Statistical significance testing and clinical trials.

The efficacy of treatments is better expressed for clinical purposes in terms of these treatments' outcome distributions and their overlapping rather than in terms of the statistical significance of these distributions' mean differences, because clinical practice is primarily concerned with the outcome of each individual client rather than with the mean of the variety of outcomes in any group of clients. Reports of the obtained outcome distributions for the comparison groups of all competently designed and executed randomized clinical trials should be publicly available no matter what the statistical significance of the mean differences among these groups, because all of these studies' outcome distributions provide clinically useful information about the efficacy of the treatments compared. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

Statistical applications for in vitro diagnostic tests and other medical device clinical trials

Some statistical methods applied to in vitro diagnostic tests for the three primary indications (screening, diagnosis, and monitoring) are discussed. Various examples with practical statistical applications are presented, including test for k by k ordered categorical matched-pair data for screening of cervical cancer, receiver operating characteristic (ROC) curve for diagnosis or screening, and the Cox time-to-event model to estimate relative risk of first cancer progression by monitoring carcinoembryonic (CEA) levels for stage IV breast cancer patients.     

Statistical reporting of clinical trials with individual changes from allocated treatment

We consider clinical trials in which information is available about subjects' treatment changes after randomization. To understand whether any difference between randomized groups in the intention-to-treat analysis can be explained by such treatment changes, we need analysis strategies which take account of treatment actually received. Selection bias is then a potentially serious problem. We relate risk in a time-dependent proportional hazards model to current treatment, with treatment combinations coded in two alternative ways. To reduce selection bias, treatment history (number of treatments dropped) and baseline covariates can be added to the model. Including current risk markers would also reduce selection bias but makes interpretation difficult. The methods are illustrated using data from the British Medical Research Council (MRC) elderly hypertension trial, with time to cardiovascular death as an outcome. Results for the comparison of diuretic and beta-blocker treatment are similar in all analyses, suggesting that selection bias is small and adding support to the hypothesis that the observed treatment differences are due to the randomized treatments themselves.     

Crossover designs for clinical trials

I discuss three-period crossover designs for an efficient comparison of two test treatments with special application to clinical trials which often have many practical limitations. In this paper I specify a subset of three-period crossover designs so that the investigators are not left with the problematic two-period two-sequence design, should the trials be terminated after the second period. I show that there is a dramatic reduction in variability for estimating the direct and residual treatment effects in three-period designs compared to two-period designs. I also show that the universally optimal design with ABB and BAA sequences is unsuitable when a complex form of residual effects is suspected, such as the second-order residual effects or treatment by period interactions. The design with ABB, BAA, AAB, and BBA sequences is relatively robust to these uncertain model assumptions. I also discuss missing data problems and conclude that, even with a large proportion of missing values, the three-period design is far more efficient than the two-period design.     

Statistical power and optimal design for multisite randomized trials.

The multisite trial, widely used in mental health research and education, enables experimenters to assess the average impact of a treatment across sites, the variance of treatment impact across sites, and the moderating effect of site characteristics on treatment efficacy. Key design decisions include the sample size per site and the number of sites. To consider power implications, this article proposes a standardized hierarchical linear model and uses rules of thumb similar to those proposed by J. Cohen (1988) for small, medium, and large effect sizes and for small, medium, and large treatment-by-site variance. Optimal allocation of resources within and between sites as a function of variance components and costs at each level are also considered. The approach generalizes to quasiexperiments with a similar structure. These ideas are illustrated with newly developed software. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Statistical analysis and optimal design for cluster randomized trials.

In many intervention studies, therapy outcome evaluations, and educational field trials, random treatment assignment of clusters rather than persons is desirable for political feasibility, logistics, or ecological validity. However, cluster randomized designs are widely regarded as lacking statistical precision. This article considers when and to what extent using a pretreatment covariate can increase experimental precision. To answer this question, the author first optimizes allocation of resources within and between clusters for the no-covariate case. Optimal sample sizes at each level depend on variation within and between clusters and on the cost of sampling at each level. Next, the author considers optimal allocation when a covariate is added. In this case, the explanatory power of the covariate at each level becomes highly relevant for choosing optimal sample sizes. A key conclusion is that statistical analysis that fully uses information about the covariate-outcome relationship can substantially increase the efficiency of the cluster randomized trial, especially when the cost of sampling clusters is high and the covariate accounts for substantial variation between clusters. Recent multilevel studies indicate that these conditions are common. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Self-designing clinical trials

I present a method of sequential analysis for randomized clinical trials that allows use of all prior data in a trial to determine the use and weighting of subsequent observations. One continues to assign subjects until one has 'used up' all the variance of the test statistic. There are many strategies to determine the weights including Bayesian methods (though the proposal is a frequentist design). I explore further the self-designing aspect of the randomized trial to note that in some cases it makes good sense (i) to change the weighting on components of a multivariate endpoint, (ii) to add or drop treatment arms (especially in a parallel group dose ranging/efficacy/safety trial), (iii) to select sites to use as the trial goes on, (iv) to change the test statistic and (v) even to rethink the whole drug development paradigm to shorten drug development time while keeping current standards for the level of evidence necessary for approval.     

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.     

Therapy and clinical trials

Little is know about intraosseous migration of nonerupting teeth, a rare natural condition of horizontal tooth movement and impaction. It occurs only in the mandible and involves primarily the second premolar or the canine. When the second premolar is the affected tooth, it always is found distal to its normal position. The origins of the second premolar intraosseous migration phenomenon are obscure and usually no treatment is recommended. Intraosseous migration involving the canine is commonly called transmigration because the affected canine moves mesially across the mandibular symphysis to the opposite side of the mandible. Analysis of 50 published cases of canine transmigration indicated higher occurrence in women and no sidedness preference. In over 80% of the studied cases, the canine remained nonerupted and, of the 24 cases receiving some treatment, all but two underwent extraction of the anomalous canine. The canine transmigration phenomenon appears to show signs of having some genetic determinants.