Quantifying the relationship between time to command-following and outcomes after TBI: the 1% rule

Samuel B Snider,Hansen Deng,Flora M Hammond,Robert G Kowalski,William C Walker,Ross D Zafonte,Joseph T Giacino,Ava M Puccio,Yelena G. Bodien

DOI: https://doi.org/10.1101/2024.06.04.24308423

2024-06-04

Abstract:Importance Recovery of command-following after traumatic brain injury (TBI) is an important prognostic indicator, however, the relationship between time to command-following and long-term functional outcome is not clear. Objective Evaluate the association between command-following and outcome 1-year after TBI. Design Cohort study of participants with moderate-severe TBI in the TBI Model Systems (TBIMS) who were followed 1-year post injury, and validation in an independent dataset from the Brain Trauma Research Center (BTRC) database. Setting TBIMS is a multi-center study of participants with moderate-severe TBI treated in an inpatient rehabilitation hospital. The BTRC database is derived from a single US level 1 trauma center and includes patients with severe TBI. Participants TBIMS: N=9,052 (mean+/-SD age 38+/-18 years, 76% male, 67% white); BTRC: N=228 (mean age 37+/-17 years, 76% male, 91% white). Participants did not follow commands on acute hospital admission and survived to discharge. Exposure Days to command-following during hospitalization. Main Outcome Glasgow Outcome Scale Extended (GOSE) score <4 (i.e., death or dependency) 1-year post TBI. Results: Participants in TBIMS were more likely than those in BTRC to follow commands during acute hospitalization (90% vs 63%; p<0.001) and had a shorter median time to command-following (5 vs 9.5 days; p< 0.001). For each additional week without command-following, the odds ratio for death or dependency at 1 year was 1.30 (95% CI: [1.26,1.35]; p<0.001) in TBIMS and 1.49 ([1.15, 1.97]; p=0.003) in BTRC. Time to command-following had an AUC of 0.61 [0.59, 0.63] in TBIMS and 0.65 [0.53, 0.76]) in BTRC. Each additional day without command-following was associated with a 1.18% (1.16%, 1.20%) increase in the proportion of participants with death or dependency at 1-year in TBIMS and 1.05% (0.99%, 1.11%) in BTRC. Conclusion: Time to command-following after moderate-severe TBI is associated with 1-year outcomes, but the predictive accuracy of absence of command-following on any single post-injury day is limited. In two independent cohorts, the likelihood of death or dependency increased by ~1% for each additional day without command-following. Clinicians should be cautious when prognosticating based on the absence of command-following in the first five weeks after TBI.