Abstract

Background and aims: In the NICE guideline on childhood urinary tract infection (UTI), it is assumed that the presence or severity of systemic symptoms, especially fever, predict for renal scarring, and different management is recommended accordingly. We aimed to test this hypothesis by retrospective case-note analysis.

Design and subjects: We reviewed the notes of children that had been referred aged under 5 years with a first UTI, from a child population of 154,000, during 1992-5, and who were all assessed for scarring by dimercapto-succinic acid (DMSA) scanning.

Main outcome criteria: Ability to predict for single or multiple scarring from age, sex, fever, vomiting or anorexia or malaise, or need for hospitalisation, within the age-bands used by NICE.

Results: The 51 scarred and 140 unscarred children had a similar sex ratio (girls 65% and 69% respectively). Fever, systemic symptoms and hospitalisation were all commoner among younger children (comparing <6 m vs 6 m–3 yr vs >3 yr; fever 0.67 vs 0.38 vs 0.38; systemic symptoms 0.78 vs 0.62 vs 0.43; hospitalisation 0.67 vs 0.29 vs 0.19; P<0.001 for all). Having vomiting, anorexia or malaise at presentation correlated weakly with single or multiple renal scarring (R2=0.03; P=0.02), but sex, age, fever or hospitalisation did not (P>0.5 for all). Sensitivity and specificity data, and plots of proportionate reduction of uncertainty (PRU) showed that none of these variables was useful to predict for any degree of scarring in children under 3 years, and that they were only weakly predictive in older children.

Conclusions: Clinical signs at presentation in childhood UTI cannot be used to predict for mild or multiple scarring, and should not be used to guide management. NICE's recommendation to do so are not justified.