Defining the role of chest radiography in determining candidacy for pectus excavatum repair.

Journal Article (Journal Article)

OBJECTIVE: The Haller index (HI), derived from computed tomography (CT) of the chest, remains the standard for determining pectus excavatum (PE) severity. The ability to accurately determine PE severity from chest radiography (CXR) may provide substantial benefits. This study proposes to establish data-driven criteria for the use of CXR as a diagnostic modality in the preoperative evaluation of patients with PE. METHODS: A database of 77 patients with PE receiving preoperative CXR and CT was created. Haller indices were obtained from CT scans. Two cardiothoracic surgeons, blinded to the CT HI scores, independently calculated CXR HIs. Interrater reliability, correlations, sensitivity, specificity, and accuracy were calculated. RESULTS: A significant correlation was found between CXR HI estimates of the two observers, r = 0.84. Computed tomography HIs were used as the standard for identifying patients meeting surgical criteria (CT HI ≥3.2). From this, sensitivities, specificities, and accuracies were calculated for the corresponding CXR HIs determined by each observer. Observer CXR HI estimates had a sensitivity of 0.94 and 0.92, specificity of 0.77 and 0.42, and accuracy of 0.88 and 0.75. CONCLUSIONS: We found a strong correlation between HIs calculated with each modality. In addition, CXR HIs demonstrated good interrater reliability. Although the sensitivity of CXR in diagnosing severe PE (CT HI ≥3.2) was high, specificity was less convincing. However, when using a cutoff of CXR HIs of 3.75 or greater, combined specificity was quite high (0.96). We recommend replacing CT with CXR in the preoperative evaluation for PE. Computed tomography should be used as a confirmatory test for CXR HIs between 3.2 and 3.75.

Full Text

Duke Authors

Cited Authors

  • Poston, PM; Patel, SS; Rajput, M; Rossi, NO; Davis, JE; Turek, JW

Published Date

  • 2014

Published In

Volume / Issue

  • 9 / 2

Start / End Page

  • 117 - 121

PubMed ID

  • 24553054

Electronic International Standard Serial Number (EISSN)

  • 1559-0879

Digital Object Identifier (DOI)

  • 10.1097/IMI.0000000000000048


  • eng

Conference Location

  • United States