Annual Meeting Recordings are now available for purchase
304132
Importance of choosing the appropriate baseline adjustment when comparing groups with pre-post data
Monday, November 17, 2014
Jeff Burton, PhD
,
Biostatistics & Epidemiology Core, Pennington Biomedical Research Center, Baton Rouge, LA
Robbie Beyl, PhD
,
Biostatistics & Epidemiology Core, Pennington Biomedical Research Center, Baton Rouge, LA
William Johnson, PhD
,
Biostatistics & Epidemiology Core, Pennington Biomedical Research Center, Baton Rouge, LA
When investigating the effects of public health interventions, often data are collected prior to implementation (baseline) and again subsequent to the intervention (follow-up). In these types of pre-post studies, adjustments to account for individual differences in baseline measurements can be applied. These adjustments are utilized to help ensure appropriate comparisons of change from baseline at follow-up between treatment groups. Here, two baseline adjustments are investigated. The first requires calculating differences of baseline and follow-up measurements and comparing these changes between groups. The second involves including baseline measurement as a covariate in a linear model and comparing follow-up measurements between groups. In a simulation study, data are generated under various configurations of treatment group intercept and slope parameter values in a linear mixed model. Following application of baseline adjustment, hypothesis tests are carried out to evaluate equality of change from baseline between treatment groups. Estimates of power and type I error rate are obtained and compared between the two adjustment scenarios. It is shown that, under certain data structures, the different baseline adjustments lead to dissimilar results.
Learning Areas:
Biostatistics, economics
Learning Objectives:
Describe how to adjust repeated measures data for individual baseline measurements. Assess properties of hypothesis tests constructed from data that has been adjusted for individual baseline measurements.
Keyword(s): Biostatistics
Presenting author's disclosure statement:Qualified on the content I am responsible for because: I have a Ph.D. in Biostatistics and a Master’s degree in Applied Statistics both from Louisiana State University. I am currently employed as an assistant professor of biostatistics at Pennington Biomedical Research Center in Baton Rouge, LA. In my position, I conduct statistical research in addition to providing statistical support to faculty of approximately 100 in numerous biomedical research fields.
Any relevant financial relationships? No
I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines,
and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed
in my presentation.
