All statistical analyses were fulfilled using IBM SPSS Statistics
software (Version 24) (IBM SPSS Statistics, Armonk, USA). The normality
of variables was established using the Kolmogorov- Smirnov test. DII
scores were categorized into quartiles. The differences in variables
across the quartiles of DII were detected using One-way ANOVA. A
comparison of the mean intake of nutrients with global mean intake per
day was performed using one-sample t-test. To compare the categorical
variables across quartiles, the Chi-squared test was applied.
Linear regression analysis was
used in 3 models (Model 0, linear regression analysis without
adjustment; Model I, linear regression analysis with adjustment for
energy intake; Model II, linear regression analysis with correction for
age, energy intake, physical activity, and education) for determination
of the Association between DII score as the independent variable with
anthropometric variables and PSQI continuous score as the dependent
variables. Moreover, the odds ratio (OR) and 95% confidence intervals
(CIs) were estimated using multivariable logistic regression in 3
different models, including Model 0: unadjusted, Model 1: adjusted for
energy intake and Model 2: adjustment for age, education, physical
activity, energy intake. Also, p-values for linear trends was used in
logistic regression to compare the OR between different quarters (Q1to
Q4). The lower quartile of DII score (Q1) was used as the reference
category. A p-value of less than 0.05 was regarded to be statistically
significant.