Send to

Choose Destination
Value Health. 2009 Mar-Apr;12(2):325-30. doi: 10.1111/j.1524-4733.2008.00420.x. Epub 2008 Jul 18.

Information on center characteristics as costs' determinants in multicenter clinical trials: is modeling center effect worth the effort?

Author information

Department of Public Health and Microbiology and Department of Statistics and Applied Maths Diego de Castro, University of Turin, Turin, Italy.



Several methodological problems arise when health outcomes and resource utilization are collected at different sites. To avoid misleading conclusions in multi-center economic evaluations the center effect needs to be taken into adequate consideration. The aim of this article is to compare several models, which make use of a different amount of information about the enrolling center.


To model the association of total medical costs with the levels of two sets of covariates, one at patient and one at center level, we considered four statistical models, based on the Gamma model in the class of the Generalized Linear Models with a log link, which use different amount of information on the enrolling centers. Models were applied to Cost of Strategies after Myocardial Infarction data, an international randomized trial on costs of uncomplicated acute myocardial infarction (AMI).


The simple center effect adjustment based on a single random effect results in a more conservative estimation of the parameters as compared with approaches which make use of deeper information on the centers characteristics.


This study shows, with reference to a real multicenter trial, that center information cannot be neglected and should be collected and inserted in the analysis, better in combination with one or more random effect, taking into account in this way also the heterogeneity among centers because of unobserved centers characteristics.

[Indexed for MEDLINE]
Free full text

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center