We found that ESR1 was only a borderline significant predictor of MD in models with stepwise exclusion of covariates.
With adjustment for the predefined predictors, on average, A1C remained a borderline significant predictor for heart failure.
In multivariate analysis with gender and conditioning intensity as predictors of outcome, female gender remained a borderline significant predictor of longer RFS, independent of the conditioning intensity (HR=0.30, 95% CI: 0.09 1.07, P=0.06).
Similar to estrogen use, pregnancy was a significant or borderline significant predictor of T4 but not TSH.
In multiple regression analysis including anti-CCP antibodies and survivin, the positivity for survivin remained a borderline significant predictor for disease development (OR = 2.26, 95% CI = 0.96 to 5.32, P = 0.061).
In the second multivariate model all (borderline) significant predictors from the univariate analyses were entered, except progressive joint damage in the first year, adjusted for possible confounders (Table 4).
In the first multivariate model, all (borderline) significant predictors from the univariate analyses were entered, adjusted for possible confounders (Table 3).
The entry p-value for forward elimination models was 0.10 and the retaining p-value for backward elimination was 0.10 in order to identify significant and borderline significant predictors.
They were then entered in a third multivariate regression analysis together with all (borderline) significant predictors from the univariate analyses and possible confounders.
Age was a borderline significant (P < 0.067) factor with a negative relationship with satisfaction, however it was a significant predictor in the regression model, in accordance with results reported elsewhere [ 41].
