 Shaw LJ, Peterson ED, Kesler K, et al. A metaanalysis of predischarge risk stratification after acute myocardial infarction with stress electrocardiographic, myocardial perfusion and ventricular function imaging. Am J Cardiol 1996;78:1327-37.
3 Underwood SR, Godman B, Salyani S, et al. Economics of myocardial perfusion imaging in Europe: the EMPIRE study. Eur Heart J 1999;20:157-66.
4 Shaw LJ, Hachamovitch R, Berman DS, et al. The economic consequences of available diagnostic and prognostic strategies for the evaluation of stable angina patients: an observational assessment of the value of precatheterisation ischaemia. J Am Coll Cardiol 1999;33:661-9.
5 Tilkemeier PL, Guiney TE, LaRaia PJ, et al. Prognostic value of predischarge low-level exercise thallium testing after thrombolytic treatment of acute myocardial infarction. Am J Cardiol 1990;66:1203-7.
6 Haber HL, Beller GA, Watson DD, et al. Exercise thallium-201 scintigraphy after thrombolytic therapy with or without angioplasty for acute myocardial infarction. Am J Cardiol 1993;71:1257-61.
7 Dakik HA, Mahmarian JJ, Kimball KT, et al. Prognostic value of exercise 201Tl tomography in patients treated with thrombolytic therapy during acute myocardial infarction. Circulation 1996;94:2735-42.
8 Miller TD, Gersh BJ, Christian TF, et al. Limited prognostic value of thallium-201 exercise treadmill testing early after myocardial infarction in patients treated with thrombolysis. Am Heart J 1995;130:259-66.
9 Travin MI, Dessouki A, Cameron T, et al. Use of exercise technetium-99m sestamibi SPECT imaging to detect residual ischemia and for risk stratification after acute myocardial infarction. Am J Cardiol 1995;75:665-9.
10 Brown KA, Heller GV, Landin RS, et al. Early dipyridamole (99m)Tc-sestamibi single photon emission computed tomographic imaging 2 to 4 days after acute myocardial infarction predicts in-hospital and postdischarge cardiac events: comparison with submaximal exercise imaging. Circulation 1999;100:2060-6.
11 Mahmarian JJ, Mahmarian AC, Marks GF, et al. Role of adenosine thallium-201 tomography for defining long-term risk in patients after acute myocardial infarction. J Am Coll Cardiol 1995;25:1333-40.
12 Peterson ED, Shaw LJ, Kesler K, et al. Clinical guideline: part II. Risk stratification after myocardial infarction. Ann Intern Med 1997;126:561-82.