The multivariate model excluding participants with a clinical diagnosis is shown in Table 4.
Even when simplifying the multivariate model, excluding margins and patient age, the number of significantly prognostic biomarkers did not increase (data not shown).
The final multivariate model excluding the early chronic stage symptoms kept the following as independent prognostic factors: CHIKV infection, comorbidity and acute stage depressed mood for both persistent arthralgia and arthritis.
In the multivariate model excluding clinical features based on examination findings, the feature most strongly associated with gut feeling was a history of convulsions (odds ratio 80.5, 95% confidence interval 6.2 to 1051).
In the multivariate model excluding patients with major risk factors for CVD, age, male sex and nonwhite race were associated with subclinical atherosclerosis in the <40 y group, while in the ≥40 y group, age, HIV viral load >10,000 copies and the use of NNRTI ORR: 7.60, 95% CI: 1.61-35.8) or PI ≥5 years (OR: 3.62, 95% CI: 0.48-26.8) were associated with subclinical atherosclerosis.
31 Backward-stepwise elimination was then used to generate a minimum adequate multivariate model; excluded covariates (p>0.05) were retested in the minimal model to confirm lack of association.
Although both larger self-reported network size and shorter length of an individual's referral chain were significantly associated with prevalence in bivariate analysis, the multivariate model excluded the length of the referral chain and did not eliminate any of the four significant individual-level variables.
The ex-smokers showed a higher risk of incident metabolic syndrome than the sustained smokers when using the multivariate regression models excluding weight change, but this finding was no longer significant after including weight changes.
When we included all 217 subjects of Cohort 1 in the multivariate models, by excluding the covariate history of hypertension from the analysis, the independent associations between telomere length and donor age (p = 0.003), history of cardiovascular events (p = 0.02), and gender (p = 0.002) remained statistically significant.
