Premium
Correlated measures in longitudinal analysis of daily drug use patterns in a general intensive care unit
Author(s) -
Stolker Joost J.,
Heerdink Eibert R.,
Zitman Frans G.,
Leufkens Hubert G. M.
Publication year - 2004
Publication title -
pharmacoepidemiology and drug safety
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.023
H-Index - 96
eISSN - 1099-1557
pISSN - 1053-8569
DOI - 10.1002/pds.948
Subject(s) - medicine , logistic regression , binomial regression , intensive care unit , odds , odds ratio , pharmacoepidemiology , longitudinal study , negative binomial distribution , drug , regression analysis , emergency medicine , statistics , demography , intensive care medicine , psychiatry , pharmacology , mathematics , poisson distribution , pathology , medical prescription , sociology
Purpose The objective of the present study is to compare two different study designs (with and without corrections for correlated measures) to identify possible determinants of psychotropic drug use in an intensive care unit (ICU). Methods In a logistic regression analysis, odds ratios (OR) were calculated for days in which patients were exposed to psychotropics compared with non‐exposed days. In order to adjust for correlated measures, logistic regression with a logistic binomial model was applied. Results We found that adjustment for correlated measures did not result in major changes in the OR. However, with more observations per patient parameter, adjustment for correlation has greater effect. Conclusions Adjustment for correlated measures may be useful in longitudinal drug analyses. Copyright © 2004 John Wiley & Sons, Ltd.