Statistical analysis
Descriptive statistics were presented as number (percentage) for categorical variables; and as mean±standard deviation for quantitative variables. The logistic regression model was used to determine the associations between patient characteristics and UMDs in univariate and multivariable analyses. The following variables were tested by univariate analysis: age, sex, medical specialty, education level, living status, health insurance, type of admission, day of admission, admitting physician experience, patient’s understanding of preadmission medications, chronic diseases ( <2 vs ≥2 ) and medications on BPMH ( ≥5 drugs vs <5 drugs ). Variables with a significant relationship ( P <0.20 ) by univariate analysis were entered into the multivariable model. Analyses were performed by SPSS 20.0. P values < 0.05 were regarded as statistically significant.