Abstract
Opioids in patients with COPD: results of mortality should be interpreted with caution regarding exclusion criteria http://ow.ly/IJbd304nC36
To the Editor:
We read with interest the study by Vozoris et al. [1] describing the risk of adverse outcomes associated with opioid use in elderly chronic obstructive pulmonary disease (COPD) patients. There is currently a strong debate on the prescription of opioids in patients with COPD. While several studies seemed to demonstrate the harmlessness of opioids in this population [2–4], two recent studies including the one by Vozoris et al. [1] challenge this result [5]. The large number of included patients and the excellent method used in these two studies strongly support the idea of an excess morbidity or mortality associated with the opioid use in COPD patients. There is, however, one limitation that needs to be taken into consideration.
Vozoris et al. [1] explain that patients receiving palliative care in the year prior to the index date were excluded. This seems appropriate since the use of morphine is recommended in the treatment of end-of-life dyspnoea. Nevertheless, we believe that this exclusion criterion is not strict enough to avoid a bias in the analysis and interpretation of mortality risk. Indeed, discussion of palliative care in COPD patients often arises in the context of acute respiratory failure [6]. Over half of pulmonologists claimed that end-of-life decisions in COPD patients occur during/after a major exacerbation [7, 8]. It is unlikely that patients receiving opioid drugs following a recent end-of-life decision were excluded from the study by Vozoris et al. [1]. This potential bias may explain why no significant association was observed between opioid use and intensive care unit admissions in both the primary analysis and the sensitive analysis. Thus, the mortality results should be interpreted with caution.
In conclusion, the debate on the safety of morphine in COPD patients will remain open until further prospective studies refute or confirm the results published by Vozoris et al. [1]. Based on table 3 in their article [1], a randomised placebo controlled study would require the inclusion of 17 664 patients to detect a significant difference (alpha=0.05 and power=0.8) in COPD or pneumonia-related mortality, while 7746 patients would be necessary to detect a difference in all-cause mortality.
Footnotes
Conflict of interest: None declared.
- Received August 12, 2016.
- Accepted August 14, 2016.
- Copyright ©ERS 2016