Table 3.

Author, date, country	Patient group	Study type	Outcomes	Key results	Study weaknesses
Rothlin et al, 1993, Switzerland	Adults (aged 15–88 years) with blunt thoracic or abdominal injuries. Ultrasound scan by surgeons compared to the patients' clinical outcome and other diagnostic techniques	Prospective study	Clinical utility of ultrasound to detect haemothorax	Sensitivity 81%	Poor gold standard.Not performed by emergency physicians.Some of the beginner surgeons initially forgot to check for haemothorax, accounting for 5 of the 11 cases of false negative reports by ultrasound. Some of the effusions developed only after the first 24 h and therefore would not have been visible on the initial ultrasound scan
Ma et al, 1995, USA	245 adult (18 years +) patients presenting to the emergency department with major blunt or penetrating torso trauma. Diagnosis with ultrasound was compared to a combination of results of other diagnostic techniques including CT scan, supine chest x ray, formal two‐dimensional echocardiography or tube thoracostomy	Diagnostic cohort	Clinical utility of rapid ultrasound assessment at detecting the presence of haemothorax	Sensitivity 96%, specificity 100%, accuracy 99%	Composite gold standard
Ma and Mateer, 1997, USA	240 adults (aged 18 years+) presenting to the emergency department with a major blunt or penetrating torso trauma. Accuracy of ultrasound compared with that of initial plain supine chest x ray for the detection of haemothorax using a combination of CT and tube thoracostomy as the gold standard	Retrospective analysis of a prior prospective study	Clinical utility of ultrasound vs initial plain supine chest x ray for the detection of haemothorax	Sensitivity 96.2% vs 96.2%,specificity 100% vs 100%,accuracy 99.6% vs 99.6%	The same patients as reference 2
Sisley et al, 1998, USA	Patients with suspected blunt or penetrating torso injury who required a chest radiograph for a complete evaluation	Prospective study	Clinical utility of ultrasound vs supine chest x ray at detecting the presence of haemothorax	Sensitivity 97.5% vs 92.5%,specificity 99.7% vs 99.7%,PPV 97.5% vs 97.4%,NPV 99.7% vs 99.1%	The results of ultrasound were compared to those of supine chest x ray rather than them both being compared to an independent “gold standard”. Emergency physicians did not perform the ultrasound scan
Abboud and Kendall, 2003, USA	Blunt trauma patients who underwent CT scan of their chest as part of their clinical assessment	Prospective study	Clinical utility of ultrasound	Sensitivity 12.5%, specificity 98.4%	None of the haemothoraces detected by CT but not on ultrasound were large enough to be clinically relevant. The interval between the ultrasound scan and the CT scan was up to 4 h
Brooks et al, 2004, UK	Patients triaged to the resuscitation room with thoracic trauma (blunt or penetrating).Ultrasound findings were compared against a combination of other diagnostic methods including supine chest x ray, intercostal drain, CT or tube thoracotomy	Prospective study	Clinical utility of ultrasound at detecting the presence of haemothorax	Sensitivity 92%,specificity 100%,accuracy 98%,PPV 100%,NPV 98%	Composite gold standard

CT, computed tomography; NPV, negative predictive value; PPV, positive predictive value.