CT Diagnosis of Rib Fractures and the Prediction of Acute Respiratory Failure

Abstract

The number of rib fractures has been reported to correlate with mortality after blunt chest trauma. These reports, however, predate routine truncal helical computed tomographic (CT) scanning and their conclusions are based on data derived from plain chest radiographs (CXR). CT scan provides better anatomic definition of chest injuries than plain CXR, and we hypothesized CT evaluation of rib fracture number and patterns would provide a better prediction of respiratory failure and mortality after chest injury than the data derived from the initial CXR. The charts on all patients of 16 years or older with one or more rib fractures after blunt trauma admitted from January 2003 through December 2005 were reviewed. Both the initial CXR and the helical CT scans were systematically re-read for the number and location of rib fractures and presence of pulmonary contusions. Anatomic fracture location (anterior, posterior, lateral) was determined using a standardized template. Outcomes data included pneumonia, respiratory failure (>/=3 ventilator days), need for trachestomy, and mortality. Logistic regression was performed to identify factors that predicted pulmonary morbidity. Three hundred and eighty eight patients had >/=1 rib fracture. The mean (+/-standard deviation) age was 44 +/- 18. injury severity score was 21 +/- 11. Mortality was 6% (22 of 388). Sixty-three (16%) patients developed respiratory failure. The mean number of rib fractures per patient was four (range, 1-23); 21% of patients had one rib fracture and 17% had six or more fractures. 208 (54%) of the initial CXRs were read as having no rib fractures. The mean number of rib fractures per patient in this group was 3.1 (CI95 2.9-3.2). In 43% (179 of 388) of patients, the CT radiology report incorrectly identified the number and location of the fractured ribs. Of these reports, 72% (129 of 179) differed from the prospective review by more than one fracture. The number of fractures was higher in patients who died (7 +/- 5 vs. 4 +/- 3; p = 0.02) and in those developing respiratory failure (6 +/- 4 vs. 3 +/- 3; p = 0.02). Any rib fracture or pulmonary contusion visible on the initial plain CXR significantly increased the incidence of pulmonary morbidity or mortality. CT determination of fracture location had no effect on respiratory failure, pneumonia, or mortality when fractures were confined to one anatomic location. The presence of rib fracture in more than anatomic region doubled the incidence of respiratory failure (24% vs. 12%; p = 0.002) but had no effect on mortality. Logistic regression identified only injury severity score and presence of a parenchymal injury on plain CXR as independent predictors of subsequent respiratory failure. Rib fracture mortality was lower than that in the previously published studies and is likely reflect the increased sensitivity of CT scan in diagnosing rib fractures. Screening CXRs miss rib fractures more than 50% of the time. Radiology reports are often not sufficiently descriptive or are incomplete with respect to the number and location fracture and reliance on these data will lead to erroneous conclusions. Using CT scanning, only the finding of rib fractures in multiple locations was associated with increased incidence of respiratory failure. In contrast, the presence of any parenchymal injury or visible rib fracture on the screening CXR significantly increases the risk for subsequent pulmonary morbidity (odds ratio, 3.8; CI95, 2.2-6.6). Although truncal CT scanning markedly improved the diagnosis and delineation of rib fractures, the screening CXR was a better predictor of subsequent pulmonary morbidity and mortality.