David J. Spiegelhalter

WinBUGS – A Bayesian modelling framework: Concepts, structure, and extensibility

WinBUGS is a fully extensible modular framework for constructing and analysing Bayesian full probability models. Models may be specified either textually via the BUGS language or pictorially using a graphical interface called DoodleBUGS. WinBUGS processes the model specification and constructs an object-oriented representation of the model. The software offers a user-interface, based on dialogue boxes and menu commands, through which the model may then be analysed using Markov chain Monte Carlo techniques. In this paper we discuss how and why various modern computing concepts, such as object-orientation and run-time linking, feature in the softwares design. We also discuss how the framework may be extended. It is possible to write specific applications that form an apparently seamless interface with WinBUGS for users with specialized requirements. It is also possible to interface with WinBUGS at a lower level by incorporating new object types that may be used by WinBUGS without knowledge of the modules in which they are implemented. Neither of these types of extension require access to, or even recompilation of, the WinBUGS source-code.

Non-invasive detection of endothelial dysfunction in children and adults at risk of atherosclerosis

Endothelial dysfunction is an early event in experimental studies of atherogenesis, preceding formation of plaques. We have devised a non-invasive method for testing endothelial function, to find out whether abnormalities are present in symptom-free children and young adults at high risk of atherosclerosis. With high-resolution ultrasound, we measured the diameter of the superficial femoral and brachial arteries at rest, during reactive hyperaemia (with increased flow causing endothelium-dependent dilatation), and after sublingual glyceryl trinitrate (GTN; causing endothelium-independent dilatation) in 100 subjects--50 controls without vascular risk factors (aged 8-57 years), 20 cigarette smokers (aged 17-62 years), 10 children with familial hypercholesterolaemia (FH; aged 8-16 years), and 20 patients with established coronary artery disease (CAD). Adequate scans were obtained in all but 6 cases. Flow-mediated dilatation was observed in arteries from all control subjects. Dilatation was inversely related to baseline vessel diameter (r = -0.81, p < 0.0001); in arteries of 6.0 mm or less, mean dilatation was 10 (SE 2)%. In smokers, FH children, and adults with CAD, flow-mediated dilatation was much reduced or absent (p < 0.001 for comparison with each relevant control group). Dilatation in response to GTN was present in all groups. Endothelial dysfunction is present in children and adults with risk factors for atherosclerosis, such as smoking and hypercholesterolaemia, before anatomical evidence of plaque formation in the arteries studied. This may be an important early event in atherogenesis.

The BUGS project: Evolution, critique and future directions

BUGS is a software package for Bayesian inference using Gibbs sampling. The software has been instrumental in raising awareness of Bayesian modelling among both academic and commercial communities internationally, and has enjoyed considerable success over its 20-year life span. Despite this, the software has a number of shortcomings and a principal aim of this paper is to provide a balanced critical appraisal, in particular highlighting how various ideas have led to unprecedented flexibility while at the same time producing negative side effects. We also present a historical overview of the BUGS project and some future perspectives.

Aging is associated with endothelial dysfunction in healthy men years before the age-related decline in women

OBJECTIVES This study assessed whether aging is associated with progressive endothelial dysfunction, whether the pattern of any age-related decline in vascular health is different in men and women and whether any gender difference is consistent with known changes in hormonal status. BACKGROUND Coronary and cerebrovascular disease are much less common in young and middle-aged women compared with men, although the gender difference in death from atherosclerosis is less marked after the menopause. Endothelial dysfunction is an early event in atherogenesis and is important in dynamic plaque stenosis in later life. The effect of aging on endothelial function in men and women, however, is not well known. METHODS We used high resolution ultrasound to study endothelium-dependent and endothelium-independent vascular responses. Brachial artery physiology was investigated in 238 subjects (103 men, 135 women; mean [+/- SD] age 38 +/- 17 years, range 15 to 72) with no known risk factors for atherosclerosis. The responses to reactive hyperemia (flow-mediated dilation, which is endothelium dependent) and to glyceryl trinitrate (an endothelium-independent dilator) were assessed for all the subjects and then for men and women separately. RESULTS On multivariate analysis for the whole group, reduced flow-mediated dilation was related to older age (r = -0.34, p < 0.0001). In men, flow-mediated dilation was preserved in subjects aged < or = 40 years but declined thereafter at 0.21%/year. In women, flow-mediated dilation was stable until the early 50s, after which it declined at 0.49%/year (p = 0.002 compared with men). In contrast, there was no significant change in the glyceryl trinitrate response with aging in either gender. CONCLUSIONS Aging is associated with progressive endothelial dysfunction in normal humans, and this appears to occur earlier in men than in women. In women, however, a steep decline commences at around the time of the menopause. This is consistent with a protective effect of estrogens on the arterial wall.

A re-evaluation of random-effects meta-analysis

Meta-analysis in the presence of unexplained heterogeneity is frequently undertaken by using a random-effects model, in which the effects underlying different studies are assumed to be drawn from a normal distribution. Here we discuss the justification and interpretation of such models, by addressing in turn the aims of estimation, prediction and hypothesis testing. A particular issue that we consider is the distinction between inference on the mean of the random-effects distribution and inference on the whole distribution. We suggest that random-effects meta-analyses as currently conducted often fail to provide the key results, and we investigate the extent to which distribution-free, classical and Bayesian approaches can provide satisfactory methods. We conclude that the Bayesian approach has the advantage of naturally allowing for full uncertainty, especially for prediction. However, it is not without problems, including computational intensity and sensitivity to a priori judgements. We propose a simple prediction interval for classical meta-analysis and offer extensions to standard practice of Bayesian meta-analysis, making use of an example of studies of ‘set shifting’ ability in people with eating disorders.

Quality of life measures in health care. I: Applications and issues in assessment

Many clinicians remain unsure of the relevance of measuring quality of life to their clinical practice. In health economics quality of life measures have become the standard means of assessing the results of health care interventions and, more controversially, the means of prioritising funding; but they have many other applications. This article--the first of three on measuring quality of life--reviews the instruments available and their application in screening programmes, audit, health care research, and clinical trials. Using the appropriate instrument is essential if outcome measures are to be valid and clinically meaningful.

Sequential updating of conditional probabilities on directed graphical structures

A directed acyclic graph or influence diagram is frequently used as a representation for qualitative knowledge in some domains in which expert system techniques have been applied, and conditional probability tables on appropriate sets of variables form the quantitative part of the accumulated experience. It is shown how one can introduce imprecision into such probabilities as a data base of cases accumulates. By exploiting the graphical structure, the updating can be performed locally, either approximately or exactly, and the setup makes it possible to take advantage of a range of well-established statistical techniques. As examples we discuss discrete models, models based on Dirichlet distributions and models of the logistic regression type.

Bayesian approaches to randomized trials

Statistical issues in conducting randomized trials include the choice of a sample size, whether to stop a trial early and the appropriate analysis and interpretation of the trial results. At each of these stages, evidence external to the trial is useful, but generally such evidence is introduced in an unstructured and informal manner. We argue that a Bayesian approach allows a formal basis for using external evidence and in addition provides a rational way for dealing with issues such as the ethics of randomization, trials to show treatment equivalence, the monitoring of accumulating data and the prediction of the consequences of continuing a study

Use and misuse of process and outcome data in managing performance of acute medical care: avoiding institutional stigma

The history of monitoring the outcomes of health care by external agencies can be traced to ancient times. However, the danger, now as then, is that in the search for improvement, comparative measures of mortality and morbidity are often overinterpreted, resulting in judgments about the underlying quality of care. Such judgments can translate into performance management strategies in the form of capricious sanctions (such as star ratings) and unjustified rewards (such as special freedoms or financial allocations). The resulting risk of stigmatising an entire institution injects huge tensions into health-care organisations and can divert attention from genuine improvement towards superficial improvement or even gaming behaviour (ie, manipulating the system). These dangers apply particularly to measures of outcome and throughput. We argue that comparative outcome data (league tables) should not be used by external agents to make judgments about quality of hospital care. Although they might provide a reasonable measure of quality in some high-risk surgical situations, they have little validity in acute medical settings. Their use to support a system of reward and punishment is unfair and, unsurprisingly, often resisted by clinicians and managers. We argue further that although outcome data are useful for research and monitoring trends within an organisation, those who wish to improve care for patients and not penalise doctors and managers, should concentrate on direct measurement of adherence to clinical and managerial standards.

Pregnancy in cyanotic congenital heart disease. Outcome of mother and fetus.

In a series of 416 women with congenital heart disease seen in the Royal Brompton National Heart and Lung Hospital, London, and the Hospital Giovanni Bosco, Torino, Italy, there were 822 pregnancies. The outcomes of 96 pregnancies in 44 patients with cyanotic congenital heart disease were studied. Patients with the Eisenmenger reaction were excluded. Patients were divided arbitrarily into groups according to the type of maternal congenital cardiac anomaly, and factors influencing maternal and fetal outcome were evaluated. The incidence of maternal cardiovascular complications was high (32%), with one death from endocarditis 2 months after delivery. Forty-one (43%) of 96 pregnancies resulted in a live birth; 15 (37%) were premature. Mean weight of full-term infants was 2575 g. Univariate analysis suggested that maternal disease, Ability Index, hemoglobin, and arterial oxygen saturation before the pregnancy were factors that discriminated between successful and unsuccessful fetal outcome, with hemoglobin and arterial oxygen saturation being the most important predictors. Women with cyanotic congenital heart disease can go through pregnancy with a low risk to themselves, with frequent treatable complications, but there is a high incidence of miscarriage, premature births, and low birth weights. An incidence of congenital heart disease in the fetus of 4.9% (2 of 41 live births) is higher than that found in the normal population.