Skip to main content
. 2018 Jul 31;17:278. doi: 10.1186/s12936-018-2430-2

Table 2.

Univariable analysis of predictors for the development of sepsis in patients admitted for severe malaria

Characteristic Category N OR 95% CI p value
1 Age Age (years) 1187 1.00 0.98, 1.01 0.749
2 Sex Sexa
 Female 288 0.37 0.24, 0.58 < 0.001
 Male 899 1.00
3 Cerebral malaria Coma (GCS < 11 or BCS < 3)
 Yes 476 1.47 0.95, 2.28 0.088
 No 711 1.00
Convulsions
 Yes 113 1.28 0.64, 2.54 0.490
 No 1074 1.00
Prostrationa
 Yes 435 0.61 0.37, 0.99 0.050
 No 752 1.00
4 Haemodynamic shock Systolic BP (mm/Hg)a,b 1177 0.98 0.97, 0.99 0.001
Diastolic BP (mm/Hg)a,b 1176 0.98 0.96, 0.99 0.002
MAP (mm/Hg) 1176 0.98 0.96, 0.99 0.001
Clinical shocka
 Yes 137 2.74 1.62, 4.65 < 0.001
 No 1050 1.00
5 Severe anaemia Haemoglobin (g/dl)a 1112 0.94 0.87, 1.00 0.052
Haematocrit (%) 1100 0.98 0.96, 1.00 0.080
6 Renal failure BUN level (mg/dl)a 1127 1.01 1.01, 1.01 < 0.001
Serum potassium (mg/dl) 1076 1.23 0.94, 1.61 0.161
7 Jaundice Jaundicea
 Yes 624 1.85 1.17, 2.94 0.009
 No 563 1.00
8 Hypoglycaemia Glycaemia (mg/dl) 1146 1.00 0.99, 1.00 0.438
9 Hyperpyrexia Temperature (°C) 1185 0.96 0.79, 1.16 0.655
10 Acidosis Venous pHa,b 1068 0.09 0.02, 0.40 0.002
Base excess (mmol/l)a 1075 0.94 0.92, 0.97 <0.001
Total COa,b2 1071 0.92 0.89, 0.96 < 0.001
Anion gap (mmol/l)a 1050 1.04 1.01, 1.06 0.002
11 Hyperparasitaemia Parasite count (/μl)a 1187 2.57 1.57, 4.19 < 0.001
Percentage parasitaemia (%)a,b 1187 1.04 1.02, 1.06 < 0.001
12 Respiratory distress Respiratory rate (cycles/min)a 1185 1.02 1.01, 1.05 0.003
Respiratory distressa
 Yes 117 3.14 1.82, 5.39 < 0.001
 No 1070 1.00

BCS Blantyre coma score, GCS Glasgow coma score, BP blood pressure, MAP mean arterial blood pressure, OR odds ratio, CI confidence interval, BUN blood urea nitrogen, mm/Hg millimetres of mercury, mmol/l millimoles per litre, g/dl grams per decilitre, mg/dl milligrams per decilitre, °C degrees centigrade; μl microlitre, min minute

aSignificant variables on univariate analysis

bVariables omitted from logistic regression model due to multicollinearity