Thursday, December 10, 2009

Canadian Study Distorts Rx-Opioid-Related Deaths

EBPM Logo Understanding Evidence-Based Pain Management (EBPM)

Reports of increasing deaths associated with prescribed opioid analgesics are of concern, but also may portray a distorted picture of the scope and urgency of the problem. This was evident in a recent report on mortality trends in Canada, which serves as a lesson in why readers must be wary of how such data are presented in the literature and the press.

Canadian investigators supported by the Institute for Clinical Evaluative Sciences (ICES) and Ontario Ministry of Health and Long Term Care reviewed opioid prescribing trends and all deaths involving alcohol, drugs, or both in Ontario during 1991 to 2004 [Dhalla et al. 2009]. Their report, published in the December 8, 2009, edition of the Canadian Medical Association Journal (CMAJ) concluded that increases in opioid analgesic prescribing have directly contributed to an alarming rise in associated mortality, and they pay particular attention to concerns about oxycodone formulations. Sensational national news headlines proclaimed, “Canada’s pain reliever deaths double”; however, a close look at the data and their presentation suggests some alternative perspectives worth considering:
  • During the 14-year time period of the study there were a total of 7,099 deaths involving alcohol, drugs, or both; of those, less than half (3,406 or 48%) implicated opioids in the cause of death. Fatalities associated with opioids increased from 13.7 deaths per million residents to 27.2 deaths per million. In perspective, however, while there indeed was nearly a doubling, the population incidence rate rose from merely 0.0014% to 0.003%, making such events a rare occurrence even at the higher extreme in 2004. (From another viewpoint, the lifetime odds of being struck by lightning, at least in the U.S., is more than 5-times greater or 0.016% [data here].)

  • Curiously, the 3,693 deaths (52%) involving alcohol (possibly also including nonopioid drugs) were not mentioned by the authors as being of concern. And, further complicating the picture, more than a third (38%) of opioid-associated deaths involved multiple drugs, including possibly heroin (which confounds the focus on prescription opioids) and nonopioids (benzodiazepines, antidepressants, and/or alcohol).

  • Most of the victims (84%) had visited their physicians or the emergency room within 2 weeks prior to death, and analysis of records revealed anxiety, depression, or drug dependence were common diagnoses in addition to pain-related complaints. And, death was attributed to suicide in nearly a quarter of all cases (24%) involving opioids. So, while the availability of opioids was a factor, mental health problems may have been a more critical precipitating factor in a significant portion of all deaths. In some cases, the undertreatment of pain could have been a contributing factor in either suicide or in opioid over-use leading to unintentional death; however, the study does not address these vital issues of concern.

  • The study authors note that the introduction of long-acting oxycodone formulations in 2000 resulted in multifold increases in prescribing and associated deaths with that drug by 2004 (at one point in the text the authors state a 416% increase in deaths, in other places they indicate it as “5-fold”). However, according to the data, the 234 oxycodone-related deaths were ranked 4th in frequency, coming after morphine or heroin (or both), codeine, and methadone; also, in 92% of the oxycodone cases one or more additional drugs were implicated. The authors concede that there is no evidence of oxycodone itself being more intrinsically dangerous than other opioids; and, we might speculate that it was found to be a particularly effective analgesic, which contributed to increased prescribing and a natural occurrence of associated fatalities as part of the overall trend. Therefore, it is peculiar that oxycodone was targeted for such attention in this study, other than the long-acting formulation being a newly introduced agent during the data gathering period.

  • The authors provide two trend graphs — one for oxycodone-related deaths and a second for opioid-related deaths — which each show declines in fatalities between 2003 and 2004. It could be important and enlightening to know if this decreasing mortality trend continued in subsequent years. For example, we had previously shown a leveling-off of opioid-related problems in the U.S. during 2005 to 2008 [see blogpost 9/16/2009].

  • There is no indication in the study data of what proportion of decedents were patients for whom the opioids were prescribed, versus being diverted to other persons, and whether death was associated with accidental misuse or intentional abuse. Such factors could make significant differences in how the purported hazards of opioids are interpreted and portrayed.

  • Although the authors actually examined prescribing trends from 1991 to 2007, they made no attempt to assess the probability risk of prescribed-opioid fatalities; for example, one way this could be calculated is the number of deaths divided by the total number prescribed doses during the time period. We have previously shown that such risks are extremely low in the U.S. [see blogpost 9/12/2009] and would likely be similar or even lower in Canada.
All of this is not to belittle legitimate concerns about the misuse or abuse of opioids and the tragedy of analgesic-related deaths. There is no question about opioids being strong medications that can be fatally toxic if improperly prescribed and/or used. However, data manipulations that exaggerate opioid problems can be a misrepresentation of reality, and it is important to recognize that there is almost always more than one way to view, analyze, and present data as evidence depending on how one wishes to slant the perspective. In a commentary accompanying the study report in CMAJ, Fischer and Rehm [2009] note that “prescription-opioid-rich” environments throughout North America have created major public health challenges. Yet, they concede that these analgesics are essential therapeutic tools and caution regarding the possible “chilling effects” of tightened regulatory controls that could result in practitioners’ unwillingness to prescribe the medications. Such mixed messages are confusing and, when combined with a “data-rich” study report that obscures rather than clarifies the problems at hand, can lead to misguided conclusions by the unwary reader and in the mass media.

Note: In fairness we must acknowledge that alternative analyses of evidence, such as presented above, are only possible when article authors are intellectually honest in reporting sufficient data from their research, as was the case in this Canadian study.

> Dhalla IA, Mamdani MM, Sivilotti MLA, et al. Prescribing of opioid analgesics and related mortality before and after the introduction of long-acting oxycodone. CMAJ (Canadian Med. Assn.). 2009(Dec 8);181(12):891-896 [
article available here].
> Fischer B, Rehm J. Deaths related to the use of prescription opioids [commentary]. CMAJ (Canadian Med. Assn.). 2009(Dec 8);181(12):881-882 [
article here].