In addition to the density histogram, the a value (the negative slope) from the log-log relationship of hole size vs percentage of holes (with hole membership defined as voxels at — 950 HU, — 910 HU, or — 850 HU) was evaluated in this study because it has been shown also to represent severity of emphysema based on the concept that as emphysema progresses, there is a tendency of small holes to mechanically destabilize a lung region, such that small holes merge into larger holes rather than additional small holes accumulating. African-American patients were matched with white patients (1:1) according to gender, height (± 12 cm), age (± 3 years), number of cigarettes smoked per day (± 10 cigarettes), and age started smoking (± 3 years). When more than one match was identified, the match that was closest in age and closest in height was chosen. further
Categorical variables were compared using x2 and Fisher Exact Test. Continuous variables (age, FEV1, TLC, RV) were compared using two-way analysis of variance to evaluate differences between gender and race, and multiple pair-wise comparisons used the Dunn-Bonferroni adjustment to maintain an experiment-wise type I error < 0.05. Prior to analysis, continuous data were tested for normality using the Shapiro-Wilk test. If the data for the continuous dependent variables were significantly nonnormal, a “normalized-rank” transformation was applied to the data prior to analysis. If, for a given measure, distribution was not normal, a Wilcoxon rank-sum test was calculated.
Analysis of demographic and pulmonary function variables of the matched patients with their respective quantitative CT variables at different cutoffs (— 950 HU, — 910 HU, — 850 HU) was done using conditional logistic regression to take account of the matching. Means and SDs are presented for all continuous variables. The null hypothesis was that there was no difference in the proportions or measured parameters among race. Differences between group proportions and means (rejection of the null hypothesis) were considered significant if the probability of chance occurrence was < 0.05 using two-tailed tests.