Tuesday, April 23, 2013

Opioid Dose Alone May Not Influence Overdose Risk

Opioid FearsThere has been much concern about the safety of long-term opioid analgesics, with some studies reporting unexpectedly high rates of overdoses and deaths as doses are increased. Now, however, preliminary data suggest a contrary perspective in the ongoing debate; that is, higher-dose opioid prescribing may not be a primary influence on overdose.

In an abstract to be presented at the 32nd Annual Scientific Meeting of the American Pain Society in New Orleans, Louisiana, May 8-12, 2013, lead author Paul Coplan and colleagues observe that several prior studies have noted up to an 11-fold increased risk of opioid-analgesic overdose among patients with pain prescribed ≥120 mg versus ≤30 mg daily morphine-equivalent dosage [Coplan et al. 2013]. However, those earlier studies were potentially biased by at least two influences:

  1. They did not account for differences in indications, formulations, and opioid substances among patients receiving high versus low opioid doses, since the studies pooled all opioids together, and

  2. Results were not adjusted for the impact of opioid abuse, since abusers prefer higher doses and abuse is under-ascertained in the claims databases used in those other studies.

The objective of this present investigation by Coplan et al. was to reassess the relationship between opioid dose and overdose risk by controlling for those biases, and with separate analyses of 3 extended-release (ER) opioids: ER morphine tablets, fentanyl patch, and buprenorphine patch. To minimize the impact of abuse, the researchers used a medical records database in the United Kingdom maintained by general practitioners, who control access to healthcare and have extended relationships with patients, which presumably helps to minimize abuse of prescribed medications.

Person-time on opioids by dose was calculated, and overdose incidents were ascertained from diagnostic codes. Poisson regression was used to calculate relative risks of overdose by dose, controlling for age, gender, cancer, mental illness, and rescue opioids.

During a 5-year period, between 2005 through 2009, 38,861 patients (287 overdoses) were prescribed ER morphine, 23,909 fentanyl patches (108 overdoses), and 20,560 buprenorphine patches (56 overdoses). The relative risk (RR) of overdose among patients prescribed ≥120 mg versus ≤30 mg of morphine equivalent dose (MED) was…

  • RR=1.44 (95% CI: 1.04-1.99) for morphine,

  • RR=1.51 (95% CI: 0.59-3.86) for fentanyl patch,

  • RR=0.78 (95% CI: 0.36-1.72) for buprenorphine patch, and

  • RR=1.18 (95% CI: 0.90–1.55) for all 3 opioids combined.

The researchers conclude that the risk of opioid overdose among UK patients prescribed ≥120mg versus ≤30mg MED was not increased for buprenorphine patches, and the increase for morphine and fentanyl patches was lower than in previously published studies. Therefore, the risk of overdose at higher opioid doses may be inaccurately increased in claims analyses of United States patients that pool all opioids and do not control for opioid abuse.

COMMENTARY: In prior UPDATES we have discussed and critiqued 3 papers that claimed strong associations of higher-dose opioid analgesics and overdoses or deaths. These included studies by Gomes et al. 2011 [here], Bohnert et al. 2011 [here], and Dunn et al. 2010 [here].

All 3 of those studies were noted to have strong limitations and multiple sources of bias, resulting in low-quality evidence. Yet, they have been cited often as demonstrating that long-term opioid dosing ≥100 mg MED is unsafe as compared with lower doses. Conceptually, it may make sense that higher doses of opioid analgesics might incur greater risks and incidences of overdose; however, without valid and reliable evidence, this notion is speculation not science.

The present investigation by Coplan and colleagues brings another perspective into consideration, helping to balance the debate about long-term opioid safety. For one thing, in terms of absolute risks, opioid-analgesic overdoses were relatively rare events in the large population studied, as was also observed in the other 3 studies.

Using data in the Coplan et al. abstract, we calculated that the combined overdose incidence rate was 1.0 per 1,000 patient-years. This assumes that all patients were followed during the entire period of the study, which may not have been the case here (or, in the other 3 studies mentioned above). [Note: Concepts of prevalence, incidence, and survival rates in pain research were recently discussed in an UPDATE here.]

Overdose RRFurthermore, constructed from data in this investigation by Coplan et al., a forest plot of Relative Risk point-estimates and Confidence Intervals — comparing high (≥120 mg) versus low (≤30 mg) dose for each opioid, and the 3 combined — is shown in the figure. This clearly demonstrates that 3 of the 4 risk effects are not statistically significant (the CIs cross 1.0, the line of no effect), and the outcome for morphine is only marginally significant. These data reinforce the notion that higher opioid-analgesic doses may not be a prime influence on overdose.

While the abstract data appear to be somewhat compelling, there also are several very important limitations of this report for consideration:

  1. So far, this study has only been presented in abstract form, so results must be considered as preliminary until more formal publication in a peer reviewed journal. Additional data in a comprehensive report would help to clarify how the study was conducted and confirm veracity of the reported outcomes.

  2. Results in the abstract were based on a data-mining approach, which has many limitations; particularly, the potential influence of confounding variables that are not recorded in or extracted from the database. We have repeated cautioned that this methodology is essentially a retrospective computer-generated observational study; outcomes may be useful for generating hypotheses worthy of further investigation, but they do not provide a high-level of evidence confirming the strength or validity of associations. Note, however, that the same concerns were raised about the other 3 investigations of this topic mentioned above.

  3. The other 3 studies exploring this topic focused on databases from within the United States; whereas, this study by Coplan et al. relied on data from the UK. It is possible that geographic differences in healthcare delivery systems, prescribing patterns, patient medication-adherence behaviors, and/or other factors may have influenced outcomes — the potential for this merits further examination.

    Also, those other 3 studies focused at the low end on doses ≤20 mg of morphine or morphine-equivalent opioids; whereas, Coplan and colleagues used ≤30 mg MED as the low-dose threshold. Reasons for this are not explained in the abstract, but this greater threshold (≤30 mg MED) might have been more inclusive of overdose incidents and decreased Relative Risk estimates as compared with using a lower threshold (≤20 mg MED).

  4. The abstract authors are attributed as representing Purdue Pharma L.P., manufacturer of a buprenorphine patch product (Butrans®); therefore, potential commercial influences — such as “confirmation bias” — should be considered. However, this alone is rarely grounds for completely discounting results of studies that otherwise appear to be of some value.

Therefore, results of this study by Coplan et al. must be considered preliminary, and there is a dire need for better research investigating this important topic; particularly, well-designed, prospective, longitudinal studies in large populations of patients with pain prescribed a variety of opioid analgesics long-term. Along with that, results should be calculated using survival analyses that yield hazard rates, taking into account differences in patient observation periods and controlling for potentially confounding variables. Until such studies can be conducted, all presently available evidence should be considered from a balanced perspective, and there now appears to be some data to suggest that higher-dose opioid prescribing, in itself, may not incur greater overdose risk.

REFERENCE: Coplan P, Wentworth C, Downing J, et al. The risk of opioid overdose among patients receiving higher versus lower doses of extended-release opioids in the UK. J Pain. 2013(Apr);14(4-suppl):S18-Abstract 168 [access by subscription].

Proviso: Pain Treatment Topics is supported in part by medical education grants from Purdue Pharma L.P. (a maker of buprenorphine products) and Teva Pharmaceuticals (a maker of fentanyl-related products). Neither of those organizations had any role in the initiation, development, or review of this UPDATE article and they are in no way responsible for its content; all facts are from the sources cited, all opinions are expressly those of the author.

eNotifications Don’t Miss Out. Stay Up-to-Date on Pain-Topics UPDATES!
Register [here] to receive a once-weekly e-Notification of new postings.