The Use of Pattern Recognition to Augment Traditional Monitoring in the Prevention of Opioid Overdose Harm.

OBJECTIVES

The aim of the study was to investigate the correlation of a pattern recognition algorithm to the opioid overdose intervention activities of trained medical staff at a safe consumption site (SCS).

METHODS

Continuous physiological data were collected using the Masimo Radius PPG pulse oximeter from volunteer users of nonprescribed, unregulated opioids at a SCS. The algorithm retrospectively calculated opioid-induced respiratory depression (OIRD) severity scores (Opioid Halo scores) were compared to interventions recorded by SCS staff.

RESULTS

The study included data prospectively collected from 167 individuals, who underwent 370 sessions of intravenous injection of nonprescribed, unregulated opioids ( Fentanyl ). Interventions were documented for 150 sessions (~41%) by the SCS staff. The remaining 220 sessions had no interventions documented. The algorithm demonstrated a strong correlation with the intervention activities (Spearman ρ = 0.80, P  < 0.001). The area under the receiver operating curve for the correlation with intervention activities (ie, supplemental oxygen or naloxone administration) was 0.94. The OIRD severity scores were significantly higher ( P  < 0.001) in sessions requiring interventions compared to nonintervention sessions.

CONCLUSIONS

In this study, the algorithm generated OIRD severity scores had a strong correlation with the intervention activities provided by SCS staff who were blinded to the study pulse oximeter and algorithm scores. This suggests that the algorithm may be useful in detecting severe opioid-induced respiratory depression for which intervention is needed.