Multidimensional reduction of multicentric cohort heterogeneity: An alternative method to increase statistical power and robustness

Abstract : Modern clinical research takes advantage of multicentric cohorts to increase sample size and gain in statistical power. However, combining individuals from different recruitment centers provides heterogeneity in the dataset that needs to be accounted for to obtain robust results. Sophisticated statistical multivariate models adjusting for center effect can be implemented, but they can become unstable and can be complex to interpret with the increasing number of covariates to consider. Here, we present a multidimensional reduction technique to identify heterogeneity in a French multicentric cohort of hematopoietic stem cell transplantations and characterize a homogeneous subgroup prior to performing simple statistical univariate analyses. The exclusion of outliers allowed the identification of two genetic factors associated with post-transplantation overall survival. We therefore provide proof-of-concept that a samplesizereduction methodcanefficientlyaccount forheterogeneity andcentereffect inmulticentric cohorts while increasing statistical power and robustness for discovery of new association signals.
Document type :
Journal articles
Complete list of metadatas

Cited literature [17 references]  Display  Hide  Download

https://www.hal.inserm.fr/inserm-02150107
Contributor : Ana Paula Dutra Azevedo <>
Submitted on : Friday, June 7, 2019 - 8:09:58 AM
Last modification on : Wednesday, August 7, 2019 - 12:16:02 PM

File

 Restricted access
To satisfy the distribution rights of the publisher, the document is embargoed until : jamais

Please log in to resquest access to the document

Identifiers

Collections

Citation

Caroline Le Gall, Julie Laurent, Nicolas Vince, Antoine Lizee, Alisha Agrawal, et al.. Multidimensional reduction of multicentric cohort heterogeneity: An alternative method to increase statistical power and robustness. Human Immunology, Elsevier, 2016, 77 (11), pp.1024-1029. ⟨10.1016/j.humimm.2016.05.013⟩. ⟨inserm-02150107⟩

Share

Metrics

Record views

39