Abstract
BACKGROUND: Knowledge of incidence and risk factors for long-term opioid prescribing is critical for surgical patients. In this retrospective cohort study, we linked information available at the time of surgery with prescription data to ascertain characteristics associated with prolonged opioid therapy. METHODS: Patients (n = 6003) with claims in the Colorado All Payer Claims Database (APCD) were matched with 20,501 encounters in a clinical database. Rates of prescription filling were defined by at least one monthly opioid claim relative to the date of surgery. Associations of variables with claims during months 2-6 post-operatively ("long-term prescription filling") were evaluated, and significant variables were jointly modeled using binomial regression. RESULTS: Rates of patients filling opioid prescriptions preoperatively [month (M) relative to date of surgery] were 22%(-3 M), 24%(-2 M), and 27%(-1 M); after surgery, opioid fill rates were 62%(1 M), 28%(2 M), 24%(3), 24%(4 M), 23%(5 M), and 22%(6 M). The majority, 71-76%, of patients filling prescriptions in months 2-6 after surgery had also filled before surgery. In the binomial regression model, long-term opioid use was associated with prior opioid use (p < 0.0001), age ≥26 to <65 relative to age ≥ 65 (p < 0.0001), orthopedic surgery (p = 0.001), colorectal surgery (p = 0.003), multiple procedures (p < 0.0001), and worse physical status classification (p < 0.0001). CONCLUSIONS: Patients who had filled opioid prescriptions preoperatively comprised the majority of the group who filled long-term prescriptions. Surgical procedures were associated with discontinuation of previous opioid prescribing in some patients. For others, surgery marked the initiation of prolonged opioid therapy. Surgical encounters should include interventions aimed to reduce long-term opioid use.