1.Because the distributions of the radiological scores were skewed to the right and transformation did not correct this, ordinal regression was used as a confirmatory analysis method.
2.The analyses were carried out in a subgroup of never-smokers only.