QJM Advance Access originally published online on January 12, 2006
QJM 2006 99(2):81-87; doi:10.1093/qjmed/hcl001
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Risk stratification in acute coronary syndromes—does the TIMI risk score work in unselected cases?
From the 1Department of Medicine, Falkirk and District Royal Infirmary, Falkirk and 2Department of Public Health, University of Aberdeen, Aberdeen, UK
Address correspondence to Dr A.D. Hargreaves, Stirling Royal Imfirmary, Livilands, Stirling FK8 2AU. email: allister.hargreaves{at}fvah.scot.nhs.uk
Received 23 November 2004 and in revised form 21 November 2005
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Background: Management of patients with an acute coronary syndrome (ACS) requires accurate risk stratification to guide appropriate therapy.
Aim: To assess the utility of the TIMI risk score in stratifying patients with possible ACS in routine clinical practice.
Design: Prospective observational study.
Methods: We recruited 869 consecutive patients with a diagnosis of possible ACS attending the acute medical receiving unit of a district general hospital. The main outcome measures were recurrent myocardial infarction, urgent revascularization, and all-cause mortality. TIMI risk score was calculated for each patient, and each was also assigned a risk group based on electrocardiogram (ECG) changes and troponin levels only. After follow-up, Cox univariate and multivariate regression was used to evaluate the influence of potential risk factors on duration of event-free survival, and likelihood ratio tests to assess the fit of the models.
Results: Increasing TIMI risk score was associated with increased risk of events (p<0.001), as was higher risk group from ECG plus troponin stratification (p<0.001). The likelihood ratio comparison favoured the TIMI risk score (difference 13.910, 5 degrees of freedom, p = 0.016).
Discussion: The TIMI risk score is a valid tool for risk stratification in unselected cases with possible acute coronary syndrome. It is superior to ECG changes and troponin alone, although this simpler method also achieves good risk stratification.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
The management of patients with an acute coronary syndrome (ACS) requires accurate risk stratification to guide appropriate therapy. However, these patients represent a heterogeneous group with a variable prognosis. In trials, high-risk patients have benefited from the use of pharmacological agents such as glycoprotein IIb/IIIa inhibitors,1,2 and early invasive management with coronary intervention.3–7 Guidelines for the treatment of ACS are therefore based on risk stratification,8,9 advocating more expensive and invasive therapies for those at higher absolute risk (and therefore those with most benefit to gain). However, the majority of patients with suspected ACS will be at low risk.
The Thrombolysis in Myocardial Infarction (TIMI) Risk Score for stratification of patients with non-ST-segment-elevation myocardial infarction (MI) was derived from the population of the TIMI 11B trial.10 The score is calculated as the sum of seven presenting characteristics, and has been validated in subsequent trial populations.4,11,12 However, there is only one small study (n = 245) evaluating the usefulness of the TIMI risk score in stratifying unselected patients with possible ACS, outside of trial or registry settings.13
The aim of this current study was to prospectively assess the applicability of the TIMI risk score in an unselected group of patients admitted to our acute medical receiving unit with possible ACS. We measured the performance of this scoring system, and compared it to a simple system using electrocardiogram (ECG) changes and serum troponin elevations alone.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Study design
This was a prospective study of consecutive patients admitted to the acute medical receiving unit of a Scottish District General hospital between October 2002 and June 2003. This hospital is in an area of high prevalence of coronary artery disease. Access to angiography is provided by well-established links to tertiary referral centres, with very few patients from the local area admitted directly to other hospitals.
All new admissions to the medical admissions or intensive care units with possible ACS were included. We pragmatically defined possible ACS as any cases that had a troponin assay within 12 h of admission, or had ACS in the differential diagnosis for the admission (as dictated by the admitting officer). There were no exclusion criteria. We therefore included cases with ST-segment-elevation MI (STEMI), non-ST-segment-elevation MI (NSTEMI), and those where ACS was a possible diagnosis.
The TIMI risk score (Table 1) was calculated for all patients, based on information recorded in the patients’ case notes. Hypercholesterolaemia as a risk factor was defined as either a pre-existing diagnosis on admission of hypercholesterolaemia or a plasma cholesterol concentration of >5 mmol/l. Family history was defined as those with a family history of premature vascular disease in a first-degree relative recorded in the case notes; we assumed for the remainder that there was no relevant family history. Patients were also stratified into a risk group by ECG and troponin results (Table 2). ECG changes were deemed to be present if there was at least 0.5 mm of ST elevation or depression, or new left bundle branch block (LBBB). All cases were entered into the study within 24 h of admission.
|
|
Outcomes and follow-up
The primary outcome measurements were recurrent MI, urgent revascularization (percutaneous coronary revascularization or coronary artery by-pass grafting), and all-cause mortality, with a composite end-point being defined as any one of these primary outcomes. Recurrent MI was defined as the presence of at least two of the following three criteria: new episode of chest pain, new ST segment elevation (or new LBBB), and troponin rise. Cases were followed up by reviewing case records on the hospital computerized clinical information system. This system records all admissions, and includes an initial discharge summary sent to the patient's general practitioner. Where this did not include all the data required, final discharge letters, or full case notes were obtained. Additionally, all patients were followed-up using a computerized patient administration system. This tracks patients’ case notes and records a date of death for all deceased patients, even if death occurs outwith the hospital.
Data analysis
The Kaplan-Meier method14 was used to obtain estimates of the event-free survival curves for the patients in each TIMI risk score category. These curves show the proportion of patients who had an event during the follow-up period. For patients who had an event during follow-up, the time to event was defined as the number of days from admission to the event; otherwise the number of days from admission to end of follow-up was used. Categories with small numbers of events were combined. These survival curves were then compared using a linear test for trend. A Cox proportional hazards model was used to obtain estimates of the hazard ratios and their 95%CIs for each category, compared to the lowest category. Similar analyses were carried out to compare survival curves using only ECG changes and a rise in troponin in combination (later referred to as the simple method). Each patient was allocated into a risk group (low, medium-low, medium-high, high) on the basis of absence or presence of ECG changes or any elevation of troponin level (Table 2).
Kaplan-Meier curves and Cox's approach were also used to evaluate the influence on the duration of event-free survival, of each individual risk factor used in the TIMI risk score. All the risk factors used in the TIMI risk score were then entered into a Cox model, and a likelihood ratio test was used to assess the fit of this model. This multivariate analysis computed hazard ratios and their 95%CIs, to measure the strength of the association between each factor and time to event, adjusted for the effects of the other factors in the model. To assess whether the additional risk factors used in the TIMI risk score explained the data significantly better than only using the two risk factors from the simple method, further multivariate Cox modelling was used. A likelihood ratio test compared the fit of two models: the first with just the two factors used in the simple method, the second with the remaining factors from the TIMI risk score added. Also, a forward stepwise Cox model was created (using a 5% level of significance), involving all the TIMI risk score factors.
Data are presented as percentages (absolute numbers) unless otherwise stated. All analyses used SPSS version 13.0.1 for Windows.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
We identified 869 cases of possible ACS. Baseline characteristics are shown in Table 3. Median length of follow-up was 71 days (range 0–310 days). A complete TIMI risk score was obtained for 96% (834) of cases. However, the presence or absence of family history was only recorded in the case notes of 63% (548) of cases, and family history was assumed to be absent if unrecorded. Data completeness was over 96% for all other variables, and any missing values were excluded from analysis. About 69% (597) of patients had experienced chest pain in the 24 h prior to admission.
|
Main outcomes
Rates of recurrent MI, urgent revascularization and death by the end of follow-up were 3% (22), 4% (35) and 8% (70), with a composite end-point rate of 13% (117). Increasing TIMI risk score was associated with increased risk of events (linear test for trend
2 = 40.210, 1 degree of freedom, p<0.001) (Figure 1a). Figure 1b shows the Kaplan-Meier survival curves for the simple method, with an increased risk of event with higher risk groups (linear test for trend, 2 = 63.080, 1 degree of freedom, p<0.001).
|
Regression analysis
Univariate Cox regression analyses revealed statistically significant increased hazard ratios at end of follow-up for increased age, the presence of 50% or more stenosis in any coronary artery, diabetes, elevated troponin level, and the presence of any ST segment deviation (Table 4). Hazard ratios increased with increasing TIMI risk score, and simple method score.
|
After stepwise multivariate Cox regression (Table 5) only age
65 years, elevated troponin level, any ST segment deviation, and the presence of 50% or more stenosis in any coronary artery remained significantly associated with increased odds of further events (after adjusting for the effects of the other factors). The likelihood ratio statistic between the TIMI risk score and the simple method was 13.910 with 5 degrees of freedom (p = 0.016). Thus, adding the five extra variables of the TIMI risk score to the simple method significantly improved the model.
|
Repeating the above analyses after excluding patients with STEMI or new LBBB MI had no significant effect on the results (data not shown but available upon request).
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
In this study, the predictive value of the established TIMI risk score was examined. In addition, we assessed the use of the ECG and troponin changes alone. Importantly, this analysis was performed in an unselected cohort of patients admitted to the acute medical receiving unit of a District General Hospital who were not part of a clinical trial, thereby being more representative of the majority of patients presenting to hospital with ACS.
There was a clear correlation between TIMI risk score and duration of event-free survival. Thus our data support the use of the TIMI risk score in unselected patients in clinical practice, and confirm the findings of Bartholomew et al.13 Figure 1 shows both methods achieving good risk stratification, although small numbers nearer the end of follow-up mean the curves can be unreliable, so the tails should be ignored. The likelihood ratio statistic also shows that the TIMI risk score is superior to considering troponin and ECG changes alone. However, we did not find similar hazard ratios among the seven TIMI criteria.
Four out of the seven TIMI risk criteria remained in the multivariate model as useful for predicting endpoints. The predictive effect of having three or more major risk factors was probably adversely affected by the poor performance of hypercholesterolaemia and family history. The former might be explained by the protective effect of statins15,16 and the increasing trend to prescribe these for secondary prevention in the absence of hypercholesterolaemia since the publication of the Heart Protection Study, or may be due to the high prevalence of hypercholesterolaemia in our population. We only coded a patient as having a positive family history if this was noted on the admission clerking form, and this may have led to under-reporting of this factor, as bias in self-reporting is well known. The original TIMI risk score only awards a point if there is more than one episode of chest pain within the 24 h prior to admission. Although this was also our intention, we found numerous ‘grey cases’ during a pilot phase, and decided it was safest to award one where there was any chest pain during the last 24 h. This limitation may explain why this factor did not perform well on regression analysis. Although there was an increased risk of events for those on aspirin, this was not statistically significant. The association between aspirin use and increased risk is not a consistent finding.17
Our study was conducted on an unselected group of patients that included obvious STEMI and NSTEMI, as well as patients with possible ACS on initial admission. We included all patients where the admitting medical team suspected this diagnosis. This all-inclusive approach was adopted, as it is often difficult to diagnose an atypical presentation of ACS. These patients form an important part of the spectrum of ACS, and are usually omitted from clinical studies. We believe that it would be desirable for any proposed prognostic tool to cater for the full spectrum of possible ACS, including those presenting without chest pain. Indeed, troponin levels are of prognostic significance in patients without ACS.18–20 Although the TIMI risk score was not developed in a population including STEMI, investigators from the global registry of acute coronary events (GRACE) have shown that risk factors for subsequent major events are similar for STEMI and NSTEMI.17
Limitations
The TIMI risk score may have under-performed in our study, due to important differences between our study and that of Antman.10 In particular, difficulties applying the original criteria for chest pain, and imperfect recording of family history may have affected the results. Our follow-up was longer, and this was a more heterogeneous population that included STEMI, although exclusion of this group had no major impact on our results. The good performance of the TIMI risk score despite these limitations suggests it is suitable for use in day-to-day practice.
The future of risk stratification
In the search for the ideal tool for risk stratification, there are other potential ‘prognostic’ factors that we have not investigated including B-natriuretic peptide level,21 high sensitivity C-reactive protein level,22 echocardiographic findings,23,24 and possibly continuous ECG monitoring.25–27 A recent study highlighted the importance of Killip Class,28 the factor most strongly associated with outcome in the GRACE study.17 The GRACE investigators have developed their own risk model, but even their simplified version relies on a computed algorithm. This may make it less accessible and practical than the methods we have assessed. Our study suggests that known coronary artery stenosis, age, elevation of plasma troponin and ECG changes should all be included in any future ‘risk stratification’ model. Outcome measures should include early recurrent MI and recurrent ischaemia requiring urgent revascularization, as including death alone may fail to identify a subgroup requiring aggressive intervention. Risk scoring systems could become increasingly thorough and complex, but should ideally be practical, easy to use and validated in day-to-day clinical practice. There is presently no risk model conforming to all the above. However, our data suggest that currently the TIMI risk score is useful in truly unselected patients treated in non-specialist surroundings.
|
Conclusions |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
We tested risk stratification strategies on a group of unselected cases of possible ACS. The TIMI risk score can be used as a tool for therapeutic decision-making in the general population presenting with possible ACS. It is superior to ECG changes and troponin alone, although this simpler method also achieves good risk stratification.
|
Acknowledgments |
|---|
This project was supported by a Minor Research Grant from Chest, Heart and Stroke Scotland (CHSS). CHSS had no involvement in study design, data collection, analysis or interpretation, or decision to publish.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
1. The Platelet Receptor inhibition in Ischemic Syndrome Management in Patients Limited by Unstable Signs and Symptoms (PRISM-PLUS) Study Investigators. Inhibition of the platelet glycoprotein IIb/IIIa receptor with tirofiban in unstable angina and non-Q wave myocardial infarction. N Engl J Med 1998; 338:1488–97.
2. The PURSUIT Trial Investigators. Inhibition of platelet glycoprotein IIb/IIIa with eptifibatide in patients with acute coronary syndromes. N Engl J Med 1998; 339:436–43.
3. Goodman D. Early-invasive or early conservative management of patients with unstable angina or non-Q-wave myocardial infarction. Adv Ther 2000; 17:45–55.[Medline]
4. Cannon CP, Weintraub WS, Demopoulos LA, et al. Comparison of early invasive and conservative strategies in patients with unstable coronary syndromes treated with the glycoprotein IIb/IIIa inhibitor tirofiban. N Engl J Med 2001; 334:1879–87.
5. Solomon DH, Stone PH, Glynn RJ, et al. Use of risk stratification to identify patients with unstable angina likeliest to benefit from an invasive versus conservative management strategy. J Am Coll Cardiol 2001; 38:969–76.
6. Solomon DH, Ganz DA, Avorn J, et al. Which patients with unstable angina or non-Q-wave myocardial infarction should have immediate catheterisation? A clinical decision rule for predicting who will fail medical therapy. J Clin Epidemiol 2002; 55:121–8.[Medline]
7. Fragmin and Fast Revascularisation during InStability in Coronary artery disease (FRISC II) Investigators. Invasive compared with non-invasive treatment in unstable coronary-artery disease: FRISC II prospective randomized multicentre study. Lancet 1999; 354:708–15.[CrossRef][ISI][Medline]
8. Boden WE. Interpreting new treatment guidelines for non-ST-segment elevation acute syndromes. Am J Cardiol 2001; 88(suppl.):19K–24K.[Medline]
9. European network for acute coronary syndrome treatment—ENACT. Eur Heart J 2000; 21:1440–9.
10. Antman EM, Cohen M, Bernink PJLM, et al. The TIMI Risk Score for Unstable Angina/Non-ST Elevation MI. A Method for Prognostication and Therapeutic Decision Making. JAMA 2000; 284:835–42.
11. Morrow DA, Antman EM, Snapinn SM, et al. An integrated clinical approach to predicting the benefit of tirofiban in non-ST elevation acute coronary syndromes. Application of the TIMI Risk Score for UA/NSTEMI in PRISM-PLUS. Eur Heart J 2002; 23:223–9.
12. Samaha FF, Kimmel SE, Kizer JR, et al. Usefulness of the TIMI Risk Score in predicting both short and long term outcomes in the Veterans Affairs Non-Q-Wave Myocardial Infarction Strategies In-Hospital (VANQWISH) Trial. Am J Cardiol 2002; 90:922–6.[CrossRef][ISI][Medline]
13. Bartholomew BA, Sheps DS, Monroe S, McGorray S, Smith K, Pepine CJ. A population-based evaluation of the thrombolysis in myocardial infarction risk score for unstable angina and non-ST elevation myocardial infarction. Clin Cardiol 2004; 27:74–8.[ISI][Medline]
14. Parmar MKB, Machin D. Survival analysis: a practical approach. Chichester, John Wiley & Son, 1995.
15. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomized, placebo-controlled trial. Lancet 2002; 360:7–22.[CrossRef][ISI][Medline]
16. Steg PG, Goldberg RJ, Gore JM, et al. Baseline characteristics, management practices, and in-hospital outcomes of patients hospitalized with acute coronary syndromes in the Global Registry of Acute Coronary Events (GRACE). Am J Cardiol 2002; 90:358–63.[CrossRef][ISI][Medline]
17. Granger CB, Goldberg RJ, Dabbous O, et al. Predictors of hospital mortality in the Global Registry of Acute Coronary Events. Arch Intern Med 2003; 163:2345–53.
18. Baillard C, Boussarar M, Fosse JP, et al. Cardiac troponin I in patients with severe exacerbation of chronic obstructive disease. Intensive Card Med 2003; 29:584–9.
19. Pruszczyk P, Bochowicz A, Torbicki A, et al. Cardiac troponin T monitoring identifies high-risk group of normotensive patients with acute pulmonary embolism. Chest 2003; 123:1947–52.
20. Ammann P, Maggiorini M, Bertel O, et al. Troponin as a risk factor for mortality in critically ill patients without acute coronary syndrome. J Am Coll Cardiol 2003; 41:2004–9.
21. Jernberg T, Stridsberg M, Venge P, Lindahl B. N-terminal pro brain natriuretic peptide on admission for early risk stratification of patients with chest pain and no ST-segment elevation. J Am Coll Cardiol 2002; 40:437–45.
22. de Winter RJ, Fischer J, Bholasingh R, et al. C-Reactive protein and cardiac troponin T in risk stratification: differences in optimal timing of tests early after the onset of chest pain. Clin Chem 2000; 46:1597–603.
23. Bodi VV, Sanchis J, Llacer A, et al. Prognostic markers of non-ST elevation acute coronary syndromes. Rev Esp Cardiol 2003; 56:857–64.[Medline]
24. Swinburn JM, Stubbs P, Soman P, et al. Independent value of tissue harmonic echocardiography for risk stratification in patients with non-ST-segment elevation acute chest pain. J Am Soc Echocardiogr 2002; 15:1031–7.[Medline]
25. Jernberg T, Lindahl B, Wallentin L. ST-segment monitoring with continuous 12-lead ECG improves early risk stratification in patients with chest pain and ECG nondiagnostic of acute myocardial infarction. J Am Coll Cardiol 1999; 34:1413–19.
26. Jernberg T, Lindahl B, Wallentin L. The combination of a continuous 12-lead ECG and troponin T; a valuable tool for risk stratification during the first 6 hours in patients with chest pain and a non-diagnostic ECG. Eur Heart J 2000; 21:1403–5.
27. Decker WW, Prina LD, Smars PA, et al. Continuous 12-lead electrocardiographic monitoring in an emergency department chest pain unit: an assessment of potential clinical effect. Ann Emerg Med 2003; 41:342–51.[Medline]
28. Khot UN, Jia G, Moliterno DJ, et al. Prognostic importance of physical examination for heart failure in non-ST elevation acute coronary syndromes: the enduring value of the Killip classification. JAMA 2003; 290:2174–81.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||