Matsumoto S, Sekine K, Funabiki T, et al. Diagnostic accuracy of oblique chest radiograph for occult pneumothorax: comparison with ultrasonography. World J Emerg Surg. 2016 Jan 13;11:5.
Single-centre, observational, cohort study.
Included: Adult (≥ 18 years) victims of blunt trauma with clinically suspected occult pneumothorax defined as an abnormality (e.g., rib fracture, permeability decay of lung field) on APXR or physical abnormalities (e.g., chest pain, bruise, subcutaneous emphysema) but without overt PTX on APXR.
Excluded: Overt PTX; refractory shock; cardiac arrest or requiring other immediate invasive interventions; transferred from another hospital.
Oblique angle chest X-ray (OXR) consisting of a mobile APXR directed over the pleural interface perpendicular to a film cassette set at 45° against a horizontal plane for each hemithorax.
All patients underwent CT scan, which was considered the gold standard for diagnosing occult PTX.
Primary: Diagnostic accuracy of OXR.
Secondary: Diagnostic accuracy of LUS; agreement between OXR and LUS.
N = 159 (318 thoraces): 70 occult PTX/318 thoraces as defined by the gold standard (CT scan).
|SS||Agreement between OXR and LUS||93.3% (k = 0.804, p < 0.001)|
|OXR||PPV = 95.6 (95% CI: 0.86 to 0.99)|
|NPV = 90.1 (95% CI: 0.87 to 0.91)|
|Sensitivity = 61.4 (95% CI: 0.56 to 0.64)|
|Specificity = 99.2 (95% CI: 0.98 to 1.00)|
|LUS||PPV = 93.6 (95% CI: 0.84 to 0.98)|
|NPV = 90.4 (95% CI: 0.89 to 0.91)|
|Sensitivity = 62.9 (95% CI: 0.57 to 0.66)|
|Specificity = 98.8 (95% CI: 0.97 to 1.00)|
AUC: area under the (receiver operating characteristic [ROC]) curve; CI = confidence interval; LR = likelihood ratio; Sig. = statistically significant; SS = statistically significant.
Risk of Bias Assessment
|1||The patients were representative of those likely to undergo testing in the ED.||Maybe||Maybe||Maybe|
|2||The patients were enrolled consecutively or in a way to ensure a representative sample.||Maybe||Yes||Yes|
|3||All patients underwent the same diagnostic evaluation.||Yes||Yes||Yes|
|4||All tests were conducted within similar time frames to preclude changes in disease status.||Maybe||Maybe||Maybe|
|5||The reference standard criteria for the candidate diagnoses are explicit and reproducible.||Maybe||Maybe||Maybe|
|6||The reference standard was applied regardless of and blinded to the index test result.||No||No||No|
|7||The assignment of the candidate diagnoses was explicit and reproducible.||Yes||Yes||Yes|
|8||Most (> 80%) patients received a diagnosis.||Yes||Yes||Yes|
|9||Undiagnosed patients received adequate clinical follow-up.||Yes||Yes||Yes|
|10||The estimates of disease probability are clinically significant.||Yes||Yes||Yes|
Funding & Conflicts of Interest
Conflicts of Interest
Potential Threats to Validity
No a priori sample size determination.
The sampling method is unclear. Since patients with overt PTX were excluded, selection bias may have occurred, whereby less severe patients were included in the study. Also, the number of patients with OPX was small due to narrow entry criteria and may not represent all cases of OPX in the ED.
Operator dependence of LUS; PTX can develop and worsen between the OXR, LUS and CT. Since the supervising attending physicians judging the CT scans (gold standard) were not blinded to the results of the OXR or LUS results, diagnostic review bias is possible.
There was incorporation bias in the interpretation of both the index and reference tests.
Worster A; Brown J; Hunter B.
Competing Interest Disclosure
Dr. Hunter - No conflicts of interest (ICMJE)
Dr. Patel - No conflicts of interest (ICMJE)