TRANSACTIONS OFTHE ROYAL SOCIETY OFTROPICAL MEDICINE AND HYGIENE (1998) 92,527-531 Clinical case definitions for malaria: clinical malaria associated with very low parasite densities in African infants David McGuinnessl, Kwadwo Koramz, Steve Bennet@, Gillian Wagnerl, Francis Nkrumahz and Eleanor Riley1 IInstitute of Cell, Animal and Population Biology, University of Edinburgh, Edinburgh, EH9 3JI; UK; 2Noguchi Memorial Institute for Medical Research, University of Ghana, Legon, Ghana; 3Department of Infectious and Tropical Dis- eases, London School of Hygiene and Tropical Medicine, London, WClE 7H7; UK Abstract In areas endemic for Plasmodium falciparum, clinical malaria is believed to be less common in infants than in older children, but specific case definitions have rarely been determined for this age group. As malaria case definitions are known to be both age- and site-specific, assessment of the risk of disease in infancy requires the development of appropriate diagnostic criteria. In southern Ghana, 154 children were re- cruited at birth and monitored for fever and malaria infection until 2 years of age. Logistic regression was used to model fever risk as a continuous function of parasite density to determine case definitions for the diagnosis of clinical malaria, and to determine age- and season-specific estimates of the fraction of fevers attributable to malaria (AF); 2360 observations were made on 154 children. For fevers defined by a meas- ured temperature >375”C, the estimated population AF was 44% (95% confidence interval 34-53). Estimates of AF varied with age and season. For infants, AF was 51% during the wet season and 22% during the dry season; for children over one year of age, AF was 89% during the wet season and 36% during the dry season.The estimated parasite density threshold for initiation of a febrile episode was 100 parasites per ltL of blood in infants, compared with 3500 parasites per lrL for children over one year of age. Using these case definitions, the incidence of clinical malaria was estimated at 0.09 cases per child-year at risk for children less than 6 months of age, 0.40 for children aged 6-11 months, and 0.69 for children aged 12-23 months. Of 66 cases of clinical malaria, only 3 were observed in children under 5 months of age. We concluded that, although most fevers in infants are not due to malaria, infant clinical malaria may occur at extremely low parasite densities. This may be indicative of a lack of anti-disease im- munity in this age group. In southern Ghana, an infant with axillary temperature >37.5”C and parasitaemia >lOO/ltL should be considered to have clinical malaria. Nevertheless, the incidence of clin- ical malaria is very low in children under 6 months of age, confirming that they are significantly protected horn clinical malaria compared to older children. Keywords: malaria, Plasmodium fulciparum, clinical case definitions, children, Ghana Introduction It is estimated that every year there are over one mil- lion deaths from malaria world-wide, mainly of children under 5 years old in Africa (WHO, 1997), but the mag- nitude of the public health problem posed by malaria in many areas of the world remains uncertain. Objective diagnostic methods with high sensitivitv and suecificitv wo;ld increase the power ‘;f research “studies&and irnl prove the outcome of interventions. Clinical malaria typically presents with fever, chills and general malaise, but these signs are not malaria-specific; viral and bacte- rial infections can present with similar clinical signs (O’DEMPSEY et al., 1993; LUXEMBURGER et al., 1998). Thus, diagnosis based on fever alone is likely to be un- reliable. In areas where malaria is endemic, and consid- erable levels of acquired immunity are present, asymptomatic infection is common and fever due to other diseases will occur concurrently with the presence of malaria parasites. Thus, diagnosis based on fever plus malaria parasitaemia may result in over-diagnosis (ROUGEMONT et al., 199 1). As high parasite counts are more likely to coincide with fever, an alternative ap- proach is to diagnose clinical malaria for fever episodes when the parasite count is above a defined cut-off value C~NOW et al.. 1988). It is to be exnected that case deti- nitions will vary with age, levels of malaria transmission and acquired immunity. In highly endemic areas, young children suffer the greatest burden of disease and clinical malaria becomes less prevalent with increasing age (GREENWOOD et al., 1987). However. it is well recognized that newborn in- fantsare relatively protected against mild clinical malar- ia (BRABIN, 1990) and severe malaria (SNOW et al., Address for correspondence: Professor E. Riley, Department of Infectious and Tropical Diseases, London School of Hy- giene and Tropical Medicine, Keppel Street, London, WClE 7HT, UK; phone +44 (0)171 636 8636, e-mail e.riley @lshtm.ac.uk 1998) compared to older children. The mechanisms of clinical protection in infants are not well understood and their investigation has been hampered by the lack of specific case definitions for this age group. In a longitu- dinal study in Ghana, we found that infants may be in- fected with malaria but that the vast majority of infections are asymptomatic (WAGNER et al., 1998). However, to determine the true incidence of clinical malaria in this age group, appropriate case definitions are needed. We have used logistic regression to model the risk of fever as a function of parasite density, to esti- mate the fraction of fever cases that are attributable to malaria (the attributable fraction, AF), and to estimate the sensitivity and specificity of case definitions using different parasite density thresholds (SMITH et al., 1994). Our results indicated that clinical malaria is truly uncommon in children under 5 months of age but that the pyrogenic threshold is very low in this age group, with clinical symptoms being induced at parasite counts as low as lOO/&. Methods Study design One hundred and tiftv-four infants, born between April 1994 and February 1997 in the village of Pram- nram (uouulation c. 7000) which is about 50 km east of Accra &iihe south coast of Ghana, were monitored for malaria infection from birth to 2 years of age. Malaria transmission in this area is perennial but peaks after the main wet season (AFAFU et al., 1996) which, for the pur- pose of this study, was defined as the months of July, August and September (WAGNER et al., 1998). Plasmo- dium falciparum is the dominant malaria species (92% of isolates), but l? malariae is also present (AFARI et al., 1996). The entomological inoculation rate is approxi- mately 8.5 infective bites per person per year and the dominant vector is Anopheles gambiae (see APPAWU et aZ., 1994).