Latent Trajectories and Predictors of Cognition: A Traumatic Brain Injury Model Systems Study

Supplementary Table 1 shows the proportion of missing data by variable stratified by class and computed maximum difference between classes and final predictors within the final backward stepwise multinomial logistic regression model. Supplementary Table 2 shows the comparison of demographic, clinical, and injury-related characteristics between participants included and excluded from the sample. Supplementary Table 3 shows the baseline demographic, clinical, and injury characteristics of 8060 participants by class trajectory membership. Supplementary Table 4 shows the model fit indices and classification quality parameters for 1-7 class model solutions from the latent class growth analysis on ordinal FIM-Cog data. Supplementary Table 5 shows the results of the backward stepwise multinomial logistic regression assessing the role of demographic, clinical, and injury-related characteristics on FIM-Cog trajectory membership.