Statistics from Altmetric.com
Commentary on the paper by Leroy et al (see page241)
Many children are investigated for vesicoureteric reflux (VUR) following a urinary tract infection (UTI), but the imaging technique of a voiding cystourethrogram (VCUG) is unpleasant for the child and not without risk. A clinical scoring system capable of confidently predicting VUR would therefore be an attractive alternative approach.
The original study by Oostenbrink and colleagues1 seemed to suggest that a score derived from the combination of clinical factors (age, sex, and positive family history), ultrasound findings and C reactive protein (CRP) result could predict the presence of VUR with high sensitivity, albeit with rather low specificity. A diagnostic test with 100% sensitivity can be useful in ruling out a condition when it is negative, even when it has low specificity (that is, cannot be relied upon when positive). This has been referred to as a “SnNout” (when a test has a high Sensitivity, a Negative result rules out the diagnosis).2 However, the lower the specificity, the smaller would be the expected proportion of children without reflux who have a negative result, and thereby avoid an unnecessary imaging investigation. A negative result, excluding VUR, was seen in 17% of children without VUR in Oosterbrink’s series.
Leroy and colleagues3 are to be commended on repeating the work of Oostenbrink. When a group of patients is used to develop a diagnostic test, the performance of that test tends to be overestimated. This is particularly …