Penetrating cardiac injuries: predictive model for outcomes based on 2016 patients from the National Trauma Data Bank

Abstract

Background

Penetrating cardiac injuries are uncommon and lethal. The objectives of this study are to examine the national profile of cardiac injuries, identify independent predictors of outcome, generate, compare and validate previous predictive models for outcomes. We hypothesized that National Trauma Data Bank (NTDB) given its large number of patients, would validate these models.

Methods

The NTDB was queried for data on cardiac injuries, using survival as the main outcome measure. Statistical analysis was performed utilizing univariate and stepwise logistic regression. The stepwise logistic regression model was then compared with other predictive models of outcome.

Results

There were 2016 patients with penetrating cardiac injuries identified from 1,310,720 patients. Incidence: 0.16%. Mechanism of injury: GSWs—1264 (63%), SWs—716 (36%), Shotgun/impalement—19/16 (1%). Mean RTS 1.75, mean ISS 27 ± 23. Overall survival 675 (33%). 830 patients (41%) underwent ED thoracotomy, 47 survived (6%). Survival stratified by mechanism: GSWs 114/1264 (10%), SWs 564/717 (76%). Predictors of outcome for mortality—univariate analysis: vital signs, RTS, ISS, GCS: Field CPR, ED intubation, ED thoracotomy and aortic cross-clamping (p < 0.001). Stepwise logistic regression identified cardiac GSW's (p < 0.001; AOR 26.85; 95% CI 17.21–41.89), field CPR (p = 0.003; AOR 3.65; 95% CI 1.53–8.69), the absence of spontaneous ventilation (p = 0.008; AOR 1.08, 95% CI 1.02–1.14), the presence of an associated abdominal GSW (p = 0.009; AOR 2.58, 95% CI 1.26–5.26) need for ED airway (p = 0.0003 AOR 1386.30; 95% CI 126.0–15251.71) and aortic cross-clamping (p = 0.0003 AOR 0.18; 95% CI 0.11–0.28) as independent predictors for mortality. Overall predictive power of model—93%.

Conclusion

Predictors of outcome were identified. Overall survival rates are lower than prospective studies report. Predictive model from NTDB generated larger number of strong independent predictors of outcomes, correlated and validated previous predictive models.

http://ift.tt/2rmwt8z