Aim To develop a clinical algorithm based on symptoms, signs and urine dipstick results to assist the identification of children who require urine sampling, antibiotic treatment and/or laboratory analysis.

Methods We conducted a diagnostic cohort study of children <5 years presenting acutely (≤28 days) unwell to primary care in the UK. We collected detailed information on the presence/absence and severity of presenting symptoms and signs, as well as socio-demographic and past medical history data. Urine was sampled by clean catch (preferred) or nappy pad, ‘dipsticked’ and sent to (i) the local NHS laboratory (the priority sample) and (ii) a reference laboratory. Blind to children’s clinical symptoms and signs, the NHS and reference laboratories processed urine samples according to their standard operating procedures.

Results (preliminary) 7,163 children were recruited with NHS and research urine sample results available for 6,328 (88%) and 5,257 (73%) respectively. Of the 5,017 children without missing data and with urine results from both laboratories: mean age was 2.2 years (s.d. = 1.4); 49% were male; 54% urines via clean catch, 45% via nappy pads and 1% via bag. UTI rates were 2.8% and 3% from clean catch and pad samples respectively. Among clean catch samples, the following were independently associated with UTI: history of UTI; parental report of smelly urine; pain/crying while passing urine; clinician’s global impression of illness severity; and on dipstick: nitrites, leukocytes and blood (area under the ROC = 0.87 (95% CI 0.82 to 0.92). Among the nappy pad samples, the factors were: female gender; age; smelly urine; darker urine; and on dipstick: nitrites, leukocytes and blood (AUROC = 0.78 (0.72 to 0.83)).

Conclusions Symptoms, signs and dipstick testing have diagnostic utility for UTI. These results will be developed into an algorithm to help clinicians select which should have: a urine sample obtained; a sample sent for laboratory culture and receive immediate antibiotic treatment.