- *The statistical inference in the logistic regression is presented as an odds ratio (OR) (95% confidence interval (CI)) and in the linear regression as a β-coefficient (β) (95% CI)
- Multivariate analyses for renal dysfunction (acute kidney injury (AKI); top third of table), liver dysfunction (maximal bilirubin; middle third of table), and mortality (bottom third of table) were performed with and without Acute Physiology and Chronic Health Evaluation (APACHE) IV (AP-IV) as an independent factor
- Regression analyses are shown between lactate quintiles (L), glucose quintiles (G), (glucose quintile – mean)2 (G–m)2, and the interaction term (lactate quintile × glucose quintile (L × G)), and outcome
- In order to accommodate for the known U-shaped relationship of glucose with outcome, the squared factor (glucose quintile – mean glucose quintile)2 is used. The term (lactate quintile × glucose quintile) was used to verify an interaction between lactate and glucose
- Note the for both for AKI and mortality, the addition of the squared factor and of the interaction term improved the contribution of glucose in the multivariate model
-
NS not significant