Thursday, April 4, 2013

Do Opioids Increase Falls in Elderly with OA?

FallingNewly published research found that elderly adults with osteoarthritis (OA) prescribed opioids for pain during a 9-year period experienced more falls and factures than patients prescribed NSAIDs or COX-2 inhibitors. Therefore, the researchers suggest that current guidelines recommending opioids in this population may be in error. However, a close look at this research reveals it is of low quality and little validity.

According to background information in the study report, 30% of all persons aged 65 and older, including half of those ≥80 years of age, experience falls at least once each year. The consequences can be devastating, including traumatic injury, loss of function and independence, and death. In many cases, analgesics being taken by elderly patients for various types of pain have been implicated in the falls; especially, those that may dull the senses, such as opioid medications in some cases.

At the same time, however, the latest guidelines for the management of persistent pain in older adults from the American Geriatrics Society (AGS) recommend that, “nonselective NSAIDs and COX-2 selective inhibitors be considered rarely, with caution, in highly selected individuals and all individuals with moderate to severe pain, pain-related functional impairment or diminished quality of life due to pain be considered for opioid therapy” [see 2012 AGS position statement here].

To help assess the continued prudence of that recommendation, a team of researchers led by Lydia Rolita, MD — from Albert Einstein College of Medicine, Bronx, NY — conducted a retrospective case-control study using data from electronic medical records compiled between 2001 and 2009 at the Geisinger Health System in Pennsylvania [Rolita et al. 2013]. They examined occurrences of falls and factures in a cohort of 13,354 community-dwelling older adults, ages 65 to 89 years, with a diagnosis of osteoarthritis (OA) and prescribed primarily either NSAIDs, COX-2 inhibitors, or opioids for their pain.

The specific oral analgesics prescribed for patients recorded in the database and used for this study included the following:

  • Opioids — hydrocodone+APAP, morphine sulfate, oxycodone+APAP, propoxyphene.
  • NSAIDs — aspirin, diclofenac, etodolac, ibuprofen, indomethacin, naproxen, piroxicam.
  • COX-2 — celecoxib, meloxicam, rofecoxib, valdecoxib.

Additionally, some patients were taking acetaminophen (APAP, paracetamol) as a sole agent, subsumed under the NSAID category. Propoxyphene was discontinued in the U.S. during the 9-year period of the study and was not included in analyses. Of special interest for study purposes, about halfway through, during 2004 to 2005, rofecoxib and valdecoxib were pulled from the market due to concerns about cardiovascular adverse effects. So, the researchers hypothesized that, after 2004, there would be a switch to more prescribing of opioid analgesics and this might incur an increased incidence of falls and fractures.

Falls and fractures were combined as a single fall/fracture event variable and, although this might occur multiple times in each patient, each event was counted only once during a discrete 6-month period so that followup care as recorded in the database was not accidentally counted as a new event. Only falls occurring in nonhospital settings were included in the study.

Results reported in the March 2013 edition of the Journal of the American Geriatrics Society indicated that 87% of the 13,354 elderly patients with OA were followed during the 9-year period and 2,462 of them experienced 3,830 fall/fracture events (approximately 31% experienced 2 or more events). Overall, patients with falls/fractures were older and had more physical comorbidities as assessed by the Charlson Index Score.

Opioids-FallsIn the figure at right (adapted from the study report), the top graph illustrates that NSAIDs were predominantly prescribed throughout the study period, but declined around 2004 along with decreases in COX-2 prescribing. At about that same time, there were increases in opioid prescribing; possibly as patients were switched to opioids from NSAIDs and COX-2 agents.

The bottom graph shows that proportions of fall/fracture events for all patients tripled over the study period (gray line) — although the absolute increase was only ≈2% — and the increase was greater in the post-COX-2 period (2005-2009) than earlier. The group treated with opioids accounted for the largest percentages of falls/fracture events (although, some of the increase preceded increases in prescribing).

Percentages of fall/fracture events associated with NSAIDs remained relatively stable throughout, despite a 25% decrease in prescribing of these agents in 2004 and thereafter. As for the COX-2 group, the researchers noted that the total number of participants was very small, leading to high variability and a sporadic pattern in the incidence proportions of events.

Along with the raw cumulative incidence data displayed in the graphs, the researchers also created a comparison cohort of individuals with osteoarthritis from the Geisinger database who did not experience falls/fractures, and these control patients were matched 3-to-1 according to age, sex, and physical comorbidity with patients who did experience falls/fractures. In a logistic regression analysis comparing the cases and control groups, the likelihood of experiencing a fall/fracture was about 3-times greater in participants prescribed opioid analgesics than those prescribed a COX-2 inhibitor (Odds Ratio [OR] = 3.3; 95% Confidence Interval [CI], 2.5-4.3), and 4-fold greater in those prescribed opioids than in patients receiving NSAIDs (OR = 4.1; 95% CI, 3.7-4.5).

The researchers conclude that, as the percentage of elderly adults with OA who were prescribed opioid analgesics increased, there was a corresponding increase in the percentage of those experiencing fall/fracture events. Moreover, individuals who were prescribed opioids accounted for most of the increase in the percentage of individuals experiencing falls and fractures after 2004.

They attribute the change in opioid prescribing as a response to concerns about COX-2 adverse effects, and this had the unintended consequence of significantly increasing morbidity (falls/factures) in this elderly population. Therefore, Rolita et al. contend that, based on this research, current AGS guidelines recommending first-line treatment with opioids in elderly patients with pain should be reevaluated. However, if anything, it is this research that deserves closer scrutiny — from an evidence-based perspective.

COMMENTARY: This study was actually first reported in 2011 by Rolita and colleagues as a conference presentation and abstract (which was not acknowledged in this current report), and we questioned its validity at that time in an UPDATE [here]. This present “official” journal publication — still described by the researchers as a “preliminary study” — provides much more detailed information to confirm our earlier skepticism.

The graphic presentations in the figure above appear to depict a clear and convincing story of how increased prescribing of opioid analgesics influences increased fall/fracture incidents — a cause-effect relationship, which associative data like this cannot confirm, is falsely implied (although the researchers do not state this directly). However, the biased illusion behind this sort of trompe-l'oeil becomes more apparent when details of the data — or more pertinently, missing data — are considered. Also, since this study is likely to be erroneously cited in future literature as evidence against the use of opioid analgesics in the elderly, some critical analysis seems appropriate:

  • Rolita et al. acknowledge that there were no data regarding the duration of analgesic use relative to fall/fracture events, patient adherence to prescribed regimens, or why/how patients were selected to receive a particular analgesics in the first place. During the 9-year course of the study there were significant changes in prescribing patterns; yet, the exact reasons why individual patients were increasingly selected to receive opioids is unknown.

  • Baseline pain levels and analgesic effectiveness in patients were not recorded in the database, nor were prior history of falls or individual patient factors that might have predisposed them to falling, including: gait, vision, living situation, disability level, neurological disorders (eg, Parkinson’s disease), obesity, or comedications (eg, antidepressants, sedatives, etc.). Any of these factors could be confounders of the data analysis, and the researchers concede that such risks “would need to be controlled for in a more detailed, thorough study.”

  • It is curious that only 3 opioid medications are considered in the analysis and all are short-acting, which might not be the best choices for treating chronic pain or from a patient adherence standpoint. It is unknown whether patients followed dosing instructions or, quite possibly, became confused by a frequent dosing regimen and were overusing the medication — resulting in adverse cognitive effects that influenced falls. Other research, albeit of limited quality, has found that, compared with longer-acting opioids, short-acting formulations incur greater fracture risk in elderly patients [see UPDATE here].

  • It also is intriguing that COX-2 agents were never extensively prescribed in this patient population despite their potentially favorable efficacy and gastrointestinal safety profiles compared with NSAIDs. Removal of 2 agents midway through the study resulted in further reduced prescribing. Was the paucity of prescribing merely due to the higher costs of COX-2s in a frugally-managed healthcare system? The lack of prescribing and the erratic pattern of fall/fracture events associated with COX-2 agents challenges the validity of the data relating to these agents.

  • The array of prescribed NSAIDs also might be of concern, since several on the list — ie, diclofenac, etodolac, and indomethacin — have been noted to have the relatively highest risks of cardiovascular adverse events [see UPDATE here]. However, it is unknown from the data if such events in the patient population were a confounding factor influencing falls/fractures.

  • Nearly 90% of all patients experiencing fall/fracture events were female, which seems like an unusually high proportion and suggests there might be some sort of bias in the data and/or the population at risk that may diminish external validity of the results.

  • In epidemiological terms, the overall incidence rate of falls/fractures spanning the 9-year period of observation in patients receiving opioids was comparable to the rate in patients receiving NSAIDs. We calculated the respective incident rates for falls/fractures as: opioids =18 per 1,000 person-years; NSAIDs = 18 per 1,000 person-years; COX-2 ≈ 1.0 per 1,000 person-years. From this perspective, risk/benefit concerns, other than falls/fractures, may be of greater importance when deciding between an opioid or NSAID prescription.

    [Note: the denominator of the incidence rate calculation took into account that there were 13,354 patients with OA in the study population at risk for falls, 87% of whom were followed over 9 years, or 104,562 person-years in total. The numerator includes total falls in each group, rather than number of patients, some of whom had multiple falls. The rate in the COX-2 group may be understated because relatively few patients were prescribed those analgesics. It would have been better if a time-to-event analysis calculating hazard ratios were possible — taking into account differential followup times and event occurrences across individual patients — but data for this were not provided in the report. Epidemiological concepts in pain research were discussed recently in an UPDATE here.]

For a balanced perspective, other researchers have found that pain itself is a significant risk factor for falls in elderly patients [Leveille et al. 2009, discussed in UPDATE here]. A prospective study of 749 adults, ≥age 70 and assessed during 18-month observation periods, demonstrated that persistent pain, even mild pain, directly influenced increasing rates of falls and associated injuries. Furthermore, purported sedating effects of analgesics, particularly opioids, had no effect on fall frequency in this study. In fact, the researchers suggested that the underuse of analgesics could contribute to falls due to the deleterious effects of unrelieved pain.

It also might be noted, as a “double-edge sword” sort of consequence, that the superior effectiveness of opioids in relieving pain might allow and encourage greater physical activity among elderly patients that results in a greater risk of falls and other injuries. The solution would not be less opioid prescribing with less adequate pain relief, but in more patient education and support focusing on fall/injury prevention.

There are two final points worth mentioning regarding the Rolita et al. study, which might seem of lesser importance, but still merit some consideration:

  1. Throughout the published report, the authors use the term “narcotic.” We have argued repeatedly [eg, UPDATE here] that use of “narcotic” in reference to legitimately prescribe opioid medications is not only old fashioned and out-of-date, but is inaccurate and carries pejorative implications. There may be no excuse for using “narcotic” in the context of a study like this unless the authors either 1) are out of touch with current language preferences, or 2) intentionally want to further disparage opioid analgesics. Apparently, the researchers may know better, since they did write “opiate analgesics” instead of “narcotics” in a single instance, and the journal peer reviewers and editors should have known better as well.

  2. We have previously challenged the use of “data-warehouses” from large healthcare systems for research purposes in general [UPDATE here], and relating to falls/fractures in the elderly in particular [eg, see UPDATE here]. There is vast potential for these data-mining exercises to produce “pseudoscience” rather than clinically relevant and externally valid evidence. Rarely are the same results replicated in other research, so they may relate more to practices — good or bad — in the particular healthcare systems than across larger populations. It seems ill-advised to accept the results of such studies as representing any meaningful level of evidentiary proof for reaching clinical judgments regarding the issues examined.

In sum, this present investigation by Rolita et al. would appear to represent low-quality, weak evidence, which says very little with any degree of reliability or validity about the relative hazards of analgesic prescribing in an elderly population when it comes to falls and factures. Furthermore, it may be time to stop using “data warehouses” from large healthcare systems for anything more than generating hypotheses, which then can be used for designing better quality and valid research that can be trusted by the pain management community.

REFERENCE: Rolita L, Spegman A, Tang X, Cronstein BN. Greater number of narcotic analgesic prescriptions for osteoarthritis is associated with falls and fractures in elderly adults. J Am Geriatr Soc. 2013 Mar;61(3):335-340 [abstract here].

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.