Calculation of power and sample size with bounded outcome scores

The two-sample Wilcoxon rank sum test is the most popular non-parametric test for the comparison of two samples when the underlying distributions are not normal. Although the underlying distributions need not be known in detail to calculate the null distribution of the test statistic, parametric assumptions are often made to determine the power of the test or the sample size. We encountered difficulties with this approach in the planning of a recent clinical trial in stroke patients. It is shown that, for power and sample size estimation, it can be dangerous to apply the classical formulae routinely, especially with outcome scores having a U-shaped or a J-shaped distribution. As an example we have taken the Barthel index, a quality-of-life outcome measure in stroke patients. Further, we have investigated alternative methods by means of Monte Carlo simulation. The distributional characteristics of the estimated powers were compared. Our findings suggest more appropriate computer software is necessary for the calculation of power and sample size when efficacy is measured by a non-parametric method.

Early and late effects of rt-PA vs placebo on left ventricular function measured by nuclear ventriculography

The aim of the study was the functional re-evaluation of 296 patients 12 to 18 months after a double-blind trial evaluating the effect of tissue plasminogen activator (rt-PA) versus placebo given within 5 h of onset of symptoms caused by an acute myocardial infarction. All patients underwent rest-stress radionuclide angiography (Egna). For each exercise level the global left ventricular ejection fraction (LVEF) was calculated together with an estimate of regional wall motion abnormalities (RWMA). A clear difference of the total workload and the peak workload was found between both therapeutic groups. Discriminant analysis evaluating four parameters (LVEF at peak exercise and at the endpoint and the workload at those levels) revealed a beneficial therapeutic effect. The RWM at rest showed only a difference in the apico-inferior region. There were less wall motion abnormalities in the treated group. Radionuclide analysis demonstrated a larger functional capacity and a better coordination of myocardial contractility during stress RNA one year after thrombolytic therapy. At rest, no major differences were found between the hospital stage and the follow-up in both therapeutic groups one year later.

Three-year follow-up of five posterior composites: SEM study of differential wear

Assessing the wear of both enamel and composite at a shared occlusal contact area offers the opportunity to determine the differential wear between enamel and composite on the same tooth. Differential wear measurements were carried out on five posterior composite materials with an accurate 3D-measuring technique. The distinct wear step between enamel and composite was visualized by means of scanning electron photomicrographs. Both investigated ultrafine midway-filled and fine compact-filled composites (Willems et al., 1992a) have a considerably higher differential wear value and are, therefore, less suitable for rehabilitation of posterior teeth than are the ultrafine compact-filled composites, which have, in this study, a very satisfactory differential wear rate and appear to be highly wear resistant materials suitable for use in stress-bearing areas.

Early and Late Effects of rt-PA vs Placebo on Left Ventricular Function Measured by Nuclear Ventriculography

The aim of the study was the functional re-evaluation of 296 patients 12 to 18 months after a double-blind trial evaluating the effect of tissue plasminogen activator (rt-PA) versus placebo given within 5 h of onset of symptoms caused by an acute myocardial infarction. All patients underwent rest-stress radionuclide angiography (Egna). For each exercise level the global left ventricular ejection fraction (LVEF) was calculated together with an estimate of regional wall motion abnormalities (RWMA). A clear difference of the total workload and the peak workload was found between both therapeutic groups. Discriminant analysis evaluating four parameters (LVEF at peak exercise and at the endpoint and the workload at those levels) revealed a beneficial therapeutic effect. The RWM at rest showed only a difference in the apico-inferior region. There were less wall motion abnormalities in the treated group. Radionuclide analysis demonstrated a larger functional capacity and a better coordination of myocardial contractility during stress RNA one year after thrombolytic therapy. At rest, no major differences were found between the hospital stage and the follow-up in both therapeutic groups one year later.

Synergistic antithrombotic properties of G4120, a RGD-containing synthetic peptide, and argatroban, a synthetic thrombin inhibitor, in a hamster femoral vein platelet-rich thrombosis model

The synergistic antithrombotic properties of G4120, a synthetic Arg-Gly-Asp (RGD) containing peptide which strongly inhibits platelet aggregation, and of Argatroban, a synthetic thrombin inhibitor, were examined in a reproducible quantitative hamster femoral vein platelet-rich mural thrombosis model. Bolus injections of G4120 and Argatroban inhibit thrombus formation in a dose-dependent way; 50% inhibition (ID50) is obtained with 11 micrograms/kg G4120 and with 2 mg/kg Argatroban. Combined bolus injections of 3 micrograms/kg G4120 with 0.5, 0.75 or 1 mg/kg Argatroban and of 1 mg/kg Argatroban with 1.5 or 3 micrograms/kg G4120 caused linear dose-dependent inhibition of thrombus formation, whereas 3 micrograms/kg G4120 or 1 mg/kg Argatroban alone had very little effect (less than 20% inhibition). ID50 was obtained with the combination of 3 micrograms/kg G4120 and 0.5 mg/kg Argatroban, corresponding to an equi-effective fractional combination of 0.62 with a 95% confidence interval of 0.50 to 0.74. Alternatively the ID50 was obtained with the combination of 1 mg/kg Argatroban and 1.3 micrograms/kg G4120, corresponding to an equi-effective fractional combination of 0.52 with a 95% confidence interval of 0.18 to 0.86. In both instances these results are indicative of a significant synergistic interaction. Bolus injection of 10 mg/kg aspirin, 100 U/kg heparin or the combination did not inhibit thrombus formation. The synergistic effect of the combination of platelet inhibiting RGD-peptides and synthetic thrombin inhibitors could be useful in the prevention of arterial occlusion with platelet-rich thrombus in patients with ischemic heart disease following thrombolytic therapy or angioplasty, although this combination is not expected to reverse platelet thrombus formation.

Effects of intravenous urokinase versus alteplase on total pulmonary resistance in acute massive pulmonary embolism: a European multicenter double-blind trial. The European Cooperative Study Group for Pulmonary Embolism

Twelve centers participated in a double-blind study in which 63 patients with angiographically documented acute massive pulmonary embolism were randomly assigned to treatment with either urokinase (4,400 U/kg as an intravenous bolus infusion, then 4,400 U/kg per h over 12 h; n = 29) or alteplase (10 mg as an intravenous bolus infusion, then 90 mg over 2 h) followed by heparin (n = 34). The primary objective was to compare the resolution of pulmonary embolism as judged by the change in total pulmonary resistance over the initial 2 h. Further objectives were to evaluate the changes in total pulmonary resistance over the next 10 h and the degree of angiographic resolution at 12 to 18 h. At 2 h, total pulmonary resistance decreased by 18 +/- 22% in the urokinase group and by 36 +/- 17% in the alteplase group (p = 0.0009). Continuous monitoring of pulmonary artery mean pressure, cardiac index and total pulmonary resistance revealed that these variables improved faster in the alteplase group, with consistently significant intergroup differences from 30 min up to 3 to 4 h. After 12 h, the decrease in total pulmonary resistance was 53 +/- 19% in the urokinase group compared with 48 +/- 17% in the alteplase group and the reduction in the angiographic severity score was 30 +/- 25% compared with 24 +/- 18%, respectively, with no significant intergroup differences. Bleeding was equally frequent in the two treatment groups, except that more urokinase-treated patients experienced hematomas at puncture sites.

Risk factors for arrhythmia and death after Mustard operation for simple transposition of the great arteries

From 1965 to 1980, 249 consecutive patients underwent a Mustard procedure for simple transposition of the great arteries. Hospital mortality was 23 of 249 (9.2%); follow-up for 24 years (mean, 11.7 +/- 6.1 yr) revealed 50 of 249 (20%) late deaths, of which seven were noncardiac, six were in the hospital, and 37 were sudden (27 patients with poor hemodynamics and 10 unexplained). Actuarial survival after 1, 10, and 20 years was 85, 75, and 67%. The instantaneous risk of death decreased rapidly after surgery but showed a second peak from 8 to 15 years postoperation. When last seen 87% were in New York Heart Association (NYHA) class I; 69% had sinus rhythm, 20% nodal rhythm, 10% atrial flutter, and 1% paced. Survival in uninterrupted sinus rhythm after 5, 10, and 20 years was 75, 62, and 47%. Nodal rhythm detected by annual ECG occurred in 69 patients. Excluding the early experience, the risk of loss of sinus rhythm was constant in time and constant throughout the experience (2.4%/yr). Atrial flutter was observed in 36 patients. Occurrence of atrial flutter was limited to two discrete surgical periods: the first operations (26 with atrial flutter of the first 50) and an ill-defined period in the middle of the experience. Patients not operated on during these two periods were very unlikely to develop atrial flutter (one of 104, p less than 0.001). As for death, the hazard function showed a bimodal curve, the second peak appearing earlier from 5 to 11 years. If nodal rhythm was noted on the yearly routine ECG, the risk for developing subsequent atrial flutter increased by a factor of 2.1 (p less than 0.05).(ABSTRACT TRUNCATED AT 250 WORDS)

Multivariate probit analysis: a neglected procedure in medical statistics

The multivariate probit model is designed to regress a vector of correlated quantal variables on a mixture of continuous and discrete predictors. Various applications can be found in the biological, economical and psychosociological literature, but the method is not yet widely used in medical applications. We reintroduce this model thereby showing its usefulness in medical problems. Software for this model is, however, not widely available. We have written a PC program to select predictors and estimate parameters in the multivariate probit framework. The performance and characteristics of the program are briefly illustrated.

Trends in the selection of dental filling materials

In 1983 and 1988 a questionnaire was sent to a random sample of Belgian dentists to compare the selection of filling materials such as amalgam and gold with the tooth-coloured filling materials, composites and glass polyalkenoate (ionomer) cements, for the restoration of Class II lesions in premolar and permanent molar teeth. Despite the increasing demand for aesthetics to be considered, the use of composite resins in the posterior region only increased moderately over the assessment period, while amalgam maintained its leading position as the restorative material of choice in premolar and molar teeth. Only in situations where aesthetics played a primary role did composite resins gain over silver-based fillings. There was no change in the use of gold as an indirect restorative material. Finally, the selection rate of glass ionomers increased slightly.

Effect of superoxide dismutase on infarct size and postischemic recovery of myocardial contractility and metabolism in dogs

The effects of superoxide dismutase treatment on infarct size, postischemic recovery of contractile function and tissue content of high energy phosphates were examined in a canine model of myocardial ischemia and reperfusion. Ischemia was induced by thrombotic occlusion of a coronary artery and reperfusion was achieved by intravenous thrombolysis. Average duration of ischemia was 90 min. Fifty closed chest anesthetized dogs were randomized to receive either superoxide dismutase (34,000 IU/min intravenously) or placebo, starting approximately 30 min before and continuing for 30 min into the reperfusion phase. Left ventricular ejection fraction and regional segmental shortening of the postischemic area were calculated from contrast angiograms after 4 h, 48 h and 1 week of reperfusion. Tissue content of high energy phosphates was determined from transmural biopsy after 4 h and 1 week. Infarct size was measured by planimetry of dye-stained heart slices. In the superoxide dismutase and placebo-treated groups, respectively, the mortality rate was 25% and 16%, collateral flow 20 +/- 10 and 23 +/- 18 ml/min per 100 g, area at risk 25 +/- 6% and 26 +/- 7% of the left ventricle and infarct size 28 +/- 19% and 36 +/- 27% of the area at risk. Multiple regression analysis failed to show any beneficial effect of superoxide dismutase treatment on infarct size. Left ventricular ejection fraction, regional segmental shortening of the postischemic area and tissue content of high energy phosphates recovered to a similar extent and at a similar rate in both treated and placebo groups up to 1 week after reperfusion. Thus, in this model of coronary occlusion and reperfusion superoxide dismutase treatment is of no benefit.

Suggestive evidence that pelvic endometriosis is a progressive disease, whereas deeply infiltrating endometriosis is associated with pelvic pain

In a 3-year prospective study of 643 consecutive laparoscopies for infertility, pelvic pain, or infertility and pain, the pelvic area, the depth of infiltration, and the volume of endometriotic lesions were evaluated. The incidence, area, and volume of subtle lesions decreased with age, whereas for typical lesions these parameters and the depth of infiltration increased with age. Deeply infiltrating endometriosis was strongly associated with pelvic pain, women with pain having larger and deeper lesions. Because deep endometriosis has little emphasis in the revised American Fertility Society classification and after analyzing the diagnoses made in each class, considerations for a simplifying revision with inclusion of deep lesions are suggested. In conclusion, suggestive evidence is presented to support the concept that endometriosis is a progressive disorder, and it is demonstrated that deep endometriosis is strongly associated with pelvic pain.

Dairy fat, saturated animal fat, and cancer risk

The relationship between the per person supply of fat from dairy products and lard and cause-specific cancer mortality was examined using 1979-1981 FAO data from 36 countries. Significant correlations (P less than 0.01) were found between dairy and lard fat intake and total, breast, prostate, rectal, colon, and lung cancer. Significant relationships between site-specific cancer mortality and ischemic heart disease mortality could also be established. Hormone-dependent cancers significantly correlate with both hormone-dependent and non-hormone-dependent cancers of the opposite sex, pointing toward other than sex-linked factors as an explanation. The level of saturated fat intake provides the most plausible link for the relationship between the cancers considered and ischemic heart disease mortality. Our findings support the concept of the important role saturated fat plays in cancer promotion.

The relationship between animal fat intake, cigarette smoking, and lung cancer

An international ecologic study on the relationship between fat intake, cigarette consumption, and age-specific lung cancer mortality rates was conducted using data obtained from 29 countries. The 1969-71 and 1984-86 mortality rates for lung cancer were related to smoking habits and to levels of animal fat intake. Highly significant independent positive correlations for lung cancer with both cigarette smoking and animal fat consumption were seen. No significant independent relationship was found between lung cancer mortality and vegetable fat consumption. In a multiple regression analysis linking lung cancer mortality to dietary intake of animal fat, cigarette smoking, and the interaction term between cigarette smoking and animal fat intake, only the latter term was significant (P less than 0.001) pointing towards a role for animal fat as a promoter of lung cancer.

The return of clinically evident ischemia after coronary artery bypass grafting

Although survival after coronary artery bypass grafting (CABG) is the most serious outcome information, the quality of life in living patients is largely determined by the freedom from ischemic events. The return of angina, acute myocardial infarct and sudden death were studied in a large (n = 5880) population of patients undergoing CABG between 1971 and 1987. The freedom from angina pectoris was 95%, 83% and 63% at 1, 5 and 10 years, respectively, after surgery. Early return of angina was related to both procedure incremental risk factors (incomplete revascularization and non-use of the internal mammary (thoracic) artery (IMA) as a conduit) and patient incremental risk factors (aggressiveness of the atherosclerotic process and severity of preCABG symptoms). Late angina return was related to patient risk factors including coexisting factors (hyperlipidemia and hypertension), preCABG symptom severity and gender (female). The freedom from an acute fatal or non-fatal postCABG myocardial infarct was 99%, 96% and 85% at 1, 5 and 10 years after surgery. The incremental risk factors for early infarction were related to incomplete revascularization, but late infarction was related to lipid levels, coexisting diseases (diabetes, positive family history) and non-use of IMA to LAD. The freedom from sudden death was 99.8%, 99% and 97% at 1, 5 and 10 years, respectively, after surgery. The incremental risk factors were dominated by the severity of the left ventricular dysfunction. The freedom from any ischemic event (any of the previous three) was 93%, 79% and 54% at 1, 5 and 10 years, respectively, after surgery. The incremental risk factors included all those cited above for the specific components. Patient-specific predictions validate the influences of these risk factors. They demonstrate that unlike the profound influence of the use of the IMA on survival, there is little benefit of the use of the IMA on return of ischemic events over and above the effect of revascularization per se. The study demonstrates that most patients will experience return of ischemic symptoms within a period of 15-20 years after surgery, but that this is most likely to be return of angina and rarely sudden death.

On the sex ratio of urinary cation excretion obtained from Intersalt and other epidemiological studies

Large differences exist within populations in the dietary intake of nutrients between men and women and these differences are still largely unknown. Using data obtained from the Intersalt study and from other relevant epidemiological studies, the sex ratio of the 24 h urinary excretion of different cations was obtained. Highly significant correlations (P less than 0.001) were found between the 24 h urinary cation and creatinine excretions of men and women. Highly variable sex ratios of 24 h urinary excretion of cations were obtained between populations, presumably due to the different levels of physical activity between men and women. The sex ratio of blood pressure did not correlate with the sex ratio of the 24 h urinary excretion of sodium, potassium, calcium or magnesium, but correlated significantly with the sex ratio of the BMI (P less than 0.01). On a global scale men had a nearly 22% higher urinary excretion of cations than women, pointing towards a 22% higher dietary intake of the cations considered.

Determinants of infarct size in non-human primates

To achieve a better understanding of the major factors that determine infarct size in non-human primates, a mathematical model was constructed using stepwise regression analysis. The model was developed on the basis of infarct size measurements, including the anatomical area at risk, regional myocardial blood flow measurements and hemodynamic determinants obtained in 23 control baboons undergoing up to 2 h of coronary artery thrombosis followed by thrombolysis. In this model, the size of the perfusion bed of the occluded coronary artery and the duration of coronary artery occlusion were found to be the only important predictors of infarct size (expressed as a percentage of left ventricular mass). R2 (square or the multiple correlation coefficient) was 70% in this model. Collateral blood flow and rate-pressure product were not identified as important predictors of infarct size. In a second group of eight baboons, atenolol (0.1 mg.kg-1) was administered intravenously 15 min after the onset of coronary artery thrombosis. Predicted infarct size (based on the mathematical model obtained in the control group) was larger than the observed infarct size in seven out of eight cases. In four instances observed infarct size was smaller than the 95% lower limit of the predicted value. It is concluded that the determinants of infarct size in non-human primates differ from those in canine models with respect to collateral flow and estimates of myocardial oxygen consumption (rate pressure product). The developed mathematical model of infarct size prediction allows the detection of cardioprotective drug effects with an acceptable efficacy.

Evaluation of the effect of thrombolytic treatment on infarct size and left ventricular function by enzymatic, scintigraphic, and angiographic methods. The European Cooperative Study Group for Recombinant Tissue Type Plasminogen Activator

In a double-blind trial of the European Cooperative Study Group, 721 patients with acute myocardial infarction of less than 5 hours' duration were given either 100 mg recombinant tissue-type plasminogen activator (rt-PA) intravenously over 3 hours or an equivalent placebo infusion. In a subset of 312 patients, infarct size was assessed by the cumulative myocardial release of alpha-hydroxybutyrate dehydrogenase (HBDH) during the first 72 hours and by planar thallium scintigraphy (index of hypoperfusion) performed 10 to 22 days after the acute event. Left ventricular ejection fraction (LVEF) was determined by contrast and nuclear angiography. The median values of HBDH during the first 72 hours were 20% lower and the median values of thallium-201 28% smaller in the rt-PA group in comparison with controls. A significant but limited improvement of angiographic LVEF (2 absolute percentage points) was also shown in the patients treated with rt-PA. A moderate but statistically significant linear association between both measurements of infarct size and LVEF was found.

Internal mammary artery: methods of use and their effect on survival after coronary bypass surgery

Hazard analysis of total and cardiac mortality after isolated primary coronary artery surgery was performed using univariate and multivariate methods with special emphasis on the importance of the use and method of use of the internal mammary artery (IMA) as a bypass graft. The clinical data of 5880 consecutive patients were studied. The sum of the real follow-up periods studied was 27,948 years. The hazard of total and cardiac mortality could be defined in three-phase parametric models with an early, a constant and a late phase. The total survival was 82% +/- 1% at 10 years and 59% +/- 3.6% at 15 years. The construction of a single IMA distal graft (using left or right IMA) had a positive influence on the hazard (P = 0.0004) in the late phase after surgery with a high estimate (-1.6). The cardiac survival was 89% +/- 0.8% at 10 years and 74% +/- 3.5% at 15 years. The use of the left IMA had a positive influence (P = 0.001) in the late phase after surgery with a very high estimate (-2.3). The generated simulation of the total survival of a median patient with an IMA graft is 97% at 5 years and 94% at 10 years; for a median patient without an IMA graft, it is 97% at 5 years and 88% at 10 years. If a patient has other risk factors reducing his life expectancy, the influence can be dissipated because of lower survival rates at 5 years after surgery, when the effect of the IMA becomes most apparent.(ABSTRACT TRUNCATED AT 250 WORDS)

The Ionescu-Shiley pericardial valve: results in 473 patients

From January 1, 1980, through December 31, 1985, 473 patients underwent valve replacement with an Ionescu-Shiley valve. Overall hospital mortality was 7.8%. Major associated procedures and preoperative New York Heart Association (NYHA) Classes IV and V influenced hospital mortality significantly. The mean follow-up was 2.6 +/- 1.3 years. Late mortality was 5.9%. Overall actuarial survival was 81% at 5 years. A chief cause of reoperation was cusp rupture of a mitral prosthesis in 5 patients (all after aortic and mitral valve replacement). The overall actuarial reoperation-free incidence was 93% at 5 years. Thromboembolic (TB) phenomena occurred at a linear incidence of 1.4 +/- 0.3% per patient-year or an actuarial thromboembolism-free incidence of 92% at 5 years. Univariate and multivariate analyses showed that postoperative NYHA Class, rhythm at follow-up, and anticoagulant therapy significantly influenced the incidence of TE phenomena.

Interpreting age-related aspects of antihypertensive treatment: statistical defects and their remedy

Much of the literature concerning age-related effects in hypertension and its treatment suffers from serious deficiencies in statistical design and analysis. The problems include the measurement of response to treatment, the lack of a suitable control group, the way response is related to baseline blood pressure, the interpretation of post hoc analyses and the pooling of data from related but different studies. We discuss these problems and attempt to identify solutions which give a satisfactory balance between scientific validity and the practical issues of conducting clinical trials in hypertension.

Pulmonary function tests after different techniques for coronary artery bypass surgery. Saphenous vein versus single versus double internal mammary artery grafts

Pulmonary function tests were measured in 33 male patients undergoing elective coronary artery bypass surgery. Three modes of surgical technique were used: Bilateral internal mammary artery graft (BIMA), single internal mammary artery graft (SIMA) and saphenous vein grafts (VS). Following parameters were recorded: patient's age, length, body weight, preoperative forced vital capacity (FVC) and forced expiratory volume at one second (FEV 1), preoperative end-diastolic pressure and function of the left ventricle, smoking habitus, the fact that the pleural cavity was entered, duration of the cardiopulmonary bypass period, perioperative fluid balance and postoperative FVC and FEV 1 on the first eight postoperative days. In the BIMA group two pleural cavities, the SIMA group one pleural sac and the VS group none of the pleural cavities was entered. The BIMA group was younger (50.1 +/- 7.6 versus 57.7 +/- 7.28 and 60.1 +/- 6.9 years (p less than 0.05)) than the SIMA and VS group. Postoperative external blood loss was lower in the VS group compared to the SIMA and BIMA groups (839 +/- 255 ml versus 1346 +/- 654 ml and 1259 +/- 396 ml (p less than 0.05)). The FVC shows a dramatic decrease especially on the second postoperative day and was most markedly diminished in the BIMA and SIMA compared to VS (31% +/- 9% and 35% +/- 8% versus 45% +/- 10% of preoperative values (p less than 0.05)).(ABSTRACT TRUNCATED AT 250 WORDS)

UV sunlight and patch test reactions in humans

No seasonal influence of UV sunlight on patch test reactions in humans over a period of 9 years of clinical practice was found in this retrospective study of almost 8000 patients. Although the mean UV monthly dose varies seasonally, up in the summer and down in the winter, no significant differences could be identified for patch test reactions, either for the mean number of positive test reactions or for the intensity of the skin reaction or for the id-like spread reactions seen in summer or winter. Furthermore, no short-term influences of UV exposure during the weekend preceding patch testing could be demonstrated. Therefore, reliable patch-test results can be expected at any season of the year from a general population, at least in Belgian-type climates.

An uncertainty measure in logistic discrimination

The effect of sampling variation on individual decisions and error rates in logistic discriminant analysis is discussed. The concept of the beta-confidence allocation rule is introduced, which allows testing of whether observations are (in)correctly assigned at a given significance level. The procedure applied to sample data adds valuable information on the sharpness and the stability of the estimated classification rule. The method also suggests that individual posterior probabilities should be associated with a credibility measure. The potential of the approach is illustrated by an example from patients with liver disease.

The value of ejection fraction as a predictor of early and late survival following aortocoronary bypass surgery in patients with moderate to severe depression of the left ventricular function

Predictability analysis of early and late survival of coronary bypass patients with reduced left ventricular function was performed using stepwise logistic regression based on preoperative and operative data. The total perioperative mortality was 3.4% and the total actuarial survival of the whole group at 42 months was 87.7%. Thirty preoperative and operative parameters of 149 patients were entered into a first mathematical model. Four parameters were selected as a consequence: ejection fraction, anginal class, obesity, septal infarct. A prediction model based on these parameters, gave a correct prediction of the operative outcome in 99.3% of the patients. However due to the only five perioperative deaths, this model should be interpreted with caution. The accuracy of ejection fraction alone was 97.3%, the sensitivity was 100% but the specificity was only 20%. The same parameters of 97 patients were entered into a second mathematical model. Four parameters were selected in this case: ejection fraction, NYHA functional classification, septal infarct, anterior infarct. A second prediction model, based on these parameters, gave a correct prediction of the late outcome at 24 months in 93.8% of the patients. The accuracy of the discrimination on the basis of ejection fraction alone was 89.7%, its sensitivity was 96.5% but its specificity was only 41.7%. Ejection fraction was the best predictor out of the studied variables but due to its low specificity not good enough as a sole parameter.

Comparison of multigroup logistic and linear discriminant ECG and VCG classification

The performance of logistic (LOG) and linear discriminant analysis (LDA) has been studied, both for the conventional 12-lead electrocardiogram (ECG) and the orthogonal Frank 3-lead electrocardiogram (VCG), using a large validated data base. Classification rules were derived form a learning set (N = 2446) and applied to a test set (N = 820) to differentiate between normal, left, right and biventricular hypertrophy, anterior, inferior and combined myocardial infarction (MI). Total accuracy of LOG, assuming no normal distribution and using population proportions as prior probabilities, was up to 3% higher than that of LDA, depending on the number of variables used. The 12- and 3-lead LOG and LDA formulas resulted in very similar accuracy rates, i.e., between 67 and 70% for the seven-group and between 77 and 84% for the five-group analysis. LDA posterior probabilities were systematically more extreme than LOG ones. Correct classification of normals' specificity by LDA was 5 to 9% higher, but sensitivity for different groups was 1.5 to 10% lower than by LOG, with sample size proportions as priors. Specificity could be improved by changing the priors at the cost of lower sensitivity and vice versa, both for the LDA and LOG models. Classification results at 95% specificity were only slightly different, except for anterior MI where LOG scored 6% better. Other measures of performance demonstrated that the LDA model was overconfident and that the LOG model fitted better the real class membership of the patients. In conclusion, logistic ECG and VCG models improve the total accuracy of classification by about 1 to 3% when compared to LDA. More importantly, reliability of classification represents the improvement we want to emphasize. These methods may enhance the diagnostic utility of the ECG and VCG in routine practice.

Comparison of the classification ability of the electrocardiogram and vectorcardiogram

Controversy exists over the classification ability of the standard 12-lead electrocardiogram (EGG) and the vectorcardiogram (VCG). In this study the diagnostic information content and classification performance of the ECG and VCG were examined using multivariate statistical techniques and a large validated data base of 3,266 cases. Logistic classification models were developed to differentiate between 7 diagnostic entities: normal (n = 538), left (n = 557), right (n = 323) and biventricular (n = 437) hypertrophy, and anterior (n = 390), inferior (n = 657) and combined (n = 364) myocardial infarction. The models were obtained from a learning sample (n = 2,446) using an optimal set of computer derived ECG and VCG measurements. They were subsequently applied to a test sample (n = 820). In the learning sample, the discrimination models resulted in a total correct classification rate of 69.6% for the ECG and 69.4% for the VCG. The total accuracy rate was slightly lower in the test set: 66.3% for the ECG and 67.1% for the VCG. The combined use of the best ECG and VCG variables did not increase total diagnostic accuracy. When cases with biventricular hypertrophy and combined infarction were deleted, accuracy rates of more than 80% were achieved for both lead systems. Differences in the classification rates for the subgroups were not statistically significant. Thus, the conventional 12-lead ECG is as good as the VCG for the differential diagnosis of 7 main entities, provided identical procedures are used in the design of the classifiers.(ABSTRACT TRUNCATED AT 250 WORDS)

Transcortin and vitamin D-binding protein levels in mouse serum

The influence of age, sex and strain on the serum concentration of transcortin (corticosteroid-binding globulin) and vitamin D-binding protein (DBP) in mice was investigated. The effect of age was studied in two strains, C57BL/6JPfd and BALB/cmHeAPfd. The concentration of transcortin and DBP increased with age. In young animals the concentration of each protein showed a significant strain difference, which disappeared in older mice for DBP, but not for transcortin. In 7-day-old animals, no sex difference was observed for either protein, but in older animals a clear sex difference was found for transcortin. Adult males tended to have somewhat higher levels of DBP than adult females, but this difference was significant only on day 70. The variation in transcortin and DBP levels was further investigated in a large number of mouse strains. The DBP concentration did not markedly vary among strains (5.98-9.65 mumol/l in males and 5.08-8.85 mumol/l in females). Transcortin, however, showed marked strain variations, ranging from 0.72 to 2.06 mumol/l in males and from 1.02 to 4.55 mumol/l in females and there was a significant correlation (r = 0.66, n = 26, P less than 0.001) between the mean transcortin levels in males and females of different strains. Interstrain variation was much higher than intrastrain variation or variation among related strains, suggesting that the transcortin concentration is largely controlled by genetically determined factors. There was a significant correlation (r = 0.82, n = 9, P less than 0.01) between the mean corticosterone and transcortin concentrations (measured at 21.00 h).(ABSTRACT TRUNCATED AT 250 WORDS)

Multivariate survival analysis for the assessment of prognostic factors and risk categories after recovery from acute myocardial infarction: the Belgian situation

Twenty-one and a half per cent of the patients with acute myocardial infarction, admitted consecutively to the coronary care unit of the University Hospitals of Leuven in the period 1973-1979, died within 28 days. The 1,669 who recovered were followed between three and nine years. The mortality rate was 13.1% during the first year and fell to below 5% in the succeeding yearly intervals. Univariate and life table analysis were performed on noninvasive, clinical data collected during the coronary care unit stay for the total population and various subgroups. The Cox proportional hazards model was applied to the data in order to determine the prognostic factors for long-term survival. Of the 34 initially selected variables, the most dominant factors were age, Killip class III and IV, peripheral vascular disease, abnormal chest x-ray findings, previous infarction, and the use of digitalis or diuretics. The model, which was validated extensively, allowed the establishment of risk categories. Mortality five years after the acute event was 8.2 times more frequent in the highest risk quintile than in the lowest one. From this study one may conclude that multivariate techniques based on noninvasive variables collected during the coronary care unit stay are valuable for the determination of the long-term prognosis of patients with myocardial infarction.

Critical evaluation of quadratic logistic discriminant analysis methods: a case study

Electrocardiographic measurements from two groups, patients with left ventricular hypertrophy and clinically normal cases, were analyzed to illustrate the performance of the linear logistic discrimination method in the case of very unequal covariance matrices. Two quadratic extensions of the linear model were critically investigated. The use of graphical methods to check the model has been stressed and illustrated.

Serum lipids and apolipoproteins A-I, A-II and B in primary hypothyroidism before and during treatment

Serum lipids and apolipoproteins (apo) A-I, A-II, and B were measured in twenty-four patients with severe primary hypothyroidism (Thyrotropin above 40 mU/l), before and during 1-thyroxine treatment. Apo A-I, A-II, and B were assayed by immunonephelometry, using monospecific antisera. The serum levels of total cholesterol (TC), of low-density lipoprotein cholesterol (LDLc), and of the major LDL apoprotein, apo B, were markedly increased in the untreated hypothyroid patients compared to the values during therapy (TC: mean +/- SD, 8.87 +/- 2.9 v. 5.48 +/- 1.6 mmol/l; LDLc: 6.66 +/- 2.6 v. 3.78 +/- 1.4 mmol/l; apo B: 1.66 +/- 0.48 v. 1.14 +/- 0.37 g/l; P less than 0.00001 for all variables). High-density lipoprotein cholesterol (HDLc) was slightly higher before than during therapy (1.58 +/- 0.7 v. 1.31 +/- 0.4 mmol/l; P less than 0.05), while the main HDL apoprotein, apo A-I, was significantly elevated (1.49 +/- 0.42 v. 1.13 +/- 0.27 g/l; P less than 0.0002). The increase of the second major HDL apoprotein, apo A-II, was less pronounced (0.33 +/- 0.1 v. 0.30 +/- 0.08 g/l; P less than 0.022). The apo A-I to apo A-II ratio, which reflects the relative concentrations of the HDL subfractions HDL2 and HDL3, was significantly higher before than during treatment (P less than 0.0006). Serum triglyceride levels were moderately elevated in the untreated hypothyroid patients (1.34 +/- 0.6 v. 0.95 +/- 0.4 mmol/l; P less than 0.002). The small decrease in body weight during therapy did not correlate with the changes of the various lipid and apoprotein parameters.(ABSTRACT TRUNCATED AT 250 WORDS)

Response of transcortin and alpha 2u-globulin to turpentine-induced inflammation in the rat: influence of corticosteroids and prolactin

Evidence is presented that (transcortin and alpha 2u-globulin react as negative acute-phase proteins in the rat. Thirty-six hours after turpentine injection, the serum concentration of these proteins showed a two- to threefold decrease. Thereafter, transcortin rapidly returned to normal values, whereas alpha 2u-globulin remained low. This reaction pattern was still present after adrenalectomy, adrenalectomy and administration of glucocorticoids, and after treatment with bromocriptine, a suppressor of prolactin secretion. It is concluded that changes in the secretion of glucocorticoids and prolactin are not required for the observed turpentine-induced decrease of transcortin and alpha 2u-globulin.

Incidence, growth and oestradiol-receptor levels of 7, 12-dimethylbenz (alpha) anthracene-induced mammary tumours in rats: effects of neonatal sex steroids and oestradiol implants

The effects of neonatally administered steroids on the sensitivity of the mammary gland to tumour induction by 7, 12-dimethylbenz (alpha) anthracene was studied as a model for delayed (de) differentiating effects of steroid hormones. Immediately after birth male and female rats were gonadectomized and treated with testosterone, oestradiol or oil. Control animals were left intact. On day 45 all the gonadectomized animals and some of the control animals received an implant which delivered continuous low levels of oestradiol. The carcinogen was administered on day 55. The administration of an oestradiol implant, which increased prolactin levels in all animals, markedly reduced tumour incidence in intact female rats and increased tumour incidence in intact male rats. Neonatal administration of testosterone or oestradiol did not significantly influence tumour incidence, histopathology or oestradiol responsiveness in neonatally gonadectomized rats but tended to decrease tumour animals suggests that the effects observed by other authors in intact rats are mediated by changes in gonadal secretions. It is concluded that the hormonal environment during and after tumour induction plays a major role in the development of 7, 12-dimethylbenz (alpha) anthracene-induced mammary carcinomas.

Serum lipids and apolipoproteins A-I, A-II, and B in hyperthyroidism before and after treatment

The serum concentrations of total cholesterol (TC), triglycerides (RG), high density lipoprotein-cholesterol (HDLc), low density lipoprotein-cholesterol (LDLc), and the apolipoproteins (apo) A-I, A-II, and B were measured in 33 hyperthyroid patients before and after treatment. The results were compared with those of healthy controls. Apo A-I, A-II, and B were assayed by immunonephelometry. The serum levels of TC (mean +/- SD, 167 +/- 36 mg/dl, HDLc (40.8 +/- 12 mg/dl), and LDLc (108 +/- 35 mg/dl) were decreased in the untreated hyperthyroid patients compared to both the values after treatment (TC: 215 +/- 54 mg/dl; P less than 0.001; HDLc: 52 +/- 14 mg/dl; P less than 0.001; LDLc: 146 +/- 47 mg/dl; P less than 0.001) and the control values (TC: 206 + 39 mg/dl; P less than 0.001; HDLc: 47.4 +/- 10 mg/dl; P les than 0.01; LDLc: 145 +/- 38 mg/dl; P less than 0.001). TG levels were not statistically different before and after treatment. The apo A-I concentrations (116 +/- 24 mg/dl) were lower before than after treatment (131 +/- 28 mg/dl; P less than 0.01), but they were not statistically different from those in the control group (115 +/- 19 mg/dl). The apo A-II levels were identical in all groups (before treatment, 35 +/- 7 mg/dl; after treatment, 37 +/- 9 mg/dl; control group, 36 +/- 9 mg/dl). The apo B levels were lower in the untreated hyperthyroid patients (86 +/- 23 mg/dl) compared to those in controls (103 +/- 19 mg/dl; P less than 0.001) and patients after therapy (103 +/- 25 mg/dl; P less than 0.001). The increase in HDLc relative to the major HDL apo A-I and A-II during treatment for hyperthyroidism was associated with changes in body weight. The apo A-I to apo A-II and LDLc to apo B ratios, however, were significantly lower before compared to those after treatment, when the influence of increasing body weight during therapy was accounted for. This study emphasizes the important regulating role of thyroid hormones on lipid and apolipoprotein metabolism.