Predicting complicated outcomes in spinal cord injury patients with urinary tract infection: Development and internal validation of a risk model

Context/objective: Patients with chronic SCI hospitalized for UTI can have significant morbidity. It is unclear whether SIRS criteria, SOFA score, or quick SOFA score can be used to predict complicated outcome. Design: Retrospective cohort study. A risk prediction model was developed and internally validated using bootstrapping methodology. Setting: Urban, academic hospital in St. Louis, Missouri. Participants: 402 hospitalizations for UTI between October 1, 2010 and September 30, 2015, arising from 164 patients with chronic SCI, were included in the final analysis. Outcome/measures: An a priori composite complicated outcome defined as: 30-day hospital mortality, length of hospital stay >4 days, intensive care unit (ICU) admission, and hospital revisit within 30 days of discharge. Results: Mean age of patients was 46.4 ± 12.3 years; 83.6% of patient-visits involved males. The primary outcome occurred in 278 (69.2%) hospitalizations. In multivariate analysis, male sex was protective (odds ratio [OR], 0.43; 95% confidence interval [CI], 0.18-0.99; P = 0.048) while Gram-positive urine culture (OR 3.07; 95% CI, 1.05-9.01; P = 0.041), urine culture with no growth (OR, 1.69; 95% CI, 1.02-2.80; P = 0.041), and greater SOFA score (for one-point increments, OR, 1.41; 95% CI, 1.18-1.69; P < 0.001) were predictive for complicated outcome. SIRS criteria and qSOFA score were not associated with complicated outcome. Our risk prediction model demonstrated good overall performance (Brier score, 0.19), fair discriminatory power (c-index, 0.72), and good calibration during internal validation. Conclusion: Clinical variables present on hospital admission with UTI may help identify SCI patients at risk for complicated outcomes and inform future clinical decision-making.