Multivariate analysis using ordinary least-squares linear regression was used to establish which patient characteristics were associated with quality of life after adjustment for other characteristics.Both stepwise and best-subsets analyses were used to create the final statistical models,and a P value of 0.05 provided a guideline for inclusion into the model. Correlation between the EQ5D utility scores and the Euroqol VAS scores was examined using Pearson and Spearman correlation coefficients. Associations between patient characteristics and each of the five EQ5D dimensions were analyzed using univariate analysis(Mann-Whitney U test, Kruskal-Wallis test,and Kendall’s T test,where appropriate) and stepwise binary and ordinal logistic regression analysis.
Factors associated with treatment satisfaction were also studied using linear regression analysis.
As a supplement to these analyses,we applied multilevel modeling to take into account the fact that the data collected originated from patients seen by 29 general practitioners(10).In this analysis,the general practitioner status is included as a random effect, and patient characteristics are included as fixed effects.This type of analysis has previously been used in diabetes research(11).
Correlations between the EQ5D utility scores, Euroqol VAS scores, and DTSQ scores were examined using both the Pearson and Spearman correlation methods.