Skip to main content
. 2019 Jul 8;17:124. doi: 10.1186/s12916-019-1359-9

Table 1.

Univariable and multivariable logistic regression models for risk of mortality in severe malaria (case definition includes the > 2500 parasites per μl threshold)

Predictors Univariable Multivariable (all variables) Multivariable (restricted) Interaction model
Odds ratio p value Odds ratio p value Odds ratio p value Odds ratio p value
Time (1989–2003) Reference
Time (2004–2008) .7 (.55 to .9) 0.005 NA NA .68 (.48 to .97) 0.03 .8 (.53 to 1.2) 0.28
Time (2009–2016) 1.09 (.88 to 1.37) 0.43 .75 (.46 to 1.24) 0.26 1.05 (.68 to 1.6) 0.83 .8 (.48 to 1.32) 0.38
Acidosis 7.07 (5.77 to 8.67) < 0.0001 3.31 (2.09 to 5.25) < 0.0001 2.75 (2.01 to 3.76) < 0.0001 2.83 (2.06 to 3.89) < 0.0001
Age (years) .96 (.93 to .99) 0.02 1.08 (.98 to 1.18) 0.11 1.05 (.98 to 1.12) 0.19 1.05 (.98 to 1.13) 0.16
Cerebral 7.63 (6.59 to 8.83) < 0.0001 3.81 (2.54 to 5.74) < 0.0001 4.21 (3.17 to 5.57) < 0.0001 4.38 (3.3 to 5.82) < 0.0001
Compensated shock 4.76 (3.4 to 6.66) < 0.0001 1.03 (.47 to 2.23) 0.94 NA NA NA NA
Hyperparasitaemia 1.19 (1.02 to 1.38) 0.03 .89 (.58 to 1.37) 0.6 NA NA .75 (.51 to 1.09) 0.13
Hypoglycaemia 8.81 (6.98 to 11.1) < 0.0001 1.93 (1.17 to 3.2) 0.01 3 (2.17 to 4.14) < 0.0001 3.2 (2.3 to 4.45) < 0.0001
Kidney injury 6.09 (4.43 to 8.37) < 0.0001 2.45 (1.29 to 4.63) 0.006 2.76 (1.79 to 4.25) < 0.0001 2.64 (1.71 to 4.09) < 0.0001
Mx convulsions 2.21 (1.67 to 2.92) < 0.0001 .92 (.56 to 1.51) 0.75 NA NA NA NA
Jaundice 2.43 (1.49 to 3.98) 0.0004 .87 (.31 to 2.49) 0.8 NA NA NA NA
Prostration 1.05 (.71 to 1.56) 0.81 NA NA NA NA NA NA
Resp. distress 9.41 (7.69 to 11.5) < 0.0001 2.23 (1.45 to 3.42) 0.0002 1.92 (1.42 to 2.6) < 0.0001 1.95 (1.44 to 2.65) < 0.0001
Severe anaemia 2.24 (1.93 to 2.6) < 0.0001 1.25 (.78 to 1.99) 0.35 1.1 (.78 to 1.54) 0.59 1.08 (.77 to 1.52) 0.66
Time (2004–2008) * Hyperparasitaemia NA NA NA NA .63 (.29 to 1.34) 0.23
Time (2009–2016) * Hyperparasitaemia NA NA NA NA 2.53 (1.07 to 5.97) 0.03

Footnote: Odds ratios (ORs) are shown with 95% confidence intervals in brackets. Interaction terms are not relevant to univariable models. “NA” is shown for cells where the model was not applicable, and “-” is shown for cells with insufficient data (occurring where multiple seizures and other covariates were not collected between 2004 and 2009). Multivariable (restricted) refers to a model where non-significant predictors were excluded from the model