Making the Connection between Centralized Biometrics & Adaptive Trial Design

Flexible study designs are gaining popularity in clinical research as sponsors search for ways to reduce study timelines and costs by making modifications along the way. Implementing Adaptive Trial Design has been one flexible design that has proven successful for many clinical trials with 20-30% of trials now using this methodology. However the degree of success of this design approach doesn’t just depend on the capabilities of your biostatistician. It can also depend on your outsourcing model.

CROS NT looks at the connections between a centralized biometrics approach and adaptive trial design success.

In an adaptive approach, at each stage and in each phase of the drug or device development cycle, the probability of success is quantified and the study team is full informed in order to evaluate risks and benefits associated with each decision. This begins with the protocol where scenario planning acts as a critical “stress test” key tool for demonstrating the value of adaptive designs.

Ways of maximizing efficiency through adaptive designs are:

– Early stopping (futility, early rejection)
– Change in treatment allocation ratios
– Alterations of hypothesis (non-inferiority vs. superiority)
– Use of different test statistics
– Sample size re-assessment
– Dropping/adding treatment arms
– Select special populations (inclusion/exclusion criteria)
– Combnation of trial phases (adaptive seamless design)

These are generally statistical concerns, therefore where does the centralized biometrics outsourcing model fit in?

It is generally agreed that the best time to involve a biostatistician in a clinical trial is at the very begninning, This allows the biostatistician to understand the study design and make suggestions on hypothesis testing and analysis and also ensures continuity throughout the study team. If one study team – including the biostatistician – is assigned to trial design, data analysis and medical communications from the start, common data standards can be applied throughout the drug development process. This is a centralized approach to data collection and analysis.

As planning is an important step in preparing and implementing an adaptive design approach, there should be a good communication plan defined between the CRO and sponsor. Agreeing on a statistical analysis plan, methodologies and programming formats can reduce the number of reviews, therefore saving time and costs.

In order for statisticians to make “go/no-go” decisions, they need access to real time data as soon as it is collected. This can be achieved through a centralized approach when all data are stored in a central warehouse and/or archive which avoids having to keep track fo multiple repositories.

The biostatistician really does collaborate with the rest of the study team, includnig Data Managers, Statistical Programmers and Medical Writers. Regarding Data Management, the biostatistician can assist with CRF development and dataset specifications. Working with statistical programmers, methodological statisticians ensure data formatting is correct and select data to be pooled. These interactions all contribute to the biostatistician’s statistical analysis that lead to adaptations in a trial such as the decision to recalculate the sample size, alter study endpoints or even terminate a study. When the study data is dispersed among various sources, it becomes more time consuming for the biostatistician to gather the data needed and calls in question the quality of the data being pooled.

Some studies have already suggested that the implications of adaptive designs could save Sponsors between $100-200 million USD per year. Those savings, coupled with the considerable 30-40% cost savings of a centralized biometrics approach, due mostly to global libraries and achieved efficiencies through one project team, can result in significant cost savings for trial Sponsors.