Construction of a risk prediction model of extended release oxycodone tablet-induced nausea and clarification of predictive factors

Nausea is a typical adverse event associated with opioids. In this study, we performed logistic regression analysis with the aim of clarifying the risk factors for nausea induced by extended-release oxycodone (EROXY). Furthermore, we constructed a decision tree (DT) model, a typical data mining method, to estimate the risk of oxycodone-induced nausea by combining multiple factors. A retrospective study was conducted on patients who newly received ER-OXY for cancer pain during hospitalization at Hokkaido University Hospital in Japan from April 2015 to March 2018. In logistic regression and DT analyses, the dependent variable was the presence or absence of nausea. Independent variables were the potential risk factors. First, univariate analyses were performed to screen potential factors associated with oxycodone-induced nausea. Then, multivariate and DT analyses were performed using factors with p-values <0.1 in the univariate analysis. Of 267 cases included in this study, nausea was observed in 30.3% (81/267). In multivariate logistic regression analysis, only female sex was extracted as an independent factor affecting nausea (odds ratio, 1.98). In the DT analysis, we additionally revealed that an age <50 years was a risk factor for nausea in female patients. Thus, our DT model indicated that the risk of ER-OXY-induced nausea was highest in the subgroup comprising females <50 years of age (66.7%) and lowest in male patients (25.1%). The DT model suggested that the factor of young women may be an increased risk of ER-OXY-induced nausea.