Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
Julia O’Neill shares strategies to accelerate development

We had the pleasure of featuring Julia O’Neill in a recent episode of Statistically Speaking. Founder and principal consultant of Direxa Consulting, she was also the Distinguished Fellow, CMC Statistics Lead, as a member of the Technical Development Leadership Team Quality by Design (QbD) at Moderna to speed the delivery of the Spikevax vaccine. She shared some of her extensive experience and best practices to deliver life-saving treatments to market faster. Our curious audience asked more questions than we had time to answer, so Julia has kindly answered them for us here.


What data and statistical analyses are expected by regulators for small-scale model qualification? I know BPOG (BioPhorum Operations Group) published a paper on this; curious to know your experience with regulators on the approaches in the BPOG paper. 


My recommendations for small-scale model qualification are very consistent with what I recommend for most development challenges: include all of the relevant data and start by plotting key results in simple comparative graphs. In some situations, for accelerated products particularly, the amount of data or number of lots needed for supply may be very small. In my experience, regulators take a reasonable, balanced approach to balancing the need for demonstration of control with practical constraints on manufacturing campaigns. There is never a one-size-fits-all answer, but BPOG usually does quite a good job summarizing current practices.


Can QbD be applied in the chemical industry, manufacturing, food, and other vital industries?


Absolutely! When I worked in specialty chemicals 20+ years ago, QbD was quite well adopted as part of the overall Total Quality and Six Sigma initiatives. I hope that has not been lost, it’s where I learned much of what I now apply in biotech and pharmaceuticals.


In regards to the accelerated products, how do you justify the small sample size when using the data? 


The responsibility is on the product developer to “make the case” for approval based on the body of evidence – this includes data generated for the specific product iteration, as well as scientific knowledge of mechanisms, and experience with other closely related products. There are certainly times when regulators do not accept small sample sizes and require the developer to go back and run more studies to strengthen the case. The numbers never tell the whole story, no matter whether there is a small sample size or a vast sample. The quantitative results always have to be connected to the scientific context. This is what makes it possible at times to move forward with relatively small sample sizes for accelerated products.


What are the best book recommendations for learning QbD? Thanks. 


This is tough to answer because there are so many books available. Personally, I like to revisit some of the classics in Six Sigma and Total Quality, including those by pioneers such as Juran and Deming, and a number of American Society for Quality publications. The ICH guidances on quality might be interesting resources even for those in other industries, since they focus on some key aspects that have remained relevant for years now.


What are good resources for someone new to QbD? 


A good starting point might be the article “Understanding Pharmaceutical Quality by Design” by Yu et al. The AAPS Journal, Vol. 16, No. 4, July 2014 (# 2014) DOI: 10.1208/s12248-014-9598-3. In 13 pages it provides an overview of QbD and a list of 44 additional references for deeper reading.


A follow-up JMP® question was asked, “I was trained on SAS in my PhD program and am used to terms like Canonical Correlation, Multicollinearity, and Durbin-Watson Statistic. Has JMP continued these descriptions in the software?" The answer is, Absolutely! Just search for the canonical, multicollinearity, or Durbin-Watson (or just about any other commonly used statistical term) in our online documentation (, click the links provided, and you’ll see when and how they’re used in JMP.


We appreciate Julia for taking the time to answer these and all the other great questions we had during the event. If you missed the livestream, the on-demand version is available for viewing at your leisure. 

Last Modified: Jun 20, 2024 8:43 AM