Effect of Longitudinal Intracranial Pressure on Ordinal Glasgow Outcome Scale Using a Joint Model Approach

• Activity Number: 257 - SPEED: Longitudinal/Correlated Data
• Type: Contributed
• Date/Time: Monday, July 30, 2018 : 2:00 PM to 2:45 PM
• Sponsor: Biometrics Section
• Author(s): Maria Laura Rubin* and Wenyaw Chan and Jose-Miguel Yamal and Claudia Sue Robertson
• Companies: The University of Texas MD Anderson Cancer Center and University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston and Baylor College of Medicine
• Keywords: Continuous-time Markov chain; Ordinal logistic regression; Longitudinal data; Joint model; Transition rates

Abstract

The association between secondary insults and long-term functional status in traumatic brain injury (TBI) has been evaluated in many research studies. Special attention has been paid to the effect of post-baseline intracranial pressure (ICP) on the 6-month Glasgow Outcome Scale (GOS), where summary measures of longitudinal ICP data have led to inconsistent results perhaps because they cannot capture the complexity of the TBI pathophysiology. Also, studies have focused on a binary form of the GOS, when this scale is originally an ordinal variable. In this study, we propose a joint ordinal logistic regression and Markov chain model, where the unobserved transitions of a two-state adjusted continuous ICP Markov chain are included as covariates in a model to predict GOS as an ordered categorical outcome. The model simultaneously examines factors associated with transitioning to and from an elevated ICP state and the effect of those transition rates on 6-month GOS. Simulation studies indicate that our method is appropriate. From our application study, we conclude that the information provided by ICP changes over time can help improve prediction of long-term ordered GOS in TBI patients.