Derivation of a quick Pitt bacteremia score to predict mortality in patients with Gram-negative bloodstream infection.

PURPOSE

This retrospective cohort study derived a "quick" version of the Pitt bacteremia score (qPitt) using binary variables in patients with Gram-negative bloodstream infections (BSI). The qPitt discrimination was then compared to quick sepsis-related organ failure assessment (qSOFA) and systemic inflammatory response syndrome (SIRS).

METHODS

Hospitalized adults with Gram-negative BSI at Palmetto Health hospitals in Columbia, SC, USA from 2010 to 2013 were identified. Multivariate Cox proportional hazards regression was used to determine variables associated with 14-day mortality.

RESULTS

Among 832 patients with Gram-negative BSI, median age was 65 years and 449 (54%) were women. After adjustments for age and Charleston comorbidity score, all five components of qPitt were independently associated with mortality: temperature < 36 °C [hazard ratio (HR) 3.02, 95% confidence interval (CI) 1.95-4.62], systolic blood pressure < 90 mmHg or vasopressor use (HR 2.40, 95% CI 1.37-4.13), respiratory rate ≥ 25/min or mechanical ventilation (HR 3.01, 95% CI 1.81-5.14), cardiac arrest (HR 5.35, 95% CI 2.81-9.43), and altered mental status (HR 3.99, 95% CI 2.44-6.80). The qPitt had higher discrimination to predict mortality [area under receiver operating characteristic curve (AUROC) 0.85] than both qSOFA (AUROC 0.77, p < 0.001) and SIRS (AUROC 0.63, p < 0.001). There was a significant difference in mortality between appropriate and inappropriate empirical antimicrobial therapy in patients with qPitt ≥ 2 (24% vs. 49%, p < 0.001), but not in those with qPitt < 2 (3% vs. 5%, p = 0.36).

CONCLUSIONS

The qPitt had good discrimination in predicting mortality following Gram-negative BSI and identifying opportunities for improved survival with appropriate empirical antimicrobial therapy.