A Prospective Study of Predictors of Long-term Opioid Use Among Patients With Chronic Noncancer Pain

Background: Chronic pain patients at increased risk of unfavorable pain and opioid misuse outcomes may be those most likely to use opioids long-term, but this has not been evaluated prospectively. Objectives: To ascertain whether pain prognostic risk, problem opioid use risk, and depression predict opioid use 1 year later among patients recently initiating opioid therapy with a moderate likelihood of long-term opioid use. Materials and Methods: Self-report and electronic health record data were collected from patients aged 45 + years who recently initiated opioid therapy (N = 762), in an integrated health care system. Logistic regression models tested whether baseline patient chronic pain prognostic risk, problem opioid use risk, depression, and expectations concerning continued opioid use independently predicted continuing use at 1 year (>= 30 d supply in the prior 4 mo). Results: At 1 year, 46% of participants continued to use opioids. Baseline problem opioid use risk score (adjusted odds ratio, 1.15; 95% confidence interval, 1.04-1.26) and expectations about continuing opioid use, but not pain prognostic risk score or depression, were significant predictors of 1-year opioid use. Compared with patients who thought continued opioid use unlikely, those who thought it was extremely or very likely had 4 times the odds of opioid use at 1 year (adjusted odds ratio, 4.05; 95% confidence interval, 2.59-6.31). Discussion: The strongest predictors of long-term opioid use were not patient-related or medication-related factors, but expectations about using opioids in the future. Asking about such expectations may be the easiest way to identify patients likely to continue opioid use long-term.