Table 5.
Adjusted models explaining mortality within 30 days in the ICU
| Multivariate explanatory models of mortality in the overall cohort | n# | OR (95% CI) adjusted* | p value § | BIC ‡ |
|---|---|---|---|---|
| Highest glucose level at admission | 833 | 1.004 (1.002–1.006) | < 0.001 | 847.026 |
| COVID-19 (ref: non-COVID-19) | 2.990 (1.767–5.060) | < 0.001 | ||
| History of diabetes (ref: no history) | 0.658 (0.435–0.995) | 0.048 | ||
| Age | 1.026 (1.013–1.038) | < 0.001 | ||
| APACHE II score | 1.057 (1.032–1.082) | < 0.001 | ||
| SOFA score in the first 24 h | 1.299 (1.201–1.405) | < 0.001 | ||
| CRP level in the first 24 h | 1.004 (1.002–1.006) | < 0.001 | ||
| Highest glucose level during ICU stay | 837 | 1.003 (1.000–1.005) | 0.021 | 519.168 |
| COVID-19 (ref: non-COVID-19) | 2.655 (1.363–5.17) | 0.004 | ||
| History of diabetes (ref: no history) | 0.816 (0.458–1.457) | 0.493 | ||
| Age | 1.020 (1.003–1.037) | 0.023 | ||
| APACHE II score | 0.997 (0.966–1.028) | 0.847 | ||
| Nosocomial infection (ref: no infection) | 0.976 (0.558–1.707) | 0.933 | ||
| Mean SOFA score | 2.327 (1.993–2.717) | < 0.001 | ||
| Mean CRP level | 1.010 (1.006–1.013) | < 0.001 | ||
| Mean glucose level during ICU stay | 837 | 1.011 (1.006–1.016) | < 0.001 | 502.875 |
| COVID-19 (ref: non-COVID-19) | 2.665 (1.336–5.315) | 0.005 | ||
| History of diabetes (ref: no history) | 0.566 (0.312–1.029) | 0.062 | ||
| Age | 1.024 (1.006–1.041) | 0.009 | ||
| APACHE II score | 0.993 (0.962–1.026) | 0.684 | ||
| Nosocomial infection (ref: absent) | 1.080 (0.630–1.851) | 0.779 | ||
| Mean SOFA score | 2.348 (2.014–2.737) | < 0.001 | ||
| Mean CRP level | 1.01 (1.006–1.014) | < 0.001 | ||
| % of days with hyperglycemia (≥ 180 mg/dL) | 837 | 1.017 (1.010–1.025) | < 0.001 | 504.526 |
| COVID-19 (ref: non-COVID-19) | 2.350 (1.192–4.633) | 0.014 | ||
| History of diabetes (ref: no history) | 0.555 (0.302–1.019) | 0.058 | ||
| Age | 1.020 (1.003–1.038) | 0.020 | ||
| APACHE II score | 0.993 (0.962–1.025) | 0.669 | ||
| Nosocomial infection (Ref: absent) | 1.010 (0.588–1.734) | 0.971 | ||
| Mean SOFA score | 2.370 (2.032–2.765) | < 0.001 | ||
| Mean CRP level | 1.010 (1.006–1.014) | < 0.001 | ||
| Mean glucose variability (CV %) during ICU stay | 762 | 1.032 (1.013–1.051) | 0.001 | 476.005 |
| COVID-19 (ref: non-COVID-19) | 3.126 (1.503–6.502) | 0.002 | ||
| History of diabetes (ref: no history) | 0.651 (0.347–1.222) | 0.181 | ||
| Age | 1.022 (1.004–1.04) | 0.015 | ||
| APACHE II score | 0.984 (0.951–1.017) | 0.329 | ||
| Nosocomial infection (ref: no infection) | 1.235 (0.716–2.13) | 0.449 | ||
| Mean SOFA score | 2.444 (2.079–2.873) | 0.000 | ||
| Mean CRP level | 1.010 (1.006–1.014) | 0.000 |
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; SOFA, Sequential Organ Failure Assessment; CRP, C-reactive protein (mg/L)
#Case numbers included in the model
* Odds ratio (OR) and 95% confidence interval (95% CI) for mortality within 30 days in the ICU, generalized linear models with multivariate binary logistic regression
§Wald test p value, with values < 0.05 indicating statistical significance
‡Bayesian information criterion (BIC) of the multivariate generalized linear model representing the quality of the model given its explanatory potential; the lower the BIC, the more the model is able to explain the dependent variable