Predicting survival in cats with first-time diabetic ketoacidosis

Abstract

BACKGROUND: Acute Patient Physiologic and Laboratory Evaluation (APPLE) scores have not been reported in cats with diabetic ketoacidosis (DKA). HYPOTHESIS: In cats with DKA, APPLE scores will be significantly higher in non-survivors compared with survivors and these scores will predict mortality. ANIMALS: Sixty-eight cats with DKA. METHODS: Retrospective study. The APPLE scores, blood glucose concentration (BG), venous pH, and ketone concentrations were compared between survivors and non-survivors. Simple logistic regression was used to determine if these variables predict the binary variable of survival or non-survival, and if they did, an empirical optimal cut point for mortality prediction was calculated. RESULTS: The APPLEand APPLEscores were significantly higher in non-survivors (30 cats;and, respectively) compared with survivors (38 cats;and; P = .01 and P = .02, respectively). The APPLE(P = .03) but not the APPLEscores (P = .06) predicted mortality. For every 1 unit increase in the APPLEscore, the odds of death increased by 1.08 (95% confidence interval [CI], 1.006-1.17; P = .03). Median BG was significantly higher in non-survivors (431 mg/dL; range, 260-832 mg/dL) compared with survivors (343 mg/dL; range, 256-738 mg/dL; P = .01) and BG predicted mortality (P = .02). For every 1 mg/dL increase in BG, the odds of death increased by 1.004 (95% CI, 1.0006-1.008). Empirical optimal cut points for APPLEand BG mortality prediction were 24.5 and 358 mg/dL, respectively. CONCLUSIONS AND CLINICAL IMPORTANCE: The APPLEscore and BG predict mortality in cats with DKA and can be used to stratify populations by risk of mortality in clinical trials of DKA in cats.