Diabetes mellitus and ultrasound findings suggestive of renal parenchymal infection in adults with suspected acute pyelonephritis: a retrospective cross-sectional study

Background Renal ultrasound is frequently requested in the emergency department for suspected acute pyelonephritis (AP), despite limited sensitivity and unstandardized indications. We examined which clinical characteristics are associated with ultrasound findings suggestive of renal parenchymal infection, using an outcome restricted to infection-related signs and separated from incidental structural findings. Methods Retrospective, single-centre, cross-sectional study of adults with suspected AP who underwent renal and urinary tract ultrasound in the emergency department of a tertiary referral centre in Cali, Colombia (February 2016 – February 2023). The outcome was the presence of at least one ultrasound finding suggestive of parenchymal infection (loss of cortico-medullary differentiation, altered parenchymal echogenicity, abscess or collection, focal Doppler perfusion alteration, or urothelial thickening >2 mm). Patients whose only abnormalities were incidental structural findings (hydronephrosis, increased renal length, urolithiasis, developmental malformation) formed a separate group and were not included in the main comparison. Candidate clinical variables were compared using Fisher's exact test with conditional exact odds ratios (OR) and exact 95% confidence intervals (CI), and the Mann–Whitney U test. Adjusted estimates were obtained with Firth penalized logistic regression using profile penalized-likelihood CIs, with covariates chosen a priori on clinical grounds. Three sensitivity analyses tested alternative outcome definitions. Results Of 170 patients, 54 (31.8%) had findings suggestive of infection, 80 (47.1%) had no abnormal findings, and 36 (21.2%) had incidental structural findings only. In the main comparison ( n = 134), diabetes mellitus was the only variable associated with the outcome (exact OR 5.38, 95% CI 1.67–20.55, p = 0.002; Firth-adjusted OR 5.64, 95% CI 1.93–18.81, p = 0.001). A history of urolithiasis, immunosuppression, a history of urinary tract anomaly, pregnancy, age, sex, presenting symptoms, vital signs, urinalysis, urine culture and renal function were not associated with the outcome. The association persisted when the outcome was restricted to objective criteria (OR 3.62, 95% CI 1.06–11.51), when the infection group was compared with the remainder of the cohort (OR 3.78, 95% CI 1.59–9.31), and when renal enlargement was reclassified as an infection criterion (OR 4.52, 95% CI 1.66–14.04). Altered parenchymal echogenicity accounted for 50 of the 54 outcome events (92.6%). Conclusion In adults imaged for suspected AP, diabetes mellitus was the only clinical variable associated with ultrasound findings suggestive of parenchymal infection. Because all patients had already been selected for imaging, because the outcome rests largely on a single subjective sonographic criterion, and because of the small number of exposed patients and wide confidence intervals, this association is hypothesis-generating and does not support an imaging decision rule.