Characterizing diagnostic heterogeneity in UTI diagnosis in resource-limited settings: Comparing clinical practice...

In resource-limited settings without access to urine culture, urinary tract infection (UTI) diagnosis relies on clinical judgment based on symptoms, dipstick chemistry, and basic urine microscopy. The extent of diagnostic variation when applying structured versus unstructured approaches to identical clinical information is unknown.

To characterize diagnostic heterogeneity by comparing routine clinical diagnoses with the systematic application of research-derived diagnostic criteria and identify factors associated with diagnostic concordance.

A retrospective cross-sectional analysis of 1,798 patient records (253 paediatric and 1,545 adult) from January 2022 to December 2024 was performed. Clinical diagnoses were compared with the classification using the research-derived Delphi consensus criteria (applying available parameters: symptoms, systemic factor; temperature, and microscopic pyuria). Agreement was assessed using Cohen's kappa and logistic regression.

Clinical diagnosis identified UTI in 92.2% of cases versus 19.7% using a structured criteria (κ=0.0129, p = 0.053; overall agreement 25.4%). Agreement decreased with age (odds ratio [OR] =0.99, p = 0.032) and female sex (OR=0.61, p < 0.001) but increased with leukocyte esterase levels (p < 0.001). Clinicians diagnosed UTI in 91.6% of cases that did not meet the structured criteria, while they agreed with 94.6% of criterion-positive cases.

Substantial diagnostic heterogeneity exists between unstructured clinical judgments and structured research-derived criteria when analyzing identical clinical information. This variation suggests potential for diagnosis improvement and has implications for antimicrobial stewardship, though prospective derivation and validation of context-appropriate clinical decision rules is needed.