Abstract

Background: Urinary Tract Infection (UTI) in diabetes mellitus patients can cause acute complications of diabetes and thus be life-threatening. The objective of this study was to propose a prediction score for UTI in diabetes mellitus patients hospitalized in the department of medicine at the Military Principal Hospital of Dakar.

Methods: We conducted a retrospective cross-sectional study from January 1, 2016, to December 31, 2019, including diabetes mellitus patients hospitalized in the study setting. We randomly sampled the study population in learning and validation datasets, and we built a prediction score for UTI from a multivariate logistic regression model. The Receiver Operating Characteristics (ROC) curve and its Area Under the Curve (AUC) was used to determine the discriminating power of the score.

Results: We enrolled 573 diabetes mellitus patients with a mean age of 63 ± 14 years and a male to female ratio of one, divided into two subgroups: a learning and a validation subgroup of 372 (65%) and 201 (35%), respectively. The prevalence of UTI was 27%. After a multivariate logistic regression, independent associated factors with UTI were: female gender (OR=1.9; 95%CI [1.1-3.3]), UTI history (OR=3.7; 95%CI (1.4-9.7)), diabetic neuropathy (OR=1.8; 95%CI (1-3.2)), urinary symptom (OR=6.9; 95%CI (3.6-13.3)), positive CRP (OR=2 .9; 95%CI (1.5-5.7)) and anemia (OR=1.8; 95%CI (1-3.2)). The AUC was 77.8% (95% CI (72.8%-82.9%)), 71.9% (95%CI (63.7%- 78.8%)) and 75.8% (95%CI (71.2%-79.7%)) on the learning, validation and total dataset, respectively.

Conclusion: A valid prediction score for UTI may help in its early diagnosis and prevent acute complications of diabetes mellitus.

Keywords: Urinary tract infection; Diabetes mellitus; Prediction score; Dakar