Biomedical subjects

S G Thompson

Publications and source records attributed to S G Thompson.

At least 19 recordsLinked to original sources

PubMed1998

Detecting and describing heterogeneity in meta-analysis.

The investigation of heterogeneity is a crucial part of any meta-analysis. While it has been stated that the test for heterogeneity has low power, this has not been well quantified. Moreover the assumptions of normality implicit in the standard methods of meta-analysis are often not scrutinized in practice. Here we simulate how the power of the test for heterogeneity depends on the number of studies included, the total information (that is total weight or inverse variance) available and the distribution of weights among the different studies. We show that the power increases with the total information available rather than simply the number of studies, and that it is substantially lowered if, as is quite common in practice, one study comprises a large proportion of the total information. We also describe normal plots that are useful in assessing whether the data conform to a fixed effect or random effects model, together with appropriate tests, and give an application to the analysis of a multi-centre trial of blood pressure reduction. We conclude that the test of heterogeneity should not be the sole determinant of model choice in meta-analysis, and inspection of relevant normal plots, as well as clinical insight, may be more relevant to both the investigation and modelling of heterogeneity.

Adult↗

PubMed1997

Investigating underlying risk as a source of heterogeneity in meta-analysis.

In a meta-analysis of clinical trials, an important issue is whether the treatment benefit varies according to the underlying risk of the patients in the different trials. The usual naive analyses employed to investigate this question use either the observed risk of events in the control groups, or the average risk in the control and treatment groups, as a measure of underlying risk. These analyses are flawed and can produce seriously misleading results. We show how their biases depend on three components of variability, the within-trial and between-trial variances of the control group risks, and the between-trial variance of the treatment effects. We propose a Bayesian solution to the problem which can be carried out using the BUGS implementation of Gibbs sampling. The analysis is illustrated for a meta-analysis of bleeding and mortality data in trials of sclerotherapy for patients with cirrhosis, and the results contrasted with those from the naive approaches. Comparisons with other methods recently proposed for this problem are also made. We conclude that the Bayesian solution presented in this paper is not only more appropriate than other proposed methods, but is also sufficiently easy to implement that it can be used by applied researchers undertaking meta-analyses.

Bayes Theorem↗

PubMed1997

Value of natriuretic peptides in assessment of patients with possible new heart failure in primary care.

BACKGROUND: The reliability of a clinical diagnosis of heart failure in primary care is poor. Concentrations of natriuretic peptides are high in heart failure. This population-based study examined the predictive value of natriuretic peptides in patients with a new primary-care diagnosis of heart failure. METHODS: Concentrations of plasma atrial (ANP and N-terminal ANP) and B-type (BNP) natriuretic peptides were measured by radioimmunoassay in 122 consecutive patients referred to a rapid-access heart-failure clinic with a new primary-care diagnosis of heart failure. On the basis of clinical assessment, chest radiography, and transthoracic echocardiography, a panel of three cardiologists decided that 35 (29%) patients met the case definition for new heart failure. ANP and NT-ANP results were available for 117 patients (34 with heart failure) and BNP results for 106 (29 with heart failure). FINDINGS: Geometric mean concentrations of natriuretic peptides were much higher in patients with heart failure than in those with other diagnoses (29.2 vs 12.4 pmol/L for ANP; 63.9 vs 13.9 pmol/L for BNP; 1187 vs 410.6 pmol/L for NT-ANP; all p < 0.001). At cut-off values chosen to give negative predictive values for heart failure of 98% (ANP > or = 18.1 pmol/L, NT-ANP > or = 537.6 pmol/L, BNP > or = 22.2 pmol/L), the sensitivity, specificity, and positive predictive value for ANP were 97%, 72%, and 55%; for NT-ANP 97%, 66%, and 54%; and for BNP 97%, 84%, and 70%. Addition of ANP or NT-ANP concentration or both did not improve the predictive power of a logistic regression model containing BNP concentration alone. INTERPRETATION: In patients with symptoms suspected by a general practitioner to be due to heart failure, plasma BNP concentration seems to be a useful indicator of which patients are likely to have heart failure and require further clinical assessment.

Adult↗

PubMed1997

The design and analysis of paired cluster randomized trials: an application of meta-analysis techniques.

The statistical issues in clinical trials where clusters, communities or groups rather than individuals are randomized are often not fully appreciated. In this paper we discuss the design and analysis of trials in which pairs of clusters are randomized in the context of one recent trial, the British Family Heart Study. Both sample size calculations and the analysis strategy need to take account of the between-cluster component of variance. The analysis can be considered as a random effects meta-analysis across cluster pairs, and can usefully be presented as such. Techniques developed in the context of meta-analysis can then be used in the analysis, for example using a profile likelihood method to derive a confidence interval for the overall treatment effect which takes into account the variability in the estimate of the between-cluster variance. The methods presented here are contrasted with previously published methods for cluster randomized trials.

Adult↗

PubMed1997

Production of C-reactive protein and risk of coronary events in stable and unstable angina. European Concerted Action on Thrombosis and Disabilities Angina Pectoris Study Group.

BACKGROUND: Inflammation is an important feature of atherosclerotic lesions, and increased production of the acute-phase reactant. C-reactive protein (CRF), is associated with a poor prognosis in severe unstable angina. We have investigated the existence and possible significance of the acute-phase responses of CRP and another sensitive reactant, serum amyloid A protein (SAA), in patients with unstable or stable angina. METHODS: We used new ultrasensitive immunoassays to measure CRP and SAA concentrations in plasma from 2121 outpatients with angina (1030 unstable, 743 stable, the rest atypical) enrolled in the European Concerted Action on Thrombosis and Disabilities (ECAT) Angina Pectoris Study. All patients underwent coronary angiography and extensive clinical and laboratory assessment at study entry, and were then followed up for 2 years. All suspected coronary events during follow-up were reviewed by an independent endpoint committee. FINDINGS: 75 individuals (41 with unstable, 29 with stable, and 5 with atypical angina) had a coronary event during follow-up. Concentrations of CRP at study entry were associated with coronary events in patients with stable or unstable angina: there was about a two-fold increase in the risk of a coronary event in patients whose CRP concentration was in the fifth quintile (> 3.6 mg/L), compared with the first four quintiles. A third of the events occurred among patients who had a CRP concentration of more than 3.6 mg/L. CRP concentrations were positively correlated with age, smoking, body-mass index, triglycerides, extent of coronary stenosis, history of myocardial infarction, and lower ejection fraction. By contrast, concentrations of SAA were not associated with risk of a coronary event. INTERPRETATION: We found that raised circulating concentrations of CRP are predictors of coronary events in patients with stable or unstable angina. The modest acute-phase responses of CRP were probably not the result of myocardial necrosis. Whatever the underlying mechanisms, the sensitive measurement of CRP as a prognostic marker may be useful in the management of coronary heart disease.

Angina Pectoris↗

PubMed1996

Multi-level models for repeated measurement data: application to quality of life data in clinical trials.

Quality of life data present considerable statistical challenges because of their longitudinal and multidimensional nature, and also because the available data are often very unbalanced through missing values. Here we exemplify the potential of multi-level models, that is, hierarchical random coefficient models, for such data. The discussion is developed in the context of analysing the quality of life data from a trial of palliative treatment in non-small-cell lung cancer. Not only do multi-level models provide a flexible modelling framework for the investigation of the underlying behaviour of response, for example, giving simple estimates of treatment effects, but they also permit a description of the differences between subjects and allow the analysis of multi-dimensional outcomes. The assumptions of Normality, homogeneity, and independence of the within- and between-subject variance components can be investigated and the models can be extended to provide explicit modelling of variance heterogeneity. It is concluded that multi-level models, for which software is now available, provide a natural and powerful approach to the analysis of longitudinal data in general, and multi-dimensional quality of life data in particular.

Carcinoma, Non-Small-Cell Lung↗

PubMed1996

Fibrinolytic factors and the risk of myocardial infarction or sudden death in patients with angina pectoris. ECAT Study Group. European Concerted Action on Thrombosis and Disabilities.

BACKGROUND: Disturbances of the fibrinolytic system that lead to decreased removal of fibrin deposits may be important risk factors for coronary thrombosis. There is as yet no consensus on the prognostic value of fibrinolytic parameters, which may be attributed in part to the choice of confounding variables controlled for. METHODS AND RESULTS: The ECAT study is a prospective multicenter study of 3043 patients with angina pectoris followed for 2 years. Baseline measurements included 10 fibrinolytic variables. The results were analyzed in relation to the subsequent incidence of myocardial infarction or sudden coronary death. They are presented before and after adjustment for clusters of confounding variables that are markers of different mechanisms: insulin resistance (body mass index, triglyceride, and HDL cholesterol), inflammation (fibrinogen and C-reactive protein), and endothelial cell damage (von Willebrand factor). An increased incidence of events was associated with higher baseline concentrations of tissue plasminogen activator (TPA) antigen (P = .0002), plasminogen activator inhibitor-1 (PAI-1) activity (P = .02), and PAI-1 antigen (P = .001). The associations of PAI-1 activity and PAI-1 antigen with risk of events disappeared after adjustment for parameters reflecting insulin resistance but were not affected by other adjustments. TPA antigen was affected to a similar extent by adjustment for parameters reflecting insulin resistance. Inflammation, or endothelial cell damage, but the risk association disappeared only after combined adjustments. CONCLUSIONS: The prognostic role of PAI-1 in predicting coronary events is related principally to insulin resistance, whereas that of TPA antigen could be explained only by its relationship with different mechanisms, including insulin resistance, inflammation and endothelial cell damage.

Adult↗

PubMed1996

The relation between treatment benefit and underlying risk in meta-analysis.

In meta-analyses of clinical trials comparing a treated group with a control group it has been common to ask whether the treatment benefit varies according to the underlying risk of the patients in the different trials, with the hope of defining which patients would benefit most and which least from medical interventions. The usual analysis used to investigate this issue, however, which uses the observed proportions of events in the control groups of the trials as a measure of the underlying risk, is flawed and produces seriously misleading results. This arises through a bias due to regression to the mean and will be particularly acute in meta-analyses which include some small trials or in which the variability in the true underlying risks across trials is small. Approaches which previously have been thought to be more appropriate are to substitute the average proportion of events in the control and treated groups as the measure of underlying risk or to plot the proportion of events in the treated group against that in the control group (L'Abbé plot). However, these are still subject to bias in most circumstances. Because of the potentially seriously flawed conclusions that can result from such analyses, they should be replaced either by statistically appropriate (but more complex) approaches or, preferably, by analyses which investigate the dependence of the treatment effect on measured baseline characteristics of the patients in each trial.

Clinical Trials as Topic↗

PubMed1996

Methods for summarizing the risk associations of quantitative variables in epidemiologic studies in a consistent form.

A major problem in reviewing the published results of different epidemiologic studies of the relation between a quantitative variable and the risk of disease is that the results are presented in many different ways. The purpose of this paper is to exemplify methods by which results expressed either as risks (or rates) according to quantlle groups of the quantitative variable or as results derived from a logistic regression analysis can be reexpressed in a uniform manner, as a mean difference in the quantitative variable between the cases of disease and the other subjects in the study. An important assumption of the methods is that the quantitative variable has an approximately normal distribution, and a way of investigating the appropriateness of this assumption is given. The methods can be applied to both prospective and case-control studies and are exemplified by a number of studies of serum albumin concentrations and mortality. In some applications, these methods can be used as a precursor to formal meta-analysis, for example, when differential control of potential confounding factors is not a problem. At the least, the methods can be useful either in quantitatively reviewing published studies before undertaking new research or in putting the results of a new study into the context of previously published ones.

Case-Control Studies↗

PubMed1996

A likelihood approach to meta-analysis with random effects.

In a meta-analysis of a set of clinical trials, a crucial but problematic component is providing an estimate and confidence interval for the overall treatment effect theta. Since in the presence of heterogeneity a fixed effect approach yields an artificially narrow confidence interval for theta, the random effects method of DerSimonian and Laird, which incorporates a moment estimator of the between-trial components of variance sigma B2, has been advocated. With the additional distributional assumptions of normality, a confidence interval for theta may be obtained. However, this method does not provide a confidence interval for sigma B2, nor a confidence interval for theta which takes account of the fact that sigma B2 has to be estimated from the data. We show how a likelihood based method can be used to overcome these problems, and use profile likelihoods to construct likelihood based confidence intervals. This approach yields an appropriately widened confidence interval compared with the standard random effects method. Examples of application to a published meta-analysis and a multicentre clinical trial are discussed. It is concluded that likelihood based methods are preferred to the standard method in undertaking random effects meta-analysis when the value of sigma B2 has an important effect on the overall estimated treatment effect.

Antihypertensive Agents↗

PubMed1996

Antithrombin III and fibrinogen as predictors of cardiac events in patients with angina pectoris.

Because measurements of hemostatic factors might aid the prediction of cardiovascular clinical events, we investigated the long-term prognostic importance of selected hemostatic factors in patients with angina pectoris. At recruitment, 209 patients underwent clinical assessment and coronary angiography, and a range of hemostatic factors were measured. During the follow-up period of 9 years, 58 patients (28%) suffered a cardiac event (acute myocardial infarction or death from cardiac causes). The risk of cardiac events was positively related to baseline measurements of fibrinogen (risk ratio per SD [RR] increase 1.29, 95% confidence interval [CI] 0.99 to 1.68, P=.06) and negatively related to antithrombin III activity measurements (RR 0.75, 95% CI 0.59 to 0.95, P=.02). No other hemostatic factor measured was significantly related to the risk of having a cardiac event. Worsening of angina in the few weeks before and ejection fraction evaluation at the initial angiography were both strongly related to the risk of cardiac events. However, the relationships of fibrinogen and antithrombin III measurements to risk remained almost unchanged after adjusting for worsening of angina and ejection fraction. Fibrinogen and antithrombin III may have an important etiologic role in the prognosis of patients with angina pectoris.

Adult↗

PubMed1996

Using a coronary risk score for screening and intervention in general practice. British Family Heart Study.

AIM: To investigate whether a risk score proposed by the British Regional Heart Study (BRHS), derived from data collected in 1978-1980, provides an appropriate basis for current coronary risk factor screening and intervention in general practice. METHODS: The BRHS risk score was applied to 1993 men aged 40-59 years and 1353 women aged 35-59 years, from 13 general practices in England, Wales and Scotland, who had health checks during 1991-1992 in the British Family Heart Study (FHS). Modifications to the BRHS risk score were made in order to identify subjects with a current high risk compared with others of the same age and sex. These were validated on 3272 men and 2229 women recruited from different general practices during 1992-1994 in the FHS. RESULTS: Only 9% of men in the FHS fell into the published top (highest risk) quintile of the BRHS score, versus an expected 20%, and 44% fell into the bottom quintile. Scores were, on average, substantially lower in the FHS men than in the BRHS men, principally because of lower measured cholesterol levels (using a Reflotron) and a lower prevalence of cigarette-smoking. The BRHS scores also tended to increase with age, disproportionately identifying older subjects, and were substantially lower in women than in men. Simple age-related modifications to the risk score were therefore devised to overcome these problems. These modifications performed well in the validation. CONCLUSIONS: The substantial difference in risk scores between the BRHS and FHS men may reflect both a real reduction in risk and changes in calibration and methodology. Current use of the BRHS risk score may therefore mislead doctors and patients in the direction of complacency. In addition, the published BRHS risk score has an age-dependence that is undesirable in terms of guiding the intensity of lifestyle intervention which should be offered to an individual patient. The simple modifications proposed provide a more appropriate basis for coronary risk factor screening and intervention in general practice, and one that can be used both for men and for women.

Adult↗

PubMed1995

Hemostatic factors and the risk of myocardial infarction or sudden death in patients with angina pectoris. European Concerted Action on Thrombosis and Disabilities Angina Pectoris Study Group.

BACKGROUND: Increased levels of certain hemostatic factors may play a part in the development of acute coronary syndromes and may be associated with an increased risk of coronary events in patients with angina pectoris. METHODS: We conducted a prospective multicenter study of 3043 patients with angina pectoris who underwent coronary angiography and were followed for two years. Base-line measurements included the concentrations of selected hemostatic factors indicative of a thrombophilic state or endothelial injury. The results were analyzed in relation to the subsequent incidence of myocardial infarction or sudden coronary death. RESULTS: After adjustment for the extent of coronary artery disease and other risk factors, an increased incidence of myocardial infarction or sudden death was associated with higher base-line concentrations of fibrinogen (mean +/- SD, 3.28 +/- 0.74 g per liter in patients who subsequently had coronary events, as compared with 3.00 +/- 0.71 g per liter in those who did not; P = 0.01), von Willebrand factor antigen (138 +/- 49 percent vs. 125 +/- 49 percent, P = 0.05), and tissue plasminogen activator (t-PA) antigen (11.9 +/- 4.7 ng per milliliter vs. 10.0 +/- 4.2 ng per milliliter, P = 0.02). The concentration of C-reactive protein was also directly correlated with the incidence of coronary events (P = 0.05), except when we adjusted for the fibrinogen concentration. In patients with high serum cholesterol levels, the risk of coronary events rose with increasing levels of fibrinogen and C-reactive protein, but the risk remained low even given high serum cholesterol levels in the presence of low fibrinogen concentrations. CONCLUSIONS: In patients with angina pectoris, the levels of fibrinogen, von Willebrand factor antigen, and t-PA antigen are independent predictors of subsequent acute coronary syndromes. In addition, low fibrinogen concentrations characterize patients at low risk for coronary events despite increased serum cholesterol levels. Our data are consistent with a pathogenetic role of impaired fibrinolysis, endothelial-cell injury, and inflammatory activity in the progression of coronary artery disease.

Adult↗

PubMed1995

Dietary change after smoking cessation: a prospective study.

A population sample of 375 men and women cigarette smokers were recruited to take part in a prospective study of smoking cessation to test the hypothesis that stopping smoking is associated with an increased consumption of the essential fatty acid linoleic acid, which explains the concomitant reduction in risk of coronary heart disease. Diet was assessed using a 10 d weighed record in 301 smokers at baseline, 153 at 4-month follow-up, of whom twenty-six had quit smoking, and 122 at 1-year follow-up, of whom twenty had quit. Compared with continuing smokers, those who had quit at the 4-month follow-up (mean 10 and 13 weeks for men and women respectively) had statistically significant increases in body weight (5%), energy intake (13%), total dietary fat (24%), all specific types of dietary fat (26% polyunsaturated fat, 26% linoleic acid, 30% eicosapentaenoic acid, 23% monounsaturated fat and 22% saturated fat) and vitamin E intake (19%). The foods which appeared to contribute to increases in energy and fat intakes at the 4-month follow-up were vegetable oils and polyunsaturated margarines, processed meats and meat pies. By follow-up at 1 year (mean time since quitting 31 and 41 weeks for men and women respectively) there were no detectable differences in energy and total fat intakes. However, intakes of eicosapentaenoic acid and pteroylglutamate (folate) were statistically significantly higher in the quitters compared with the continuing smokers (37% for eicosapentaenoic acid and 16% for folate). We conclude that the short-term increase in dietary intake of linoleic acid, which is not sustained by 1 year, cannot explain the reduction in risk of coronary disease following smoking cessation.

Adult↗

PubMed1995

Dose-response relationships of serum lipid measurements with the extent of coronary stenosis. Strong, independent, and comprehensive. ECAT Angina Pectoris Study Group.

Serum lipids, lipoproteins, and more recently apolipoproteins and lipoprotein(a) [Lp(a)] have been shown to be independent risk factors for coronary vessel disease and its prognosis. However, the relationships between serum lipid levels and the extent of coronary artery disease (CAD) have not been consistently shown. Twenty-five hundred male and female patients with suspected angina pectoris were recruited from 18 European medical centers. The independent relations of total cholesterol, triglycerides, HDL cholesterol, LDL cholesterol, apo A-I and B, and Lp(a) with the presence and extent of CAD, as assessed by coronary angiography, were investigated. All of the lipid measures showed strong relations P < .0001) with the presence of CAD, defined by the existence of at least one > or = 50% coronary vessel stenosis. Total cholesterol, LDL cholesterol, apo B, triglycerides, and Lp(a) were substantially higher and HDL cholesterol and apo A-I lower in patients with CAD. The odds ratio of CAD, in the high-risk tertile of each lipid's distribution compared with the low-risk tertile, was in the range 1.5 to 2.3. Each of total cholesterol (or LDL cholesterol or apo B), HDL cholesterol (or apo A), and Lp(a) had an independent effect in predicting the presence of CAD. In addition, all lipids showed a strong association (P = .0006 for triglycerides, P < .0001 otherwise) with the extent of CAD as defined by the number of stenosed coronary vessels.(ABSTRACT TRUNCATED AT 250 WORDS)

Arteriosclerosis↗

PubMed1995

Haemostasis factors in angina pectoris; relation to gender, age and acute-phase reaction. Results of the ECAT Angina Pectoris Study Group.

The ECAT Angina Pectoris Study is a European multicentre study with the aim of investigating the pathogenetic and predictive role of haemostatic factors in the progression of coronary heart disease. It is the largest study performed up to now with regard to both the number of patients with angina pectoris (n = 3043) and the number of haemostasis assays (n = 23) included. The present paper presents baseline cross-sectional data with particular reference to the relationship of haemostatic factors with each other and with the coronary risk factors age, gender and acute-phase reaction (1). Two clusters of haemostatic factors could be distinguished in which each variable was correlated (P < 0.001) to every other variable: (a) Eight fibrinolysis assays including t-PA, PAI-1 and euglobulin clot lysis time (ECLT), for which PAI-1 appeared to be the dominating factor; (b) antithrombin III, protein C, alpha 2-antiplasmin and plasminogen, the interdependence of which has no obvious explanation. (2). Twelve out of the 23 haemostasis assays were associated (P < or = 0.01) with age. Except for alpha 2-antiplasmin, these relationships indicated an increased tendency to thrombosis with increasing age. (3). Gender differences found in 14 haemostasis parameters do not indicate a consistent difference in the tendency to thrombosis between men and women. Eight haemostasis parameters were on average higher in female than in male patients in the age group over 50 years. (4). C-reactive protein, an acute-phase reactant, was positively correlated (P < 0.001) with fibrinogen, factor VIIIc, von Willebrand factor, the fibrinolysis assays t-PA, PAI-1, ECLT and plasminogen.(ABSTRACT TRUNCATED AT 250 WORDS)

Acute-Phase Reaction↗

PubMed1995

Comparing non-nested regression models.

A method for comparing the fits of two non-nested models, based on a suggestion of Davidson and MacKinnon (1981), is developed in the context of linear and nonlinear regression with normal errors. Each model is regarded as a special case of an artificial "supermodel" and is obtained by restricting the value of a mixing parameter gamma to 0 or 1. To enable estimation and hypothesis testing for gamma, an approximate supermodel is used in which the fitted values from the individual models appear in place of the original parametrization. In the case of nested linear models, the proposed test essentially reproduces the standard F test. The calculations required are for the most part straight-forward (basically, linear regression through the origin). The test is extended to cover situations in which serious bias in the maximum likelihood estimate of gamma occurs, simple approximate bounds for the bias being given. Two real datasets are used illustratively throughout.

Biometry↗

PubMed1995

Readiness for lifestyle advice: self-assessments of coronary risk prior to screening in the British family heart study. Family Heart Study Group.

BACKGROUND: Where health professionals and patients hold similar views of a problem, health outcomes may be better. AIM: The aims of this paper were to document how attenders at primary care cardiovascular screening clinics perceived their risks of coronary heart disease prior to screening; the degree of similarity between perceived level of risk and an epidemiologically derived risk score; and the relative importance assigned to individual risk factors by subjects compared with those assigned by the risk score. METHOD: These issues were investigated in 3725 middle aged men and women who accepted an invitation to attend health screening as part of the British family heart study. RESULTS: Overall, there was a tendency for subjects to be optimistic (37%) rather than pessimistic (21%) when judging their risk of coronary heart disease. Nevertheless, there were strong significant associations between perceived risk and the levels of individual risk factors, particularly personal and family medical history and body mass index. There was also a strong association with the overall risk score though a large minority (31%) held views of their risk of coronary heart disease that were quite different from those based upon the epidemiologically derived index of risk. Respondents accorded greater importance to smoking and parental death from coronary heart disease and less importance to cholesterol level and blood pressure than did the risk score. CONCLUSION: Possible explanations for the observed disagreement are over-optimism or the relative importance given to individual risk factors. The relationships between patients' perceptions of risk and the epidemiological indices likely to be espoused by health professionals are important in understanding the difficulties in communication that might arise in offering lifestyle advice after screening for cardiovascular risk.

Adult↗