Dipsticks are one of the most commonly used near-patient tests in primary care, but few clinical or dipstick algorithms have been rigorously developed.
To confirm whether previously documented clinical and dipstick variables and algorithms predict laboratory diagnosis of urinary tract infection (UTI).
Design of study
A total of 434 adult females with suspected lower UTI had bacteriuria assessed using the European Urinalysis Guidelines.
Sixty-six per cent of patients had confirmed UTI. The predictive values of nitrite, leucocyte esterase (+ or greater), and blood (haemolysed trace or greater) were confirmed (independent multivariate odds ratios = 5.6, 3.5, and 2.1 respectively). The previously developed dipstick rule — based on presence of nitrite, or both leucocytes and blood — was moderately sensitive (75%) but less specific (66%; positive predictive value [PPV] 81%, negative predictive value [NPV] 57%). Predictive values were improved by varying the cut-off point: NPV was 76% for all three dipstick results being negative; the PPV was 92% for having nitrite and either blood or leucocyte esterase. Urine offensive smell was not found to be predictive in this sample; for a clinical score using the remaining three predictive clinical features (urine cloudiness, dysuria, and nocturia), NPV was 67% for none of the features, and PPV was 82% for three features.
A clinical score is of limited value in increasing diagnostic precision. Dipstick results can modestly improve diagnostic precision but poorly rule out infection. Clinicians need strategies to take account of poor NPVs.
Keywords: algorithms, clinical scoring; diagnosis, urinary tract infection; primary care; urinalysis