A comparison of various estimators of a treatment difference for a multi-centre clinical trial

When a clinical trial is conducted at more than one centre it is likely that the true treatment effect will not be identical at each centre. In other words there will be some degree of treatment-by-centre interaction. A number of alternative approaches for dealing with this have been suggested in the literature. These include frequentist approaches with a fixed or random effects model for the observed data and Bayesian approaches. In the fixed effects model, there are two common competing estimators of the treatment difference, based on weighted or unweighted estimates from individual centres. Which one of these should be used is the subject of some controversy and we do not intend to take a particular methodological position in this paper. Our intention is to provide some insight into the relative merits of the indicated range of possible estimators of the treatment effect. For the fixed effects model, we also look at the merits of using a preliminary test for interaction assuming a 10 per cent significance level for the test. In order to make comparisons we have simulated a 'typical' trial which compares an active drug with a placebo in the treatment of hypertension, using systolic blood pressure as the primary variable. As well as allowing the treatment effect to vary between centres, we have concentrated on the particular case where one centre is out of line with the others in terms of its true treatment difference. The various estimators that result from the different approaches are compared in terms of mean squared error and power to reject the null hypothesis of no treatment difference. Overall, the approach that uses the fixed effects weighted estimator of overall treatment difference is recommended as one that has much to offer.     

Refined crystal structure of methylamine dehydrogenase from Paracoccus denitrificans at 1.75 A resolution

We extend the random intercept logistic model to accommodate negative intracluster correlations for bivariate binary response data. This approach assumes a single random effect per cluster, but entails separate affine transformations of this random effect for the two responses of the pair. We show this approach works for two data sets and a simulation, whereas other mixed effects approaches fail. The two data sets are from a crossover trial and a developmental toxicity study of the effects of chemical exposure on malformation risk among rat pups. Comparisons are made with the conditional likelihood approach and with generalized estimating equations estimation of the population-averaged logit model. Simulations show the conditional likelihood approach does not perform well for moderate to strong negative correlations, as a positive intracluster correlation is assumed. The proposed mixed effects approach appears to be slightly more conservative than the population-averaged approach with respect to coverage of confidence intervals. Nonetheless, the statistical literature suggests that mixed effects models provide information in addition to that provided by population-averaged models under scientific contexts such as crossover trials. Extensions to trivariate and higher-dimensional responses also are addressed. However, such extensions require certain constraints on the correlation structure.     

Peer preferences in a desegregated school: A round robin analysis.

Analyzed children's sociometric ratings of their classmates in a desegregated school by both conventional and round robin ANOVAs. 49 Black male, 51 Black female, 36 White male, and 27 White female 6th graders indicated how much they would like to play and work with each of their classmates. Unlike conventional ANOVA, which aggregates each S's ratings of the members of a group, the round robin procedure permits the assessment of the effects of individual dyadic relationships on expressed preferences, and can therefore provide additional information on the processes involved in preference formation. The impact of both race and sex on sociometric choices was explored using these 2 techniques. Although conventional ANOVA showed strong same-race preferences, round robin ANOVA revealed that individual relationships were more important than race in forming peer preferences. A high degree of reciprocity of Ss' ratings of each other was found both within and between racial groups. Both conventional and round robin analyses found strong same-sex preferences, and much less reciprocity of ratings between the sexes than within the sexes. The complementary uses of conventional and round robin analyses of sociometric data are discussed. (36 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Sequential Bayes' strategy and mechanism of decision making in the OMIS software

The paper describes the method of image recognition in which a useful element of Bayes' approach, numerous alternatives, is united with the productive idea of heterogeneous sequential analysis--ordering of signs by their informative value for decision making. The idea of the method--a sequential Bayes' algorithm--consists in the following: the a priori probabilities are not determined before-hand but specified in turn, depending on the empirical material. The author proves that algorithms conforming to the Neumann-Pierson or Wald strategies, specifically, the heterogeneous sequential algorithm, are not to be referred to the Bayes' algorithms, as is usually done in analyses of clinical laboratory data. Ideologically the heterogeneous sequential algorithm was developed as a method for analysis of empirical data, whereas the Bayes' approach is a deductive method. Confusion of the Bayes' approach with the algorithms based on the probability ratios is methodologically unjustified at least because the Neumann-Pierson and Wald's approaches are the greatest statistical achievements of the twentieth century and have nothing to do with the Bayes' formula. The methodology of constructing the probability measure in the clinical laboratory signs space is described in detail, as are the new objects, interval and binary structures, which emerge in the course of this construction. These objects help improve the diagnostic significance of clinical laboratory information even in cases when the results of analyses are apparently normal. The sequential Bayes algorithm is compared with the traditional Bayes approach to certain clinical problems. The author concludes that the sequential Bayes algorithm is a serious alternative to algorithms for making multiple-alternative decisions in the solution of clinical tasks.     

Service-learning for graduate students

OBJECTIVES: The European Commission funded EUROTOLD Project sought to examine the legal and ethical implications of living donor organ transplantation within Europe, facilitated by a multi-centre study acquiring data on practices, laws, policies, attitudes and decision-making processes. METHODS: Methods involving primary sources included interviews with clinical staff and past and present organ donors and recipients, and questionnaire surveys of transplant centres, individual clinical staff and legal experts. These strategies were supplemented by an examination of secondary sources such as official reports, transplantation literature, etc. RESULTS: The surveys generated substantial new evidence relating to transplant centre policies and practices, and the attitudes of physicians toward living donation generally and the use of certain specific classes of donor. The latter was facilitated by the use of case scenarios, providing a window upon factors influencing judgments in this sphere. The interview data confirmed earlier findings about donor decision-making. CONCLUSIONS: Responding centres and clinicians displayed a fairly liberal attitude toward living donation but substantial diversity nonetheless exists with regard to living donor transplant volumes between centres and surrounding acceptable waiting times for transplant. Further research is required to identify precisely the reasons underpinning such disparity.     

Making sense of randomization; responses of parents of critically ill babies to random allocation of treatment in a clinical trial

Randomized controlled trials (RCTs) are widely accepted by the scientific community as the most rigorous way of evaluating interventions in health care. Although their central feature, random allocation of treatment, is generally seen as methodologically appropriate, its application has caused much debate amongst health professionals and ethicists. This paper describes the views of parents who consented that their critically ill newborn baby should be enrolled in a neonatal trial. In-depth interviews were used to determine their response to the trial and randomization. The nature of the trial was often poorly understood. The random basis of the allocation of treatment and the rationale behind this approach were also problematic issues. Some parents did not perceive a random element in the process at all. These findings advance understanding of the perceptions of trial participants and raise important issues for those concerned with RCTs. Greater understanding of participants' views provides the potential to improve the management of future trials and so the experience of those agreeing to take part.     

The population risk as an explanatory variable in research synthesis of clinical trials

The population risk, for example the control group mortality rate, is an aggregate measurement of many important attributes of a clinical trial, such as the general health of the patients treated and the experience of the staff performing the trial. Plotting measurements of the population risk against the treatment effect estimates for a group of clinical trials may reveal an apparent association, suggesting that differences in the population risk might explain heterogeneity in the results of clinical trials. In this paper we consider using estimates of population risk to explain treatment effect heterogeneity, and show that using these estimates as fixed covariates will result in bias. This bias depends on the treatment effect and population risk definitions chosen, and the magnitude of measurement errors. To account for the effect of measurement error, we represent clinical trials in a bivariate two-level hierarchical model, and show how to estimate the parameters of the model by both maximum likelihood and Bayes procedures. We use two examples to demonstrate the method.     

Some observations on two methods in psychology.

Makes explicit distinctions between the experimental and the observational (or correlational) approaches. It is suggested that the observational approach involves a distinct set of problems for which controlled experimentation is inadequate. Basic statistical models are briefly considered. It is suggested that fixed models are not generally appropriate for the observational method and that random models do not normally yield useful information in a controlled experiment. In particular, it is contended that the magnitude of variances of a random model and of the error variance of both random and fixed models is generally arbitrary in experimental studies. It is suggested that a variety of coefficients associated with analysis of variance and regression models are inappropriate in an experimental setting and that accuracy of prediction, in the usual sense, is not a proper objective of experimentation. It is concluded that constructs developed through an observational approach are not likely to be useful in an experimental science. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Process and outcome considerations in juror evaluation of eyewitness testimony.

Responds to comments by G. L. Wells (see record 1985-20040-001) on the present authors' (see record 1984-06612-001) argument that current empirical findings on perception and memory do not justify a role for psychologists in evaluating eyewitness testimony. The present authors argue that Wells's statements on process and outcome confuse the outcome of an individual trial and trials in the aggregate. The question of whether jurors tend to overbelieve eyewitness testimony is discussed. (4 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Predictive model selection for repeated measures random effects models using Bayes factors

The random effects model fit to repeated measures data is an extremely common model and data structure in current biostatistical practice. Modern data analysis often involves the selection of models within broad classes of prespecified models, but for models beyond the generalized linear model, few model-selection tools have been actively studied. In a Bayesian analysis, Bayes factors are the natural tool to use to explore these classes of models. In this paper, we develop a predictive approach for specifying the priors of a repeated measures random effects model with emphasis on selecting the fixed effects. The advantage of the predictive approach is that a single predictive specification is used to specify priors for all models considered. The methodology is applied to a pediatric pain data analysis.     

Meta-analysis of multiple outcomes by regression with random effects

Earlier work showed how to perform fixed-effects meta-analysis of studies or trials when each provides results on more than one outcome per patient and these multiple outcomes are correlated. That fixed-effects generalized-least-squares approach analyzes the multiple outcomes jointly within a single model, and it can include covariates, such as duration of therapy or quality of trial, that may explain observed heterogeneity of results among the trials. Sometimes the covariates explain all the heterogeneity, and the fixed-effects regression model is appropriate. However, unexplained heterogeneity may often remain, even after taking into account known or suspected covariates. Because fixed-effects models do not make allowance for this remaining unexplained heterogeneity, the potential exists for bias in estimated coefficients, standard errors and p-values. We propose two random-effects approaches for the regression meta-analysis of multiple correlated outcomes. We compare their use with fixed-effects models and with separate-outcomes models in a meta-analysis of periodontal clinical trials. A simulation study shows the advantages of the random-effects approach. These methods also facilitate meta-analysis of trials that compare more than two treatments.     

Does anti-estrogen therapy after estrogen receptor analysis have value in palliative therapy of hepatocellular carcinoma?

The best way to establish the place of interventions including pharmaceuticals, physical therapies, counselling, and surgery in general practice is by a trial. When undertaking such trials, care must be taken to ensure the reference population is appropriate, that recruitment strategies and the conduct of the trials do not introduce systematic bias, that adequate numbers of subjects are included to prove or disprove the question being researched and that the trial is ethical.     

A case for Bayesianism in clinical trials

This paper describes a Bayesian approach to the design and analysis of clinical trials, and compares it with the frequentist approach. Both approaches address learning under uncertainty. But they are different in a variety of ways. The Bayesian approach is more flexible. For example, accumulating data from a clinical trial can be used to update Bayesian measures, independent of the design of the trial. Frequentist measures are tied to the design, and interim analyses must be planned for frequentist measures to have meaning. Its flexibility makes the Bayesian approach ideal for analysing data from clinical trials. In carrying out a Bayesian analysis for inferring treatment effect, information from the clinical trial and other sources can be combined and used explicitly in drawing conclusions. Bayesians and frequentists address making decisions very differently. For example, when choosing or modifying the design of a clinical trial, Bayesians use all available information, including that which comes from the trial itself. The ability to calculate predictive probabilities for future observations is a distinct advantage of the Bayesian approach to designing clinical trials and other decisions. An important difference between Bayesian and frequentist thinking is the role of randomization.     

Newborn organ weight and spontaneous hypertension: recombinant inbred strain study

This study investigated the effects of neonatal hippocampal ablation on the development of spatial learning and memory abilities in rats. Newborn rats sustained bilateral electrolytic lesions of the hippocampus or were sham-operated on postnatal day 1 (PN1). At PN20-25, PN50-55, or PN90-95, separate groups of rats were tested in a Morris water maze on a visible "cue" condition (visible platform in a fixed location of the maze), a spatial "place" condition (submerged platform in a fixed location), or a no-contingency "random" condition (submerged platform in a random location). Rats were tested for 6 consecutive days, with 12 acquisition trials and 1 retention (probe) trial per day. During acquisition trials, the rat's latency to escape the maze was recorded. During retention trials (last trial for each day, no escape platform available), the total time the rat spent in the probe quadrant was recorded. Data from rats with hippocampal lesions tested as infants (PN20-25) or as adults (PN50-55 and PN90-95) converged across measures to reveal that 1) spatial (place) memory deficits were evident throughout developmental testing, suggesting that the deficits in spatial memory were long-lasting, if not permanent, and 2) behavioral performance measures under the spatial (place) condition were significantly correlated with total volume of hippocampal tissue damage, and with volume of damage to the right and anterior hippocampal regions. These results support the hypothesis that hippocampal integrity is important for the normal development of spatial learning and memory functions, and show that other brain structures do not assume hippocampal-spatial memory functions when the hippocampus is damaged during the neonatal period (even when testing is not begun until adulthood). Thus, neonatal hippocampal damage in rats may serve as a rodent model for assessing treatment strategies (e.g., pharmacological) relevant to human perinatal brain injury and developmental disabilities within the learning and memory realm.     

A comparative analysis of quality of life data from a Southwest Oncology Group randomized trial of advanced colorectal cancer

Longitudinal quality of life measurements from an advanced-stage cancer clinical trial are analysed using a variety of methods, and the results compared. The methods used require different assumptions about the mechanism that produces the missing data. They include analyses that require the data to be missing completely at random; fixed-effects models and weighted generalized estimating equations, which require missing at random data; and a fully parametric approach where the outcomes and the missingness mechanism are jointly modelled, allowing non-ignorable missing data. The data show evidence of non-random missingness, but a formal test of non-ignorable missing data is not significant.     

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.     

Role of clinical trials in memory clinics

The main points of a workshop on the place of clinical trials in the memory clinics in France are reproduced schematically. The efficiency of these centres has been tested at the time of the numerous clinical trials performed in France on Alzheimer's disease. In parallel, several difficulties have been pointed out: inhomogeneity of the technical approaches, official role of psychologists in the hospital teams, integration of clinical research to individual care. A particular effort should be oriented towards specific research on the different types of dementia, on early stages, on information for both media and authorities as well as on specific technical and methodological problems.     

Abdominal wall metastasis following surgical removal of colorectal carcinomas

Abdominal wall metastases after laparoscopic resection of colorectal cancer have been reported by various authors. It appeared that abdominal wall metastases occur more frequently after laparoscopic than after conventional, open resection of colorectal cancer. However, the frequency of abdominal wall metastases after laparoscopic surgery varies from only 0 to 1.9% in centres with sufficient relevant experience, whereas after conventional resections the frequency is 0.8-3.3%. A randomized clinical study comparing laparoscopic with conventional resection of colon cancer is necessary to assess the optimal surgical approach to colon cancer. Such a trial has been set up.     

A Bayesian hierarchical survival model for the institutional effects in a multi-centre cancer clinical trial

In randomized clinical trials comparing treatment effects on diseases such as cancer, a multi-centre trial is usually conducted to accrue the required number of patients in a reasonable period of time. While we interpret the average treatment effect, it is necessary to examine the homogeneity of the observed treatment effects across institutions, that is, treatment-by-institution interaction. If the homogeneity is confirmed, the conclusions concerning treatment effects can be generalized to a broader patient population. In this paper, a Bayesian hierarchical survival model is used to investigate the institutional effects on the efficacy of treatment as well as on the baseline risk. The marginal posterior distributions are estimated by a Markov chain Monte Carlo method, that is, Gibbs sampling, to overcome current computational limitations. The robustness of the inferences to the distributional assumption for the random effects is also examined. We illustrate the methods with analyses of data from a multi-centre cancer clinical trial, which investigated the efficacy of immunochemotherapy as an adjuvant treatment after curative resection of gastric cancer. In this trial there is little difference in the treatment effects across institutions and the treatment is shown to be effective, while there appears to be substantial variation in the baseline risk across institutions. This result indicates that the observed treatment effects might be generalized to a broader patient population.     

Summing up evidence: one answer is not always enough

Are meta-analyses the brave new world, or are the critics of such combined analyses right to say that the biases inherent in clinical trials make them uncombinable? Negative trials are often unreported, and hence can be missed by meta-analysts. And how much heterogeneity between trials is acceptable? A recent major criticism is that large randomised trials do not always agree with a prior meta-analysis. Neither individual trials nor meta-analyses, reporting as they do on population effects, tell how to treat the individual patient. Here we take a more rounded approach to meta-analyses, arguing that their strengths outweigh their weaknesses, although the latter must not be brushed aside.