Mudumbai S, COIN-Palo Alto; Unger-Hu K, COIN-Palo Alto; Oliva E, COIN-Palo Alto; Lewis E, COIN-Palo Alto; Trafton J, COIN-Palo Alto; Mariano ER, VA Palo Alto; Stafford RS, Stanford University; Wagner T, COIN-Palo Alto; Clark D, VA Palo Alto;
Objectives:
Despite widespread use of perioperative opioids, their standards of care remain poorly defined. We hypothesized that preoperative use would better predict time-to-cessation of postoperative use than pain, substance-use disorder, or affective disorder diagnoses.
Methods:
We examined data for all VA patients who underwent surgery and discharged after stays > = 1 day in fiscal year 2011. We excluded metastatic cancer patients and included patients with > 1 opioid prescription within 90 days of discharge. We based assessment of preoperative use on opioid prescriptions taken within 180 days before admission. We categorized patients' opioid use as: 1) opioid-naïve, 2) tramadol only, 3) short acting(SA) on an acute or intermittent basis ( < = 90 days), 4) short-acting on a chronic basis, ( > 90 days) or 5) any long-acting(LA). We calculated postoperative time-to-opioid-cessation, from Day 1 to Day 365. We defined cessation as 90, consecutive, opioid-free days. We calculated descriptive statistics and developed a Cox regression model with a priori identified predictors including time-dependent covariates for adjunctive, postoperative pharmacotherapeutics . Sensitivity analyses used cessation definitions of 30 or 180 consecutive days.
Results:
Our overall sample (n = 64,391) consisted primarily of men 55-65 years old (47.6%), white (79.9%), urban, married, who underwent diverse surgeries and had significant comorbidity burdens (Charlson score 2+ = 38.5%). Many had chronic pain (56.9%) and took adjunctive pain pharmacotherapy (e.g., non-opioid analgesics; 49.7%) following discharge. Approximately 75% ceased postoperative opioids by one year. Although median time-to-cessation was 43 days, greater preoperative levels of opioid were associated with longer times-to-cessation postoperatively. Preoperative use of SA opioids on an acute/ intermittent basis was associated with substantially greater risk than taking no opioids (HR = 0.51; 95% CI = 0.50-0.52,p < 0.001), but had lower risk than taking tramadol only (HR = 0.33; 95% CI = 0.32-0.34,p < 0.001), SA opioids on a chronic basis (HR = 0.11; 95% CI = 0.11-0.12,p < 0.001), or LA opioids (HR = 0.11; 95% CI = 0.10-0.12,p < 0.001).
Implications:
Preoperative opioid use better predicts extended postoperative use than pain, substance-use or affective disorder diagnoses.
Impacts:
If preoperative opioid status may be a modifiable risk factor for extended opioid use, focus on the preoperative management of surgical patients' opioid status may be necessary.