PubMed Health⌕ Search

Biomedical subjects

Simon G Thompson

Publications and source records attributed to Simon G Thompson.

At least 19 recordsLinked to original sources
PubMed2007

Multilevel models for estimating incremental net benefits in multinational studies.

Multilevel models (MLMs) have been recommended for estimating incremental net benefits (INBs) in multicentre cost-effectiveness analysis (CEA). However, these models have assumed that the INBs are exchangeable and that there is a common variance across all centres. This paper examines the plausibility of these assumptions by comparing various MLMs for estimating the mean INB in a multinational CEA. The results showed that the MLMs that assumed the INBs were exchangeable and had a common variance led to incorrect inferences. The MLMs that included covariates to allow for systematic differences across the centres, and estimated different variances in each centre, made more plausible assumptions, fitted the data better and led to more appropriate inferences. We conclude that the validity of assumptions underlying MLMs used in CEA need to be critically evaluated before reliable conclusions can be drawn.

Cost-Benefit Analysis↗
PubMed2006

Regression dilution methods for meta-analysis: assessing long-term variability in plasma fibrinogen among 27,247 adults in 15 prospective studies.

BACKGROUND: Within-person variability in measured values of a risk factor can bias its association with disease. The extent of this regression dilution bias for plasma fibrinogen was investigated using repeat measurement data collected at varying time intervals on 27 247 adults in 15 prospective studies. METHODS: Regression dilution ratios (RDRs) were estimated from a linear regression of repeat measurements on baseline values in each study and for each time interval, and pooled allowing for within- and between-study heterogeneity. RDRs were estimated both without and with adjustment for confounders, and factors were investigated that might influence the RDRs. RESULTS: The unadjusted overall RDR was 0.51 (95% CI: 0.47, 0.55), which decreased to 0.46 (95% CI: 0.42, 0.49) after adjustment for age, sex and measured values of other established vascular risk factors. The RDR did not vary materially by assay method, age, sex or smoking status, but decreased at higher levels of baseline fibrinogen. CONCLUSION: It is appropriate to use an RDR of 0.5 to correct approximately for regression dilution bias in plasma fibrinogen values; however, this correction factor may produce somewhat conservative hazard ratios in adjusted analyses, at higher fibrinogen concentrations and in follow-up beyond a decade. More generally, the methods described in this report have widespread applicability to quantifying regression dilution bias in repeatability data from multiple prospective studies.

Adult↗
PubMed2006

Constructing intervals for the intracluster correlation coefficient using Bayesian modelling, and application in cluster randomized trials.

Studies in health research are commonly carried out in clustered settings, where the individual response data are correlated within clusters. Estimation and modelling of the extent of between-cluster variation contributes to understanding of the current study and to design of future studies. It is common to express between-cluster variation as an intracluster correlation coefficient (ICC), since this measure is directly comparable across outcomes. ICCs are generally reported unaccompanied by confidence intervals. In this paper, we describe a Bayesian modelling approach to interval estimation of the ICC. The flexibility of this framework allows useful extensions which are not easily available in existing methods, for example assumptions other than Normality for continuous outcome data, adjustment for individual-level covariates and simultaneous interval estimation of several ICCs. There is also the opportunity to incorporate prior beliefs on likely values of the ICC. The methods are exemplified using data from a cluster randomized trial.

Bayes Theorem↗
PubMed2006

Addressing the issues that arise in analysing multicentre cost data, with application to a multinational study.

Differences in the mean, spread and skewness of cost data collected from different countries present problems for analysis and interpretation. Here we develop generalised linear multilevel models to estimate the effects of patient and national characteristics on costs. Using gamma distributions and multiplicative effects for patient characteristics fitted the data better than models which assumed normal distributions or estimated additive effects. A multilevel gamma model is employed to allow for heterogeneity in the effects of patient case-mix across centres. Analysis of multinational cost data must recognise differences in mean, spread and skewness across centres, as well as the data's hierarchical structure.

Costs and Cost Analysis↗
PubMed2006

The importance of varying the event generation process in simulation studies of statistical methods for recurrent events.

Statistical methods for the analysis of recurrent events are often evaluated in simulation studies. A factor rarely varied in such studies is the underlying event generation process. If the relative performance of statistical methods differs across generation processes, then studies based upon one process may mislead. This paper describes the simulation of recurrent events data using four models of the generation process: Poisson, mixed Poisson, autoregressive, and Weibull. For each model four commonly used statistical methods for the analysis of recurrent events (Cox's proportional hazards method, the Andersen-Gill method, negative binomial regression, the Prentice-Williams-Peterson method) were applied to 200 simulated data sets, and the mean estimates, standard errors, and confidence intervals obtained. All methods performed well for the Poisson process. Otherwise, negative binomial regression only performed well for the mixed Poisson process, as did the Andersen-Gill method with a robust estimate of the standard error. The Prentice-Williams-Peterson method performed well only for the autoregressive and Weibull processes. So the relative performance of statistical methods depended upon the model of event generation used to simulate data. In conclusion, it is important that simulation studies of statistical methods for recurrent events include simulated data sets based upon a range of models for event generation.

Computer Simulation↗
PubMed2006

Modelling multivariate outcomes in hierarchical data, with application to cluster randomised trials.

In the cluster randomised study design, the data collected have a hierarchical structure and often include multivariate outcomes. We present a flexible modelling strategy that permits several normally distributed outcomes to be analysed simultaneously, in which intervention effects as well as individual-level and cluster-level between-outcome correlations are estimated. This is implemented in a Bayesian framework which has several advantages over a classical approach, for example in providing credible intervals for functions of model parameters and in allowing informative priors for the intracluster correlation coefficients. In order to declare such informative prior distributions, and fit models in which the between-outcome covariance matrices are constrained, priors on parameters within the covariance matrices are required. Careful specification is necessary however, in order to maintain non-negative definiteness and symmetry between the different outcomes. We propose a novel solution in the case of three multivariate outcomes, and present a modified existing approach and novel alternative for four or more outcomes. The methods are applied to an example of a cluster randomised trial in the prevention of coronary heart disease. The modelling strategy presented would also be useful in other situations involving hierarchical multivariate outcomes.

Artificial Intelligence↗
PubMed2006

Meta-analysis of studies of alcohol and breast cancer with consideration of the methodological issues.

OBJECTIVE: To give an up-to-date assessment of the association of alcohol with female breast cancer, addressing methodological issues and shortfalls in previous overviews. METHODS: Meta-analysis of studies (any language) providing original data on incidence of first primary breast cancer and alcohol. Two reviewers independently extracted data. Study quality assessed by objective criteria including degree of control for confounding; funnel plots examined for publication bias; meta-regression techniques to explore heterogeneity. Risks associated with drinking versus not drinking and dose-response not constrained through the origin estimated using random effects methods. RESULTS: Ninety-eight unique studies were included, involving 75,728 and 60,653 cases in drinker versus non-drinker and dose-response analyses, respectively. Findings were robust to study design and analytic approaches in the meta-analyses. For studies judged high quality, controlled for appropriate confounders, excess risk associated with alcohol drinking was 22% (95% CI: 9-37%); each additional 10 g ethanol/day was associated with risk higher by 10% (95% CI: 5-15%). There was no evidence of publication bias. Risk did not differ significantly by beverage type or menopausal status. Estimated population attributable risks were 1.6 and 6.0% in USA and UK, respectively. CONCLUSIONS: Taking account of shortcomings in the study base and methodological concerns, we confirm the alcohol-breast cancer association. We compared our results to those of an individual patient data analysis, with similar findings. We conclude that the association between alcohol and breast cancer may be causal.

Alcohol Drinking↗
PubMed2005

Plasma fibrinogen level and the risk of major cardiovascular diseases and nonvascular mortality: an individual participant meta-analysis.

CONTEXT: Plasma fibrinogen levels may be associated with the risk of coronary heart disease (CHD) and stroke. OBJECTIVE: To assess the relationships of fibrinogen levels with risk of major vascular and with risk of nonvascular outcomes based on individual participant data. DATA SOURCES: Relevant studies were identified by computer-assisted searches, hand searches of reference lists, and personal communication with relevant investigators. STUDY SELECTION: All identified prospective studies were included with information available on baseline fibrinogen levels and details of subsequent major vascular morbidity and/or cause-specific mortality during at least 1 year of follow-up. Studies were excluded if they recruited participants on the basis of having had a previous history of cardiovascular disease; participants with known preexisting CHD or stroke were excluded. DATA EXTRACTION: Individual records were provided on each of 154,211 participants in 31 prospective studies. During 1.38 million person-years of follow-up, there were 6944 first nonfatal myocardial infarctions or stroke events and 13,210 deaths. Cause-specific mortality was generally available. Analyses involved proportional hazards modeling with adjustment for confounding by known cardiovascular risk factors and for regression dilution bias. DATA SYNTHESIS: Within each age group considered (40-59, 60-69, and > or =70 years), there was an approximately log-linear association with usual fibrinogen level for the risk of any CHD, any stroke, other vascular (eg, non-CHD, nonstroke) mortality, and nonvascular mortality. There was no evidence of a threshold within the range of usual fibrinogen level studied at any age. The age- and sex- adjusted hazard ratio per 1-g/L increase in usual fibrinogen level for CHD was 2.42 (95% confidence interval [CI], 2.24-2.60); stroke, 2.06 (95% CI, 1.83-2.33); other vascular mortality, 2.76 (95% CI, 2.28-3.35); and nonvascular mortality, 2.03 (95% CI, 1.90-2.18). The hazard ratios for CHD and stroke were reduced to about 1.8 after further adjustment for measured values of several established vascular risk factors. In a subset of 7011 participants with available C-reactive protein values, the findings for CHD were essentially unchanged following additional adjustment for C-reactive protein. The associations of fibrinogen level with CHD or stroke did not differ substantially according to sex, smoking, blood pressure, blood lipid levels, or several features of study design. CONCLUSIONS: In this large individual participant meta-analysis, moderately strong associations were found between usual plasma fibrinogen level and the risks of CHD, stroke, other vascular mortality, and nonvascular mortality in a wide range of circumstances in healthy middle-aged adults. Assessment of any causal relevance of elevated fibrinogen levels to disease requires additional research.

Adult↗
PubMed2005

Intensive case management for severe psychotic illness: is there a general benefit for patients with complex needs? A secondary analysis of the UK700 trial data.

The UK700 trial failed to demonstrate an overall benefit of intensive case management (ICM) in patients with severe psychotic illness. This does not discount a benefit for particular subgroups, and evidence of a benefit of ICM for patients of borderline intelligence has been presented. The aim of this study is to investigate whether this effect is part of a general benefit for patients with severe psychosis complicated by additional needs. In the UK700 trial patients with severe psychosis were randomly allocated to ICM or standard case management. For each patient group with complex needs the effect of ICM is compared with that in the rest of the study cohort. Outcome measures are days spent in psychiatric hospital and the admission and discharge rates. ICM may be of benefit to patients with severe psychosis complicated by borderline intelligence or depression, but may cause patients using illicit drugs to spend more time in hospital. There was no convincing evidence of an effect of ICM in a further seven patient groups. ICM is not of general benefit to patients with severe psychosis complicated by additional needs. The benefit of ICM for patients with borderline intelligence is an isolated effect which should be interpreted cautiously until further data are available.

Critical Care↗
PubMed2005

Adjusting for partially missing baseline measurements in randomized trials.

Adjustment for baseline variables in a randomized trial can increase power to detect a treatment effect. However, when baseline data are partly missing, analysis of complete cases is inefficient. We consider various possible improvements in the case of normally distributed baseline and outcome variables. Joint modelling of baseline and outcome is the most efficient method. Mean imputation is an excellent alternative, subject to three conditions. Firstly, if baseline and outcome are correlated more than about 0.6 then weighting should be used to allow for the greater information from complete cases. Secondly, imputation should be carried out in a deterministic way, using other baseline variables if possible, but not using randomized arm or outcome. Thirdly, if baselines are not missing completely at random, then a dummy variable for missingness should be included as a covariate (the missing indicator method). The methods are illustrated in a randomized trial in community psychiatry.

Data Interpretation, Statistical↗
PubMed2005

Randomized trial to determine the effect of nebivolol on mortality and cardiovascular hospital admission in elderly patients with heart failure (SENIORS).

AIMS: Large randomized trials have shown that beta-blockers reduce mortality and hospital admissions in patients with heart failure. The effects of beta-blockers in elderly patients with a broad range of left ventricular ejection fraction are uncertain. The SENIORS study was performed to assess effects of the beta-blocker, nebivolol, in patients >/=70 years, regardless of ejection fraction. METHODS AND RESULTS: We randomly assigned 2128 patients aged >/=70 years with a history of heart failure (hospital admission for heart failure within the previous year or known ejection fraction </=35%), 1067 to nebivolol (titrated from 1.25 mg once daily to 10 mg once daily), and 1061 to placebo. The primary outcome was a composite of all cause mortality or cardiovascular hospital admission (time to first event). Analysis was by intention to treat. Mean duration of follow-up was 21 months. Mean age was 76 years (SD 4.7), 37% were female, mean ejection fraction was 36% (with 35% having ejection fraction >35%), and 68% had a prior history of coronary heart disease. The mean maintenance dose of nebivolol was 7.7 mg and of placebo 8.5 mg. The primary outcome occurred in 332 patients (31.1%) on nebivolol compared with 375 (35.3%) on placebo [hazard ratio (HR) 0.86, 95% CI 0.74-0.99; P=0.039]. There was no significant influence of age, gender, or ejection fraction on the effect of nebivolol on the primary outcome. Death (all causes) occurred in 169 (15.8%) on nebivolol and 192 (18.1%) on placebo (HR 0.88, 95% CI 0.71-1.08; P=0.21). CONCLUSION: Nebivolol, a beta-blocker with vasodilating properties, is an effective and well-tolerated treatment for heart failure in the elderly.

Adrenergic beta-Antagonists↗
PubMed2005

Methods for incorporating covariate adjustment, subgroup analysis and between-centre differences into cost-effectiveness evaluations.

BACKGROUND: Overall assessments of cost-effectiveness are now commonplace in informing medical policy decision making. It is often important, however, also to investigate how cost-effectiveness varies between patient subgroups. Yet such analyses are rarely undertaken, because appropriate methods have not been sufficiently developed. METHODS: We propose a coherent set of Bayesian methods to extend cost-effectiveness analyses to adjust for baseline covariates, to investigate differences between subgroups, and to allow for differences between centres in a multicentre study using a hierarchical model. These methods consider costs and effects jointly, and allow for the typically skewed distribution of cost data. The results are presented as inferences on the cost-effectiveness plane, and as cost-effectiveness acceptability curves. RESULTS: In applying these methods to a randomised trial of case management of psychotic patients, we show that overall cost-effectiveness can be affected by ignoring the skewness of cost data, but that it may be difficult to gain substantial precision by adjusting for baseline covariates. While analyses of overall cost-effectiveness can mask important subgroup differences, crude differences between centres may provide an unrealistic indication of the true differences between them. CONCLUSIONS: The methods developed allow a flexible choice for the distributions used for cost data, and have a wide range of applicability--to both randomised trials and observational studies. Experience needs to be gained in applying these methods in practice, and using their results in decision making.

Bayes Theorem↗
PubMed2005

Using multilevel models for assessing the variability of multinational resource use and cost data.

Multinational economic evaluations often calculate a single measure of cost-effectiveness using cost data pooled across several countries. To assess the validity of pooling international cost data the reasons for cost variation across countries need to be assessed. Previously, ordinary least-squares (OLS) regression models have been used to identify factors associated with variability in resource use and total costs. However, multilevel models (MLMs), which accommodate the hierarchical structure of the data, may be more appropriate. This paper compares these different techniques using a multinational dataset comprising case-mix, resource use and cost data on 1300 stroke admissions from 13 centres in 11 European countries. OLS and MLMs were used to estimate the effect of patient and centre-level covariates on the total length of hospital stay (LOS) and total cost. MLMs with normal and gamma distributions for the data within centres were compared. The results from the OLS model showed that both patient and centre-level covariates were associated with LOS and total cost. The estimates from the MLMs showed that none of the centre-level characteristics were associated with LOS, and the level of spending on health was the centre-level variable most highly associated with total cost. We conclude that using OLS models for assessing international variation can lead to incorrect inferences, and that MLMs are more appropriate for assessing why resource use and costs vary across centres.

Age Factors↗
PubMed2005

Analyzing the duration of recurrent events in clinical trials: a comparison of approaches using data from the UK700 trial of psychiatric case management.

In studies of chronic disease the outcome measure may be based upon the duration of recurring illness. Our example is the UK700 trial of psychiatric case management, where the total number of days spent in hospital over a 2-year follow-up was the primary outcome. Investigations of treatment effect modifiers were undertaken using an analysis of that primary outcome, a comparison of length of hospitalizations, and also a multi-state modeling approach. The days in hospital outcome was relatively straightforward to analyze, and allowed the complete randomized treatment groups to be compared. In contrast, the comparison of length of hospitalizations included only hospitalized patients, with censored observations not being well accommodated. The multi-state model provided separate treatment effect estimates for admission and discharge, this being more informative about how any reduction in days spent in hospital is achieved. Estimation of the treatment effects through the use of proportional hazards regression allowed appropriate incorporation of censored observations. However, with the multi-state model approach treatment effect estimates are not based upon comparisons of complete randomized treatment groups, as individuals are removed from the risk set for admission whilst at risk for discharge, and vice versa. We conclude that total duration is an appropriate primary outcome for clinical trials, but that multi-state models deserve greater use as an informative secondary analysis.

Case Management↗
PubMed2005

Prior distributions for the intracluster correlation coefficient, based on multiple previous estimates, and their application in cluster randomized trials.

Numerous estimates for the intracluster correlation coefficient (ICC) are available in research databases and publications. When planning a cluster randomized trial, an anticipated value for the ICC is required; currently, researchers base their choice informally on the magnitude of previous ICC estimates. In this paper, we make use of the wealth of ICC information by formally constructing informative prior distributions, while acknowledging the varying relevance and precision of the estimates available. Typically, for a planned trial in a given clinical setting, multiple relevant ICC estimates are available from each of several completed studies. Our preferred model allows for the imprecision in each ICC estimate around its underlying true value and, separately, allows for the similarity of ICC values from the same study. The relevance of each previous estimate to the planned clinical setting is considered, and estimates corresponding to less relevant outcomes or population types are given less influence. We find that such downweighting can increase the precision of the anticipated ICC. In trial design, the prior distribution constructed allows uncertainty about the ICC to be acknowledged, and we describe how to choose a design that provides adequate power across the range of likely ICC values. Prior information on the ICC can also be incorporated in analysis of the trial data, when taking a Bayesian approach. The methods proposed enable available ICC information to be summarised appropriately by an informative prior distribution, which is of direct practical use in cluster randomized trials.

Bayes Theorem↗
PubMed2005

The use of random effects models to allow for clustering in individually randomized trials.

BACKGROUND: We describe different forms of clustering that may occur in individually randomized trials, where the observed outcomes for different individuals cannot be regarded as independent. We propose random effects models to allow for such clustering, across a range of contexts and trial designs, and investigate their effect on estimation and interpretation of the treatment effect. METHODS: We apply our proposed models to two individually randomized trials with potential for clustering, a trial of teleconsultation in hospital referral (the main outcome being offer of a further hospital appointment) and a trial of exercise therapy delivered by physiotherapists for low back pain (the outcome being a back pain score). Extensions to the methods include the possibility of explaining heterogeneity between clusters using cluster level characteristics and the potential dilution of cluster effects due to noncompliance. RESULTS: In the teleconsultation trial, the odds ratio was significant (1.52, 95% CI 1.27 to 1.82) when clustering was ignored, but smaller and nonsignificant (1.36, 95% CI 0.85 to 2.13) when clustering by hospital consultant was taken into account. The 95% range of estimated treatment effects across consultants was from 0.21 to 8.76. This variability was only partially explained by the specialty of the consultant. In the back pain trial, although there was an overall benefit of exercise (change of - 0.51 points on the back pain score) and little evidence of clustering, the estimated treatment effects for different physiotherapists ranged from -1.26 to +0.26 points. CONCLUSIONS: Clustering is an important issue in many individually randomized trials. Ignoring it can lead to underestimates of the uncertainty and too extreme P-values. Even when there is little apparent heterogeneity across clusters, it can still have a large impact on the estimation and interpretation of the treatment effect.

Back Pain↗
PubMed2005

Meta-analysis of individual patient data from randomized trials: a review of methods used in practice.

BACKGROUND: Meta-analyses based on individual patient data (IPD) are regarded as the gold standard for systematic reviews. However, the methods used for analysing and presenting results from IPD meta-analyses have received little discussion. METHODS: We review 44 IPD meta-analyses published during the years 1999-2001. We summarize whether they obtained all the data they sought, what types of approaches were used in the analysis, including assumptions of common or random effects, and how they examined the effects of covariates. RESULTS: Twenty-four out of 44 analyses focused on time-to-event outcomes, and most analyses (28) estimated treatment effects within each trial and then combined the results assuming a common treatment effect across trials. Three analyses failed to stratify by trial, analysing the data is if they came from a single mega-trial. Only nine analyses used random effects methods. Covariate-treatment interactions were generally investigated by subgrouping patients. Seven of the meta-analyses included data from less than 80% of the randomized patients sought, but did not address the resulting potential biases. CONCLUSIONS: Although IPD meta-analyses have many advantages in assessing the effects of health care, there are several aspects that could be further developed to make fuller use of the potential of these time-consuming projects. In particular, IPD could be used to more fully investigate the influence of covariates on heterogeneity of treatment effects, both within and between trials. The impact of heterogeneity, or use of random effects, are seldom discussed. There is thus considerable scope for enhancing the methods of analysis and presentation of IPD meta-analysis.

Data Interpretation, Statistical↗