A pair of recently published investigations found that elderly patients with chronic pain taking opioid analgesics had higher risks of serious adverse events (AEs) compared with those taking NSAIDs or coxibs, and the opioids varied among themselves in AE potential. This would appear to caution against opioid use in these patients and contradict current guidelines; however, a closer look at the research evidence casts some doubt on its validity and clinical relevance.
Acute and chronic pain are prevalent conditions in older patients and prior surveys have found that undertreated pain or inadequate pain relief are common in these patients. And, as we discussed [here], this situation worsens as individuals approach later years of life, with unrelieved pain affecting a quarter of patients during the last two years of life, increasing to affect nearly half of them during the last month of life. We further reported [here] that persons older than 65 years of age are less likely to receive strong analgesics for moderate to severe pain than younger patients complaining of comparable pain.
The American Geriatrics Society recommended in updated guidelines that nonselective nonsteroidal antiinflammatory drugs (NSAIDs) and the more selective COX-2 inhibitors (coxibs) should be rarely considered in older populations [AGS Panel 2009]. Rather, they propose that all elderly patients with moderate to severe pain, pain-related functional impairment, or diminished quality of life due to pain be considered for opioid therapy. They state, “… evidence that use of NSAIDs and COX-2 inhibitors may result in serious and life-threatening gastrointestinal and cardiovascular adverse events or gastrointestinal bleeding has shifted attention to opioids, especially for older patients who may be at particular risk for NSAID-related adverse effects.” We have previously discussed evidence for the various harms associated with NSAIDs and coxibs in our ongoing series on NSAID safety [here].
Therefore, the recent publication of two articles in the December 13/27, 2010 issue of the Archives of Internal Medicine by researchers at Brigham and Women’s Hospital, Boston, was surprising. The investigators, led by Daniel H. Solomon, MD, MPH, found that opioid analgesics are associated with more adverse events (AEs) among older adults with arthritis than either NSAIDs or coxibs — including cardiovascular complications, bone fractures, gastrointestinal bleeding, hospitalizations, and all-cause mortality — and that different types of drugs within the opioid class incur different risks of those serious adverse events.
Data for both articles were derived from medical records of Medicare beneficiaries diagnosed with conditions requiring prescribed analgesics in two states, Pennsylvania and New Jersey, and covering a 7-year period (January 1999 — December 2005). This provided a vast pool of subjects >65 years of age and, using a sophisticated computerized methodology, the researchers were able to generate large, well-matched cohorts of subjects in each treatment group for a retrospective analysis and comparison of data.
In the first report [Solomon et al. 2010A], researchers examined the comparative safety of NSAIDs, coxibs, and opioids among 12,840 Medicare beneficiaries diagnosed with osteoarthritis (roughly 90% of subjects) or rheumatoid arthritis (10%), and who were prescribed one of those types of analgesics (exactly 4,280 subjects per group). Opioid users experienced 68% higher rates overall of serious AEs. For example, the risk of fractures among opioid users was 5 times greater than with coxibs and 4 times greater than with NSAIDs. Opioids and coxibs appeared to be associated with greater cardiac risks than NSAIDs, and opioid use was associated with a higher incidence rate of hospitalization due to AEs and greater all-cause mortality than either coxibs or NSAIDs. Unexpectedly, opioids and NSAIDs exhibited equivalent incidence rates of gastrointestinal AEs, and these rates were nearly twice that of coxibs.
In the second report [Solomon et al. 2010B], the same team of investigators focused on those Medicare beneficiaries who specifically received opioids for any type of noncancer pain between 1996 and 2005. They compared the rates of adverse events after 30 and 180 days among 31,375 patients taking either codeine, hydrocodone, oxycodone, propoxyphene, or tramadol (n=6,275 in each group). The risk of gastrointestinal AEs was low and similar for all 5 of the opioids studied, and 30 days after beginning opioid therapy the risk of cardiovascular AEs also was similar across the 5 agents. However, after 180 days, the risk of cardiovascular AEs was increased in those taking codeine. Using hydrocodone as the reference point, risk of fracture was 79% lower in subjects taking tramadol and 46% lower with propoxyphene. Also compared with those taking hydrocodone, hospitalization due to an opioid-related AE was greatest with codeine, and death from any cause was more than twice as likely among those taking codeine or oxycodone. Of interest, the observed risks for any opioids were not explained by the dosage levels being prescribed and, curiously, there actually were some decreases in relative risks at the highest doses.
Findings of these studies appear to provide new evidence supporting the relative safety of NSAIDs, contradicting other research in this regard, while raising safety concerns about opioid analgesic therapy in older patients with nonmalignant pain conditions. Furthermore, the data “do not agree with a commonly held belief that all opioids are associated with similar risk,” the authors write [Solomon et al. 2010B]. They propose that the differential risks of opioids, when translated into numbers needed to harm, are substantial and might be considered as clinically significant. Of particular concern were unexpected increased AE risks with codeine.
CAVEATS & COMMENTARY: At first glance, these two studies would appear to bode poorly for opioids and muddle recommendations for what is safest to prescribe for pain in elderly patients. Upon closer examination from an evidence-based pain management (EBPM) perspective, this line of research is of interest but also may be misleading. The authors themselves concede that their research produced strong statistical associations of selected AEs with the analgesics examined but the data cannot be used to infer causation. This, and other aspects of the methodology, cast some doubt on the clinical validity and applicability of the findings.
We have noted before that retrospective studies of this sort have the advantage of tapping into a vast reservoir of data that lends a high level of statistical power to the analyses. At the same time, however, the Medicare beneficiary database was not designed to answer the specific questions (hypotheses) asked of the researchers; so, the subsequent “data mining” process to extract seemingly pertinent information for drawing conclusions raises questions of validity. The researchers correctly note that prospective, randomized, controlled trials would be the most valid approach for assessing relative analgesic harms in the elderly, but this approach would be exceedingly difficult and probably unethical. Therefore, at the least, the findings of these present studies must be considered cautiously for the following reasons:
- The cohorts in both studies (8 groups in all) were extremely well-matched; however, all subjects were roughly 80 years of age on average, predominantly female (generally over 80%), and white race (87%-92%). A majority of subjects in each group also had preexisting hypertension (range 59%-71%). Furthermore, the subjects were drawn from a limited region; ie, two adjacent eastern states. All of these factors considered, whether or not the composition of the groups is representative of the overall population of elderly patients across the U.S., or in any typical clinical practice, might be questioned.
- In both studies, data were unavailable for inclusion in the analysis regarding critical variables, such as: concurrent over-the-counter (OTC) analgesic or other medication use, body mass index, alcohol and/or tobacco use, pain levels and functional status of the subjects, or cause of death in the mortality cases. Any of these could have been confounding factors that may not have been equally distributed across groups and, thereby, biased the results — a deficiency that the authors duly acknowledge in their reports.
- In the case of opioids, the authors further concede that “sicker patients” — having worse pain for longer periods of time — may have been selectively prescribed the stronger analgesics. This greater morbidity, itself, may have contributed to increased risks for AEs. Furthermore, a companion editorial to the articles, by Becker and O’Connor , observes that is it highly likely that many subjects in the opioid groups also were taking OTC NSAIDs, since these are routinely recommended along with opioids for arthritis; hence, the opioid cohort was essentially ‘contaminated’ in the study comparing opioids with NSAIDs and coxibs, and there may have been unknown confounding factor across the opioid groups in the second study.
- Even known factors may have been unevenly distributed across groups and inadequately controlled. For example, in the first study, the authors observe that a history of falls and fractures was more common in persons taking opioids than among those taking coxibs or NSAIDs. Therefore, based on history alone, it might be expected that falls/factures would be more prevalent in the opioid-using group — which, indeed, was reflected in the data.
- In their editorial Becker and O’Connor  raise the question of biologic plausibility when considering differential effects of opioids on AEs. For example, there is no known unique biologic effect of codeine that would account for increased cardiotoxicity, as was found in the second study. On the other hand, propoxyphene was recently removed from marketing in the U.S. due to cardiac concerns, as we recently discussed [here]; however, it is interesting to observe that the incidence rates of cardiovascular AEs with propoxyphene in the Solomon et al. study [2010B] were not remarkably different from the other opioids examined.
- Neither of the investigations present normative data; that is, how would findings in the study groups compare with a similar cohort of patients who were not taking any opioids, NSAIDs, or coxibs during the 7-year time period in question? While locating an “analgesic-free” group of elderly patients might be difficult, without such comparisons it is unknown if the use of certain analgesics actually might reduce the incidences of certain AEs that normally occur in elderly patients.
[NOTE: A Guest Editorial based on this UPDATE was published in the March 2011 edition of the journal, Practical Pain Management (Vol. 11, Issue 2, pp 70-72) and is available online HERE.]
> AGS Panel on Persistent Pain in Older Persons. Pharmacological Management of Persistent Pain in Older Persons. American Geriatrics Society. J Am Geriatr Soc 2009 [access available here].
> Becker WC, O’Connor PG. The Safety of Opioid Analgesics in the Elderly: New Data Raise New Concerns [Comment on "The Comparative Safety of Opioids for Nonmalignant Pain in Older Adults"]. Arch Intern Med. 2010;170:1986-1988 [extract here].
> Solomon DH, Rassen JA, Glynn RJ, et al. The Comparative Safety of Analgesics in Older Adults With Arthritis. Arch Intern Med. 2010A;170:1968-1978 [abstract here].
> Solomon DH, Rassen JA, Glynn RJ, et al. The Comparative Safety of Opioids for Nonmalignant Pain in Older Adults. Arch Intern Med. 2010B; 170:1979-1986 [abstract here].