Recently I was fortunate to be asked by Bioprocess International to give an analyst’s perspective on how the biopharmaceutical industry provides assurance and control over the quality of its biologics.  This was captured in an article that was published in an special edition titled The State of the Industry as a celebration of BioProcess International’s 15th year of publication.  The reason I was asked to comment was that I had previously written an article for the journal in 2012 called balancing the statistical tightrope.  This discussed a perspective of how statistics could be used effectively by scientist to promote data driven development.

Therefore, it was good to follow up five years on to understand whether my perspective or anything else had changed in the past five years.   One of the questions was how companies were working with statisticians.  Were they hiring them or using consultants?   My comment was that CMC statistics can be quite complex and that you need specialist statisticians who have experience in this particular area.  While pharmaceutical and biopharmaceutical companies, particularly multinational pharma, often have internal statisticians, they are most clinical statisticians.  I have at times encountered views that a ‘statistician is a statistician’ and can be applied generally.  This isn’t my view – they simply can’t be lifted from clinical analysis and parachuted into CMC – it really just doesn’t work, at least it want decent analysis of data performed.  It can eventually be done, but it takes time, experience and training.  Consequently companies are both hiring statisticians and working with external statisticians.

Another question was on whether scientists working within companies have enough training to perform statistical analysis themselves.  My view on this is it depends.  There are a few scientists who are very good at statistical analysis, particularly if they are doing if regularly and routinely.  The reality is that most scientists perform statistical analysis only sporadically – they may end up dredging up notes from college or dusting off a training manual from 12 months ago and applying it to what they’re doing.  This often isn’t the best approach and brings risks of errors, delays and elevated costs.  In most circumstances, it is best to get advice, either internally or externally, on what they plan to do and what they want to achieve.  It may be that the plan is sound and only a bit of tweaking is required.  However, if may be something more substantial.  My advice, was and remains, to seek the right advice from the beginning and work in partnership with a good statistician.

Finally, I was asked about the implementation of Quality by Design (QbD) in the industry.  Most people who know me understand that I’m a big advocate of QbD.  Not only have the regulators been pushing for this for years now, but the cost benefits to the business can be huge by gaining control of processes and avoiding failed batches.   When you look at the cost of a failed batch, it really can be very significant. QbD to me simply means taking a smart long-term approach and working systematically and consistently.  Some elements of QbD, such as Designed Experiments (DoE) use statistics, but much of it is about carefully thinking through the science, using your target product profile (TPP) to help you identify what the critical quality attributes (CQAs) are for your product and then using quality and risk management systems to help manage your process to meet them.  At times, it can get frustrating trying to get the message over that a QbD approach should be taken – particularly when talking to senior people in organisations, as there can be an impression that QbD complicates things, will cost them a fortune and extends timelines.  The truth is, if done right, it has the opposite effect on all counts!

Anyway, please read Cheryl Scott’s article as it contains lots of views on things that affect us all within the industry, (and not just my views).  I’d be interested to hear any feedback on any of the topics covered in the article, so please do leave comments if you feel so inclined.