Comparison of the multivariate analysis and CADENZA systems for determination of the probability of coronary artery disease

Influence of 99mTc-tetrofosmin SPECT myocardial perfusion imaging on the prediction of future adverse cardiac events

BACKGROUND

The impact of myocardial perfusion imaging (MPI) on prediction of risk for future cardiac events was examined by comparing predictions based on clinical information alone and in combination with MPI findings.

METHODS AND RESULTS

A 3-cardiologist Cardiac Event Prediction Panel (CEPP) estimated three-year cardiac event (non-fatal MI; aborted sudden cardiac death; cardiac death) risk based on clinical data (C) for 371 subjects. CEPP repeated this estimation after receiving Summed Stress Scores (SSS) and Summed Rest Scores (SRS) from blinded reading of rest-stress (99m)Tc-tetrofosmin MPI SPECT studies (C + MPI). The C and C + MPI estimates were then compared with three-year and total event rates. MPI was normal (SSS 13) in 77 (21%). Eighteen cardiac events occurred within 3 years and 33 occurred during a mean follow-up of 3.9 years. C + MPI estimates of low risk (<1-1.5%/year) identified significantly more patients who did not have events than C. C + MPI three-year event-rate predictions were more accurate than those made with C (P < .01). C + MPI categorization also provided better delineation of incremental risk on time-to-event analyses.

CONCLUSIONS

(99m)Tc-tetrofosmin MPI single photon emission computed tomography (SPECT) findings significantly improve accuracy of cardiac event rate prediction compared to those based on clinical information alone.

Significance of lead strength during exercise testing

BACKGROUND

The aim of the study was to evaluate the contribution of relative lead strengths to exercise-induced ECG changes (ST depression) to predict the degree of myocardial ischemia as compared to the angiograms. This was accomplished by comparing the magnitude of ST depression to the ST/R ratio. Studies have shown that the diagnostic strength of a lead is directly related to the R wave amplitude and that sensitivity is significantly improved.

METHODS

Three hundred patients, who underwent treadmill exercise testing and coronary angiography revealing significant coronary narrowing (> or = 70% luminal diameter narrowing), were studied, along with 150 patients clear of significant coronary artery disease (<70% luminar diameter narrowing). Our goal was to determine the correlation between the relative lead strengths, using a constructed ST/R ratio, to exercise induced ECG changes (ST depression) to predict the presence of myocardial ischemia as compared to angiographic findings. Using a cutoff of 0.1 for the ST/R ratio, our data were compared to the sensitivity and specificity of 1.0 mm ST depression.

RESULTS

Overall sensitivity was improved for the ST/R ratio (84% vs 78%), while specificity was slightly decreased (81% vs 92%) in comparison to standard ST depression. When differentiating between R wave amplitudes, those with R wave < or = 10 mm showed significantly improved sensitivity (88% vs 54%) and a minor decrease in specificity (90% vs 92%). In those with R wave > or = 20 mm, the sensitivity of ST depression was higher (88% vs 71%) but the ST/R ratio was much more specific (88% vs 46%). No significant difference was observed when differentiating between male and female patients.

CONCLUSION

We found that the correction of ST depression for R wave amplitude results in improved sensitivity in patients with low R waves and specificity in patients with very tall R waves (R > or = 20 mm).

Diagnostic accuracy of predicting coronary artery disease related to patients' characteristics

Patients' demographic and clinical characteristics may affect diagnostic accuracy of cardiologists. We asked a group of experienced cardiologists from three institutions to estimate the pretest probability of coronary artery disease in 257 patients referred for diagnostic coronary angiography and with no history of previous myocardial infarction nor valvular heart disease. Physicians pretest estimates were compared with the diagnostic findings of coronary angiography. We tested the influence of five variables on the accuracy of the pretest estimates: age, sex, chest pain characteristics, rest electrocardiogram and electrocardiographic exercise test result. Cardiologists tended to overestimate the presence of coronary artery disease and this tendency was particularly remarkable in the group of patients showing a negative exercise test. Pretest diagnostic accuracy was 0.72 when the test result was negative and 0.85 when the test result was positive (95% confidence interval of the difference 0.03 to 0.23; p < 0.001). The diagnosis of coronary artery disease was also more accurate for male than for female patients (0.81 vs 0.70; 95% confidence interval of the difference 0.02 to 0.21; p < 0.02). Characteristics of chest pain, age and rest electrocardiogram did not affect the level of pretest diagnostic accuracy. Cardiologists should be cognizant of correctly interpreting a negative exercise test and the clinical data of female patients; in both cases, they should move circumspect of the diagnosis of coronary artery disease.

Development and validation of a logistic regression-derived algorithm for estimating the incremental probability of coronary artery disease before and after exercise testing

OBJECTIVES

Our goals were to develop and validate a multivariate algorithm for estimating the incremental probability of the presence of coronary artery disease.

BACKGROUND

Multivariate methods, including logistic regression analysis, have been extensively applied to diagnostic exercise testing. However, few previous studies have included both an incremental design and external validation.

METHODS

A retrospective collection of clinical, exercise test and catheterization data was performed involving four U.S. referral medical centers. All patients had no prior history of coronary disease and had undergone coronary angiography < or = 3 months after exercise stress testing. An algorithm was developed in one center (590 patients with a 41% prevalence of coronary artery disease) with the use of logistic regression analysis and was validated in the other three centers (1,234 patients, 70% prevalence). The algorithm incorporated pretest variables (age, gender, symptoms, diabetes, cholesterol), exercise electrocardiographic (ECG) variables (mm of ST segment depression, ST slope, peak heart rate, metabolic equivalents [METs], exercise angina) and one thallium variable. Discrimination was measured with receiver operating characteristic curve analysis. Calibration (that is, reliability) was assessed from a comparison of probability estimates and the actual prevalence of disease.

RESULTS

The overall incremental receiver operating characteristic curve areas for the validation group were pretest, -0.738 +/- 0.016; postexercise ECG, 0.78 (SE 0.017); and postthallium, 0.82 (SE 0.016); p < 0.01 for both increments. Within the three validation institutions, the institution with a disease prevalence closest to that of the derivation institution had the best incremental receiver operating characteristic curve areas. There was a stepwise incremental improvement in calibration especially from exercise ECG to thallium testing.

CONCLUSIONS

An incremental multivariate algorithm derived in one center reliably estimated disease probability in patients from three other centers. The incremental value of testing was best demonstrated when the derivation and validation groups had a similar disease prevalence. This algorithm may be useful in decision making that relates to the diagnosis of coronary disease.

Comparison of logistic regression and Bayesian-based algorithms to estimate posttest probability in patients with suspected coronary artery disease undergoing exercise ECG

Two multivariate methods, a logistic regression-derived algorithm and a Bayesian independence-assuming method (CADENZA), were compared concerning their abilities to estimate posttest probability of coronary disease in patients with suspected coronary disease. All patients underwent exercise testing within 3 months prior to coronary angiography. Coronary disease was defined as the presence of one or more vessels with greater than or equal to 50% luminal diameter narrowing. A group of 300 patients (disease prevalence = 37%) was used to derive the algorithm. Another group of 950 patients was used to validate the algorithm and compare it to CADENZA. Seven variables (age, sex, symptoms, diabetes, mm ST depression, ST slope, and peak heart rate) were used to generate posttest probabilities for each method. The receiver operating characteristic curve area for the logistic regression method (0.81 +/- 0.01) was significantly higher than CADENZA (0.75 +/- 0.01; p less than 0.05). There was, however, no difference in the calibration of the two methods. When given equivalent variable information, the logistic regression algorithm had better discrimination than CADENZA for estimating the probability of coronary disease following exercise electrocardiography.

The significance of lead strength on ST changes during treadmill stress tests

The contribution of relative lead strengths to ST depression during exercise was evaluated in 334 patients who had both a treadmill stress test and an angiogram. Patients were referred for exercise testing for the evaluation of suspected or known coronary artery disease. This was accomplished by comparing the magnitude of ST-segment depression to a constructed ST/R ratio. Using a cutoff of 0.1 for the ST/R ratio, the data were compared to the sensitivity and specificity of the 1 mm criteria for ST depression. There was only a slight increase in sensitivity (59% vs 63%) and specificity (60% vs 78%) for the ST/R ratio in comparison to the standard ST depression. However, when these two criteria were reevaluated for patients with less than or equal to 10.0 mm of R wave amplitude, the 0.1 ST/R ratio had a small decrease in specificity (94% vs 80%) when compared to 1 mm of ST depression and a marked increase in sensitivity with 31% for the standard ST depression and 82% using the ST/R ratio. In those with an R wave greater than 20 mm, 1 mm of ST depression was much more sensitive than the ST/HR ratio (95% vs 59%), but the ratio was more specific than the conventional ST depression (78% vs 59%). It is concluded that ST depression should be corrected for R wave amplitude in patients with R waves less than 10 mm and over 20 mm.

Logistic discriminant analysis improves diagnostic accuracy of exercise testing for coronary artery disease in women

BACKGROUND

Diagnostic accuracy of the exercise electrocardiogram in women has been shown to be limited for the detection of coronary artery disease. New diagnostic methods based on computer analysis of the exercise electrocardiogram and multivariate analysis have improved the diagnostic value of exercise testing in male subjects. The aim of the present study was to assess whether the diagnostic value of exercise testing can be enhanced in women by using multivariate analysis of exercise data.

METHODS AND RESULTS

Between 1978 and 1984, 135 infarct-free women underwent exercise testing and coronary angiography. Significant coronary artery disease was present in 41% of the patients. In this first group, maximal exercise variables were submitted to a stepwise logistic analysis. Work load, heart rate, and ST60X were selected to build a diagnostic model. The model was tested in a second group of 115 catheterized women (significant coronary artery disease in 47%) and of 76 volunteers. We compared the present model with conventional analysis of the exercise electrocardiogram, with ST changes adjusted for heart rate, and with a previously described analysis. In both groups, sensitivity was better with the present model (66% and 70%) than by conventional (68% and 59%) and by the previously described analysis (57% and 44%) without a loss of specificity (85% and 93%). Receiver-operator characteristic curves showed also a better diagnostic accuracy with the present model.

CONCLUSIONS

In women, logistic analysis of exercise variables improves the diagnostic value of exercise testing. It yields a significantly better sensitivity without a loss of specificity.

A comparison of methods of analysing exercise tests for diagnosis of coronary artery disease

The diagnostic accuracy of the following methods of analysing exercise tests were evaluated: (a) the cumulative area of ST segment depression during exercise normalised for workload and heart rate (exercise score); (b) discriminant analysis of electrocardiographic exercise variables, workload, and symptoms; and (c) ST segment amplitude changes during exercise adjusted for heart rate. Three hundred and forty five men without a history of myocardial infarction were studied. One hundred and twenty three were apparently healthy. Less than half (170) had coronary artery disease. All had a normal electrocardiogram at rest. A Frank lead electrocardiogram was computer processed during symptom limited bicycle ergometry. The accuracy of the exercise score (a) was low (sensitivity 67%, specificity 90%). Discriminant analysis (b) and ST segment amplitude changes adjusted for heart rate (c) had excellent diagnostic characteristics (sensitivity 80%, specificity 90%), which were little affected by concomitant use of beta blockers. Both methods seem well suited for diagnostic application in clinical practice.

Comparison of three Bayesian methods to estimate posttest probability in patients undergoing exercise stress testing

To determine whether recent refinements in Bayesian methods have led to improved diagnostic ability, 3 methods using Bayes' theorem and the independence assumption for estimating posttest probability after exercise stress testing were compared. Each method differed in the number of variables considered in the posttest probability estimate (method A = 5, method B = 6 and method C = 15). Method C is better known as CADENZA. There were 436 patients (250 men and 186 women) who underwent stress testing (135 had concurrent thallium scintigraphy) followed within 2 months by coronary arteriography. Coronary artery disease ([CAD], at least 1 vessel with greater than or equal to 50% diameter narrowing) was seen in 169 (38%). Mean pretest probabilities using each method were not different. However, the mean posttest probabilities for CADENZA were significantly greater than those for method A or B (p less than 0.0001). Each decile of posttest probability was compared to the actual prevalence of CAD in that decile. At posttest probabilities less than or equal to 20%, there was underestimation of CAD. However, at posttest probabilities greater than or equal to 60%, there was overestimation of CAD by all methods, especially CADENZA. Comparison of sensitivity and specificity at every fifth percentile of posttest probability revealed that CADENZA was significantly more sensitive and less specific than methods A and B. Therefore, at lower probability thresholds, CADENZA was a better screening method. However, methods A or B still had merit as a means to confirm higher probabilities generated by CADENZA (especially greater than or equal to 60%).(ABSTRACT TRUNCATED AT 250 WORDS)

Application of probability analysis in the diagnosis of coronary artery disease

The accuracy and applicability of probability analysis to the diagnosis of coronary artery disease is still an open question. Although earlier criticisms are well taken, much of the resistance to the application of probability analysis is based on tradition, rather than logic. Probabilistic algorithms, like any new technology, must be researched and developed and then withstand the test of time. They should not be dismissed simply because they are not traditional. On the other hand, probability analysis in the diagnosis of coronary artery disease must not be accepted just because it is attractive or because it appears to simplify clinical decisions. Application of probabilistic approaches should depend on their accuracy. There is evidence that results of tests and clinical data are not statistically independent. There is also evidence that sensitivities and specificities derived from pooled literature cannot be appropriately applied to just any patient in a particular institution. This is due to variability in the population of patients, a lack of standardization of testing methods, and methodologic problems in reporting results of sensitivities and specificities. In a large institution, where probabilistic formulae can be derived with some degree of confidence, probability analysis has an application today. Discriminant functions will be more accurate than Bayesian formulas, but whatever method one chooses, one must be certain that the parameters used are appropriate. Where the institution is not large enough to generate such a data base, there is presently no accurate approach to the estimation of the probability of coronary disease.

Clinician decisions and computers

The clinician's decisions are subject to numerous distorting influences. Computer decision aids can help avoid these distortions by placing the clinician's limited personal experience into broader perspective through comparison with a larger repository of clinically relevant information; by making explicit the assumptions implied by his or her decisions; and by alerting the clinician whenever the decisions made do not appear consistent with these assumptions, with the available information or with the conventional rules of logic. Practical standards of performance with respect to the development, validation and clinical application of these decision aids are still in evolution, however, and a variety of ethical and legal issues have yet to be addressed. Despite the promise of computer decision aids, it remains to be seen whether their diffusion into medical practice will improve the quality and cost of health care.

The changing role of the exercise electrocardiogram as a diagnostic and prognostic test for chronic ischemic heart disease

The exercise electrocardiogram has been the subject of intense research over the last 50 years, as both a diagnostic and prognostic method to assess patients with chronic ischemic heart disease. In 1986, the strengths and limitations of the technique to predict coronary and multivessel disease in clinical patient subsets are understood. The diagnostic accuracy of the test is improved by consideration of Bayesian theory, multivariate models and new non-ST segment criteria. Post-test coronary disease risk estimates are best reported in terms of a conditional probability, rather than statements of "positive" or "negative." The value of exercise testing in prognostic risk stratification is considerably enhanced by recent reports of long-term follow-up data in asymptomatic and symptomatic patients. Powerful prognostic information can be obtained when the clinical, electrocardiographic and physiologic data from the exercise test are used to formulate the post-test risk of a cardiac event, even in patients whose coronary anatomy is known. The changing role of the exercise electrocardiogram as a diagnostic and prognostic test is reviewed, with emphasis on the strengths and limitations of the procedure.

Computer-assisted diagnosis of acute azotemia

We designed a computer program to make diagnostic suggestions in cases of acute azotemia. The performance of the program was tested in a retrospective manner on 63 patients with acute azotemia. The program correctly identified 90 of 94 diagnoses and made 20 false positive diagnostic suggestions. A revised version of the program correctly identified 94 of 94 diagnoses and made 4 false negative suggestions. The justification for this kind of diagnostic aid is discussed.

Detection of high risk coronary artery disease by thallium imaging

One hundred and three patients who underwent coronary arteriography were studied by thallium imaging and the results analysed by Bayesian principles to assess the usefulness of semiquantitative stress thallium imaging for predicting the presence or absence of multivessel coronary disease. Significant disease was found in 80 patients, of whom 77 had abnormal thallium scans (sensitivity 96%). Thallium images were normal in 15 of 23 patients with no significant disease (specificity 65%). Multiple thallium segmental defects were found to be 90% sensitive and 65% specific for multivessel coronary artery disease and were present in 80% of patients with left main stem disease and in 93% of patients with triple vessel disease. A single thallium defect or normal scan excluded multivessel, left main, and triple vessel disease with 81%, 94%, and 91% predictive accuracy respectively. By Bayesian analysis the predictive accuracy for excluding multivessel disease was greater than 90% in patients with a pretest probability of multivessel disease of less than or equal to 40%. Coronary arteriography to exclude multivessel disease is therefore unnecessary in a high proportion of patients with known or suspected coronary artery disease.