Evaluation of a diagnostic algorithm in children infected with malaria using signs and symptoms observed by their mothers in western Kenya
Oluoch, Patricia Rose
MetadataShow full item record
The use of health facilities by mothers with sick children having malaria is hampered by poor recognition of the condition and logistical reasons. Mothers often resort to inappropriate use of over the counter drugs, which are readily available as well as traditional medicine to treat illnesses suspected to be malaria. The aim of this study was to evaluate an algorithm designed for use by health workers to see its validity when used by mothers in an area of intense and perennial transmission of malaria to predict clinical malaria. In a cross sectional study, 3,495 mothers with children 5 years were interviewed about their children's health status in the previous two weeks using a standardized tool of signs and symptoms. Each child had the axillary temperature taken and also thick blood smear by finger prick for parasitaemia determination. Clinical malarial was defined as parasitaemia level the age specific critical density for the study area and temperature37.5°C. The prevalence of parasitaemia in this population was 62%, the prevalence of parasitaemia above the age specific critical density was 19.3%, while those with temperature 37.5°C were 6.7%. Logistic regression was performed on the signs and symptoms to find the best predictors of malaria and this gave in descending order: fever, vomiting, pale eyes, difficulty in breathing, loss of weight, white nails and absence of running nose. The prospective algorithm had an optimum sensitivity of 54.4% and specificity of 30.7% at a cut off score combination of signs and symptoms of 7. This is a starting point and demonstrates that an algorithm of signs and symptoms can be developed for use by mothers with children, 5 years as a better predictor of malaria than the use of fever alone as is the current recommendation and practice.
- MST-Zoological Sciences