Search PubMed⌕ Search

Biomedical subjects

S J Pocock

Publications and source records attributed to S J Pocock.

At least 55 records · Page 3Linked to original sources
PubMed1993

On the design and analysis of randomized clinical trials with multiple endpoints.

This paper considers some methods for reducing the number of significance tests undertaken when analyzing and reporting results of clinical trials. Emphasis is placed on designing and analyzing clinical trials to examine a composite hypothesis concerning multiple endpoints and combining this multiple endpoint methodology with group sequential methodology. Four methods for composite hypotheses are considered: an ordinary least squares and a generalized least squares approach both due to O'Brien (1984, Biometrics 40, 1079-1087), a new modification of these, and an approximate likelihood ratio test, due to Tang, Gnecco, and Geller (1989, Biometrika 76, 577-583). These are extended for group sequential use. In particular, simulation is used to generate critical values and sequences of nominal significance levels for the approximate likelihood ratio test, which is not normally distributed. An example is given and the relative merits of the suggested approaches are discussed.

Asthma↗
PubMed1992

Repeated measures in clinical trials: analysis using mean summary statistics and its implications for design.

This paper explores the use of simple summary statistics for analysing repeated measurements in randomized clinical trials with two treatments. Quite often the data for each patient may be effectively summarized by a pre-treatment mean and a post-treatment mean. Analysis of covariance is the method of choice and its superiority over analysis of post-treatment means or analysis of mean changes is quantified, as regards both reduced variance and avoidance of bias, using a simple model for the covariance structure between time points. Quantitative consideration is also given to practical issues in the design of repeated measures studies: the merits of having more than one pre-treatment measurement are demonstrated, and methods for determining sample sizes in repeated measures designs are provided. Several examples from clinical trials are presented, and broad practical recommendations are made. The examples support the value of the compound symmetry assumption as a realistic simplification in quantitative planning of repeated measures trials. The analysis using summary statistics makes no such assumption. However, allowance in design for alternative non-equal correlation structures can and should be made when necessary.

Analysis of Variance↗
PubMed1992

Prediction of short-term survival with an application in primary biliary cirrhosis.

Many long-term follow-up studies for survival accumulate repeated measurements of prognostic factors. Survival models which include only covariate values at baseline do not use all available information, and do not relate to survival predictions for times other than at that baseline. Time-dependent covariate models (which update covariate values as measurements occur through time) might be used, though limitations of software for estimating the underlying hazard functions and difficulty in relating hazard function changes to survival prediction present serious drawbacks. By dividing each patient's follow-up into successive intervals of equal length (using a length of interest for prediction) and with measurements available at the start of each, we describe how an analysis taking person-intervals as the observation units can be undertaken using readily available software to produce short-term survival models. We show that this approach is related to both the baseline and time-dependent covariate models. The method is illustrated using data from a long-term study of patients with primary biliary cirrhosis, where interest is in short-term survival predictions to aid the decision when to undertake liver transplantation.

Humans↗
PubMed1992

Within-subject diastolic blood pressure variability: implications for risk assessment and screening.

Because of variability in diastolic blood pressure within an individual, repeated measurements increase precision in assessing an individual's underlying mean pressure and so also aid risk classification. Data from a cohort of 11,299 middle-aged men is used to model the variability in diastolic pressure between annual measurements. A simple model with pressure normally distributed about an underlying mean with standard deviation increasing with level fits the data very well. In modelling risk of cardiovascular mortality, a strong association is found with observed diastolic pressure level but not to trends in or variability between observed values. The effect of regression dilution is clear with the risk relationship appearing greater as one uses the mean of an increasing number of measurements. A method of adjusting for this regression dilution is described so giving an estimate of the relationship with underlying mean diastolic pressure. Using this survival model and the model for blood pressure variability, a method is presented for estimating both underlying mean pressure and absolute risk of cardiovascular disease given a sequence of blood pressure measurements from screening. This allows a sequential strategy for determining whether (a) antihypertensive intervention is desirable, (b) no further screening is necessary, or (c) further screening would aid the assessment, and emphasizes the need to consider blood pressure in the context of multiple risk factors.

Adult↗
PubMed1992

Effects of 22 months of treatment with inhaled corticosteroids and/or beta-2-agonists on lung function, airway responsiveness, and symptoms in children with asthma. The Dutch Chronic Non-specific Lung Disease Study Group.

In a randomized double-blind multicenter clinical study, 116 children with asthma were randomly assigned to treatment with an inhaled beta-2-agonist (salbutamol 0.2 mg) plus an inhaled corticosteroid (budesonide 0.2 mg) three times a day (BA+CS) or to an inhaled beta-2-agonist (salbutamol 0.2 mg) plus a placebo three times a day (BA+PL). After a median follow-up time of 22 months, 26 patients receiving BA+PL (45%) had withdrawn from randomized treatment, mainly because of asthma symptoms, compared with three withdrawals in the patients receiving BA+CS (p less than 0.0001). The FEV1, expressed as a percentage of the predicted value for age, sex, and height, showed an absolute increase of 7.0% after 2 months of BA+CS compared with a decrease of 4.0% after 2 months of BA+PL. This 11% difference in percent predicted FEV1 (95% confidence interval, 7 to 15%; p less than 0.0001) was then maintained after a median follow-up period of 22 months. Postbronchodilator FEV1 showed an absolute increase of 3.7% predicted within 2 months in patients receiving BA+CS and an absolute decrease of 1.1% predicted in children receiving BA+PL (p = 0.0005). Thereafter, this difference between the two treatment groups was maintained. Average peak expiratory flow rate (PEFR) increased from baseline by 36.6 L/min in the BA+CS group compared with 3.7 L/min in the BA+PL group (p = 0.003). This difference then remained for the median follow-up time of 22 months.(ABSTRACT TRUNCATED AT 250 WORDS)

Administration, Inhalation↗
PubMed1992

The impact of stopping rules on heterogeneity of results in overviews of clinical trials.

This paper explores the extent to which application of statistical stopping rules in clinical trials can create an artificial heterogeneity of treatment effects in overviews (meta-analyses) of related trials. For illustration, we concentrate on overviews of identically designed group sequential trials, using either fixed nominal or O'Brien and Fleming two-sided boundaries. Some analytic results are obtained for two-group designs and simulation studies are otherwise used, with the following overall findings. The use of stopping rules leads to biased estimates of treatment effect so that the assessment of heterogeneity of results in an overview of trials, some of which have used stopping rules, is confounded by this bias. If the true treatment effect being studied is small, as is often the case, then artificial heterogeneity is introduced, thus increasing the Type I error rate in the test of homogeneity. This could lead to erroneous use of a random effects model, producing exaggerated estimates and confidence intervals. However, if the true mean effect is large, then between-trial heterogeneity may be underestimated. When undertaking or interpreting overviews, one should ascertain whether stopping rules have been used (either formally or informally) and should consider whether their use might account for any heterogeneity found.

Analysis of Variance↗
PubMed1991

Can meta-analyses be trusted?

The enthusiasm for meta-analyses (or overviews) expressed by their proponents is not always shared by the broader medical community. To encourage constructive debate, we adopt a critical perspective on the conduct and interpretation of meta-analysis. We focus particularly on some of the statistical issues, especially heterogeneity between studies, and also on the extrapolation of meta-analysis findings to clinical practice. We conclude that meta-analysis is not an exact statistical science that provides definitive simple answers to complex clinical problems. It is more appropriately viewed as a valuable objective descriptive technique, which often furnishes clear qualitative conclusions about broad treatment policies, but whose quantitative results have to be interpreted cautiously.

Bias↗
PubMed1991

Does plasma cholesterol concentration predict mortality from coronary heart disease in elderly people? 18 year follow up in Whitehall study.

OBJECTIVE: To explore the extent to which the relation between plasma cholesterol concentration and risk of death from coronary heart disease in men persists into old age. DESIGN: 18 year follow up of male Whitehall civil servants. Plasma cholesterol concentrations and other risk factors were determined at first examination in 1967-9 when they were aged 40-69. Death of men up to 31 January 1987 was recorded. SUBJECTS: 18,296 male civil servants, 4155 of whom died during follow up. MAIN OUTCOME MEASURES: Cause and age of death. Cholesterol concentration in 1967-9 and number of years elapsed between testing and death. RESULTS: 1676 men died of coronary heart disease. The mean cholesterol concentration in these men was 0.32 mmol/l higher than that in all other men (95% confidence interval 0.26 to 0.37 mmol/l). This difference in cholesterol concentrations fell 0.15 mmol/l with every 10 years' increase in age at screening. The risk of raised cholesterol concentration fell with age at death. Compared with other men cholesterol concentration in those who died of coronary heart disease was 0.44 mmol/l higher in those who died aged less than 60 and 0.26 mmol/l higher in those aged 60-79 (p = 0.03). For a given age at death the longer the gap between cholesterol measurement and death the more predictive the cholesterol concentration, both for coronary heart disease and all cause mortality (trend test p = 0.06 and 0.03 respectively). CONCLUSION: Reducing plasma cholesterol concentrations in middle age may influence the risk of death from coronary heart disease in old age.

Adult↗
PubMed1991

Risk factors for stroke in middle aged British men.

OBJECTIVE: To determine the risk factors for stroke in a cohort representative of middle aged British men. DESIGN: Prospective study of a cohort of men followed up for eight years. SETTING: General practices in 24 towns in England, Wales, and Scotland (the British regional heart study). SUBJECTS: 7735 men aged 40-59 at screening, selected at random from one general practice in each town. MAIN OUTCOME MEASURE: Fatal and non-fatal strokes. RESULTS: 110 of the men had at least one stroke; there were four times as many non-fatal as fatal strokes. The relative risk of stroke was 12.1 in men who had high blood pressure (systolic blood pressure greater than or equal to 160 mm Hg) and were current smokers compared with normotensive, non-smoking men. Diastolic blood pressure yielded no additional information, and former cigarette smokers had the same risk as men who had never smoked. Heavy alcohol intake was associated with a relative risk of stroke of 3.8 in men without previously diagnosed cardiovascular disease. Men with pre-existing ischaemic heart disease had an increased risk of stroke, but only when left ventricular hypertrophy on electrocardiography was also present. CONCLUSIONS: Systolic blood pressure, cigarette smoking, and left ventricular hypertrophy on electrocardiography in men with pre-existing ischaemic heart disease were found to be the major risk factors for stroke in middle aged British men. Heavy alcohol intake seemed to increase the risk of stroke in men without previously diagnosed cardiovascular disease. A large proportion of strokes should be preventable by controlling blood pressure and stopping smoking.

Alcohol Drinking↗
PubMed1991

A perspective on the role of quality-of-life assessment in clinical trials.

This article summarizes some of the methodologic and practical issues relevant to health related quality-of-life (HRQL) assessment in randomized clinical trials. Particular attention is given to 1) the reasons for undertaking HRQL assessment, 2) the types of clinical trials particularly suitable for HRQL assessment, 3) the selection of appropriate HRQL measures, 4) statistical issues in the design and analysis of HRQL studies, and 5) the need to coordinate progress in HRQL studies in clinical research.

Forecasting↗
PubMed1990

Estimation issues in clinical trials and overviews.

There is a general move towards greater emphasis on point and interval estimates of treatment effect in reporting of clinical trials, so that significance testing plays a lesser role. In this article we examine a number of issues which affect the use and interpretation of conventional estimation methods. Should we accept or avoid the stereotypes of 95 per cent confidence? Should the abstract of a trial report include confidence intervals for major endpoints? Are frequentist confidence intervals being interpreted correctly, and should Bayesian probability intervals be more widely used in trial reports? Does the timing of publication, such as early stopping because of a large observed treatment difference, lead to exaggerated point and interval estimates? How can we produce realistic estimates from subgroup analyses? Is publication bias seriously affecting our ability to obtain unbiased estimates? Is the emphasis on estimation methods a powerful tool for encouraging larger sample sizes? Can we resolve the controversy concerning fixed or random effects models for estimation in overviews of related trials? Our arguments are illustrated by results from recent trials in cardiovascular disease.

Bayes Theorem↗
PubMed1990

Prognostic scores for detecting a high risk group: estimating the sensitivity when applied to new data.

The sensitivity of a prognostic scoring system will tend to be exaggerated if the scoring system is both derived and validated on the same data. This paper provides, by analogy to regression with error in an explanatory variable, an intuitive basis for the methodological results of Copas which seek to estimate the degree of such exaggeration. There was good agreement between Copas' results and those achieved in a series of cross-validation exercises where logistic regression models predicting the risk of ischaemic heart disease were derived using data from the prospective British Regional Heart Study. When truly important variables were included, the exaggeration of the sensitivity increased as the number of cases of disease available decreased. It is concluded that Copas' method, which is easy to implement in practice, may be helpful in realistically anticipating the extent of such exaggeration, and that it can be usefully employed before pursuing a scoring system on newly collected data.

Adult↗
PubMed1990

The variability of serum cholesterol measurements: implications for screening and monitoring.

The reliability of screening for high serum total cholesterol is adversely affected by the variability of cholesterol levels over time. This problem is investigated using data on repeated cholesterol measurements for 14,600 men and women in the MRC Mild Hypertension Trial. For measurements 1 year apart, the within-person coefficient of variation (CV) is 7%, which is substantial compared with the between-person CV of 15%. In a screening programme, this within-person variability may lead to the misclassification of individuals and inappropriate intervention. For example, 28% of middle-aged British men with a single cholesterol measurement above 6.9 mmol/l have a long-term average cholesterol below that value even without intervention. Using averages of several cholesterol measurements reduces, but does not eliminate, these problems. Furthermore, monitoring the effect of interventions in individuals by sequential cholesterol measurement may be unhelpful or even misleading. These problems cast serious doubt on the value of general population screening for high cholesterol levels.

Adult↗