Determining Mortality Likelihood of Opioid Drug Combinations using Decision Tree Analysis

Abstract The opioid epidemic is taxing medical infrastructure in terms of patient care and costs. Being able to prioritize care based on expected outcomes could improve patient safety and reduce costs. To investigate, we constructed a decision tree to examine what opioids in conjunction with heroin were most likely to lead to patient death versus their ability to survive. Our model demonstrated that heroin and fentanyl contributed to 77.60% of patient deaths, however, when heroin and fentanyl combinations included oxycodone, patient deaths decreased to 34.78% versus an increase to 81.94% without oxycodone. We speculate that patients already taking oxycodone have built a resistance to opioids allowing them to better survive. Patients taking heroin and fentanyl without oxycodone were more likely to experience cardiac, toxicity and respiratory symptoms whereas these symptoms decreased with the addition of oxycodone.