2018
DOI: 10.1002/pst.1919
|View full text |Cite
|
Sign up to set email alerts
|

Subgroup analysis and interpretation for phase 3 confirmatory trials: White paper of the EFSPI/PSI working group on subgroup analysis

Abstract: Subgroup by treatment interaction assessments are routinely performed when analysing clinical trials and are particularly important for phase 3 trials where the results may affect regulatory labelling. Interpretation of such interactions is particularly difficult, as on one hand the subgroup finding can be due to chance, but equally such analyses are known to have a low chance of detecting differential treatment effects across subgroup levels, so may overlook important differences in therapeutic efficacy. EMA … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
6
0
3

Year Published

2019
2019
2024
2024

Publication Types

Select...
6
1

Relationship

0
7

Authors

Journals

citations
Cited by 16 publications
(9 citation statements)
references
References 29 publications
0
6
0
3
Order By: Relevance
“…To test whether subtype moderated the effects of APs on remission, a multiple logistic regression analysis was conducted that included AP subgroup (3 levels), subtype group (3 levels), and subtype-by-AP subgroup interaction. That said, it is challenging and unstable to select variables in regression models and subgroup analysis because different conclusions can be reached depending on the selection method applied (e. g., backward or forward selection) [31]. In the present study, the bootstrapping method (β = 1000 bootstrap samples) [32] was used, and the selection process was repeated on a series of bootstrapped samples from the original data to account and adjust for selection bias.…”
Section: Discussionmentioning
confidence: 99%
“…To test whether subtype moderated the effects of APs on remission, a multiple logistic regression analysis was conducted that included AP subgroup (3 levels), subtype group (3 levels), and subtype-by-AP subgroup interaction. That said, it is challenging and unstable to select variables in regression models and subgroup analysis because different conclusions can be reached depending on the selection method applied (e. g., backward or forward selection) [31]. In the present study, the bootstrapping method (β = 1000 bootstrap samples) [32] was used, and the selection process was repeated on a series of bootstrapped samples from the original data to account and adjust for selection bias.…”
Section: Discussionmentioning
confidence: 99%
“…However, resampling does not allow to account for prior beliefs in homogeneous or heterogeneous populations and the level of bias reduction seems to vary. Following the argumentation of Dane, Spencer, Rosenkranz, Lipkovich, and Parke (), one can assume that the quality of adjustment depends on how often the most influential data points are included in the bootstrap samples. Thus, the level of adjustment rather reflects whether the observed effects are based on a few or on many observations.…”
Section: Methodsmentioning
confidence: 99%
“…В 2015 г. рабочая группа Управления по контролю за качеством продуктов питания и лекарственных средств США (Food and Drug Administration, FDA) подготовила технический документ по анализу подгрупп в подтверждающих исследованиях [1], в 2019 г. Европейское агентство по лекарственным средствам (European Medicines Agency, EMA) выпустило руководство по методологии и интерпретации результатов анализа подгрупп 1 , содержание которого практически полностью соответствует вышеупомянутому документу FDA. Еще один концептуальный документ, основные выводы которого в целом согласуются с позицией регуляторных органов, был подготовлен совместной рабочей группой Европейской федерации статистиков в фармацевтической промышленности (European Federation for Statisticians in the Pharmaceutical Industry, EFSPI) и профессиональной организацией статистиков в фармацевтической промышленности (Statisticians in the Pharmaceutical Industry, PSI) PSI/EFSPI [2], представляющей интересы фармацевтических компаний -разработчиков лекарственных средств.…”
unclassified
“…), хотя имеющиеся разночтения в деталях еще не урегулированы и на практике могут приводить к серьезным спорным ситуациям. В частности, одна из центральных идей документа EMA 2 , касающаяся поискового анализа подгрупп при изучении отношения ожидаемой пользы к возможному риску применения лекарственных препаратов (соотношение «польза -риск» ) в подтверждающих КИ, заключается в методологическом подходе «генерирования сигналов». Под «сигналом» подразумевается вероятность обнаружения подгруппы, в которой эффект лечения отличается по величине/знаку от значений, наблюдаемых в остальной популяции исследования, при этом номинальный уровень значимости теста взаимодействия p должен составлять >0,05.…”
unclassified
See 1 more Smart Citation