Clinical Trials

The OPTions In Management with Antiretrovirals (OPTIMA) Trial, a collaboration between three governmental agencies in the USA, UK and Canada is a large-scale, multicenter, randomized controlled trial designed to compare the relative efficacy of different therapeutic strategies in HIV disease. The collaboration of three coordinating centers introduced unique data management issues including: a) use of different data systems for managing “country” trial data; b) two-way data transfer between the coordinating centers and the center where OPTIMA data is merged and analysis files are generated; and c) translation of certain data forms (mainly patient completed questionnaires) into French and Spanish. The involvement of three data centers provided a challenge in planning, designing and executing data management procedures in OPTIMA. Processes were implemented to ensure the trial's successful execution at the trinational level, without disregarding countryspecific requirements and regulations. The OPTIMA model required a significant time and resource investment on behalf of all agencies involved, but allowed for autonomy and a sense of joint ownership of data in different countries.

Background The quality of the decisions reached by data monitoring committees (DMCs) is crucial. The aim of this paper is to identify factors that may make errors more or less likely in small, task-oriented, decision-making expert groups and to consider the implications of these factors for data monitoring committees.

Methods A systematic overview was carried out of reviews of empirical studies of small group processes and decision errors in small, task-oriented decision-making groups in laboratory or real-world settings, published between 1950 and 2002 (n = 57 included reviews).

Results These reviews suggest that a number of factors may increase the likelihood that small groups will make poor and potentially erroneous decisions. The most important of these, in terms of empirical support, are: biased or overly directive leadership, expression of a limited range of opinions during group discussion, poor procedures for identifying or appraising the available information, and presentation of the available information in a way that is likely to result in biased perception of it.

Conclusions The main implications for DMCs relate to membership, the role of the chairperson, the information provided for DMCs and training for DMC members. Selection methods that encourage a degree of diversity within the DMC are recommended. Chairs of DMCs should be experienced members, who have the skills to facilitate a discussion, can manage conflict effectively and can be impartial. Adherence to a prespecified analysis plan is recommended to reduce the risk of error associated with strong evidence or excess information. Training in the use of methodical decision-making procedures, education about the factors that influence decision quality and an opportunity to participate in mock DMC discussions may be of benefit for new members.

Group-randomized study designs are useful when individually randomized designs are either not possible, or will not be able to estimate the parameters of interest. Blocked and/or stratified (for example, pair-matched) designs have been used, and their properties statistically evaluated by many researchers. Group-randomized trials often have small numbers of experimental units, and strong, geographically induced between-unit correlation, which increase the chance of obtaining a “bad” randomization outcome. This article describes a procedure – random selection from a list of acceptable allocations – to allocate treatment conditions in a way that ensures balance on relevant covariates. Numerous individual- and group-level covariates can be balanced using exact or caliper criteria. Simulation results indicate that this method has good frequency properties, but some care may be needed not to overly constrain the randomization. There is a trade-off between achieving good balance through a highly constrained design, and jeopardizing the appearance of impartiality of the investigator and potentially departing from the nominal Type I error.

Protective vaccine efficacy, VE _S, is measured as one minus the incidence rate ratio (IRR) or the relative risk (RR) in the vaccinated group compared with the unvaccinated group. In this paper, we systematically present Bayesian estimation of protective vaccine efficacy based on the Poisson and binomial distributions. We also propose a new tool, the vaccine efficacy acceptability curve, to represent the uncertainty for the estimate of the vaccine efficacy graphically. It is very useful, especially when there is no universal agreement on the acceptable vaccine efficacy. The vaccine efficacy acceptability curve is defined as the posterior probability that the measure of vaccine efficacy VE _S ≥k for each acceptable value k. When a vaccine is highly efficacious, the number of vaccinated susceptibles being infected is likely to be very small or even zero. Then the assumptions of normality and log-normality of IRR or RR usually do not hold well. Although frequentist exact methods provide good estimates of the confidence interval, they are overly conservative and are computationally difficult to extend to estimate the vaccine efficacy acceptability curve. In this paper, our focus is on Bayesian estimation of protective vaccine efficacy, its highest probability density credible set, and the vaccine efficacy acceptability curve through Markov chain Monte Carlo (MCMC) methods. We illustrate the methods using the data from two pertussis vaccine studies and the H. influenza Type B preventive trial.

The presence of possible treatment interaction when designing a factorial study can be either greeted with dismay (for an antagonistic or negative interaction) or can be welcome (in the case of a synergistic effect). The type of potential interaction may greatly influence the choice of study design. Depending on the magnitude of an expected additive or multiplicative interaction, a three-arm study, instead of a 2 × 2 factorial, may yield greater statistical power. A graphical aid for examining the loss of power due to the presence of such interaction is developed. The technique can also be applied to designing studies where prespecified subgroup analyses are of particular interest. Tests for interaction effects between subgroups are usually underpowered, even if the subgroups have been prespecified in the protocol. The technique can be used to determine the appropriate sample sizes in the subgroups to ensure adequate power to detect potential interaction effects. The method is illustrated with respect to a published 2 × 2 factorial study. In this study, interaction reduced the power of the final analysis to detect significant main effects.

The National Emphysema Treatment Trial (NETT) was a multicenter randomized controlled trial comparing medical treatment plus lung-volume-reduction surgery (LVRS) to medical treatment alone for the treatment of severe emphysema. The primary outcomes specified for the trial were mortality from all causes and change in functional status as indicated by the change in maximum exercise capacity measured two years after randomization. A secondary objective of the trial was to define criteria to identify subgroups of patients at risk of harm or benefit from LVRS. Stopping guidelines for safety and efficacy based on 30-day mortality and a combination of overall mortality and functional status at two years were specified at the inception of the trial. Although specific subgroups of patients likely to benefit were not identified in advance, several clinical factors were specified as likely to be important in defining subgroups with differential outcome. In May 2001, with 40% of expected deaths accrued, the Data and Safety Monitoring Board determined that a subgroup of patients was at significantly higher risk of 30-day mortality from LVRS without counterbalancing evidence of functional benefit, and recommended that the protocol be modified to exclude further randomization of such patients. The trial's sponsor, the National Heart, Lung and Blood Institute, accepted the recommendation, which was rapidly communicated to participating clinics. This paper describes the operational aspects of identification of the subgroup and implementation of the recommendation to continue the trial, but to terminate enrollment of new patients in the subgroup. These aspects include notification of the investigators, the institutional review boards, the Research Group, the patients and the medical community. We also describe the repercussions of the publication and the misinterpretations of the results based on media coverage.

A simple framework for assessing and reporting data availability in an ongoing clinical trial is described. Protocol requirements, visit schedules and data availability are combined into a simple report to track the progress of a study.