Non-malaria fevers in a high malaria endemic area of Ghana

Background: The importance of fevers not due to malaria [non-malaria fevers, NMFs] in children in sub-Saharan Africa is increasingly being recognised. We have investigated the influence of exposure-related factors and placental malaria on the risk of non-malaria fevers among children in Kintampo, an area of Ghana with high malaria transmission. Methods: Between 2008 and 2011, a cohort of 1855 newborns was enrolled and followed for at least 12 months. Episodes of illness were detected by passive case detection. The primary analysis covered the period from birth up to 12 months of age, with an exploratory analysis of a sub-group of children followed for up to 24 months. Results: The incidence of all episodes of NMF in the first year of life (first and subsequent) was 1.60 per child-year (95 % CI 1.54, 1.66). The incidence of NMF was higher among infants with low birth weight [adjusted hazard ratio (aHR) 1.22 (95 % CI 1.04-1.42) p = 0.012], infants from households of poor socio-economic status [aHR 1.22 (95 % CI 1.02-1.46) p = 0.027] and infants living furthest from a health facility [aHR 1.20 (95 % CI 1.01-1.43) p = 0.037]. The incidence of all episodes of NMF was similar among infants born to mothers with or without placental malaria [aHR 0.97 (0.87, 1.08; p = 0.584)]. Conclusion: The incidence of NMF in infancy is high in the study area. The incidence of NMF is associated with low birth weight and poor socioeconomic status but not with placental malaria. A prospective birth cohort study to explore the relationship between placental malaria and malaria in infancy was conducted between 2008 and 2011 in the Brong-Ahafo region of Ghana; this study is described in detail elsewhere [19]. Malaria transmission in the study area is high (entomological inoculation rate - 269 infective bites/person/year) and perennial, but transmission peaks between April and October [20]. The health system in the study area is basic and includes public and private health facilities. Infant mortality rate is relatively high, estimated at 52 deaths per 1,000 live births in 2010 [21] and about 40 per 1000 live births in 2013 (Kintampo Health Research Centre, 2015 Report). Laboratory investigations for non-malarial infections are limited to bacterial cultures, which are available in only one health facility. The study procedures have been reported in detail elsewhere [19]. In summary, forty-two communities where good follow-up could be obtained were selected from within the Kintampo Health and Demographic Surveillance System (KHDSS). All pregnant women resident in the selected communities were identified using vital registers collated by community key informants or by staff of the KHDSS [21] who made home visits. At enrolment, demographic, socio-economic and obstetric characteristics of study women were recorded by trained fieldworkers using a standard questionnaire. Study women were followed throughout pregnancy until delivery and, whenever possible, a placental sample was obtained. The malaria status of the placenta was defined as showing either 1) an acute infection (parasites present with minimal pigment), 2) a chronic infection (parasites and substantial pigment present) 3) a past malaria infection (substantial pigment only) or 4) no evidence of malaria infection [22, 23]. Newborns of mothers who had been enrolled in the study prior to delivery were included in the infant cohort study. All infants recruited to the study (with the exception of those who died, migrated or were lost to follow-up) were followed for a minimum of 12 months. However, because recruitment to the study was gradual and children remained in follow-up until the end of the study in May 2011, some children were followed up to 24 months of age. Episodes of illness were detected passively at study clinics. To maximise capture of fevers, families were provided with health insurance for the duration of the study and encouraged to attend clinics whenever an infant was unwell. Furthermore, community-based fieldworkers facilitated transportation of sick infants to see a study clinician for clinical evaluation. On evaluation by a study clinician, a history of fever within the 48 h prior to the clinic visit was recorded and an axillary temperature was measured with a digital thermometer. Infants’ illnesses were investigated and managed according to the Ghana National Treatment Guidelines. Rapid diagnostic tests were used to diagnose malaria prior to treatment at the clinic. Thin and thick peripheral blood smears were also made and read subsequently, following the methods described by Swysen et al. [24]. Blood culture, serological or molecular assays for other infectious agents were not done routinely. Non-malaria fevers, the focus of this study, were defined as 1) the presence of fever (a history of fever in the last 48 h prior the clinic visit OR a measured axillary temperature ≥ 37.5 °C) and 2) no malaria parasitaemia detected by microscopy. The cause of death among the study cohort was assessed using verbal post mortem. Cleaned data were analyzed using STATA 13 (StataCorp, College Station, TX.). Principal component analysis of women’s durable assets was used to derive quintiles of socio-economic status (SES), as described previously [19, 25–27]. Cox regression models were used to determine hazard ratios for multiple episodes of NMF, using a robust standard error to account for within-child correlation. The Efron method was used for tied failure times. Potential risk factors including household characteristics: place of residence (urban, rural), household size (<5, 5–9, ≥10 residents), socio-economic status, roof construction (thatched or other); maternal characteristics: number of courses of intermittent preventive therapy in pregnancy (IPTp) received and infant characteristics: sex, birth weight (low, < 2.5 kg; normal), the season of birth (wet, April-November; dry, December-March), and ITN use. ITN use was assessed as tertiles (high, medium or low) based on scores of ITN use in the previous night of scheduled home visits made to access the presence of participants in the study area during follow up. The primary analysis covered the period from birth up to the age of 12 months. A previous study of non-malaria fevers in Benin [18] investigated non-malaria fevers by subtype in children less than 18 months of age. Since respiratory and gastrointestinal diseases are common among young children, the analysis was repeated to determine the incidence of NMFs accompanied by gastrointestinal and respiratory symptoms., Since, depending on the date of enrolment, some children were followed for longer than 12 months, we also conducted exploratory analyses of incidence patterns in the period 0–18 months and 6–18 months of age as in the Benin study. The study was approved by the ethics committees of the Kintampo Health Research Centre (KHRC), Ghana Health Service, London School of Hygiene & Tropical Medicine and Noguchi Memorial Institute for Medical Research. Written informed consent was sought from all study women.