Question

A company has developed a new dipstick test for urinary tract infection (UTI). The test shows either “positive” or “negative” (there are no equivocal results). You wish to test this in your A&E, where, on average, 15% of your patients under 1 year of age with fever have a true UTI, as diagnosed by a positive urine culture (> 10⁵ cfu/ml) when obtained under sterile conditions (e.g. suprapubic aspirate or catheter-obtained urine).

Over the next 6 months, you evaluate the dipstick on 100 patients less than 1 year of age presenting with fever, and obtain the following results:

		N
Urine Culture +	Dipstick +	10
Urine Culture +	Dipstick -	2
Urine Culture -	Dipstick +	11
Urine Culture -	Dipstick -	77

Your colleague, Dr Bayes, in a different A&E decides, on the basis of your evaluation, to adopt the dipstick test. Her A&E has a different catchment area to yours, and has an estimated UTI rate of 8% in febrile infants.

What is the probability that an infant presenting to her A&E who tests dipstick positive actually has a UTI?Is this the same probability as for a “dipstick positive” infant in you’re A&E, above? If so, why… and if not, why not?

Answer 1