Statistical Considerations in Setting Product Specifications

Abstract

According to ICH Q6A (1999), a specification is defined as a list of tests, references to analytical procedures, and appropriate acceptance criteria, which are numerical limits, ranges, or other criteria for the tests described. For drug products, specifications usually consist of test methods and acceptance criteria for assay, impurities, pH, dissolution, moisture, and microbial limits, depending on the dosage forms. They are usually proposed by the manufacturers and subject to the regulatory approval for use. When the acceptance criteria in product specifications cannot be pre-defined based on prior knowledge, the conventional approach is to use data from a limited number of clinical batches during the clinical development phases. Often in time, such acceptance criterion is set as an interval bounded by the sample mean plus and minus two to four standard deviations. This interval may be revised with the accumulated data collected from released batches after drug approval. In this article, we describe and discuss the statistical issues of commonly used approaches in setting or revising specifications (usually tighten the limits), including reference interval, (Min, Max) method, tolerance interval, and confidence limit of percentiles. We also compare their performance in terms of the interval width and the intended coverage. Based on our study results and review experiences, we make some recommendations on how to select the appropriate statistical methods in setting product specifications to better ensure the product quality.

Performance characteristics of the VIDAS® SARS-COV-2 IgM and IgG serological assays

“…The test is interpreted as negative when i < 1.00 and positive when i ≥ 1.00. The positivity cut-off values for the IgM and IgG tests were determined from a healthy pre-pandemic cohort (259 [IgM test] and 120 [IgG test] samples collected prior to August 2019), using non-parametric 99 th percentile because of normality rejection for the IgM positivity cut-off and using the (99,99) tolerance intervals approach after Box-Cox transformation (24) for the IgG positivity cut-off (99 th percentile index values at a 99% confidence level) (data not shown).…”

Thresholds for Arterial Wall Inflammation Quantified by 18F-FDG PET Imaging

“…Differences in 18 F-FDG uptake between the different groups were assessed using a multivariate model to account for age, sex, hypertension (systolic blood pressure >140 mm Hg, diastolic blood pressure >90 mm Hg, or use of antihypertensive medication), body mass index, smoking, drug use (statins, ezetimibe, angiotensin-converting enzyme inhibitors, acetylsalicylic acid, beta-blockers), lipid profile, and glucose. We estimated SUV and TBR upper threshold values based on the tolerance interval (35) using the 95th and 90th percentiles of log-normal SUV and TBR in the healthy control subjects.…”

Strategies for Setting Patient-Centric Commercial Specifications for Biotherapeutic Products

“…Appropriate statistical methods that may be used to characterize historical data and aid in the establishment of acceptance criteria are described widely in pharmaceutical statistical literature. Dong et al (2015) describe and compare the performance of limits based on reference interval, (Min,Max) methods, tolerance intervals (TI), and confidence limits (CL) on percentiles. 39 Kurtulik et al (2007) consider how limits set using mean ± 3SD, TI, and historical minimum/ maximum value compare and how these limits are affected by historical sample size.…”

“…Dong et al (2015) describe and compare the performance of limits based on reference interval, (Min,Max) methods, tolerance intervals (TI), and confidence limits (CL) on percentiles. 39 Kurtulik et al (2007) consider how limits set using mean ± 3SD, TI, and historical minimum/ maximum value compare and how these limits are affected by historical sample size. 40 Coffey and Yang (2018) describe extension of such statistical methods by estimating random variance components and incorporating consideration of stability trends.…”