These predictors included certain geodemographic characteristics (male sex, southern residents), health-risk behaviours (smoking and history of drunk driving) and adverse health diagnoses (depression).
These predictors included positive and negative feedback during each of the four experimental conditions: low versus high actual difficulty crossed with "EASY" versus "HARD" difficulty label.
These predictors included: locations expressed as easting and as northing, predicted 3-day average HSI, year, and a yes/ no binary variable that was an indicator for the presence of a heat wave.
These predictors included BMI and arthritis which may have little to no effect on fracture risk; older age which may be associated with a small increase in fracture risk; and BMD and glucocorticoid use which may be associated with moderate to large increases in fracture risk.
These clinical predictors include demographic variables, (comorbid) psychiatric symptoms, cognitive and psychological patient characteristics, and the quality of the therapeutic alliance.
These shared predictors include: maladaptive beliefs and attitudes supporting a criminal lifestyle [ 12, 13] and poor social problem-solving ability [ 14- 17].
These inconsistent predictors include some individual level variables (the woman's age at the birth of the last child, ethnicity, the notion of ideal family size, and approval of family planning), a community-level variable (prevalence of the small family norm in the community), and a state-level variable (ratio of PHC to the population).
Therefore the predictors included for each outcome variable are different.
Then the predictors included were tested for significance using stepwise selection.
Interestingly, four of the predictors included in the screening instrument were not specific for ICU patients.
