Variable
|
Cut-off
|
Number
a
|
Odds ratio
|
95% Confidence interval
|
P
|
|---|
White cell count | >11 × 109/L | 195 (30%) | 2.74 | 1.53 to 4.89 | 0.001 |
Plasma creatinine | >265 μmol/L | 176 (27%) | 2.40 | 1.3 to 4.43 | 0.005 |
Plasma lactate | >4Â mmol/L | 260 (41%) | 2.25 | 1.26 to 4.02 | 0.006 |
Total bilirubin | >51 μmol/L | 313 (57%) | 2.26 | 1.13 to 4.50 | 0.02 |
Platelet count | <50 × 109/L | 328 (51%) | 1.75 | 0.96 to 3.18 | 0.07 |
Reticulocyte count | <0.5% | 190 (37%) | 1.19 | 0.63 to 2.26 | 0.59 |
- aNumber (%) of patients with a value meeting the cut-off. The denominator for the calculation of the percentage was the number of patients in the dataset with a measurement available (not all patients had data for every variable). Multivariable logistic regression with death as the dependent variable; explanatory variables chosen on the basis of a significant association with outcome in univariate analysis (with P <0.01 allowing for multiple comparisons, see Table 2). The cut-offs were chosen based on the WHO severity criteria cut-off (for those that are WHO severity criteria: creatinine, lactate and bilirubin). For those that are not WHO severity criteria, the cut-offs were based on the reference range (white blood cell count and reticulocyte count) and previous studies (platelet count). Base deficit was not included in the model as there were fewer data for this variable; acidosis was represented by plasma lactate.
