Urinary tract infection (UTI) is a relatively common but often missed diagnosis in young children with fever. One possible explanation for this is that UTI may be overlooked as a source of fever. Another difficulty is the technical challenge of obtaining a clean-catch urine specimen from a young child without performing urethral catheterization. Many studies that attempt to correlate UTI with clinical findings such as fever, vomiting, diarrhea and poor feeding in children younger than one year have been inconsistent. In other studies, however, UTI has been consistently associated with factors such as age, race and level of fever. Gorelick and Shaw conducted a prospective study of febrile young girls in an effort to identify specific clinical factors predictive of a positive urine culture and to use those factors to distinguish patients at low risk of UTI from those who need further evaluation.
All girls younger than two years who presented to an urban emergency department over a 12-month period with a temperature of at least 38.3°C (100.9°F) without a definitive or unequivocal focus of infection were eligible for the study. Inclusion criteria consisted of nonspecific symptoms of upper respiratory tract infection. Children with a confirmed bacterial infection such as pneumonia, meningitis, cellulitis or streptococcal pharyngitis, or a specific viral infection, such as varicella or Coxsackie disease, were excluded from the study. Catheterized urine specimens were obtained on all patients enrolled in the study, and the examining physician completed a questionnaire to record the patient's signs and symptoms. The primary outcome was a positive urine culture, defined as pure growth of at least 104 colonies per mL of a pathogenic species of bacteria. Through a variety of statistical techniques, including the calculation of univariate relative risks with 95 percent confidence intervals (CI), the authors assessed a variety of clinical factors to develop a predictive model for UTI with fever.
Urine cultures were obtained from a total of 1,469 girls. Of these, 63 were positive. Sixty-eight percent of the children were described as “well-appearing,” and 77 percent had a potential source of fever on examination. The clinical factors assessed were age, race, duration of fever, presence of gastrointestinal or urinary symptoms, past history of UTI, temperature of at least 39°C (102.2°F), an ill appearance and abdominal tenderness. Of these, only the following five factors were considered predictive of UTI in this population: (1) age less than 12 months; (2) temperature of at least 39°C; (3) duration of fever for two days or more; (4) white race; and (5) absence of another source of fever on examination. The presence of two or more of these variables was 95 percent sensitive and 31 percent specific for a positive urine culture. The likelihood ratio for a positive culture (95 percent CI 1.21 to 1.43) was 1.35, whereas it was 0.18 for a negative result (95 percent CI 0.06 to 0.49). By retrospectively applying these variables, only three girls with a UTI would have been missed, but a total of 349 urine cultures (30 percent of the total) would have been eliminated.
The authors believe that this clinical decision model accurately predicts a positive urine culture in 95 percent of children younger than two years who have two or more of the identified variables. This model also would eliminate about one third of unnecessary urine cultures.