This value was chosen as univariate P values ≥0.20 were considered unlikely to contribute to a model containing other potential determinants of suboptimal iron status.
In these regressions the effect of each factor (present versus absent) was calculated in a model containing all potential predictors and confounders.
In multivariate analysis, young age (≤ 40 years) remained a significant predictor of relapse when entered into a model containing all potential demographic, pathologic and immunohistochemical variables (Table 4. Hazard Ratio (HR), 1.5; confidence interval, 1.0 – 2.2; p = 0.0352).
In multivariate analysis, young age (<35 years) remained a significant predictor of recurrence when entered into a model containing all potential demographic, pathologic and immunohistochemical variables (Table 3. Hazard Ratio (HR), 1.7; 95% confidence interval, 1.1 2.6; p = 0.010).
Regional-scale habitat models were used to predict, based on coarse-scale geographic information system (GIS) data, spatial distribution of areas containing potential habitat of rare plant species and the probability of encountering potential rare plant habitats.
The backward model selection procedure was initiated with the saturated model containing all the potential confounding factors (Rothman & Greenland, 1998).
The aim was not to develop a complex Wnt signaling model containing all the potential protein-protein interactions and transport, synthesis and degradation processes.
A meta-regression model containing both of these covariates identified as potential effect modifiers could not be estimated due to perfect collinearity.
A backward elimination procedure was used to determine the final model containing only the infection group, together with significant covariates and potential confounders.
Before training regression models using the custom PBM data, we filtered out any sequence that contained potential TF binding sites in the regions flanking the CACGTG core, to ensure that each sequence contains one and only one TF binding site.
