Correlation of six different cystic fibrosis chest radiograph scoring systems with clinical parameters

In past decades, several chest radiograph scoring systems for cystic fibrosis were developed. This study was performed to compare interobserver variability of six different radiograph scores and to correlate them with clinical parameters. Thirty chest radiographs of 30 patients with cystic fibrosis were scored according to Shwachman-Kulczycki scoring, Chrispin-Norman scoring, adjusted Chrispin-Norman scoring, Brasfield scoring, Wisconsin scoring, and the Northern scoring system by two independent observers. Data on clinical parameters such as lung function, nutritional status, and infectious exacerbation rate, obtained simultaneously with the chest radiograph, were reviewed. Interobserver variability was low (Pearson's correlation coefficients, 0.76-0.84; all P < 0.01), and scores had good limits of agreement (Bland and Altman). Correlation of radiograph score with clinical parameters was good for most pulmonary function test data (correlation coefficients from 0.72-0.78 for percent of forced expired volume in 1 sec (FEV(1)%) predicted and from 0.69-0.74 for FVC% predicted) and for infectious exacerbation rate (correlation coefficients from 0.68-0.73). All six radiograph scoring systems, especially the Chrispin-Norman score, showed a low interobserver variability and correlated well with lung function tests, especially FEV(1)% predicted and infectious exacerbation rate, and moderately with maximum work capacity and thoracic mobility.