We presented the prognostic model as a chart that cross tabulates these predictors with each of them recoded in several categories.
Table F in appendix 1 shows the unadjusted associations of each of the predictor variables with each of the outcomes.
The proportions of diarrhea-related visits predicted to be attributable to norovirus, rotavirus, background, or secular trends for each month were estimated by multiplying the monthly values for each predictor with its corresponding coefficient in the regression model.
For example, of the six articles that used a Cox model, only Terwee et al. [ 34] gave the beta (log hazard ratio) estimates for each included predictor, with others mainly reporting hazard ratios for each variable.
As fixed effects, we entered predictors and interaction terms of each predictor with the 'Protection' variable into the model.
The second metric we adopted, investigated the percent contribution of each predictor with respect to the full model.
VIF is a positive value representing the overall correlation of each predictor with all others in a model.
For all three experiments, we examined if time since water removal had a significant effect on isotope ratios using MANOVA, with δ2H and δ18O isotopes as responses and time as the predictor, with a block representing each experimental run.
Two-way interactions of each predictor with this classification factor were also included.
Firstly, Model 1 was fitted for each predictor with only a basic adjustment for gender and age.
