Work

Longitudinal and Machine-Learning Analysis of Concentration-Dependent Clinicopathologic Changes During Phenobarbital Treatment in Dogs

Year: 2026
Type: article
Abstract: Phenobarbital (PB) is widely used for long-term management of canine epilepsy, but treatment may alter hepatic biochemical variables and, less commonly, cause idiosyncratic hematological reactions. This retrospective observational repeated-measure study evaluated associations between serum PB concentration and concurrent hepatic and hematological findings and explored their predictive value. The dataset comprised 233 laboratory visits from 67 dogs. Repeated-measures generalized estimating equations, within-dog and longitudinal analyses, receiver operating characteristic analyses, and dog-grouped machine-learning models were applied. Serum PB concentration showed its clearest association with alkaline phosphatase (ALP). Each 10 μg/mL increase in PB was associated with a 37.6% increase in ALP activity (95% confidence interval [CI], 11.6–69.7%; p = 0.0028); after adjustment for age, sex, neutering status, and diagnostic category, the increase was 34.9% (95% CI, 10.2–65.0%; p = 0.0036) and remained significant after false-discovery-rate correction. No comparable concentration-dependent relationship was identified for the principal hematological outcomes. A PB concentration > 30 μg/mL was relatively specific but poorly sensitive for hepatic biochemical abnormalities, and machine-learning models showed only modest or poor discrimination. Serum PB concentration should therefore be interpreted together with serial biochemical and hematological measurements rather than as a stand-alone indicator of treatment safety.
Language: English
Field: Medicine
Open Access status: gold
Estimated APC paid: $1,930
OpenAlex ID: W7233825816
FWCI:
Cites: 51
Cited by:
Related to: 10