Cohort study of bacterial species causing urinary tract infection and urinary tract abnormalities in children

BMJ 1999; 318 doi: (Published 20 March 1999) Cite this as: BMJ 1999;318:770
  1. Olli Honkinen, research fellowa,
  2. Olli-Pekka Lehtonen, laboratory directorb,
  3. Olli Ruuskanen, professora,
  4. Pentti Huovinen, laboratory directorc,
  5. Jussi Mertsola, consultant in paediatric infectious diseases (jussi.mertsola{at}
  1. aDepartment of Paediatrics, Turku University Hospital, 20520 Turku, Finland
  2. bLaboratory of Clinical Microbiology, Turku University Hospital
  3. cNational Public Health Institute, Turku Department, Turku, Finland
  1. Correspondence to: Dr Mertsola
  • Accepted 9 November 1998

Children with urinary tract infection often have anatomical or functional abnormalities in their urinary tract. Imaging studies are therefore recommended, especially for young children, after the first urinary tract infection.13 There is, however, considerable variation in clinical practice and some resistance to diagnostic imaging for childhood urinary tract infection.4 We conducted this study to find out whether an association exists between the bacterial species causing the first urinary tract infection and abnormal findings in subsequent imaging studies.

Patients, methods, and results

We examined retrospectively the hospital records of all children with a positive urine culture from a sample obtained by suprapubic aspiration (any growth) or catheterisation (growth of at least 103 colony forming units/ml) during January 1980 to December 1994. Asymptomatic children and those who had been in hospital for more than 2 days before the specimen was obtained were excluded. Children with proved infection had intravenous pyelography (1980-3) or ultrasonography (1984-94) plus voiding cystourethrography (radiographic in boys and nuclear in girls) 1-2 months after the acute infection.

Age, sex, imaging results, and operations in children according to cause of first urinary tract infection. Values are numbers (percentages) unless stated otherwise

View this table:

We identified 1237 positive urine samples. Of these, 942 (76%) were obtained by suprapubic aspiration and 295 (24%) by bladder catheterisation. There were 982 (79%) cultures positive for Escherichia coli, 66 (5.3%) for Enterococcus sp, 55 (4.4%) for Klebsiella sp, 47 (3.8%) for coagulase negative staphylococcus, 39 (3.2%) for Proteus sp, and 48 for other species. Of the 207 patients with enterococcal, klebsiella, coagulase negative staphylococcal, or proteus infections, hospital records were available for 201. Forty one patients were excluded (20 had asymptomatic bacteruria, seven had chronic urinary problems such as neurogenic bladder, and in 14 cases one or both of the urinary tract imaging studies were lacking). The urinary tract infection was the first in 92 of the remaining 160 children with infections other than E coli.

We compared the results of these children with those of 92 age and sex matched children whose first urinary tract infection was caused by E coli. Student's ttest was used to compare the mean serum C reactive protein values in the two groups. The frequency of abnormalities in the urinary tract was compared by Pearson's χ2 test or Fisher's exact test if the number of expected observations was five or less in at least one cell.

The table shows the characteristics of the 184 patients and findings in the imaging studies. There were significantly more abnormalities among children with enterococcal, klebsiella, or coagulase negative staphylococcal infections than in children infected with E coli. The rate of corrective surgical procedures was also substantially higher in these patients. Only two abnormalities were diagnosed in children with proteus infection.


Our results agree with earlier studies showing that about 80% of urinary tract infections are caused by E coli and that vesicoureteral reflux is found in about a third of children with infections caused by E coli.3 We also found that when the first infection was caused by klebsiella or enterococcus the rate of vesicoureteral reflux was almost double and that surgical procedures were four times more common in children with infections caused by these organisms than in those with infections caused by E coli.

E coli lacking putative virulence factors such as gal-gal specific adhesins are overrepresented in patients with pyelonephritis associated with vesicoureteral reflux. 3 5 Similarly, in children with acute pyelonephritis, renal scarring is significantly more common in those infected with non-E coli organisms3 and with E coli strains that do not possess certain virulence determinants.5 The bacteria probably do not need special virulence properties if the host has an abnormality such as vesicoureteral reflux.


Contributors: OH, the principal investigator for the study, was involved in the data collection, documentation and analysis. O-PL and PH organised the collection and analysis of the microbiological data. OR initiated the research and participated with JM in the analysis and interpretation of the clinical data. The paper was written jointly by OH, O-PL and JM. OH and JM are guarantors of the study.

Funding: Paediatric Research Foundation and Academy of Finland.

Competing interests: None declared.


View Abstract