PMU31 Defining Clinically Relevant Threshold for Chronic Opioid Use Using NOVEL Modelling Techniques

Abstract

Multiple efforts from policy makers and public health agencies have been made to stem the opioid epidemic. However, the variability in the literature in defining chronic use highlights the lack of empirical evidence on the duration of opioid therapy that may represent an increase in harm. The goal of this study was to determine the duration of opioid use that is associated with an increased risk of opioid-related emergency department (ED) visits, re-admission or death using novel modelling techniques to define a clinically relevant threshold for high risk opioid use. A prospective cohort of hospitalized patients was followed one year post-discharge. Marginal structural Cox PH models (MSM Cox) were used to determine the association between time-varying opioid use and the risk of the outcome.Non-linear, time-dependent effects as well asflexible weighted cumulative exposures (WCE) were modeled. Results from conventional MSM models showed current opioid use to be associated with a 66% (aHR): 1.66, 95% CI (1.17 – 2.36) increased risk of opioid-related ED visits/ hospitalizations or death. In comparison, the WCE models showed more than a two-fold increase in the risk (aHR: 2.30, 95% CI 1.20 - 4.93). The shape of the function suggests that the risk of opioid-related healthcare events is the highest in the first 30 to 40 days of opioid use with exposure beyond that period having no impact on someone’s current risk of the outcome. The non-linear modelling suggests largest increases in the risk with duration to up to 50 days with a pleating effect in the risk of harm with durations above 80 days. The method of recency-weighted cumulative opioid use allowed us to assess how careful consideration of modeling duration of exposure may improve the model’s fit and enhanced our understanding of the mechanism underlying potential adverse events of exposure to opioids.