Objective: To assess the relative risks and benefits of 10 potential urine testing strategies (compared with no testing) involving urinalysis and urine culture for children aged 3 to 24 months with fever but no focus of bacterial infection.
Design: Decision analysis based on the literature. The 10 testing strategies consist of five pairs; within each pair of strategies, one calls for urinalysis and urine culture of a clean-voided (bag) specimen, and urine culture, and in the other, the urine specimen is sent for culture only if the result of the urinalysis is abnormal. The five pairs differ in selectivity for testing: all children, girls only, temperature > or = 39 degrees C only, fever only (no respiratory or gastrointestinal symptoms), or temperature > or = 40 degrees C only. The results of the decision analysis are expressed as the preventive fraction (the proportion of cases prevented) for end-stage renal disease (ESRD) and hypertension, and as two risk/benefit (RB) ratios: the number of children tested per case of ESRD prevented (RB1), and the number of children with false-positive diagnosis and treatment of urinary tract infection per case of ESRD prevented (RB2).
Results: On the basis of the available evidence, none of the testing strategies succeeds in preventing the majority of cases of ESRD and hypertension (preventive fraction = 0.10 to 0.50), and all are associated with high ratios of children tested (RB1 = 4167 to 12,500) and false-positive diagnosis and treatment (RB2 = 563 to 1800) per case of ESRD prevented. A strategy of combined urinalysis and urine culture in children with temperature > or = 39 degrees C is associated with the most favorable RB profile: preventive fraction = 0.45, RB1 = 5556; RB2 = 776. Sensitivity analyses indicate that the relative ranking of the strategies is relatively robust in regard to alterations in the estimates of the sensitivity or specificity of the urinalysis, the relative risk of renal scarring associated with delayed diagnosis and treatment, and the risk of scarring-induced hypertension or ESRD.
Conclusions: Up to 50% of the long-term sequelae of occult urinary tract infections in young febrile children appear preventable by urine testing, but even the most favorable strategies require testing of thousands of children, and unnecessarily treating hundreds, for every case prevented. Our analysis reveals those strategies with more favorable RB profiles and emphasizes the need for rapid and convenient urine tests with much higher sensitivity and specificity or the need for less aggressive management strategies for febrile infants and young children with urinary tract infection.