Predictors of persistent opioid use in non-cancer older adults: a retrospective cohort study.

Abstract

Long-term opioid use and associated adverse outcomes have increased dramatically in recent years. Limited research is available on long-term opioid use in older adults. We aimed to determine the incidence and predictors of long-term or persistent opioid use (POU) amongst opioid-naïve older adults without a cancer diagnosis. This was a retrospective cohort study using five national administrative healthcare databases in New Zealand. We included all opioid-naïve older adults (≥65years) who were initiated on opioid therapy between January 2013 and June 2018. The outcome of interest was POU, defined as having continuously filled ≥1 opioid prescription within 91-180days after the index opioid prescription. Multivariable logistic regression was used to examine the predictors of POU. The final sample included 268,857 opioid-naïve older adults; of these, 5,849(2.2%) developed POU. Several predictors of POU were identified. The use of fentanyl (adjusted odds ratio (AOR) = 3.61; 95% confidence interval (CI) 2.63-4.95), slow-release opioids (AOR = 3.02; 95%CI 2.78-3.29), strong opioids (AOR = 2.03; 95%CI 1.55-2.65), Charlson Comorbidity Score ≥ 3 (AOR = 2.09; 95% CI 1.78-2.46), history of substance abuse (AOR = 1.52; 95%CI 1.35-1.72), living in most socioeconomically deprived areas (AOR = 1.40; 95%CI 1.27-1.54), and anti-epileptics (AOR = 2.07; 95%CI 1.89-2.26), non-opioid analgesics (AOR = 2.05; 95%CI 1.89-2.21), antipsychotics (AOR = 1.96; 95%CI 1.78-2.17) or antidepressants (AOR = 1.50; 95%CI 1.41-1.59) medication use were the strongest predictors of POU. A significant proportion of patients developed POU, and several factors were associated with POU. The findings will enable healthcare providers and policymakers to target early interventions to prevent POU and related adverse events.