Inhalation

INH1024

Issue link: https://www.e-digitaleditions.com/i/1527415

Contents of this Issue

Navigation

Page 25 of 38

24 OCTOBER 2024 Inhalation based on a sample from that batch) and Descrip- tive Statistics (a quantitative summary of a sam- ple, with no immediate connection to the larger population). Figure 2 highlights that symptoms which are easier to identify and observe are, often, the result of much more complex and unseen root causes, but to truly address the symptoms, these underlying root causes must be addressed holistically, as an industry. In particular, the misuse of statistical methods, such as incorrect application of descriptive statistics to draw statistical inferences about the larger popula- tion, stems from a combination of diverse factors that together contribute to this ongoing problem. In categorizing these diverse factors, observable symp- toms or outcomes were brainstormed then grouped by common themes. Table 1 provides an example of this brainstorming process for symptoms that could be observed related to the misuse of descriptive statis- tics as an inferential release test. A better way to embed quality Industry would be well served to approach the qual- ity framework and the statistical methods support- ing that framework more holistically. e bene ts of such a re-thinking are numerous because they would: • Demonstrate strong commitment to quality, e - cacy and safety, as integrated components of good business practices. or experience, rather than on a quanti ed patient need as well as objectively demonstrated product and process performance. However, this current way of working can not only be ine cient, but could also lead to unintended con- sequences and undesirable outcomes. For example, reliance on historical ways of working for setting product speci cations could, over time, contribute to a belief that those limits are required for clinical outcomes when that may not be the case, or limits could be set such that there is incorrect rejection of t-for-purpose batches. is, in turn, could require resources for root-cause investigations even though the true product quality may be acceptable, and this may also contribute to drug shortages. e issues that can be observed from this (labeled as "Symptoms" in Figure 2) have impeded industry from reaching an ideal state and are due to a number of factors: • Culture (e.g., silos between clinical statisticians and CMC statisticians; preference for doing things as they have "always" been done to minimize reg- ulatory risk, etc.) • Lack of In Vitro/In Vivo Correlation • Quality Framework • Statistical Framework • Quality Standard, and • Misunderstanding of the di erence between Sta- tistical Inference (making decisions or predic- tions about the larger population, such as a batch, Figure 2 Factors Contributing to Current Challenges in Setting Meaningful Speci cations Inference vs. Descriptive Symptoms Statistical standards referenced but often not utilized No well-de ned in vitro/ in vivo correlation Limit-setting based on intuition Common terms have different interpretations Lack of statistical clarity from USP Lack of deep statistical capabilities consistently across manufacturing and CMC Industry, USP and FDA inadvertently used pharmacopeial criteria as batch release criteria Quality has been de ned by sample data not population characteristic Problem Causes Culture IVIVC Quality Framework Quality Standard Statistical Framework

Articles in this issue

view archives of Inhalation - INH1024