ICH E9 states that the role and responsibilities of the biostatistician, in collaboration with the rest of the team, is to ensure that statistical principles are properly applied in clinical trials. Therefore, the statistician must have a balanced combination of training and experience to properly implement these principles.
It is common thinking that a good statistician is an expert in statistical methods and techniques, but there are other equally or more important aspects that are essential to ensure a good statistical analysis. These skills are not acquired quickly, but learned and developed over time, with training, experience and above all with communication and discussion with colleagues. The main attributes of a skilled statistician are as follows:
- Critical Thinking
Henry Ford, founder of Ford automobile company, said that "thinking is the hardest job there is, and probably the reason why so few people practice it." Building, reasoning and acting with a mindset based on critical thinking is indeed a difficult task. Critical thinking consists of analyzing information with the aim of clarifying its veracity, ignoring possible biases, and it is essential to be able to deal with data analysis in an objective manner. It is important that the statistician participates from the earliest stages of the study, in its approach and design. Critically questioning the source of the data, the target population, the methods of sample selection, and the way of assigning the treatment under study are fundamental aspects that will undoubtedly have an impact on the validity of the results (internal validity) and their possible extrapolation to the target population (external validity).
- Communication and teamwork
Ryunosuke Satoro, Japanese writer wrote "Individually, we are a drop. Together, we are the sea." The statistician must work and collaborate closely with all members involved in the study. Communication and collaboration with the sponsor are essential to understand the ultimate purpose and objectives of the study, and to plan the data analysis accordingly. Communication with the study team is vital and, although it is often thought that statistics is something that is done at the end of the study, in reality, the statistics team starts planning the analysis from the very beginning. The statistics team must have a close communication with the data management team alerting about any inconsistency or data issue on a regular and immediate basis, thus strengthening the quality assurance and integrity of the data (that will later be analyzed). Additionally, the relationship between the statistician and the programmer is a fundamental tandem that must go hand in hand throughout the study. The programmer is responsible for bringing to reality the analysis that the statistician himself has left embodied in the statistical analysis plan.
- Analytical and logical thinking skills
The ability of analysis and logical thinking consists of being able to systematically and logically examine the information obtained, deconstructing it into basic components in order to understand its structure and relationships. This is one of the skills that the statistician develops throughout his/her usually intensive training in mathematical sciences. The statistician must make important decisions oriented to perform a correct and fair analysis. An example in our work would be when deciding the best way to deal with missing data (missing imputation strategy). In this case, the first thing to do is to understand what the cause of the missing data is, and to evaluate whether it could be related in any way to the efficacy or safety of the study. There are missing data points that do not generate bias because they occur randomly in the study. It is said that these missing data follow a MCAR (Missing Completely at Random) pattern, occurring for reasons unrelated to the relevant variables or results of the study (for example, missing data caused by a transportation issue that prevents patients from attending the visit). But there are other types of missing data that can introduce bias, so it is important to be able to discern which scenario we are in and to be able to apply the appropriate imputation strategy.
- Learning capacity
Statistics is a very extensive science, with multiple theories and methods that are applicable to the health science industry. It is also common for regulatory agencies to propose certain techniques of recent use that are considered more appropriate for the study in question. This is why the statistician must be open to learning and assimilating new concepts and be able to implement them accordingly. He/she must also keep up to date with emerging technologies or tools such as new data analysis systems and languages or artificial intelligence, which will gradually be implemented more and more in the health science industry.
- Responsibility and Integrity
Because of the important role statistics plays in validating research results, Vardeman and Morris (2003) point out that integrity and ethical principles are vital in this role. Statisticians are key in identifying outliers, non-compliance, or research misconduct. They must work rigorously, following established processes with honesty and transparency, while documenting the analysis process adequately. This must be done throughout all stages of the study, thus ensuring scientific credibility and compliance with GCP principles.
It is therefore crucial for any CRO to have not only an expert statistical team with the appropriate technical acumen but also one having these key attributes which take time to develop and are fundamental to carrying out quality work.
Author:
Eduardo Sobreviela
Director, Biostatistics - Linical