Development of a novel scoring system to predict mortality in canine patients with infection

Abstract

Objectives

To develop a novel illness severity scoring system to identify canine patients with infection that are at higher risk of mortality.

Key Findings

Mentation (Modified Glasgow Coma Scale and the Acute Patient Physiologic and Laboratory Evaluation mentation score), heart rate, pH, P_vco₂, potassium, and total plasma protein were found to be associated with mortality on univariate analysis. Logistic regression found that mentation, heart rate, and P_vco₂ combined to form the following score: [(100 × mentation score) + (3 × heart rate) + (10 × P_vco₂], which was predictive of mortality (area under the receiver operating characteristic curve: 0.88). A cutoff of 900 provided a sensitivity of 80% and a specificity of 80%. If the cutoff was changed to 800, the sensitivity was 100% and the specificity was 50%.

Significance

A novel scoring system was developed to predict mortality in hospitalized dogs with confirmed or suspected infection. Further prospective evaluation in a larger patient cohort is necessary to validate this score.