Managing cancer pain at the end of life with multiple strong opioids: a population-based retrospective cohort study in primary care

PLoS One. 2014 Jan 27;9(1):e79266. doi: 10.1371/journal.pone.0079266. eCollection 2014.


Background: End-of-life cancer patients commonly receive more than one type of strong opioid. The three-step analgesic ladder framework of the World Health Organisation (WHO) provides no guidance on multiple opioid prescribing and there is little epidemiological data available to inform practice. This study aims to investigate the time trend of such cases and the associated factors.

Methods: Strong opioid prescribing in the last three months of life of cancer patients were extracted from the General Practice Research Database (GPRD). The outcome variable was the number of different types of prescribed non-rescue doses of opioids (1 vs 2-4, referred to as a complex case). Associated factors were evaluated using prevalence ratios (PR) derived from multivariate log-binomial model, adjusting for clustering effects and potential confounding variables.

Results: Overall, 26.4% (95% CI: 25.6-27.1%) of 13,427 cancer patients (lung 41.7%, colorectal 19.1%, breast 18.6%, prostate 15.5%, head and neck 5.0%) were complex cases. Complex cases increased steadily over the study period (1.02% annually, 95%CI: 0.42-1.61%, p = 0.048) but with a small dip (7.5% reduction, 95%CI: -0.03 to 17.8%) around the period of the Shipman case, a British primary care doctor who murdered his patients with opioids. The dip significantly affected the correlation of the complex cases with persistent increasing background opioid prescribing (weighted correlation coefficients pre-, post-Shipman periods: 0.98(95%CI: 0.67-1.00), p = 0.011; 0.14 (95%CI: -0.85 to 0.91), p = 0.85). Multivariate adjusted analysis showed that the complex cases were predominantly associated with year of death (PRs vs 2000: 1.05-1.65), not other demographic and clinical factors except colorectal cancer (PR vs lung cancer: 1.24, 95%CI: 1.12-1.37).

Conclusion: These findings suggest that prescribing behaviour, rather than patient factors, plays an important role in multiple opioid prescribing at the end of life; highlighting the need for training and education that goes beyond the well-recognised WHO approach for clinical practitioners.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Analgesics, Opioid / therapeutic use*
  • Chronic Pain / drug therapy*
  • Chronic Pain / etiology
  • Cohort Studies
  • Drug Utilization / statistics & numerical data*
  • Humans
  • Models, Statistical
  • Neoplasms / complications*
  • Practice Patterns, Physicians' / statistics & numerical data
  • Primary Health Care / methods*
  • Retrospective Studies
  • Terminal Care / methods*


  • Analgesics, Opioid