Determining the optimal Holter monitoring duration for detecting ventricular arrhythmia in dogs: a Bayesian approach.

Abstract

INTRODUCTION/OBJECTIVES: The objective of this study was to determine the optimal duration of ambulatory electrocardiography (Holter) monitoring for detecting ventricular arrhythmias (VAs) and to assess day-to-day spontaneous variability in dogs using seven-day recordings. ANIMALS, MATERIALS AND METHODS: Medical records of dogs undergoing the consecutive seven-day Holter monitoring were retrospectively reviewed. Dogs were included if the Holter monitoring was performed for collapse, arrhythmia-related symptoms, or documented VAs on prior electrocardiography. Dogs undergoing routine breed screening or evaluated for bradyarrhythmias or supraventricular tachycardia were excluded. Day-to-day variability in VA burden was assessed using the percent coefficient of variation (CV%). Bayesian probability modeling was used to analyze arrhythmia detection from days one through seven, identifying the point at which additional monitoring provided diminishing diagnostic returns. RESULTS: Of 124 Holter recordings reviewed, 87 met inclusion criteria. The percent coefficient of variation (CV%) for daily ventricular premature complex (VPC) frequency ranged from 8 to 188%. Bayesian analysis showed that detection of >1000 VPCs, >500 VPCs, or ventricular tachycardia episodes increased within the first three days of monitoring, with minimal improvement thereafter. Detection of >100 VPCs plateaued after two days. R-on-T VPCs exhibited continued detection increases through days four to five. Dogs receiving anti-arrhythmic drugs plateaued by day 3, whereas untreated dogs showed continued detection increases up to day 5. LIMITATIONS: The study's retrospective nature, modest sample size for treated dogs, and variability in anti-arrhythmic protocols may limit generalizability. CONCLUSIONS: A three-day Holter monitoring duration is recommended for detecting most VAs in dogs, while four days may be required for R-on-T event detection.