Statistical Assessment of the Impact of Parasitological Diagnosis of Malaria in Febrile Children Aged under 5 Years in Kisumu County

Abstract

In malaria endemic areas, fever is commonly caused by malaria infection but it may be a manifestation of several childhood diseases for example bacterial and viral illnesses. Malaria microscopy is the gold standard for diagnosis of malaria although other diagnostic platforms do exist for example rapid diagnostic tests. The World Health Organization recommends parasitological diagnosis by microscopy or rapid diagnostic test for all children under the age of 5 years. However, in malaria endemic areas, parasitological diagnosis is not always available and presumptive treatment based on clinical diagnosis is practised. It is believed that clinical diagnosis based on integrated management of childhood illnesses has resulted in over-diagnosis of malaria, neglect of other co-infections and escalation of cost of medical management of malaria in endemic areas. This retrospective study carried out statistical assessment of the impact of parasitological diagnosis of malaria by microscopy on: mortality, malaria incidence rate, over-diagnosis of malaria and cost of medical management of illnesses in febrile children aged under 5 years. Linear, logistic and Cox regression models were applied. Software R version 2.15.0 and GraphPad Prism version 5 were used in the analysis. In-patient records reviewed were 400 and 626 for the years 2009 and 2011 respectively. The results showed that malaria incidence rate significantly declined from 60.29% in 2009 to 52.14% in 2011 (P = 0.0147, OR=0.72). Malaria infection was associated with anaemia (P=0.0002, OR=2.982). Furthermore, a logistic regression model demonstrated that malaria was significantly associated with chills (P=0.00519, OR=18.59), vomiting (P=0.02463, OR=1.5771), convulsion (P=0.015077, OR=1.7653) and splenomegaly (P=0.008044, OR=3.0105). It was unlikely to present with diarrhoea (P=0.000685, OR=0.4958), oedema (P=0.013662, OR=0.2032), abdominal pain (P=0.002901, OR=0.1482), malnutrition (P<0.0001, OR=0.1206) and other infections (P<0.0001, OR=0.1258), but highly likely in malnourished children given other infections (P=0.002212, OR=4.8984). It was independent of pallor (P=0.115991, OR=1.4023), dyspnoea (P=0.0751, OR=0.654) and chest retraction (P=0.0735, OR=0.3955). Malaria over-diagnosis decreased from 50.64% in 2009 to 22.43% in 2011 in overall (P<0.0001, OR=0.2818) and insignificantly in children with negative microscopy (P=0.2865, OR=0.8388). A Cox model indicated that death hazard was greater in 2011 than in 2009 (P=0.0393, HR=1.9391). The hazard declined with increasing age (P=0.023, HR=0.729) and lower among malaria in-patients (P = 0.0119, HR=0.5146). Cost of medical management in 2011 increased (P<0.0001) and reduced for malaria treatment (P=0.0295). This study concluded that parasitological diagnosis was associated with reduced malaria incidence rate, over-diagnosis and treatment cost resulting in better survival experience of malaria in-patients. However, mortality risk and medical management cost were elevated in 2011. The results provide valuable insight into the rapidly evolving dynamics of malaria epidemiology, quality of clinical care afforded by diagnosis-based treatment and a clinical model for prediction of malaria in endemic areas lacking diagnostic equipments.