Agreement between definitions of pharmaceutical opioid use disorders and dependence in people taking opioids for chronic non-cancer pain (POINT): a cohort study

Background: Classification of patients with pharmaceutical opioid use disorder and dependence varies depending on which definition is used. We compared how WHO’s ICD-10 and proposed ICD-11 and the American Psychiatric Association’s DSM-IV and DSM-5 classifi ed individuals in a community-based sample of Australians with chronic non-cancer pain for which opioids have been prescribed.

Methods: We studied participants in the Pain and Opioid IN Treatment (POINT) cohort, a 2 year prospective cohort study of 1514 people prescribed pharmaceutical opioids for their chronic pain who were recruited in 2012–13 from community-based pharmacies across Australia. After giving patients the Composite International Diagnostic Interview about their opioid use, we assessed which patients would be categorised as having disorders of pharmaceutical opioid use by ICD-10, the draft ICD-11, DSM-IV, and DSM-5. We examined agreement between classification systems, and tested the unidimensionality of the syndrome with confirmatory factor analysis.

Findings: We included 1422 participants (median time of pain disorder 10 years [IQR 5–20]; median length of strong opioid prescription 4 years [IQR 1·5–10·0]; mean age 58 years). Similar proportions of individuals met lifetime criteria for dependence with DSM-IV (127; 8·9%), ICD-10 (121; 8·5%), and ICD-11 (141; 9·9%). Criteria in DSM-5 classified 127 (8·9%) participants with moderate or severe use disorder. There was excellent agreement between ICD-10, ICD-11 and DSM-IV dependence (κ>0·90). However, there was only fair to moderate agreement between ICD-10 and DSM-IV dependence diagnoses, and DSM-5 use disorder (mild, moderate, or severe). There was only good agreement between moderate to severe use disorder in DSM-5 and the other definitions. Criteria for all definitions loaded well on a single factor; the best model fit was for the definition for dependence in the draft ICD-11, the worst was in DSM-5.

Interpretation: Classification of problematic pharmaceutical opioid use varies across editions of ICD and DSM. The much lower levels of agreement between DSM-5 and other definitions than between other definitions might be attributed to DSM-5 containing an increased number of criteria and treating dependence and problematic use as a continuum. The more parsimonious ICD-11 dependence definition showed excellent model fit and excellent agreement with previous classificatory systems.