Novel application of latent class analysis to outcome assessment in traumatic brain injury with multiple injury subtypes or poly-TBI

Objective: The aim of this study was to stratify poly-traumatic brain injury (poly-TBI) patterns into discrete classes and to determine the association of these classes with mortality and withdrawal of life-sustaining treatment (WLST).

Methods: The authors performed a single-center retrospective review of their institutional trauma registry from 2018 to 2020 to identify patients with traumatic brain injury (TBI). Patients were included if they had moderate to severe TBI, defined as Glasgow Coma Scale score ≤ 12 and Abbreviated Injury Scale (AIS) head score ≥ 3, and the presence of more than one TBI subtype. TBI subtypes were defined as subdural hemorrhage (SDH), subarachnoid hemorrhage (SAH), intracerebral hemorrhage (ICH), and epidural hemorrhage (EDH). Latent class analysis was used to identify patient classes based on TBI subtypes and Rotterdam CT (RCT) scores. The authors then evaluated class membership in relation to categorical outcomes of in-hospital mortality and WLST by using Lanza et al.'s method.

Results: A total of 125 patients met inclusion criteria for poly-TBI. Latent class analysis yielded 3 poly-TBI classes: class 1-mixed; class 2-SDH/SAH; and class 3-EDH/SAH. Class 1-mixed had a higher likelihood of SDH, SAH, and ICH, and a lower likelihood of EDH. Class 2-SDH/SAH had a higher likelihood of only SDH and SAH. Class 3-EDH/SAH had a higher likelihood of EDH and SAH, and a lower likelihood of SDH and ICH. Class 1-mixed was relatively more likely to have an RCT score of 2. Class 2-SDH/SAH was relatively more likely to have an RCT score of 2, 3, and 4. Class 3-EDH/SAH had a higher likelihood of an RCT score of 3, 4, and 5. Class 1-mixed had significantly lower mortality (χ2 = 7.968; p = 0.005) and less WLST (χ2 = 4.618; p = 0.032) than Class 2-SDH/SAH. Class 2-SDH/SAH had the highest probability of death (0.612), followed by class 3-EDH/SAH (0.385) and class 1-mixed (0.277). Similarly, class 2-SDH/SAH had the highest WLST probability (0.498), followed by class 3-EDH/SAH (0.615) and class 1-mixed (0.238).

Conclusions: Distinct poly-TBI classes were associated with increased in-hospital mortality and WLST. Further research with larger datasets will allow for more comprehensive poly-TBI class definitions and outcomes analysis.