Abstract
Chronic prenatal exposure to opioids often causes fetal opioid dependence that leads to neonatal opioid withdrawal syndrome (NOWS) shortly after delivery. Rat models of NOWS often require quantifying neonatal withdrawal behaviors using time-consuming, labor-intensive manual scoring methods. The goal of this study was to automate quantification of opioid withdrawal in neonatal rat pups. Accordingly, we used the animal behavior software Ethovision® XT to analyze archived videos of rat pups subjected to precipitated opioid withdrawal testing on postnatal day 0. We compared results obtained from Ethovision® XT with those previously obtained from manual scoring. Two endpoints reported by Ethovision® XT, Distance Moved (cm) and Movement Duration (s), had strong positive linear relationships with manually derived global withdrawal scores (GWS; R2 > 0.73). Sensitivity and specificity of each endpoint to discriminate presence and absence of low-grade withdrawal were assessed by receiver operator characteristic curve analysis, which indicated that Distance Moved and Movement Duration had excellent accuracy (AUC > 0.90). Finally, we analyzed main and interaction effects of prenatal treatment (with vehicle or mu opioid receptor full agonists) and postnatal challenge (with saline or an opioid receptor antagonist) on each endpoint and determined they were similar for the manual and automated methods. These results show that Ethovision® XT software can reliably quantify opioid withdrawal in neonatal rat pups with non-inferiority to manual scoring even in videos that were not originally purposed and optimized for Ethovision® XT analysis. This faster and less labor-intensive method of analysis is expected to accelerate progress in preclinical studies of NOWS.