Use of a combination of routine hematologic and biochemical test results in a logistic regression model as a diagnostic aid for the diagnosis of hypoadrenocorticism in dogs.

OBJECTIVE To assess the discriminatory value for corticosteroid-induced alkaline phosphatase (CiALP) activity and other variables that can be measured routinely on a CBC and biochemical analysis for the diagnosis of hypoadrenocorticism in dogs. SAMPLE Medical records of 57 dogs with confirmed hypoadrenocorticism and 57 control dogs in which hypoadrenocorticism was suspected but ruled out. PROCEDURES A retrospective case-control study was conducted. Dogs were included if a CBC and complete biochemical analysis had been performed. Dogs with iatrogenic hypoadrenocorticism and dogs treated previously with glucocorticoids were excluded. Cortisol concentration for dogs with hypoadrenocorticism was <= 2 mu g/dL both before and after ACTH administration. Cortisol concentration for control dogs was > 4 mu g/dL before or after ACTH administration. RESULTS Area under the receiver operating characteristic (ROC) curve for CiALP activity was low (0.646; 95% confidence interval, 0.494 to 0.798). Area under the ROC curve for a model that combined the CiALP activity, Na-to-K ratio, eosinophil count, activity of creatine kinase, and concentrations of SUN and albumin was high (0.994; 95% confidence interval, 0.982 to 1.000). Results for this model could be used to correctly classify all dogs, except for 1 dog with poadrenocorticism and no electrolyte abnormalities. CONCLUSIONS AND CLINICAL RELhyEVANCE CiALP activity alone cannot be used as a reliable diagnostic test for hypoadrenocorticism in dogs. Combined results for CiALP activity, Na-to-K ratio, eosinophil count, creatine kinase activity, and concentrations of SUN and albumin provided an excellent means to discriminate between hypoadrenocorticism and diseases that mimic hypoadrenocorticism.