Feb 07

Predicting the need for imaging in febrile urinary tract infections

Journal reference: van Nieuwkoop C, Hoppe BPC, Bonten TN, et al. Predicting the need for radiologic imaging in adults with febrile urinary tract infection. Clinical Infectious Diseases 2010;51(11):1266–1272 [1]

Link: http://dx.doi.org/10.1086/657071

Evidence cookie says...

A simple clinical prediction rule may predict patients with febrile urinary tract infection (UTI) who would not benefit from radiologic imaging.

Those with:

  • no history of urolithiasis, and
  • urinary pH < 7.0, and
  • eGFR > 40 mL/min/1.73 m3

… are very unlikely to have an urgent urologic disorder diagnosed on radiologic imaging.

Note: this study has some significant limitations.  See below for details.

More details:

Article details

Study design:

prospective multi-centre cohort study

Study aim:

to develop a clinical rule to predict the need for imaging in the setting of febrile urinary tract infection

Methods summary:

  • Derivation cohort of 346 patients followed for 3 months: underwent usual management for febrile urinary tract infection as directed by treating physician
  • Primary end point: clinically relevant urologic disorder as detected by radiologic imaging
  • Outcome after three months classified as “urgent urological disorder”, “ non urgent urological disorder” and “clinically irrelevant findings” and “ incidental non-urological disorders” based upon radiologic imaging findings
  • Model development: logistic regression model fitted whole data set of risk factors to predict probability of receiving radiological imaging
  • Multivariate analysis to develop clinical prediction rule
  • Validation cohort of 131 patients to determine negative predictive value (NPV), positive predictive value (PPV), sensitivity and specificity.

Results summary:

  • 71% of 346 patient in derivation cohort underwent radiologic imaging
  • Radiologic imaging was significantly associated with female sex, history of urolithiasis, absence of indwelling urinary catheter, and presence of flank pain or chills
  • Derivation of prediction rule:
    • history of urolithiasis, and/or
    • urinary pH ≥ 7.0, and/or
    • renal insufficiency (eGFR ≤ 40 mL/min/1.73 m3)
  • Validation of prediction rule had an 89% negative predictive value (NPV) for any clinically relevant radiologic finding and NPV of 100% for any urgent radiologic finding.
  • Using the prediction score, a potential reduction of 40% (relative) or 28% (absolute) in imaging studies would be achieved

Study conclusion:

Radiologic imaging can selectively be applied in adults with febrile UTI without loss of clinically relevant information by using a simple clinical prediction rule.


  • Consecutive patients attending 8 emergency departments in a single region of Netherlands between January 2004 and November 2008 with:
    • age ≥ 18 years
    • fever ≥ 38.0 C and/or history of fever and chills within 24 hours
    • symptoms of UTI (dysuria, frequency, urgency, perineal pain, flank pain, or costovertebral tenderness)
    • positive leukocyte esterase dipstick test or the presence of > 5 leukocytes/high powered field in urinary sediment
  • Exclusion criteria:
    • current treatment for urolithiasis or hydronephrosis
    • pregnancy
    • known allergy to fluoroquinolones
    • receipt of haemodialysis or peritoneal dialysis
    • history of kidney transplantation
    • known presence of polycystic kidney disease
  • Mean participant age was 70 years (interquartile range 49-80)
  • 41% male
  • Over of third were on antibiotic UTI treatment at baseline
  • 60% were categorised as having “acute complicated UTI”

Methodological weaknesses

  • Exclusion of 8 patients who died and did not undergo radiologic imaging which might have influenced outcome of non radiologic imaging group
  • It was assumed that patients who did not undergo imaging had normal radiologic outcomes if their clinical course was unremarkable. This assumption might not be true; this would not detect asymptomatic non-urgent urological disorders.
  • The exclusion of participants receiving current treatment for urolithiasis puts the use of “history of urolithiasis” as part of the prediction rule into question; i.e., it is likely that any history of urolithiasis is a stronger predictor than suggested by the prediction rule in the general population.
  • Study does not provide evidence for cost savings from radiologic imaging
  • Participant demographics limit how far the results can be generalised; elderly homogenous Western population and emergency department presentation only.
  • The exclusion of pregnant women, those with various types of renal disease (currently treatment for urolithiasis and hydronephrosis, dialysis, kidney transplant, polycystic kidney disease) complicates the prediction rule with numerous clinically relevant caveats.

Methodological strengths

  • Selection of consecutive patients presenting to ED increase the likelihood that the cohort was representative of patients presenting with febrile UTI
  • Clinical management and radiologic decision making was left to the treating physician
  • Multiple centres were included in the study and encompassed patients with common comordbitiies
  • Several imputation models were tested to account for methodological weaknesses and were not significantly different from original model

Biases and conflicts of interests

None declared and nil seem obvious.

Clinical relevance to primary health care

This study demonstrates that a relatively simple rule in the emergency department management of febrile urinary tract infections (UTIs) can substantially reduce the quantity of radiologic investigation performed.  Almost no patients with urgent urological disorders will be inappropriately excluded from imaging when it is restricted to those with:

  • a history of urolithiasis, and/or;
  • a urinary pH ≥ 7.0, and/or;
  • an eGFR ≤ 40 mL/min/1.73 m3

At face value, the findings are not immediately applicable in primary health care as the study was performed in the emergency department setting.  The clinical prediction rule cannot be validly generalised for patients presenting with febrile UTIs in the community general practice.

Nevertheless, the negative predictive value of the tool remains applicable; indeed it may be even stronger given the likely lower prevalence of urgent urological disorders in general practice.  That is, patients with a febrile UTI with no history of urolithiasisand a urinary pH < 7.0and eGFR > 40 mL/min/1.73 m3 are very unlikely to have an urgent urological disorder diagnosed on radiologic imaging.

Note: care should be taken in considering the demographics of the participants of this study when utilising its results.  The participants were older and ethnic Western European.  Participants with a number of important renal and urological conditions were excluded.


  1. van Nieuwkoop C, Hoppe BPC, Bonten TN, et al. Predicting the need for radiologic imaging in adults with febrile urinary tract infection. Clinical Infectious Diseases 2010;51(11):1266–1272

Editor: Michael Tam

Print Friendly, PDF & Email

Leave a Reply

%d bloggers like this: