Statistics in Pharma; Boon or Bane
The pharmaceutical industry is widely known for its use of statistics in different areas of its enclosure. Statistics has now been an integral part of the drug lifecycle, starting from drug development, technology transfer, commercial manufacturing and till product discontinuation. The main objective of any pharmaceutical industry providing drugs is answering two questions i.e. ‘Is the Drug Effective?’ And ‘Is the Drug Safe?’ To answer these questions, it requires a ton of numerical data, which needs to be inferred and validated. The statistical maneuver may not necessarily answer the questions directly but provide ways to interpret the problem. Earlier the pharmaceutical industry relied on a casual approach towards the interpretation of vital data to meet the scrutiny of the regulators. Obviously things did not work out between the two ends. It became increasingly difficult for the pharmaceutical industry in complying with the quality standards as the so-called casual approach was not based on any science. Hence, Statistics became the savior in ensuring the drug quality and the Pharma were obliged to obey it. Recent regulatory guidelines from the Food and Drug Administration, the European Medicines Agency and the International Conference on Harmonization have encouraged scientifically-based approaches to quality and compliance. Implementing the conceptions exemplified in those guidelines will require new, more statistically rigorous and risk based ways of doing things. Statistics is a science based approach and it’s the best we have currently but there are always two sides of a coin. The flip side is ghastly and if not acted upon can lead to dismal outcomes.
The quest for longer life expectancy and desire for a higher quality of life amongst people has become a virtue of the pharmaceutical industry. A drug has consistently demonstrated its prominence in today’s world. Statistics helps to ascertain reasonable and targeted inferences from available information and helps to make rational decisions in
a pharmaceutical organization. Earlier the clinicians made the decisions by generalizing from the experiences of patients. Now, most of the process and product development includes statistical concepts such as screening experiments, optimization studies, regression modeling, process optimization, and robustness studies. When the product hits the commercial phase, various methods need to be developed to prove the safety and efficacy of the drug substance. In these analytical method development; statistical evaluation such as ANOVA, variance component studies and method ruggedness studies are used widely. Also, in manufacturing, statistical process control (SPC) is used extensively to monitor and improve processes. In pharmaceutical operations, statistical techniques such as Six Sigma, Lean Manufacturing, Process Analytical Technology and Quality by Design are used to improve its consistency. When a new drug is developed even the stability or the shelf life is also predicted by using statistical regression analysis. Statistical thinking and its application now plays an important role in the entire drug life cycle and definitely is a leap towards total quality management.
But, the pharmaceutical organization is undergoing massive changes. With the onset of technology, the obsolete systems are now getting upgraded. There is a constant process adjustment happening in all the pharmaceutical companies across the world. The way data is now realized has also been changed. The Pharma companies are facing numerous challenges and it is becoming increasingly difficult to comply with all the regulations and ensure profit. For example, it may take 10 years for the drug to get approved for marketing, from the drug discovery to its clinical approval. This 10 years already covers 50% of the patent exclusivity period. So, in these adversities, pharmaceutical heads have come with enigmatic ways to bend the system and statistics is one of them. Statistical approach has its own assumptions. Statistics does not work on heterogeneous data. In layman terms, a linear regression can be applied only when the data is linear or the sigma control chart works best only when the data is normally distributed. Sample size plays an important role in application of any statistical methods. Any statistical interpretation is true only for average. If there is any manipulation in collecting, analyzing and interpreting the data, statistical results can be misleading. This impacts patient safety.
The stakes are very high. The drug is intended for ‘life saving’. Statistics which was initially introduced to provide a risk based approach in solving and interpreting data may pose a risk in itself. Not having a Statistical approach is also not a viable option at the moment, but proper controls and understanding is required.
If there is any manipulation in collecting, analyzing and interpreting the data, statistical results can be misleading