Considering CDC’s Proposed 2016 Guideline for Prescribing Opioids for Chronic Pain: A Good Start that Needs Improvement
A summary of this essay was published on Huffington Post (January, 2016). Below is an extended version of the same piece. It offers a deeper review of the challenges raised by the CDC’s Draft Guideline.
President Obama began his 2016 State of the Union address by referencing an American epidemic of opioid overdoses. He was right to do so. The Centers for Disease Control report that 2014 saw a record of 18,893 deaths related to opioid overdose, a category that includes both medications and heroin. Given a rising tide of deaths, it is only sensible to look closely at how opioids come into distribution. There is more than one path. Doctors write prescriptions, and the pills may be consumed properly or improperly. Or they can be sold, given to friends, or stolen. Heroin is incredibly cheap and potent these days. It’s often laced with other drugs and can cause overdose in ways that users can’t predict.
A major portion of the public health response has focused on doctors and their prescriptions (disclosure: I’m a primary care doctor trained in internal medicine and addictions). Most public health authorities believe a major contributor to the rising tide of overdoses has something to do with the prescriptions for opioids we write. Our tendency to write prescriptions for pills like hydrocodone or morphine rose precipitously from 2000 to 2011.
Everyone knows a story of someone who wheedles pills out of credulous physicians. Barring a few so-called pill mills (which alone cannot account for the rise in prescriptions), most doctors writing prescriptions for opioid pills do so in response to a patient with severe chronic pain. There are an estimated 100 million Americans with chronic pain, and between 5 and 8 million take opioids for that pain.
It stands to reason that among the patients who have received opioid prescriptions, surely some (or many) should not have received them. Many doctors have decided to prescribe less, starting in 2012, according to national data. If prescribing went down while overdoses went up these past few years, it’s fair to say that there is room for argument about precisely how doctor’s prescriptions relate to overdoses. But few would argue there is no relationship at all. Thus, great hopes are pinned on the notion that getting doctors to prescribe differently (and less) for their patients with pain will be key.
Last year the Centers for Disease Control, after consulting an extensive array of experts and interest groups, prepared a draft guideline for doctors on prescribing opioids. In December they placed notice in the Federal Register seeking public commentary. By deadline on January 13, over 4300 comments were received.
There is a reason this document excites so much passion. In part, organizations such as the American Cancer Society project this guideline will not be voluntary, but will carry force of law.
The hope is to prevent development of addiction and overdose that devastates countless families. Yet, there are those 5 to 8 million patients who receive opioids, some of whom believe that they are at risk of losing access to a crucial medication that is helping manage their pain, improve their quality of life and overall function. As medical boards, insurers and government agencies enforce this guideline, prescribing differently from the topline recommendations is likely to become onerous, leaving many patients in the lurch.
If you listen to this conversation between this 70-year old coal miner who suffered 18 major injuries, and a chief advocate (an addiction specialist) for the reduction of opioid prescribing, you feel the tension. You will hear the distress of a man who fears being confined to bed from his pain, and the concern of an addiction doctor who believes opioid pills have done harm, not good, even perhaps to the man to whom he is speaking.
The experts convened by the CDC include many I know and respect. They have taken a fairly strong stand. They conclude that the literature shows no evidence of enduring benefit from opioids, and that measurable harms are tied closely to dose. They urge careful assessment of risk and benefit. They urge aggressive use of urine drug testing to identify patients who take opioid medication differently from intended or use illicit drugs.
In 56 pages, they say a lot more. My primary care patients include several with chronic pain, and my practice lines up pretty closely with precisely what the guideline recommends. And despite that, I feel this guideline is not yet ready, not given the power we project it to have.
For reasons I shared with the CDC (shown below), I think it reaches a bit beyond the available science in some places, neglects it in others, and misconstrues how best to translate it in the care of our patients. It risks making opioids less available to patients who are benefiting from them. It is not far from where it needs to be, but it needs improvement.
Friends, some of them national leaders in primary care, addiction and pain medicine, have urged me to publish this concern broadly.
I offer it below to show that that one can take a different stand without rejecting the science or the underlying public health commitment that I fully share with the honorable drafters of the CDC’s draft Guideline. For the readers who believe I am right, or perhaps have also misconstrued the science, I welcome your thoughts.
Opinions expressed are solely my own and do not represent positions of any agency of the U.S. Federal Government or the State of Alabama
Follow me on Twitter @StefanKertesz
— — —
(Below follows the full text, as submitted to the Centers for Disease Control public docket on January 13, 2016. I have made minor edits and added a hyperlink, relative to the submitted document)
Docket # CDC-2015–0112
Full Comment Regarding CDC Guideline for Prescribing Opioids for Chronic Pain
January 13, 2016 Stefan G. Kertesz, MD, Associate Professor, University of Alabama at Birmingham School of Medicine
The Centers for Disease Control and Prevention (CDC) has requested comments on its Guideline for Prescribing Opioids for Chronic Pain, having recently published a full docket of supportive materials and the guideline itself. As both a physician who cares for patients with pain and addiction, and as a federally funded researcher who studies addiction and primary care, I share the CDC’s fundamental goals of improving the quality of life for patients while minimizing the risks of opioid overdose and addiction.
Within the guideline itself, I see substantial good work, marred by flaws, some of which simply could not be foreseen, and others of which are remediable. I wish to offer criticisms of the Guideline’s scientific review and of how it is translated into recommendations. However, there is one crucial matter that must come first, because it may not have been considered by the guideline writers at inception. It concerns the powerful legal and regulatory import of this document.
This is a guideline like no other. It is unique because of the particular sensitivity of its topical focus, and because of the legal and regulatory force it will carry. Its sensitivity is obvious: its guidance will affect the immediate well-being of millions of Americans with chronic pain, and it may yet help us to remediate an ongoing, cataclysmic public health emergency of opioid overdoses.
Beyond this sensitivity is the unique regulatory force of this guideline, which CDC implicitly acknowledged by entering this document into the Federal Register and posting a full docket for public commentary. Use of the Federal Register for public comment is required only for rules, not guidelines. The option to use the rule-making apparatus reflects CDC’s own appreciation of the power that will attach to this document. As if to underscore the point, Congress gave this guideline force of law within the Department of Veterans Affairs, in its Omnibus Appropriations Law passed at the end of 2015 (HR 2029), as shown in Appendix J, 44th pdf page. Congress explicitly requires “adoption of the safe opioid prescribing practices” as laid out by CDC. Guideline authors could be forgiven, last summer, for assuming that a “guideline” implies a degree of clinical judgment about how and when to apply it. But in fact, Congress did not order “consideration” of a “guideline” by professionals. It required, explicitly, adoption of the practices. This throws all the legal and disciplinary weight to this document’s top-line recommendations, even when the underlying text works hard to soften them and acknowledge multiple evidentiary shortcomings.
In light of the special sensitivity and legal power applicable to this guideline, some criticisms are required despite the fact that I agree with many points.
I am a physician certified in addiction medicine with a history of research support from the National Institute on Drug Abuse. I run a primary care clinic where I see pain patients for whom opioid prescription has arisen either as a therapeutic proposition or as a clinical problem. Our day-to-day practice aligns with nearly all principles laid out in the guideline. We attempt non-pharmacologic therapies first. We carefully negotiate treatment goals with opioid decision-making calibrated to risk and benefit. We watch patients closely and use urine testing frequently. We keep doses quite low for most patients. If this “guideline” were merely written to reflect current clinical wisdom and did not carry the enormous public and regulatory import that it does in fact carry, I would sign on as supportive.
However, this guideline has enormous potential regulatory impact nationally. It has guaranteed legal impact on all clinicians practicing in the US Department of Veterans Affairs. And despite the enormous work that obviously went into it, there are serious flaws in how this Guideline considered research evidence and translated them into clinical guidance. As a result, it fails to offer meaningful and direct support for primary care providers who must, of necessity, make decisions daily in the care of actual patients.
I will note four.
1. Inappropriately broad conclusions are drawn from untestable propositions and a problematic literature review.
Recommendation 1’s topline recommendation offers a global assessment that “nonpharmacologic therapy and nonopioid pharmacologic therapy are preferred for chronic pain.”
In medicine, the evidentiary standard for favoring one treatment over another requires comparative study. It would be extraordinary in any guideline to declare that every conceivable therapy not including a named pharmacologic agent (i.e. opioids) is generally superior to all therapies that do include that agent. Such a statement is not a scientific proposition, and it is by its nature not a testable proposition.
I offer this criticism with deep appreciation for the clinician’s instinct to try non-opioids (not just first, but repeatedly!), because that is what I do as a doctor.
Nevertheless I would urge avoidance of broad, untestable conclusions in a guideline. A far more circumscribed and appropriate topline recommendation is buried within the text beneath Recommendation 1:
“experts agreed that opioids should not be considered first-line or routine therapy for chronic pain”.
I urge that this more appropriate statement replace the excessively broad one that serves as the topline for Recommendation 1.
Beneath the overbroad conclusion offered by Recommendation 1 is a problematic literature review in which the Guideline applies unequal standards for assessing the potential benefits of opioids versus the potential benefits of all other therapies.
With regard to opioids in particular, the guideline asserts that only studies documenting benefit for longer than one year could be considered informative. Given this threshold, the authors found “insufficient evidence” to determine if there is sustained pain relief or improvement of function with opioid treatment. Problematically, the evidentiary window was opened far wider in relation to all non-opioid therapies, where studies of any duration were counted. Several studies (including the study of exercise for knee pain, citation 87) report that there was NO long-term benefit from the intervention. It is hard to connect such forthright statement of limited short-term results to a broad statement of preference for the infinite category of “nonpharmacologic therapy and nonopioid pharmacologic therapy” over opioids. It is an imbalanced review that mars the evidentiary conclusions.
When considering potential benefits of two alternatives, there is no scientific reason to allow short-term benefit studies for one, and to prohibit consideration of short-term benefits for the other. If this guideline applied a similar evidentiary standard for benefits of both opioid and non-opioid therapies, then it would have also considered plausible benefits from opioids, form shorter studies, such as those in this 2013 Cochrane Review by Chapparro.
A challenging reality for all of this: we lack long-term studies regarding most of the therapies we may wish to consider, and clinicians and patients have to base their reasoning from experience.
Acknowledging the true state of the literature, in a balanced way, would not have foreclosed the option for this guideline to present a defensible topline recommendation: “opioids should not be considered first-line or routine therapy for chronic pain”.
2. In Recommendation 5 concerning opioid dosage, there is a lack of any effort to discuss or consider absolute (as opposed to relative) risks, and a failure to consider the importance of risk and protective factors that can mitigate or exacerbate risk related to dosage. These combined flaws results in an unduly narrow focus on dosage alone (“should generally avoid increasing dosage to >90 MME/day”). This narrow dose-focused top-line recommendation should be amended to encourage acknowledgment of the other risk and protective factors that are documented in the literature and known to clinicians, and to incorporate consideration of absolute as opposed to relative risk.
By word of explanation, I wish to underscore the problematic nature of relying solely on relative risk (and similar calculations of hazard ratio) when applying guidelines to the care of individual patients. The most recent guideline for management of high cholesterol requires calculation of absolute risk in order to assess the likely benefit of prescribing high-potency statins. The cholesterol guideline approach is the right one.
The literature on opioid overdose risk routinely reports absolute risk, and thus it is a flaw that such data are ignored in this guideline. Knowing the underlying absolute risk completely changes the true impact of prescribing a lower or higher dose, and must be considered by doctors.
An excellent study of US veterans receiving opioids for non-chronic cancer pain, published in 2011, found that doses ranging from 50–100 mg Morphine Milligram Equivalents (MME) was associated with overdose rate of 0.24 per 1000 person-months (arithmetically, one can convert this roughly to the following statement: treating 83 persons for a year at this dose would result in one having an overdose, with or without death).
That absolute risk may or may not be acceptable for a certain patient given the overall level of pain-related dysfunction and the actual benefits they obtain from opioids.
Certainly, reliance on relative risk information, absent knowledge of the absolute risk applicable to patients, has been appropriately discouraged in the last 2 decades, and the newest American Heart Association cholesterol guideline underscores the value of absolute risk when considering care for individual patients. Thus it is unfortunate to see that this evidence review turn aside such a central element in health care decision-making.
Perhaps equally problematic, research data show that the risk related to opioid dose should not be considered in isolation from all the other information a clinician may know. The aforementioned study of veterans also showed that other characteristics, including age and race, had very strong effects that could eliminate or greatly amplify the dose-related risk. Being black versus white, in this sample, reduced risk by 2/3. Being over 60 years old (versus 18–29) reduced risk by 82%. Clinicians often know even more about their patients than these factors.
Thus, the topline recommendation’s focus on dose alone truly obscures a great deal of scientific evidence that has already been published, and which should be central to well-informed prescribing. Given the regulatory and legal force that has already been applied to this Guideline, this narrow consideration of the literature on opioid dosage invites actions by regulators, pharmacies and insurers. Those actions could sharply constrain the ability of physicians to prescribe on the basis of a full and thorough consideration of their patient.
3. The top-line recommendations for opioid prescribing lack a sufficient, explicit top-line endorsement of clinical decision-making based on a clinical relationship and knowledge of the patient’s overall functioning and life context.
It has been shown that a key element of good decision-making for any chronic condition depends substantively on the capacity of caregivers to understand what is going on in the life of the patient (see works by Saul J. Weiner, including the book Listening for What Matters, 2016 and this article). Nowhere does this consideration apply with more force than in the care of patients with chronic pain, for which the desired outcomes are only partially measurable, and typically transpire outside of the office. The decision to prescribe an opioid, or to escalate doses, must formally include that consideration.
I believe the guideline document offers clues that the writers had some notion of this issue, but the final product ultimately fell short in expressing it. Thus top-line Recommendations 1 and 2 refer to benefit versus risk, and to “treatment goals”, both of which are crucial concepts.
What’s lacking is a guidance that squarely urges clinicians (and more important, health care organizations) to solicit and listen for the information about life functioning, social functioning and daily activities that help one to determine if a prescription fits well or fits poorly in the life context of a given patient. The same guidance should serve to caution health care enforcers (administrators, licensing boards, and federal Departments) against punishing prescribers based on dosage or duration alone.
4. Recommendations endorsing urine drug testing and holding treatment for acute pain to 3 days’ duration are not strongly grounded in science. Given the regulatory and legal force applied to this guideline, they either should be jettisoned or significantly reconfigured.
Both recommendations have the benefit of reflecting a plausible clinical consensus emerging from weak to nonexistent data. As a practicing clinician, we use urine drug testing, although my fellow physicians and I spend a great deal of time trying to interpret the results, because they are often very complicated.
The guideline itself reviews the weakness of the research data to support urine drug testing. The guideline does not sufficiently consider cost, which will in fact be a prohibitive barrier for many indigent patients. This was the case when I worked in a community health center. To support drug testing as a precondition to opioid therapy (top-line recommendation 10), knowing full well that many patients can’t afford to pay for it, poses a very serious problem of equity and justice. It is inappropriate, especially when the supporting data are explicitly acknowledged to be weak, to endorse a precondition that will eliminate the poor from access to a treatment that they could otherwise afford.
The recommendation to limit opioid prescription for acute injuries to 3 days is offered tentatively, as it should be. I was unable to find a single citation of a scientific study to support this recommendation, however.
If this document presented itself as merely a statement of current expert opinion, and not a document having legal and regulatory force, I would basically agree with it, pending better data. But given the projected use of this guideline for health regulations and review of professionals across multiple contexts, I believe the top-line recommendations should explicitly reference the optional or guarded nature of the clinical wisdom being offered.
In closing, I wish to thank CDC for convening a guideline writing process and to credit the extraordinary work of the guideline writers. As I noted at top, my practice in most ways would be seen as conforming to the clinical thinking offered by this draft guideline.
However, given this guideline’s projected impact on public policy and professional regulation, I would recommend that problems in the scientific evidence review be addressed, and that the top-line recommendations be enhanced to reflect what we truly know, and where our scientific knowledge falls short.
Stefan G. Kertesz, MD, MSc
Associate Professor, University of Alabama at Birmingham School of Medicine
This public comment represents my personal opinion. It does not reflect the views of any agency of the United States Federal Government, including the U.S. Department of Veterans Affairs, or the State of Alabama.