Multi-Institutional Retrospective Analysis of Prognostic Scoring Systems for Dogs With Acute Pancreatitis (504 Dogs)

Abstract

Background

Acute pancreatitis (AP) in dogs has a broad clinical presentation and variable progression, making prognostication challenging.

Hypothesis/Objectives

(i) To compare predicted prognosis for death and for prolonged (≥ 5 days) hospitalization across scoring schemes for AP in dogs and (ii) to predict concordance of each scoring scheme with death and for prolonged hospitalization.

Animals

Five hundred four client-owned dogs.

Methods

Multi-institutional retrospective study. Data extracted from medical records included: signalment, history, physical examination findings, diagnostic results, length of hospitalization, and death. Five prognostic schemes (OS, CSI, APPLE_full, CAPS, MCAI) were calculated for each dog.

Results

Overall concordance was low. Only APPLE_full (p = 0.004) and MCAI (p = 0.01) scores were significantly different between survivors and non-survivors. Overall, APPLE_full had the greatest concordance (0.632, 95% CI: 0.592–0.672) with length of hospitalization. Of the other more pancreatitis-specific schemes, MCAI had the greatest concordance (0.576, 95% CI: 0.567–0.635) with length of hospitalization, while CSI had the lowest concordance with length of hospitalization (0.525, 95% CI: 0.494–0.556).

Conclusions and Clinical Importance

On a population level, APPLE_full and MCAI had the greatest predictive discrimination between dogs of normal and prolonged hospitalization. If an individual dog has any of the 5 prognostic score schemes above the proposed cut-off for death, it should be interpreted with caution because of the low case fatality rate.