Predictors of Abnormal CT Findings

Continuous and categorical variables were presented as median (IQR) and n (%), respectively. SPSS for Windows software package, version 20.0 (SPSS Inc, Chicago, IL) was used. Mann-Whitney U test or χ² test was used to compare differences between the normal CT group and the abnormal CT group. To explore the risk factors associated with abnormal CT findings, univariate and multivariate logistic regression models were used. By referring to the relevant literature, we selected age, gender, body temperature, underlying diseases, symptoms, white blood cell count, neutrophil count, lymphocyte count, and C-reactive protein as variables for stepwise logistic regression analysis (20 (link)). At the same time, parameters of odds ratios that were difficult to estimate were excluded due to the small number of cases as well as the parameters that had collinearity with the underlying disease. A P-value < 0.05 was considered statistically significant.