CC BY-NC-ND 4.0 · Arquivos Brasileiros de Neurocirurgia: Brazilian Neurosurgery 2018; 37(S 01): S1-S332
DOI: 10.1055/s-0038-1673047
E-Poster – Trauma
Thieme Revinter Publicações Ltda Rio de Janeiro, Brazil

Incremental prognostic value of the activated partial thromboplastin time and creatinine in addition to the CRASH score in traumatic brain injury patients

Robson Luís Oliveira de Amorim
1   Universidade de São Paulo (USP)
,
Davi Jorge Fontoura Solla
1   Universidade de São Paulo (USP)
,
Manoel Jacobsen Teixeira
1   Universidade de São Paulo (USP)
,
Wellingson Silva Paiva
1   Universidade de São Paulo (USP)
› Author Affiliations
Further Information

Publication History

Publication Date:
06 September 2018 (online)

 

Introduction: The CRASH score is one of the main validated prognostic models for Traumatic Brain Injury (TBI). This and other TBI scores have underevaluated laboratory variables, including coagulopathy markers, which have been implicated on trauma outcomes. We have previously identified the Activated Partial Thromboplastin Time (aPTT) and Creatinine (Cr) as independent predictors of TBI mortality after adjustment for the CRASH score.

Objective: To verify the incremental prognostic value of the aPTT and Cr in addition to the CRASH score to predict 14-day mortality outcome in TBI patients.

Methods: This is a prospective cohort that included consecutive TBI patients admitted to a Trauma ICU of a tertiary university hospital from March 2012-January 2015. Clinical, laboratory and radiological data were registered. A new model, CRASH-aPTT-Cr, was created by logistic regression adjustment and compared to the original CRASH score (CT model) regarding calibration (Hosmer-Lemeshow goodness-of-fit test [HL] and Brier scores), discrimination (area under the receiver operating characteristic curve [AUC-ROC] and integrated discrimination improvement [IDI]) and clinical utility (net reclassification index [NRI]).

Results: A total 519 patients were included (mean age 41.4 [±18.2] years, 85.1% male, median admission GCS 8 [quartiles 6–13]). Neurosurgery was performed on 44.9%. The 14-day mortality was 22.8%. The CRASH-aPTT-Cr score outperformed the CRASH score on calibration, as assessed by the H-L test or the Brier scores (0.118 ± 0.216 vs 0.132 ± 0.228, mean difference 0.015, 95% CI 0.004–0.026, p = 0.006). The CRASH-aPTT-Cr score also had a better discrimination: AUC-ROC 0.844 ± 0.024 (vs 0.813 ± 0.024), p = 0.067; and IDI 0.32 ± 0.07 (vs 0.23 ± 0.06), p = 0.004. The NRI favored the CRASH-aPTT-Cr, with a net correct reclassification of 11.5% (95%CI 5.2–17.5%).

Conclusion: The addition of the aPTT and Cr to the CRASH score increased its accuracy. The aPTT may be an early marker of coagulopathy and the Cr a marker of basal frailty. Additional studies are required to externally validate this finding and further investigate its implications for TBI management.