Assessment of potential confounders when imaging pectus excavatum with chest radiography alone

Chest radiography (CXR) has emerged as an attractive alternative imaging option for objective pre-operative assessment of pectus excavatum (PE) with comparable accuracy, reduced cost, and less radiation exposure when compared to computed tomography (CT). This study asked whether image quality, scoliosis, and asymmetry of the PE deformity would decrease the accuracy of CXR as compared to CT.

A database of PE patients receiving preoperative CXR and CT was created, and Haller-indices (HI) and correction-indices (CI) were calculated using each imaging modality. Each potential confounding variable were analyzed using Spearman correlations the Fisher r-to-z transformation test.

The database was comprised of 77 patients. Image quality, scoliosis and the 'eccentric type' of asymmetry did not demonstrate any significant worsening of measurement accuracy. However, the correlation coefficients for CIs for those with and without the 'unbalanced type' of asymmetry were 0.593 and 0.890, respectively, with a Fisher r-to-z of 2.16 (p=.031).

The accuracy of CXR-derived pectus indices remains quite favorable despite the heterogeneity from radiographic quality, scoliosis and chest wall asymmetry. Nonetheless, the unbalanced type of chest wall asymmetry did emerge as a significant confounder. As such, use of CXR alone in cases of gross chest wall asymmetry should be cautioned.