However, the regression models using age, years in present position and work place only explained a relatively small part of competency, which indicates a multidimensional situation including other factors of importance.
We also ran models using age at baseline applying Cox proportional hazard models which, again, did not alter the results.
The goodness of fit for different models using age, period and cohort effects was evaluated by calculating AIC.
As a sensitivity analysis, we also fitted the models using age, BMI, age at menarche, and physical activity as continuous variables.
Others have found that age is a more important predictor, and comorbidity offers little prediction benefit compared to models using age alone [ 4- 6].
We estimated the overall incidence of mortality and studied risk factors associated with death using Cox proportional hazard models using age as the underlying time variable.
Instead, we ran conditional logistic regression models using age in months as a matching variable, rather than individual case-to-control matchings.
Risks of incident CHD were quantified as hazard ratios (HRs) from Cox proportional hazards models using age as time scale [ 16].
Because of the known association between aging and methylation, we also repeated the models using age, as an additional covariate to saturate the model for an age effect and found no difference in our results (data not shown).
aMultivariable Cox proportional hazards regression models using age (in days) as the time metric stratified by age (in years) and adjusted for race (white, black), study site (Detroit, Los Angeles), tumor stage (localized, non-localized).
