by Brigitte Scott in discussion with Tim Grant PhD
Core value of biostatistics
Biostatistics has a critical role in clinical research to ensure that clinical trials are conducted in a scientifically and statistically rigorous way and the data produced are a true reflection of the clinical trial. Biostatisticians draw on their expertise in biostatistical analysis to guarantee all data are correctly captured and analysed, to maximise the information that can be derived from the data, and to help interpret the results of the clinical trial.
The core value of biostatistics in the clinical research process is in maintaining internal and external validity, ensuring that results are stochastically unbiased.
Internal validity
When comparing an active treatment versus a control in a clinical trial, applying a randomisation scheme means there is an equal chance of a clinical trial subject being in the active treatment group or in the control group , and each group has an equal chance of having an advantage. The results produced will therefore be stochastically unbiased.
Here is an example where four clinical trial subjects outperform the other subjects.
- Ideally, each group would gain two of the four subjects
- If all four subjects were randomised to the same group, this happened by chance
- If all four subjects do well in the clinical trial and they are all in the active treatment group, this would overstate the efficacy of the active treatment
- If all four subjects do well and they are all in the control group, this would understate the efficacy of the active treatment
Internal validity requires comparison between two equal groups – like comparing apples to apples – and randomisation maintains internal validity. Stochastic bias is based on the principle that, on average, a trial will be unbiased. An individual trial may have a bias but the process is still stochastically unbiased provided that each group has an equal chance of having an advantage.
Healthcare professionals (e.g., general practitioners or consultants) may pick certain patients to enter a clinical trial. This preferential selection creates bias. Using randomisation will ensure that there is an equal chance that the patients will go into the active
treatment group or the control group, thus negating this selection bias.
External validity
External validity means the trial population is representative of the patient population. This means that clinical trial subject populations should represent the patient population of interest. In epidemiology or behavioural studies, this is protected by random sampling. When a trial has external validity, observations made about the clinical trial subjects are an estimate of what would be
observed in the whole patient population of interest. In clinical trials, the patient population is defined by those individuals who have a condition and are seeking treatment. In addition, trials have inclusion/exclusion criteria that further reduce the patient population. These conditions lessen the need for random sampling, which rarely happens in clinical trials as the selection process ensures a patient sample that matches the target population.
Early biostatistics input is essential in clinical trial design
All clinical trials have a biostatistics element. The design of a clinical trial is crucial to the progress and success of a clinical research programme. Early biostatistics input is essential in clinical trial design and the creation of the trial protocol to guarantee trial quality, ensure regulatory compliance, maximise the usefulness of data, and provide adequate statistical power to enable conclusions to be
drawn. A well-designed trial provides a platform to investigate important clinical research questions efficiently and accurately. Biostatisticians ensure that the clinical trial hypothesis is translated into a statistically logical and valid set of trial-specific statistical tests and analyses that meet the criteria for acceptance by the FDA/relevant regulatory authority. Progress at the early stages of clinical trial development requires close collaboration, effective communication and transparency between different specialist teams, including biostatistics, data management, clinical, and medical writing.
Types of biostatistics input
Biostatistics input includes:
- Rationale for sample size and sampling strategies
- Development of the randomisation schedule*
- Creation of endpoints
- Contribution to inclusion /exclusion criteria*
- Development of ICH-compliant statistical analysis plans (SAPs) with mock tables, figures and listings (TFLs)
- Blind dataset testing of statistical programming to validate the implementation of SAP procedures and assess data quality and validity
- Blind data review to highlight and resolve errors in the data
- Data analysis, with descriptive statistics forming the “lion’s share” of the calculated output
*To protect internal and external validity
The ongoing need for biostatistics in clinical research
Biostatistics will always be needed in clinical research. The landscape of clinical trials is continually evolving, with the evaluation of new drugs, devices and procedures requiring trial designs of varying complexity. The biostatistician is well placed to provide value in terms of strategy, approach and analysis to enable optimal design of even the most complex clinical trials.
StatisticaMedica comprises a high-calibre team of biostatisticians and statistical programmers with expert knowledge of advanced statistical methods, clinical trial design approaches and global regulatory requirements. The StatisticaMedica team provides comprehensive, high-quality biostatistics and high-level advisory and consultancy services for the success of your clinical trial and clinical development strategy.