Predictors of transitioning to incident chronic opioid therapy among working-age adults in the United States

J. Douglas Thornton, Nilanjana Dwibedi, Virginia Scott, Charles D. Ponte, Douglas Ziedonis, Nethra Sambamoorthi, Usha Sambamoorthi

Research output: Contribution to journalArticlepeer-review

11 Scopus citations

Abstract

BACKGROUND: Opioids have been prescribed and used for chronic noncancer pain at prolific rates in the United States during the past 2 decades. Patients who transition to incident chronic opioid therapy are at increased risk for significant negative health consequences, including cardiovascular risk, endocrine disorders, opioid use disorder, and death. OBJECTIVE: To identify the leading predictors associated with transitioning to incident chronic opioid therapy among working-age adults without cancer. METHODS: This retrospective observational cohort study is based on medical and pharmacy claims of a nationally representative sample of adults enrolled in commercial health insurance plans. Standard parametric (logistic regressions) and nonparametric methods based on a decision tree were used for prediction. To facilitate comparison with the available published literature, we also present adjusted odds ratios (AORs) and 95% confidence intervals (CIs). The 10% random sample of 491,442 patients included in the study who were working-age adults (age, 28-63 years) were insured in a commercial health plan, did not have cancer, and initiated opioid therapy between January 2007 and May 2015. Transition to incident chronic opioid therapy was defined as having claims for at least a 90-day supply of opioids within 120 days after the index date (ie, initiation of opioid therapy). Predictive models used for the analysis comprised a comprehensive list of factors available in the claims data, including opioid regimen characteristics, pain conditions, physical and mental health conditions, concomitant medications use (ie, benzodiazepine, stimulants, nonopioid analgesics, and polypharmacy), patient characteristics, and health insurance type. RESULTS: In our sample, the transition to incident chronic opioid therapy was 1.3% and pain-specific diagnoses were documented for only one-third (31.7%) of patients. The 4 leading predictors of chronic opioid therapy were opioid duration of action (AOR, 12.28; 95% CI, 8.06-18.72), the parent opioid compound (eg, tramadol vs codeine; AOR, 7.26; 95% CI, 5.20-10.13), the presence of conditions that are very likely to cause chronic pain (AOR, 5.47; 95% CI, 3.89-7.68), and drug use disorders (AOR, 4.02; 95% CI, 2.53-6.40). CONCLUSIONS: The initial opioid regimen’s characteristics are powerful predictors of chronic opioid therapy. Predictive algorithms created from readily available claims data can be used to develop real-time predictions of the future risk for a patient’s transition to chronic opioid use.

Original languageEnglish
Pages (from-to)12-21
Number of pages10
JournalAmerican Health and Drug Benefits
Volume11
Issue number1
StatePublished - Feb 2018

Keywords

  • Chronic opioid use
  • Chronic pain
  • Databases
  • Decision tree
  • Noncancer pain
  • Opioid regimen
  • Pharmacoepidemiology
  • Predictive modeling

Fingerprint Dive into the research topics of 'Predictors of transitioning to incident chronic opioid therapy among working-age adults in the United States'. Together they form a unique fingerprint.

Cite this