Table 3.

Summary of articles with models predicting mortality in adult severe malaria

N	Study	Year	Period of participant recruitment	Country	Type of study	Sample size	Statistics used	Name of model	Method internal of validation	Age profiles	Sex profiles	Outcome predicted	Variables used	Diagnostic properties	External validation	Use in clinical settings
Mortality
1	Wilairatana and Looareesuwan41	1995	July 1991 to May 1993	Thailand	Cohort	72	Validation of APACHE II model (Original APACHE II score use clinical judgement and physiological relationships to assign weightings)	APACHE II score58	ROC analysis	Mean age: 29.9	Females—33.3%	Mortality in adult patients with cerebral falciparum malaria	MAP, temperature, heart rate, respiratory rate, arterial pH, PaO2, haematocrit, WBC count, creatinine, sodium, potassium and Glasgow coma score	Predicted mortality with 95.8% accuracy	None	NE
2	Dondorp et al17	2004	NC	Vietnam	Cohort	268	Logistic regression	None	Hosmer-Lemeshow goodness-of-fit test	15–79 years	Females—19%	Mortality in adults with severe falciparum malaria	Plasma lactate, plasma strong anion gap and plasma creatinine	AUROC: 0.81	None	NE
3	Mishra et al24	2007	NC	India	Cohort	212	Linear regression	MSA (malaria score for adults)	Not done	NC	NC	Mortality in adults with severe malaria	severe anaemia, acute renal failure, respiratory distress, cerebral malaria	Sensitivity: 89.9%, specificity: 70.6%, positive predictive value: 94.1% with cut-off of 5/10	Yes43	NE
								MPS (malaria prediction score)	Not done	NC	NC	Mortality in severe malaria	Age, serum creatinine level, haemoglobin level, cerebral malaria, presence of a pregnancy, use of a ventilator	NE	Yes43	NE
4	Hanson et al18	2010	June 2003 to May 2005	Bangladesh, India, Indonesia and Myanmar	Retrospective analysis of a randomised control trial	789	Logistic regression	CAM (coma acidosis malaria) score	Hosmer-Lemeshow goodness-of-fit	NC	NC	Mortality in adults with severe malaria	Coma and acidosis (base deficit	AUROC: 0.81 (95% CI 0.77 to 0.84)	Yes59	NE
5	Mohapatra and Das26	2009	January 200 to December 2004	India	Cohort study	2089	Logistic regression	MSS (Malaria severity score)	Hosmer-Lemeshow goodness-of-fit (internal validation by splitting data—2089 vs 509)	18–71 years	Female—34.6%	Mortality in adult patients with severe falciparum malaria	Neurological, renal, haematological, hepatic, respiratory, cardiovascular, and metabolic organ systems	AUROC: 0.9	None	NE
6	Newton et al29	2013	1986–2002	Thailand	Retrospective analysis	988	Logistic regression	MPI (malaria prognostic index)	ROC curve analysis and internal validation by data splitting	15–74 years	Females—43%	Mortality in adult severe falciparum malaria	Glasgow coma scale, parasitaemia, plasma lactate, serum bilirubin, pigmented parasites and treatment with ACT	AUROC: 0.97	None	NE
7	Mohapatra et al25	2014	NC	India	Cohort	112	NC	GCBRS (GCS, creatinine, respiratory rate, bilirubin and systolic BP) score	NC	Mean: 35.8±15.1 years	Females—16.1	Mortality in severe falciparum malaria	Cerebral malaria, renal failure, respiratory distress, jaundice and shock	Sensitivity: 85.3%. Specificity: 95.6%	None	NE
8	Hanson et al19	2014	1996–2013	Bangladesh, India, Indonesia, Vietnam and Myanmar	Randomised control trials and cohort studies	1801	Logistic regression	None	Hosmer-Lemeshow goodness-of-fit	21–45	Females—24.4	48 hours survival and survival to discharge in patients with severe malaria	Shock, oligo-anuria, dysglycaemia, respiratory rate, Glasgow Coma Score and absence of fever	PPV for 48 hour-survival: 99.4% (95% CI 97.8 to 99.9). PPV for survival to discharge: 96.9% (95% CI 94.3 to 98.5)	None	NE

ACT, artemisinin combined therapy;AUROC, area under the receiver operating curve; NC, not clear; NE, no evidence; PPV, positive predictive value; WBC, white blood cells.