A principal component analysis of coagulation after trauma.

BACKGROUND

Clotting factor abnormalities underlying acute traumatic coagulopathy are poorly understood, with application of traditional regression techniques confounded by colinearity. We hypothesized that principal components analysis (PCA), a pattern-finding and data reduction technique, would identify clinically predictive patterns in the complex clotting factor milieu after trauma.

METHODS

Plasma was prospectively collected from 163 critically injured trauma patients. Prothrombin; factors V, VII, VIII, IX, X; D-dimer; activated and native protein C; and antithrombin III levels were assayed and subjected to nonlinear PCA to identify principal components (PCs).

RESULTS

Of 163 patients, 19.0% were coagulopathic on admission. PCA identified 3 significant PCs, accounting for 67.5% of overall variance. PC1 identified global clotting factor depletion; PC2 the activation of protein C and fibrinolysis; and PC3 factor VII elevation and VIII depletion. PC1 score correlated with penetrating injury and injury severity, predicting coagulopathy (odds ratio [OR], 4.67; p < 0.001) and mortality (OR, 1.47; p = 0.032). PC2 score correlated with injury severity, acidosis, and shock, and significantly predicted ventilator-associated pneumonia (OR, 1.59; p = 0.008), acute lung injury (OR, 2.24; p < 0.001), multiorgan failure (OR, 1.83; p = 0.002), and mortality (OR, 1.62; p = 0.006) but was not associated with international normalized ratio (INR)-based or partial thromboplastin time (PTT)-based coagulopathy (p > 0.200). PC3 did not significantly predict outcomes.

CONCLUSION

PCA identifies distinct patterns of coagulopathy: depletion coagulopathy predicts mortality and INR/PTT elevation, while fibrinolytic coagulopathy predicts infection, end-organ failure, and mortality, without detectable differences in INR or PTT. While depletion coagulopathy is intuitive, fibrinolytic coagulopathy may be a distinct but often overlapping entity with differential effects on outcomes.