Statistical analysis
All statistical analyses were performed using IBM SPSS Statistics 25.0 (IBM Corp., NY, USA). Continuous data were expressed as means with standard deviations or medians with interquartile ranges (depending on data distribution). The Mann-Whitney U test was used to compare continuous variables and a post-hoc analysis with the Dunn-test and the chi-square test were used to compare categorical variables. After adjustment for gender and body mass index (BMI), correlations between lipid level and the SCORAD score were evaluated by partial correlation analysis. Generalized linear models with the logit function or gamma function (depending on data distribution) were used to determine the odds ratios or beta values for the relationship between lipid parameters and atopic dermatitis or SCORAD score.
Propensity scores were used to mitigate biases arising from nonrandom treatment assignment. Thus, for each pairwise comparison, matched cohorts were constructed using 1:4 propensity score matching with a caliper of 0.25 on the logit scale.23,24 A Cox proportional hazard model was then fit to the matched cohorts using the Cohort Method R package (https://github.com/OHDSI/CohortMethod), and the hazard ratio (HR) and 95% confidence interval (CI) was calculated for each outcome of interest, a procedure that reduces residual bias and variance.25