The Association Between Preoperative Opioid Exposure and Prolonged Postoperative Use

Charles Katzman, BS; Emily C. Harker, BS; Rizwan Ahmed, BS; Charles A. Keilin, BS; Joceline V. Vu, MD; David C. Cron, MD, MS; Vidhya Gunaseelan, MS, MHA; Yen-Ling Lai, MS, MPH; Chad M. Brummett, MD; Michael J. Englesbe, MD; Jennifer F. Waljee, MD, MPH


Annals of Surgery. 2021;274(5):e410-e416. 

In This Article


Our cohort included 191,043 patients who underwent an elective surgical procedure and filled an initial postoperative opioid prescription during the study period, of which 46,206 (24%) underwent a major procedure and 144,837 (76%) underwent a minor procedure. Of the patients included in our study, 134,715 (71%) were female, 137,152 (72%) were white, 21,393 (11%) were African American, and 20,720 (11%) were Hispanic. Tobacco use ranged from 25.1% among individuals who were opioid-naïve to 49.7% among patients with recent intermittent use. Two percent of individuals who were opioid-naïve had alcohol or substance use disorders, 5% of patients with remote intermittent opioid use and 10% of individuals with recent intermittent opioid use. Pain disorders were more prevalent among individuals with recent intermittent use with 80% had arthritis and 61% with back pain, compared to opioid-naïve who only 46% had arthritis and 24% had back pain. Similarly, there was a higher prevalence of mood disorders among recent intermittent users compared to opioid-naïve patients (40% vs 16%, respectively). Patient characteristics in the context of varying levels of preoperative opioid use are detailed in Table 1.

Preoperative use Classification Groups

In this cohort, 78,145 (41%) filled an opioid prescription in the year before surgery. Of these, 60,024 (31%) were minimal users, 18,121 (10%) were intermittent users, with 11,264 (6%) with remote intermittent use and 6,857 (4%) with recent intermittent use. The specific characteristics related to dose, recency, duration, continuity, number of opioid prescription fills for preoperative opioid users are described in Table 2. Patients with minimal use (n = 60,024, 31%) had low dose exposure with a median total dose of 180 [interquartile range (IQR) 200] OMEs over a median duration of 1 (IQR 0) month of varying recency. Individuals with remote intermittent (n = 11,264) use had low dose, remote, short-duration use, with a median total dose of 650 (IQR 820) OMEs over a median duration of 3 (IQR 2) months of opioid use. In contrast, those with recent intermittent use had slightly higher total doses with a median dose of 2125 (IQR 2850) OMEs, more recent fills, with medium-course use over median of 6 (IQR 3) total months with some discontinuity of fill.

Predictors of Postoperative Persistent Opioid use

Table 3 details the multivariable analysis describing the patient-level covariates associated with prolonged postoperative opioid use. Preoperative opioid exposure group was independently associated with prolonged use after controlling for patient- and procedural- characteristics. For example, patients with minimal exposure were at higher likelihood of developing prolonged use [odds ratio (OR) = 1.96; 95% confidence interval (CI): 1.89–2.03] compared with opioid-naïve patients. Similarly, patients with remote intermittent use (OR = 4.69; 95% CI: 4.46–4.93) and recent intermittent use (OR = 12.17; 95% CI = 11.49–12.90) were at substantially higher risk than patients who were opioid-naïve before surgery. Other factors significantly associated with higher prevalence of persistent opioid use included preoperative pain disorders, particularly arthritis (OR 1.52, 95% CI 1.46–1.57, P < 0.001); preoperative benzodiazepine use (OR 1.32, 95% CI 1.28–1.37, P < 0.001), history of tobacco use (OR 1.30, 95% CI 1.26–1.34, P < 0.001), alcohol or substance abuse disorders (OR 1.28, 95% CI 1.18–1.38, P < 0.001), mood disorders (OR 1.19, 95% CI 1.15–1.24, P < 0.001), and an initial postoperative dose of OME ≥300 (1.18, 95% CI 1.13–1.25, P < 0.001). When looking at the type of opioid prescribed during the preoperative period, we observed that tramadol was associated with a higher odds of developing persistent use (OR 1.37, 95% CI 1.30–1.44, P < 0.001), whereas oxycodone was correlated with a lower odds of persistent use (OR 0.88, 95% CI 0.84–0.92, P < 0.001). Hydrocodone was found to have a slight increased risk for the development of persistent use (OR = 1.09; 95% CI 1.04–1.15, P < 0.001) and there was no significant change in risk for development of persistent use for the group of other opioids (OR 1.03, 95% 0.97–1.09, P = 0.316). Notably, 25% of the cohort filled opioids of multiple types (Table 4).

In this cohort, the adjusted risk of developing persistent opioid use was significantly related to preoperative opioid use (Figure 1). Notably, the adjusted likelihood of developing persistent opioid use among individuals with even minimal opioid use in the year before surgery was higher than that of the opioid-naïve patients (12% vs 6%, P < 0.001). Similarly, patients with remote intermittent use and recent intermittent had higher adjusted rates of prolonged use compared with opioid-naïve patients (29% and 75% respectively, P < 0.001). Finally, postoperative average daily dose exceeded their preoperative average daily dose for 10% of the patient cohort.

Figure 1.

Likelihood of persistent opioid use (%) by preoperative use (opioid-naïve, minimal use, remote intermittent, recent intermittent).