Predictors of initial management failure in traumatic hemothorax: A prospective multicenter cohort analysis

Background Traumatic hemothorax is common, and management failure leads to worse outcomes. We sought to determine predictive factors and understand the role of trauma center performance in hemothorax management failure. Methods We prospectively examined initial hemothorax management (observation, pleural drainage, surgery) and failure requiring secondary intervention in 17 trauma centers. We defined hemothorax management failure requiring secondary intervention as thrombolytic administration, tube thoracostomy, image-guided drainage, or surgery after failure of the initial management strategy at the discretion of the treating trauma surgeon. Patient-level predictors of hemothorax management failure requiring secondary intervention were identified for 2 subgroups: initial observation and immediate pleural drainage. Trauma centers were divided into quartiles by hemothorax management failure requiring secondary intervention rate and hierarchical logistic regression quantified variation. Results Of 995 hemothoraces in 967 patients, 186 (19%) developed hemothorax management failure requiring secondary intervention. The frequency of hemothorax management failure requiring secondary intervention increased from observation to pleural drainage to surgical intervention (12%, 22%, and 35%, respectively). The number of ribs fractured (odds ratio 1.12 per fracture; 95% confidence interval 1.00–1.26) and pulmonary contusion (odds ratio 2.25, 95% confidence interval 1.03–4.91) predicted hemothorax management failure requiring secondary intervention in the observation subgroup, whereas chest injury severity (odds ratio 1.58; 95% confidence interval 1.17–2.12) and initial hemothorax volume evacuated (odds ratio 1.10 per 100 mL; 95% confidence interval 1.05–1.16) predicted hemothorax management failure requiring secondary intervention after pleural drainage. After adjusting for patient characteristics in the logistic regression model for hemothorax management failure requiring secondary intervention, patients treated at high hemothorax management failure requiring secondary intervention trauma centers were 6 times more likely to undergo an intervention after initial hemothorax management failure than patients treated in low hemothorax management failure requiring secondary intervention trauma centers (odds ratio 6.18, 95% confidence interval 3.41–11.21). Conclusion Failure of initial management of traumatic hemothorax is common and highly variable across trauma centers. Assessing patient selection for a given management strategy and center-level practices represent opportunities to improve outcomes from traumatic hemothorax.