Introduction
Transparency, freedom from bias, and accountability are, in principle, hallmarks of
taxpayer-funded institutions. Unfortunately, it seems that at least one institution,
the Centers for Disease Control and Prevention (CDC), continues to struggle with all
three. What began with a prescribing guideline created in secrecy has now evolved
to the use of statistical data and public statements that fail to capture not only
the complexity of the problem but also the distinction between licit and illicit opioids
and their relationship to the alarming increase in unintentional overdose. This is
unfortunately consistent with Mark Twain’s assertion that “There are lies, there are
damn lies, and then there are statistics.”1
For instance, when the CDC was in the process of drafting guidelines for the use of
long-term prescription opioids to treat chronic pain, the identities of the project’s
Core Expert Group members remained a secret until they were leaked.2,3 When its members
were eventually identified, many were concerned that the group’s composition was not
balanced and had an inherent bias against the use of prescription opioids to treat
pain.4 Then, when the time came for public input on the draft of the prescribing guideline,
the CDC’s invitation for meaningful comment can be best described as somewhere between
a charade and a comedy of errors. For example, the only way the public could view
the draft was during a one-time internet webinar. The actual guideline itself was
not publicly available and was not posted on the CDC website, and admission to the
webinar was limited. Those fortunate enough to gain entry were sometimes able to see
the actual recommendation as it flashed on the screen, and while attendees were permitted
to ask questions, the CDC stated that they would not provide any answers. At the end
of the webinar, attendees could then email their comments to the CDC, but they only
had 25 hours to do so.5 Due to the many technical problems associated with the first
webinar (which was reminiscent of Get Smart’s Cone of Silence),6 the CDC decided to
repeat the webinar on the following day and allow an additional 24-hour comment period.
4 But the controversy did not end there. Following the webinar, allegations of unlawful
behavior by the CDC in the creation of the guideline were made, specifically that
the CDC had violated the Federal Advisory Committee Act.7 The United States Congress
eventually stepped in, and the CDC caved under pressure and permitted a 30-day open
comment period8 during the Christmas holidays. Although the new open comment period
yielded several thousand comments in the Federal Register, there was little change
between the draft guideline and the final guideline. And while the CDC had asserted
that they would be open to revising the guideline, and one of their own consultants
had notified them that the guideline may be negatively impacting prescribing and pain
treatment, there is no evidence that the CDC intends to do so.9 Remarkably, the actions
of the CDC in the creation and publication of the prescribing guideline appears to
violate every single standard that the Institute of Medicine recommended whenever
clinical practice guidelines are created.10 Consequently, the authors of this commentary
fear that the CDC’s earlier lack of transparency, freedom from bias, and accountability
in the creation of a prescribing guideline has now infected the way they characterize
statistical data to the public regarding the relationship between opioids (licit and
illicit) and unintentional overdose.
Simplicity and data conflation is making the problem worse
The United States is in the midst of an opioid crisis and prescribers and policymakers
continue to struggle with how to effectively reduce the incidence of harm from prescription
drugs while at the same time ensuring appropriate access. While accurate measurement
of the degree by which various actors or variables have contributed to the initiation
and prolongation of the current crisis remains challenging, it is fair to say that
prescribers, policymakers, patients, non-patients, the pharmaceutical industry, the
insurance industry, regulators, illicit opioids, prescription opioids, dosage,11 and
the addiction community have all played a role, whether in isolation or in combination
with the above. While we do not believe that all actors or variables contributed equally
to the present crisis, nor do we believe that the unintended negative outcomes flowing
from their actions were intended, there is understandably one universal contributor
to the increase in unintentional overdose that continues to be singled out by policymakers
and the CDC: opioids. But the more important question becomes: about what opioids
are they speaking? All opioids in general, some opioids in particular, illicit opioids
such as heroin and illicitly manufactured fentanyl, or prescription opioids? These
distinctions matter. According to the CDC, the “majority of drug overdose deaths (more
than six out of ten) involve an opioid,”12 and “as many as 1 in 4 people who receive
prescription opioids long term for noncancer pain in primary care settings struggles
with addiction.”13 Absent accompanying qualification of the statistics associated
with these alarming statements, are they facially accurate or do they reflect a particular
agenda – a campaign that fails to recognize the complexity of the issues, the factors,
sources, and problems that once identified could help forge effective solutions? We
believe it is the latter. The following commentary examines some of the recent statistical
claims asserted by the CDC and argues that not only are these statements inaccurate,
but they also impede the ability of the public and policymakers to understand the
complexity of the problem and create solutions that are balanced and effective. In
brief, while prescription opioids continue to play a part in the crisis, illicit opioids
such as heroin and illicitly manufactured fentanyl, not prescription opioids and overprescribing,
are currently the driving forces behind the increase in unintentional overdose deaths
in the United States. This critical distinction is often ignored or underappreciated
by the press and policymakers, and is a distinction that needs to be emphasized by
the CDC. The failure to do so has far-reaching consequences in terms of policy, pain
treatment, substance abuse prevention, and reduction of unintentional overdose.
Prescribing: is the problem under-or over-prescribing?
The Institute of Medicine has estimated that over 100 million Americans suffer from
chronic, long-term pain.14 Pain is individualized, and so should be its treatment.
Opioids can be effective in treating pain, and while there may be a variety of alternatives
to opioids, some may not be as effective or covered by insurance reimbursement. Yet
despite the millions of people who suffer from chronic pain in the United States,
there is scant evidence that the CDC considers chronic pain a serious public health
problem. For instance, the CDC maintains an A–Z index on their public website, an
index of “topics with relevance to a broad cross-section of CDC.gov’s audiences” that
“are representative of popular topics […] or have critical importance to CDC’s public
health mission.”15 A search using the A–Z index under the letter “C” found references
to chronic conditions, but chronic pain was not one of them; a search using the letter
“P” found only two topics containing the word “pain”: Pain Killer OD, and Pain Killer
Overdose.16 While undertreated pain does not seem to be of critical importance to
the CDC, that has not prevented them from creating a prescribing guideline to treat
pain, recommending against the use of prescription opioids, and asserting that nonpharmacologic
therapy is preferred.17
But whether pain is treated with or without prescription opioids, one thing seems
clear: it has been undertreated for decades. Although myriad reasons for undertreatment
exist, decades ago many well-intentioned health care providers responded to this public
health crisis by increasing their prescribing of opioids. While more prescription
drugs were made available to treat pain, it undoubtedly increased the likelihood of
leftover medication which in turn increased the potential for misuse, abuse, and diversion.
In fact, studies continue to provide evidence that a small percentage of individuals
who started using heroin began by misusing prescription drugs (whether they were the
drugs they were originally prescribed or someone else’s), so regardless of how one
characterizes over- and underprescribing, there should always be efforts aimed at
reducing the amount of leftover medication since the common source of diverted pharmaceuticals
is a friend or relative. 18
So, what amount of prescribing is appropriate? This represents a difficult question
since one size does not fit all. A particular type or dose of one medication may be
appropriate for one patient and condition and wholly inappropriate for someone else.
Yet despite the medical necessity of tailoring treatments to the individual, the tendency
today is for an across-the-board reduction in prescription opioid availability. We
can certainly understand the calls for reduction in the surgical context if one considers
emerging evidence that a substantial portion of prescription pain medication goes
unused by patients in the perioperative care setting.19 But is prescription pain medication
the driving force in today’s overdose epidemic, and if not, does an across-the-board
reduction in opioid availability and prescribing make sense? We recognize that an
excessive supply of prescription opioids can create serious public health problems,
much like the overprescribing of antibiotics can have micro and macro health concerns.
But are prescribers prescribing more, or less than they used to? According to the
CDC and several states’ Departments of Health, less opioids are being prescribed,
which in turn indicates that there may be something other than just prescription opioids
that are contributing to the escalation in unintentional overdose.
Data from the CDC indicate that between 2010 and 2015, the amount of opioids prescribed
in the United States actually decreased by more than 18%, with a 13.1% decrease reported
between 2012 and 2015 alone.20 Survey data also suggest that more than half of physicians
in the United States have reduced their opioid prescribing, with nearly 10% having
stopped prescribing opioids altogether.21 Although more recent federal data on the
decrease in opioid prescribing are not available, state data are helpful. For example,
in Tennessee between 2015 and 2016, the total amount of opioids prescribed decreased
by another 9.2%.22 In New Mexico, the Department of Health reported that the amount
of opioids prescribed decreased by 5% from the first quarter of 2016 to the first
quarter of 2017.23 In terms of prescriptions filled, Michigan reported a 9.7% decrease
between 2015 and 2016, Maine reported an 11.9% decrease, Massachusetts reported a
12.7% decrease, West Virginia reported a 15.6% decrease, and so on.24 In fact, perusal
of the data from the study24 indicated that every state (with the exception of Florida)
and the District of Columbia reported a decrease in the number of opioid prescriptions
written between 2015 and 2016. Consequently, there is little doubt that the amount
of opioids being prescribed is decreasing dramatically, so it would be disingenuous
to suggest otherwise. While the rate of prescribing continues to decline, what about
the trends in prescription opioid mortality? Are they declining as well? This question
is a more complex one, and thus needs to be considered carefully.
Measuring and reporting prescription opioid mortality
What constitutes “a prescription opioid death?” It appears that there is some disagreement
and controversy regarding this issue,25,26 a disagreement that may be agenda-driven
or subject to bias. In 2009, to its credit, the CDC reported that the tremendous variation
between states’ rates of prescription opioid overdose deaths “should be interpreted
with caution as there is some variation in the reporting of substances on death records.”27
This is an understatement. Determining therapeutic levels and causes of unintentional
overdose can be challenging, and postmortem drug redistribution has been described
as a toxicological nightmare.28 Although there is a need for more consistency between
coroners and medical examiner offices across the United States when opioid-related
deaths are involved,29,30 many states utilize death certificate data, which often
do not include the source of a drug, the purpose for which the drug was used, the
level of opioid tolerance in the decedent, and even the specific type or the name
of the drug(s) that were involved.31 A recent study32 looked at “fentanyl deaths”
in southeastern Massachusetts, with the authors of that study determining that 82%
of the fentanyl deaths over a 6-month period from 2014 to 2015 were likely due to
illicitly manufactured fentanyl, with only 4% attributed to legal, pharmaceutical
fentanyl. It is important to note that carfentanil, a common contaminate, has 100×
the potency of fentanyl. Extrapolating from the reported figures, only a mere 5% of
all “fentanyl” overdose deaths were due to a legal, pharmaceutical product, with the
remaining 14% due to “an unknown source of fentanyl.” It is important to note, however,
that Rudd et al33 warned in a 2016 article that “illicitly manufactured fentanyl cannot
be distinguished from prescription fentanyl in death certificate data,” suggesting
that this problem is widespread. In other words, if only 5% of all “fentanyl overdose
deaths” are due to pharmaceutical fentanyl, and fentanyl (defined broadly) is now
responsible for approximately 79% of all “prescription opioid overdose” and “fentanyl”
deaths (extrapolating from 2016 Maryland data),34 could the number of actual prescription
opioid deaths be only a small percentage of the overall opioid overdose statistics?
This is not only a measurement problem; this becomes a policy and solution problem
if the public and policymakers continue to fixate on prescription opioids as the problem
and the primary factor involved in unintentional overdose.
In order to more accurately assess the actual number of prescription opioid analgesic
deaths, it was necessary for the first author (MES) to go to individual states’ Departments
of Health (DOH). Looking at the data from Illinois, what were listed as deaths due
to “Opioid Analgesics” reportedly increased from 589 to 1233 between 2015 and 2016.35
This 107% increase appeared extremely curious given the aforementioned decrease in
opioid prescribing. In speaking to one of DOH’s statisticians, he confirmed our suspicion:
This dramatic increase was almost completely driven by illicit fentanyl and its potent
analogs, not by legitimate pharmaceutical opioids used to treat pain. Furthermore,
he noted that irrespective of the number of dangerous substances that are found in
toxicology reports, the state’s death certificate data categorize any death in which
even an iota of a prescription opioid (or a nonpharmaceutical fentanyl product) is
found as a “prescription opioid death” (J Tharp, personal communication, September
22, 2017). A number of other states (eg, Maryland, New Hampshire) have recently taken
the initiative to delineate between fentanyl (generally illicit) and other prescription
opioid deaths. So, for example, New Hampshire’s Department of Health data characterizes
opioid deaths as attributable to specific legitimate prescription opioids, prescription
opioids in combination with other licit and illicit drugs (including benzodiazepines,
cocaine, methamphetamine and alcohol; methadone, oxycodone, hydrocodone, morphine,
tramadol, and oxymorphone) (S Watkins, personal communication, September 22, 2017).
A total of 59 deaths involving a legitimate prescription opioid were recorded, with
32 involving oxycodone. However, 72% of the deaths involving oxycodone included alcohol,
a benzodiazepine (or both alcohol and a benzodiazepine), kratom, methamphetamine,
or another prescription opioid (which may or may not have been prescribed concurrently).
These data strongly suggest not a “prescription opioid crisis” but rather a “polypharmacy
crisis.”
The role of polypharmacy in “prescription opioid over-dose deaths” was recently elucidated
empirically.36 The investigators determined that more than half of decedents with
an opioid-positive toxicology had alcohol in their systems, and the average number
of drugs identified in the toxicologies was six. The second most commonly used drug
associated with mortality in New Hampshire was methadone, which was involved in 18
deaths – more than half of which involved other substances as well. The relatively
high number of methadone-associated deaths is not surprising, certainly given the
literature supporting the high level of potential lethality of this drug and its questionable
suitability for use in treating chronic noncancer pain, especially for the inexperienced
clinician that continues or initiates methadone, most particularly due to complex
pharmacokinetics and unanticipated drug interactions.37 In fact, many in the pain
community continued to express concern over the placement of methadone on Washington
State’s formulary list to treat pain. Washington State officials defended the placement
of methadone on the preferred drug list to treat pain, but it was not until the Seattle
Times examined the escalation in overdose and its ties to methadone that the state
revisited the matter.38 Consequently, a deeper analysis of the existing data helps
reveal the complexity of the issue and what is really behind the current opioid crisis.
Yet another cause of the unfortunate overestimation of opioid mortality in the United
States (and its link to prescription pain medication) relates to difficulties in distinguishing
between heroin and morphine in postmortem samples. Heroin is twice as potent as morphine,
has a half-life of only 6–25 minutes prior to its metabolism to morphine in the liver,39
resulting in studies suggesting that heroin deaths are also underestimated while morphine
deaths are overestimated in their prevalence.40,41 Given that the prescription opioid
epidemic of the past is progressively being replaced by a surge in heroin use,42 the
negative implications for inaccurate reporting that misidentifies a commonly prescribed
opioid that has analgesic efficacy when used appropriately are clear. Compounding
the problem is that heroin is frequently laced with illicit, nonpharmaceutical fentanyl
products,43 resulting in coroners’ reports now often yielding inaccurate data suggesting
that addicts are succumbing to the effects of not just one prescription opioid but
two.
Nevertheless, despite the data and their shortcomings, the CDC continues to claim
that we are in the midst of a prescription opioid crisis. As their most recent published
data suggest that “In 2015, more than 15,000 people died from overdoses involving
prescription opioids,”44 one of the authors [MES] submitted a formal request for more
recent data. The CDC responded by stating that it would be “addressed in a timely
manner.” As of the time of writing of the present commentary, this still has not occurred.
The concern is this: for the CDC to suggest that more than 15,000 died in that year
from “prescription opioids” when a closer examination of the data indicates that illicit
opioids and/or polypharmacy were involved is not only inaccurate and disingenuous,
it can negatively impact patients who are well-managed on long-term opioid therapies
and have no effective safe alternatives that are covered by insurance, negatively
impact health care providers who seek to relieve suffering, and negatively impact
people who are suffering from substance use disorders.
The necessity of balance
In medicine, as in life, there are risks and benefits. Prescription opioids can bring
enormous benefits to those who are suffering from acute or chronic long-term pain.
Prescribed appropriately, prescription pain medication has provided relief to millions
of Americans; it has increased their quality of life, improved their function, provided
an option to those for which other analgesics are contraindicated due to certain medical
conditions, and reduced their suffering. At the same time, however, prescription opioids
by their chemical nature are susceptible to abuse, misuse, and physical harm, particularly
among certain subgroups of the population. Yet in today’s environment, the narrative
is not about how to balance the need to ensure access while preventing abuse, nor
is it about the people suffering from acute or chronic pain requiring long-term opioids.
Rather, it is about opioids, their misuse and abuse, and the increase in unintentional
overdose. The proposed solution by policymakers to this complex problem has often
been simple: just say no to drugs, and those who are struggling with addiction simply
need to be enrolled in a substance abuse program so that they can become clean again.
But things are not so easy. Despite decreases in the prescribing of prescription opioids,
we continue to see an increase in the rates of unintentional overdose. While there
is clear evidence that this increase is driven by the use of illicit opioids such
as heroin or illicitly manufactured fentanyl derivatives, this particular fact continues
to get lost in the shuffle and results in knee-jerk reactions calling for the reduction
of both the supply and use of prescription opioids. People who are not adequately
treated for their pain will seek out alternatives, often harmful alternatives that
can lead to addiction, unintentional overdose, or even suicide.
Conclusion
It is easy to demonize and point fingers at industry, prescribers, or anyone who calls
into question the newest battle in the never-ending war on drugs. While we would agree
that anyone involved in the distribution of illicit drugs such as heroin and illicitly
manufactured fentanyl derivatives should be stopped from harming others, and the misuse
and abuse of prescription opioids have played a role in the problems we see today,
in the right hands prescription opioids can help eliminate human suffering. Prescription
opioids are not the panacea, but they have medicinal benefits, unlike tobacco. For
instance, according to the CDC, “cigarette smoking is responsible for more than 480,000
deaths per year in the United States, including more than 41,000 deaths resulting
from secondhand smoke exposure.”45 In contrast to appropriately prescribed and administered
opioids, we are unaware of any legitimate medical treatments involving the use of
cigarettes, a product that continues to be available to anyone over 18 years without
a prescription.
Is there anything that can be done to help solve the opioid crisis? We believe so.
First, we would suggest that the CDC refrain from making alarmist statements that
cite statistics that are not supported by the evidence without qualification. As noted
earlier, a recent web-based publication by the CDC cited an alarming statistic in
support of its campaign to reduce addiction and unintentional overdose: “As many as
1 in 4 people who receive prescription opioids long term for noncancer pain in primary
care settings struggles with addiction.”46 While the study cited by the CDC (appearing
as Footnote number 7 in the CDC publication) was an important contribution to the
field, the authors of the study, unlike the CDC publication, correctly noted the study’s
limitations and the difficulty in generalizing the results:
Study limitations include that our diagnostic data were based on patient self-report;
that our survey completion rate was less than optimal, thus study estimates may be
biased; and, as patients were drawn from a predominately Caucasian population in one
US region, it may not be possible to generalize these findings.47
Second, the CDC should recognize that chronic pain impacts millions of people in the
United States and should, at a minimum, create an entry for pain on their website’s
A–Z index. We are concerned that the absence of information about the millions of
Americans who suffer from chronic pain sends the message that pain does not “have
critical importance to CDC’s public health mission.” Creating the link and providing
information on this serious health problem will go a long way to bringing about balance.
Finally, we need to find ways to work together, instead of against each other, emphasizing
civil discourse instead of finger pointing. We are concerned that some people who
are intent on blaming prescribers, patients, and the pharmaceutical industry for the
problem without offering solutions (other than perhaps eliminating prescription opioids)
are making it more about them than the people they are actually trying to help. We
have many problems, but there are also many solutions.
And while we are also concerned that the message about the opioid crisis is not accurately
depicted in the media, we close with an interesting observation from a journalist
that may be helpful to all of us. Malcolm Gladwell, a Canadian journalist, recently
commented that the “one thing that continues to baffle me about American society is
how Americans love to accentuate how they are different from each other. Whereas in
Canada, all we do is celebrate what we have in common.”48
Maybe we should give that a try and find that common ground that unites all of us
in achieving better outcomes not only in reducing abuse and harm but also in treating
pain.