Estimating malaria attributable fraction using quantitative PCR in a longitudinal cohort in Eastern Uganda.

Persistent, asymptomatic Plasmodium infections are common in areas of high transmission due to acquired immunity. When asymptomatically infected individuals seek care for a fever caused by something other than malaria parasites, they may test positive for parasites and be incorrectly diagnosed as having clinical malaria. This study used distributions of qPCR parasite densities to estimate the fraction of fever attributable to malaria (malaria attributable fraction, MAF) in a cohort of 659 individuals followed for up to 3 years from three geographically distinct zones in eastern Uganda. Prevalence of P. falciparum by qPCR ranged from 47-63% in the three zones, with over 95% of cohort members parasitemic at least once. Overall MAF across the three zones ranged from 54-64%. MAF was highest in under-five year-olds (72%), next highest in 5-15 year-olds (56%) and lowest in adults over 16 (45%). Notably, nearly 50% of fevers with low to moderate parasite density (10 - 100 parasite/ microliter) were attributed to malaria. MAF-corrected incidence was higher than the definition of clinical malaria used in many vaccine field-studies (fever and parasite density ≥ 5000/microliter) and the difference varied by age group: MAF-corrected incidence was 18% higher in children under five, 7% higher in 5-15 year olds, and 70% higher in adults. These results suggest parasite density thresholds commonly used to define primary study outcomes will underestimate the true incidence of clinical malaria. Studies aiming to precisely estimate intervention effects on incidence should consider estimating MAF in their study population and incorporating it into their design.