Просмотр новости
Найдите то, что Вас интересует
Спутники
РН
SpaceX
Роскосмос
NASA
СпутникиРНSpaceXРоскосмосNASABackground. In clinical bioequivalence (BE) studies, the need to split the overall sample of volunteers into several groups (cohorts) often arises due to bed-capacity constraints. The inclusion of the “Group” factor in the analysis of variance (ANOVA) model remains a matter of debate: historically stringent FDA requirements are being replaced by more flexible ICH M13A approaches, but the regulatory documents of the Eurasian Economic Union (EAEU) still contain uncertainties and certain risks related to the fact that including this factor may widen the confidence interval and thus hinder the demonstration of BE.Objective. To analyze the evolution of regulatory requirements for accounting for the “Group” factor and to statistically assess its actual impact on the width of the 90 % confidence interval and on residual variance in BE studies.Methods. A critical review was conducted of FDA guidance (2001, 2022), ICH M13A (2024), EMA (2010) and the EAEU Rules (Decision No. 85). A retrospective analysis of 17 BE studies (34 pairs of comparisons between standard and modified ANOVA models) with a 2×2×2 design, performed at a single center during 2010–2025, was carried out. The relative change in confidence interval width (ΔW, %) and in residual variance (ΔMSE, %) upon adding group factors was evaluated. Stratification by intra-subject variability level (low, moderate, high) and correlation analysis with the time gap between cohorts (ΔT, days) were performed.Results. No statistically significant differences were found between the standard and modified models (p = 0.351). The mean ΔW was −0.37 % (median 0.55 %; range −9.12 to +3.18 %). No correlation was observed between ΔW and the time gap (ranging from 2 to 10 days) (ρ = –0.039; p = 0.825). In the stratified subgroups, the change in MSE was not significant: for low variability –0.66 % (p = 0.962), for moderate −3.44 % (p = 0.118), and for high +1.65 % (p = 0.119). The Bland–Altman plot confirmed good agreement between the models (95 % limits of agreement from −5.93 % to +5.19 %).Conclusion. When the time lag between cohorts is up to 10 days and procedures are strictly standardized, inclusion of the “Group” factor in the ANOVA model does not lead to a clinically meaningful change in the confidence interval width or residual variance, irrespective of the drug’s inherent variability. This allows studies to be conducted with group separation without the risk of unjustified widening of the confidence interval. The EAEU regulatory framework needs to be updated in line with the current ICH M13A recommendations. The “Group” factor is primarily an organizational component imposed by regulators, rather than a factor that can seriously affect the BE conclusion.