Acute kidney injury and death in severe falciparum malaria

BackgroundSerum or plasma concentrations of creatinine and urea are the usual laboratory parameters measured to assess kidney function and disease severity in patients with suspected severe malaria. Creatinine is preferred for estimation of the glomerular filtration rate, but the current threshold for defining severe malaria (> 265 umol/L) is based on adult values and does not reflect the large difference between children and adults in normal values resulting from differences in muscle mass. We reevaluated these thresholds and their prognostic significance in severe malaria using individual patient data from large prospective studies. MethodsWe pooled individual patient data from studies of severe malaria. Patients were included in the primary analysis population if their age (if missing, imputed from weight or height), coma status on admission, and either an admission serum or plasma creatinine or urea measurement (blood urea nitrogen, BUN), were all recorded. The primary outcome was mortality by day 28. The secondary outcome was the admission plasma Pf HRP2 concentration (a measure of parasite biomass). Bayesian penalised spline regression models were used to characterise the age-specific relationships (interaction with age) between the outcome and the absolute creatinine, the creatinine fold change relative to the age/weight predicted baseline, and the absolute urea. The models adjusted for coma status and calendar year, with nested random additive effects by study site and study. We estimated threshold plasma or serum creatinine and urea concentrations associated with >5% mortality with probability > 0.9. ResultsWe pooled individual patient data from 22,663 patients recruited into 14 studies of severe malaria of whom 16,441 were included in the primary analysis population. The majority (84%; 13,793/16,441) were children under 15 years of age. Current thresholds for renal impairment were associated with mortalities substantially greater than 5%, especially in children. A creatinine increase of more than three times the age/weight predicted baseline value was associated with 0.9 probability of mortality >5%. A urea of 10 mmol/L had approximately similar prognostic value. A urea >10 mmol/L was 3.5 times more common in adults than children. In children <15 years of age urea was a better prognostic indicator of death than creatinine and, unlike creatinine, correlated with the plasma Pf HRP2 concentration. InterpretationAdmission serum or plasma creatinine >3 times the age/weight expected baseline value (KDIGO stage 3) can be used to define severe malaria. In children urea is a better predictor of the true infection biomass and a better prognostic indicator for death. We recommend the following as defining criteria for severe malaria; either the simple urea (BUN) threshold of 10 mmol/L for both children and adults, or a >3 times fold change in creatinine from predicted baseline.