Stata v 11

We first compared cases and controls. We then compared groups of patients with and without major interventions. We assessed the relevance of WHO severity criteria by determining whether they were associated with at least one major intervention being performed. We also assessed the relevance of anaemia and thrombocytopenia thresholds by determining whether they were associated with at least one major intervention being performed and/or with any other WHO severity criteria (excluding anaemia and blood product transfusion). A descriptive analysis was performed. Continuous variables were described with medians and IQRs (medians (first; third quartile)), or means and SDs (mean±SD) for variables with a Gaussian distribution. χ 2 test and Fisher's exact tests were used to analyse categorical variables and Wilcoxon rank sum tests were used for continuous data, as appropriate. We considered p values <0.05 to be statistically significant. If the observed rate for controls was 0, we calculated an approximate OR by adding 0.5 to each cell. 11 (link) Thrombocytopenia was analysed dichotomously after setting a relevant threshold by screening sensitivityspecificity pairs by receiver operating characteristics (ROC) curve analysis. Missing data were not imputed. Statistical analyses were performed with Stata V.11.2 software (STATA, College Station, Texas, USA).