USP <1010> Analytical Data – Interpretation and Treatment provides information regarding the statistical treatment of data generated for chemical and other analyses, with a goal of sound decision-making throughout pharmaceutical development and production. The chapter touches on evaluating data sets, treatment of outliers, and comparison of analytical procedures. The chapter does not provide an exhaustive list of acceptable statistical approaches, but rather presents some suggestions for best practices.
At PTL, USP <1010> is most commonly referenced for the comparison of analytical procedures. This often takes the form of comparing the results generated by a legacy instrument to those of a comparable model or analogous technology – for example, comparing the performance of the Malvern Mastersizer 3000 laser diffractor to its predecessor, the Mastersizer 2000. A comparison study may also be conducted to evaluate performances across several units of the same instrument – for example, comparing the performances across two Mastersizer 3000 units to claim statistical equivalency.
Why Might You Need a Comparison Study?
Why does PTL periodically replace old instrumentation? Consider the way in which your cell phone becomes obsolete. After several years, it still makes calls and sends text messages, but slows down and becomes incompatible with software updates. Upgrading your cell phone periodically is necessary in order to stay up-to-date with the latest technology.
In a similar way, replacing obsolete laboratory instrumentation with new or otherwise equivalent models is PTL’s standard practice to ensure our clients’ materials are analyzed using the most current technology.
Statistical comparison across instruments is part of PTL’s internal risk assessment work, conducted when upgrading instrumentation or expanding in-house testing capabilities. PTL’s internal work relies on analyzing model materials. However, PTL also offers a similar comparison service to clients interested in transferring their analytical method from a legacy instrument to the newest instrument model or analogous technique. Such a study can provide insight as to how the transfer to the new technology may affect the results generated for their particular material.
What Does a Comparison Study Entail?
Accuracy and precision are the focal points. Every analytical method will have some inherent variability. When you conduct several replicates of an analysis, you generate a set of data with a mean value and a standard deviation. Comparison of accuracy seeks to determine how close the new procedure’s mean is to the old. Are the two procedures, on average, producing equivalent results?
Furthermore, comparison of precision seeks to determine if the new procedure’s variance (i.e., the standard deviation squared) is not significantly greater than the variance of the old procedure. Are the results produced by the new procedure more variable or less variable than the results produced by the old? These two components of the comparison study, accuracy and precision, work together to paint a picture of the expected differences between two procedures.
Prior to conducting a comparison study, acceptance criteria must be determined: namely, the largest acceptable difference between two methods that, if achieved, still concludes that the methods are equivalent. This value, along with the sample size for the study, must be selected such that the probability of drawing an incorrect conclusion is minimized.
A false positive result in this scenario would conclude that the two methods are equivalent, when in fact they are not. Conversely, a false negative would conclude that the two methods do not perform equivalently, when in actuality they do. Minimizing the probability of an incorrect conclusion is critical, especially for clients testing materials against specification limits.
PTL Is Here to Help
For those interested in pursuing a comparison study referencing USP <1010>, PTL will be your partner throughout the entire process. We will recommend acceptable limits for the comparisons of accuracy and precision and suggest an appropriate sample size to minimize the probability of error. Such recommendations are based on an established body of work using the material of interest. While tracking the frequency of out-of-specification results is outside the scope of PTL’s USP <1010> package, clients are welcome to bring their own considerations and historical data to the discussion to ensure a well-planned study.
Contact us today to get started!
By Kayla Siemann, Particle Characterization Chemist II.