Saturday, November 3, 2012

Who Receives Higher-Dose Opioids for Back Pain?

OpioidsBenefits and risks associated with long-term opioid analgesic therapy for chronic noncancer pain are inadequately understood, and there have been recent controversies surround this practice. A new study examines the extent of such prescribing, along with factors that appear to influence higher opioid analgesic dosing. However, this study contributes little to the debate regarding the efficacy and safety of long-term opioids for chronic noncancer pain at any dose.

Researchers at Oregon Health & Science University and the Kaiser Permanente Center for Health Research, Portland, OR, constructed an analysis to examine the prevalence of high-dose opioid use as well as associated demographic, clinical, and health service utilization correlates among patients with low-back pain [Kobus et al. 2012]. Data were derived from extensive electronic medical records of 470,000 patients served by the Kaiser Permanente Northwest (KPNW) region.

Writing in the November edition of the Journal of Pain, the researchers report that data collection focused on patients who had presented with musculoskeletal low-back pain during a single year, 2004 — total N=26,014. Of that total, 41% (N=10,543) received short-term opioid therapy and 39% (N=10,184) did not receive any opioid medication for their condition.

This left 5,287 patients who received longer-term (≥90 days) opioid analgesics, representing 20% of all patients with low-back pain. Of these patients, 453 or about 9% were prescribed higher doses — defined as ≥100 mg/day morphine equivalent dose (MED) — whereas, most patients (4,815 or 91%) were prescribed lower doses (1-99 mg/d MED, ≥90 days).

In sum, approximately 3% of all patients prescribed opioid medication for low-back pain for any length of time received higher doses (≥100 mg/d MED; 453/15,830). Among patients specifically receiving longer-term opioid therapy (≥90 days), roughly 9% of were prescribed higher doses (453/5,287). The median higher dose amount was 180.0 mg/d MED.

The researchers found that compared with the no-opioid group, higher-dose opioid users reported poorer health. Additionally, compared with no opioids or lower-dose opioids, patients prescribed higher doses had greater rates of mental health and substance use disorders, concurrent sedative-hypnotic use, and health services utilization.

After adjusting for select covariates via logistic regression, several factors were associated with a greater likelihood (odds) of patients receiving high-dose opioids. These included male gender, greater comorbidity, Medicare coverage, any mental health or substance use diagnosis, co-prescription of sedative-hypnotics, and more emergency department and specialty pain clinic visits.

In conclusion, the researchers express concerns about the potential safety of longer-term, higher-dose opioid therapy. They suggest that further research is needed to determine the balance of benefits and harms of this practice for chronic non-cancer pain, as well as factors that may progressively lead to higher-dose opioid use.

COMMENTARY: If there are inherent and specific harms of long-term opioid use for chronic noncancer pain, particularly at what might be considered higher doses, it would be critical to understand them. Research to date on this subject has been of low quality and/or inconclusive, and this present investigation by Kobus and colleagues may do little to help clarify matters.

Perhaps of greatest concern, the retrospective data for this present study were a subset of data stemming from an earlier-reported investigation by this same research team [Deyo et al. 2011] and representative only of a single period of time roughly 8 years ago at a localized health care system. So, the study might be merely of limited historic or baseline interest, depicting a very select population of healthcare providers and their patients.

The selection of 2004 for data collection is unexplained by the researchers, but they acknowledge that the data do not reflect current practices. Possibly most disconcerting, the authors note that since the time their study data were collected the KPNW region implemented new risk mitigation strategies for improving the safety of opioid prescribing. So, it would be much more enlightening to know of the situation today following enactment of those changes.

Still, since this present study is likely to be construed and cited as evidence for potential risks and associated patient factors regarding long-term higher-dose opioid therapy for chronic pain, it is worthwhile examining some key points in the journal report:

  • To begin, it seems that the prevalence of higher-dose opioid therapy overall (≈3%) and specifically in patients receiving long-term treatment (≈9%) may not have been of unusually large proportions. Especially, considering that the data do not portray the clinical etiology of how and why patients might have been titrated up to those doses.

    The report authors concede that they did not have data on functionality or pain severity either at baseline or during opioid therapy. Furthermore, they could not determine if opioids were prescribed solely for back pain or due to greater levels of comorbidity in those persons who received higher amounts of opioids. So, the higher doses may have been quite appropriate and necessary for a proportion of patients.

  • An important question is, “How high a dose is high.” The authors state that they selected 100 mg/d MED as their threshold for high dose because of its “substantial risks for side effects, overdose, and mortality associated with this dosage or higher.” However, as evidence for this they cite studies that have been previously critiqued and found to be of low quality in Pain-Topics UPDATES; for example, Dunn et al. 2010 [here] and Gomes et al. 2011 [here].

  • Data regarding the distribution of opioid dosages across the entire population — including median and interquartile ranges — are not provided, but high doses ranged up to 2,160 mg/d MED. The median of all “higher doses” (≥100 mg/d) in the study was 180 mg/d MED. So, there seems some question as to whether the rather arbitrary threshold of 100 mg/d MED was sufficiently high to characterize individuals receiving truly high dosing. Setting a higher threshold might have been more valid for describing a patient population potentially at risk of adverse effects, and less biased.

  • As for the greater levels of mental health disorders (eg, depression, anxiety, PTSD, substance use disorder) in patients on higher opioid doses, the researchers relied on clinician diagnoses entered in the medical records, rather than standardized assessment measures; hence, they concede that their findings cannot “infer causation or the direction of causality.” For example, patients with persistent, possibly unrelieved, pain typically have greater levels of depression and anxiety, although which came first — pain or psychological distress — is often unknown, and the relationship to higher opioid dose requirement is unclear in this present study.

  • Higher-dose patients were more frequent users of medical services, but this might be expected of a group of patients who are basically in poorer health and require more frequent visits for medication refills and consultations regarding physical and mental health issues. Across both higher- and lower-dose groups there were few hospitalizations, but the frequency was greater in the higher-dose group — which, again, might be expected among sicker patients.

  • Via sophisticated statistical procedures, the researchers identified several independent predictors for distinguishing higher-dose opioid users — such as male gender, higher comorbidity, Medicare coverage, any mental health or substance use diagnosis, and co-prescriptions of sedative-hypnotics — with Odds Ratios ranging from 0.95 to 1.75. However, these were only small-to-medium effect sizes and the importance of these factors for influencing higher dosing may be minimal [see UPDATE discussing effect sizes here]

    One predictive factor, pain clinic utilization (OR 2.30; 95% CI: 1.80-2.94), appeared to be of more significance, and this seems logical since pain specialists might be expected to appropriately prescribe higher analgesic doses as necessary. Only a minority (23%) of higher-dose patients were seen in specialty pain clinics, but this was twice the number as in the lower-dose group.

  • Most patients (88%) prescribed higher-dose opioids in the study received unspecified long-acting formulations, rather than short-acting opioids alone, which is consistent with guideline recommendations for long-term opioids in chronic pain. A quite troubling finding of the study was the co-prescribing of sedative-hypnotics (mostly benzodiazepines) in 61% of the higher-dose opioid patients, which may pose a safety hazard.

    However, the researchers do not specify, or even comment on, any incidences of overdoses or deaths in their study population — so the safety impact of sedative-hypnotic prescribing, if any, is unknown. Avoidance of such prescribing was likely one of the risk-mitigation strategies later adopted by the KPNW region healthcare system.

  • Finally, the researchers observed that being male, white, and having Medicare were significantly associated with higher-dose opioid use. At the same time, black patients were less likely to be receiving higher doses, which is consistent with other research suggesting that black populations are generally less often prescribed opioid analgesics. As this relates to their study, the researchers equivocally note, “whether receiving less opioids for chronic pain treatment reflects better or worse care is unclear.”

    Similarly, after close examination of this study report, readers might be wondering whether higher opioid dosing might have actually represented a better or a worse standard of care for the select population of patients with chronic lower back pain. The conclusion seems unsettled, but it is possible that higher doses were probably necessary and appropriate in the relatively small numbers of patients who received them.

We have repeatedly cautioned in these UPDATES [eg, here] that data mining exercises such as this study by Kobus et al., relying on what the researchers themselves described as a “virtual data warehouse,” represent a low level of evidence. There almost always is a lack of adequate data to fully explore a subject, the approach can lead to faulty associations and assumptions, and more questions are usually raised than adequately answered.

In fact, discovering new questions, or hypotheses, which can then be subjected to further, more rigorous scrutiny in targeted clinical trials is a major strength of data mining approaches. However, interpreting results from these studies as if they represent any level of proof beyond “reasonable suspicion” would be fallacious.

To their credit, Kobus and colleagues do not present their findings as strong evidence, but the underlying premise of this study seems to imply that higher-dose opioid prescribing for chronic pain is of questionable benefit and possibly harmful. Their evidence does not support this contention and it is hoped that future researchers and writers on this subject will not misinterpret the results of this study as indicating otherwise.

> Deyo RA, Smith DHM, Johnson ES, et al. Opioids for Back Pain Patients: Primary Care Prescribing Patterns and Use of Services. J Am Board Fam Med. 2011;24:712-727 [
article here].
> Kobus AM, Smith DH, Morasco BJ, et al. Correlates of Higher-Dose Opioid Medication Use for Low Back Pain in Primary Care. J Pain. 2012(Nov);13(11):1131-1138 [

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.