Eur Radiol (2008) 18: 538–547 DOI 10.1007/s00330-007-0810-z
Anastasia Oikonomou John Tsanakas Elpis Hatziagorou Fotios Kirvassilis Stavros Efremidis Panos Prassopoulos
Received: 23 November 2006 Revised: 28 September 2007 Accepted: 17 October 2007 Published online: 27 November 2007 # European Society of Radiology 2007 A. Oikonomou (*) . P. Prassopoulos Department of Radiology, University Hospital of Alexandroupolis, Democritus University of Thrace, Dragana, 68100 Alexandroupolis, Greece e-mail:
[email protected] Tel.: +30-2551076803 Fax: +30-2551030473 J. Tsanakas . E. Hatziagorou . F. Kirvassilis 3d Department of Pediatrics, Ippokratio Hospital of Thessaloniki, Aristotle University of Thessaloniki, Konstantinoupoleos 49, 54642 Thessaloniki, Greece S. Efremidis Department of Radiology, University Hospital of Ioannina, University of Ioannina, Leoforos Panepistimiou 1270, 45500 Ioannina, Greece
CHEST
High resolution computed tomography of the chest in cystic fibrosis (CF): is simplification of scoring systems feasible?
Abstract The purpose of this study was to simplify HRCT scoring systems (SS) for CF by selecting representative HRCT parameters. Forty-two consecutive patients with CF underwent baseline and follow-up chest HRCT. Three radiologists evaluated 84 HRCTs employing five SS. “Simplified” HRCT SS were formed by selection of parameters exhibiting statistically significant relations with FEV1. Pulmonary function tests (PFTs) and nutrition (IBW%) were recorded. Regression analysis, Pearson correlation and T-test were used for statistical analysis. Three HRCT parameters were selected for the formation of “simplified” HRCT SS (severity of bronchiectasis, bronchial wall thickening, atelectasis-consolidation) using regression analysis. There was excellent correlation between each “simplified” and corresponding complete score (0.8924 –
Absent Subsegmental
Segmental/lobar
–
*The three parameters selected for the “simplified” Bhalla HRCT score are marked in bold
Selection of variables for simplified scores All the parameters of each complete score (baseline and follow-up) were compared with the corresponding FEV1 as dependent variable using stepwise regression analysis. The parameters that were not statistically significant were withdrawn. Among the parameters that remained statistically significant from all the regression analyses, only those that were common in all scores were selected to form the “simplified” HRCT scoring systems.
Wilcoxon-signed ranks test was applied to each one of HRCT parameters, which were included in the complete scores, to determine those solitary HRCT parameters that changed significantly between the baseline and follow-up examination. Also, baseline FEV1, FVC, FEF25–75 and IBW% were compared with the corresponding follow-up ones, using paired-samples T-test to investigate statistically significant deterioration. A p value of less than 0.05 was considered to be statistically significant. A software package (SPSS, version 11; SPSS, Chicago, IL) was used for statistical analysis.
Statistical analysis All distributions were checked for normality employing the Kolmogorov-Smirnov test. Each “simplified” HRCT score was correlated with its corresponding complete score. Moreover, each one of the “simplified” scores was correlated with the remaining four complete scores (baseline and follow-up). Each one of the five baseline and follow-up “simplified” scores was correlated to the others. Baseline and follow-up PFTs and IBW% were correlated with the corresponding complete scores and with the “simplified” scores. All the above-mentioned correlations were undertaken using Pearson correlation. The baseline complete and "simplified” HRCT scores were compared with the follow-up ones, using paired samples T-test to check if there was statistically significant deterioration of lung disease.
Results Three common parameters in all five complete HRCT scoring systems were proven using linear stepwise regression analysis to fit statistically significantly against FEV1: “severity of bronchiectasis” (p
