Finally, fasting glucose levels were similarly associated with CKD or CVD (odds ratio 1.011, 95% CI 1.005-1.017, P < 0.001) when they were entered into the multivariate model instead of the HbA1c concentration.
Since this group was small, adding it to the multivariate model instead of 'Contact with the police or child welfare committee' did not change the results concerning the risk of convictions for violent crime and psychiatric diagnoses, we decided to use the broader construct in accordance with most violence risk assessment instruments.
The alternative multivariate model including monocytes, instead of neutrophils, yielded a similar result (likelihood ratio χ of 87.7 vs 86.6).
Furthermore, we further improved the method by using a multivariate linear regression model instead of a univariate model (Hong et al. 2004; Jin et al. 2004; Uno et al. 2005) to enable the adjustment of BMD/ BMDL for potential confounders.
We also noted a progressive increase in risk of all PCs with increasing height in univariate analysis, for this reason we also ran a multivariate model using height instead of BMI to examine any confounding effect, however, the associations of cholesterol and height with PC incidence remained consistent after adjusting for smoking and socioeconomic status (Table 3).
Where possible, results from adjusted multivariate models were extracted instead of single variable model results.
Nevertheless, alternative multivariate models considering TFR instead of contraceptive use revealed two opposite effects of TFR on mortality ratios: while displaying an inverse relationship with MMR and MMRAO, TFR showed a direct association with iAMR, explaining 17.2% of the difference in abortion-related mortality among states.
Results were also unchanged when we included BMI at age 18 years and weight change since age 18 years, instead of current BMI, in the multivariate model, or when we included time since last birth instead of parity and age at first birth (data not shown).
For simplicity, the FA model was implemented as a standard multivariate model using G with a FA structure, instead of an extended FA model discussed by Thompson et al. [ 6].
Using only univariate analysis to assess the relationship between religion and depression, instead of adopting a more comprehensive multivariate model that controls for potentially intervening covariates, may not accurately reflect the relationship between religion and depression.
However, when PMN-E was added as a continuous instead of a dichotomised variable to the multivariate model shown in Table 2, its contribution was not statistically significant (P=0.31).
