Predictors for a positive blood culture in African children with pneumonia.

Background. Blood culture isolation rates for children with pneumonia are generally low. Therefore it would be helpful if epidemiologic studies could identify children who have a higher than average chance of yielding a positive culture. Methods. Simple clinical and laboratory predictors of a positive blood culture were investigated in 1222 children younger than 5 years of age with pneumonia who presented at rural dispensaries in The Gambia, West Africa. Results. A high temperature (chi square for linear trend, 24.1 ; P < 0.0001), a rapid respiratory rate (chi square for linear trend, 6.33 ; P = 0.010), dehydration (odds ratio, 2.33 ; P < 0.001), nasal flaring (odds ratio, 2.06 ; P = 0.001), grunting (odds ratio, 4.01 ; P < 0.001), dullness to percussion (odds ratio, 2.31 ; P < 0.001), bronchial breathing (odds ratio, 2.61 ; P < 0.001) and diminished breath sounds (odds ratio, 2.07 ; P < 0.001) were positive predictors for a positive blood culture. Wheezing (odds ratio, 0.16 ; P < 0.001) and malaria parasitemia (odds ratio, 0.26 ; P = 0.008) were negative predictors. A combination of these findings were used to assess how the number of cultures taken might be reduced without substantially reducing the yield of positive cultures. For example it was found that exclusion of children with a temperature of <38.0°C and/or a respiratory rate of <50/min and/or wheezing would have reduced the number of cultures taken by 55.6% but would have led to a loss of only 31.3% of positive cultures. Conclusion. Careful selection of children investigated by blood culture could help to reduce the work required during the preparations for and conduct of pneumococcal vaccine trials.