OBJECTIVES : The purpose of this study was to develop a predictive model for opioid-induced respiratory depression (OIRD) and oversedation to standardize risk assessment and improvement of patient care.

METHODS : A total of 75 hospitalized patients experiencing OIRD/oversedation were identified between 2010-2015. A case-control study was performed with 1: 2 ratios of cases and controls randomly selected from the electronic health record; 224 patients were included in the analysis. A sum score of non-opioid medications was developed with each sedative medication scoring one, and each medication with a sedative and respiratory effect scoring two. Diagnosis (cancer, surgery, other medical conditions), total number of long-acting opioids during the hospitalization (0, 1, 2), sex, age, administration routes of as needed opioids (oral, intravenous push, none), creatinine level (normal, low, high), sleep apnea (treated sleep apnea, untreated sleep apnea, without sleep apnea), sum score, and marital status (married, not married) were included in the model development. A stepwise binary logistic regression analysis was employed using SAS software 9.4.

RESULTS : A total of eight predictors was identified in the best model (Max-rescaled R-square: 0.4637; c: 0.852; Hosmer-Lemeshow test: p= 0.4829). Females (OR: 2.289, 95% CI: 1.073-5.108, p= 0.0364], using more long-acting opioid during the same hospitalization period (1 opioid: OR: 2.960; 95% CI: 1.37-6.549, p= 0.9179; 2 opioids: OR: 7.888, 95% CI: 1.438-41.494, p= 0.0627), high creatinine level (OR: 4.956, 95% CI: 2.063-12.429, p= 0.0004), untreated sleep apnea (OR: 48.448, 95% CI: 6.997->999.999, p= 0.0094), and unmarried status (OR: 2.390, 95% CI: 1.147-5.134, p= 0.022) put opioid-users at high risk of OIRD and oversedation.

CONCLUSIONS : Due to the complexity of medical conditions, co-medications, and lack of clear guidelines to follow-up, OIRD and oversedation remains a major issue in hospitals. This predictive model will assist clinical care providers to identify high-risk populations and prevent OIRD and oversedation.