Reliability of a standardized protocol to calculate cross-sectional chest area and severity indices to evaluate pectus excavatum

Abstract 

Purpose

In evaluating the impact of surgical repair of pectus excavatum, the Haller index developed for preoperative decision-making purposes may be inadequate to quantify postoperative changes in shape of the chest. Individual patients may also have chest characteristics that impact the success of repair, many of which would be unlikely to be measured by the Haller index alone. We have developed a protocol that measures the cross-sectional chest area and the asymmetry index along with the Haller index to more completely quantify the nature of the deformity. The purpose of this study was to determine the reliability of this protocol in the interpretation of chest computed tomography images from multiple sites. The protocol was developed as part of a multicenter study of clinical outcomes after surgical repair of pectus excavatum.

Methods

Two radiologists independently selected 5 images from each of 32 computed tomography scans from multicenter study participants according to the protocol. A digitizer was used to measure the diameters and cross-sectional areas of the images selected; these results were used to calculate the Haller and asymmetry indices. The protocol was tested for intradigitizer and interradiologist reliability. Using the Haller and asymmetry indices, we also assessed agreement between radiologists classifying patients as abnormal.

Results

Agreement was uniformly high for all comparisons (all Lin's concordance coefficients >0.99 and all Cohen's κ's >0.85, all agreement on classification of patients >95%) indicating almost perfect agreement. Disagreement on classification of patients using the Haller and asymmetry index was at the cut points for abnormality.

Conclusion

The protocol was found to be a highly reliable method for deriving the cross-sectional area of the chest and the Haller and asymmetry indices and for classifying patients for surgical eligibility. Borderline cases should be examined carefully to determine the appropriateness of surgical intervention. Cross-sectional area can be measured reliably using this protocol and thus may be useful in quantifying the success of surgical intervention.