Long-term opioid therapy in chronic noncancer pain. NIH – Feb 2015 – a systematic review and meta-analysis of efficacy, tolerability, and safety in open-label extension trials with a study duration of at least 26 weeks (that’s 1/2 year).
This study confirms what pain patients all know: long-term opioids are effective for long-term pain, require few increases, and only rarely result in “opioid use disorder”.
The efficacy and safety of long-term (≥ 6 months) opioid therapy (LtOT) in chronic noncancer pain (CNCP) is under debate.A systematic review with meta-analysis of the efficacy and harms of opioids in open-label extension studies of randomized controlled trials (RCTs) has not been conducted until now.
We screened MEDLINE and clinicaltrials.gov (through to December 2013), as well as reference sections of systematic reviews of long-term RCTs of opioids in CNCP.
We included open-label extension trials with a study duration ≥ 26 weeks of RCTs of ≥ 2 weeks duration.
Using a random effects model, pooled estimates of event rates for categorical data and standardized mean differences (SMD) for continuous variables were calculated.
We included 11 open-label extension studies with 2445 participants with nociceptive (low back, osteoarthritis) and neuropathic (radicular, polyneuropathy) pain. Median study duration was 26 (range 26-108) weeks.
- Four studies tested oxycodone,
- two studies tramadol and buprenorphine;
- hydromorphone, morphine, oxymorphone and tapentadol were each tested in one study.
Of the patients randomized at baseline,
- 28.5 % (95 % confidence interval, CI, 17.9-39.2 %) finished the open-label period;
- 53.5 % (95 % CI 38.1-68.2 %) of patients entering the open-label period finished the open-label period.
- the total loss was 71.5 % (95 % CI 60.9-83.1 %) of all patients primarily included into the RCT.
- A total of 4.9 % (95 % CI 2.9-8.2 %) of patients dropped out due lack of efficacy;
- 16.8 % (95 % CI 11.0-24.8 %) dropped out to due adverse events (AE) in the open-label period and
- 0.08 % (95 % CI 0.001-0.05 %) of patients died during the open-label period.
I cannot make sense of these statistics: the first two seem identical. This article is a 9-page PDF file and well worth reading if you’re interested:
482_2014_1452_MOESM1_ESM.pdf (653 kb) English version PDF
Only one study systematically assessed aberrant drug behavior of the patients:
- 5.7 % (95 % CI 3.4-9.6 %) showed aberrant drug behavior in the opinion of the investigators and
- 2.6 % (95 % CI 1.2-5.8 %) were judged to show aberrant drug behavior by independent expert assessment.
I find it troubling that the “independent experts” found “aberrant drug behavior” in less than half of those that the “investigators” did.
I attribute this to the researchers trying so hard to find “aberrant drug behavior” that they labeled any case that came anywhere near it.
This kind of biased research exists only to make opioids look bad, yet it’s currently “standard practice” for research – probably because that’s the only kind being funded. (See Opioids Blamed for Side-Effects of Chronic Pain)
There was no significant change (p = 0.50) in pain intensity between the end of the randomized period and the end of open-label phase (SMD 0.19 [- 0.03, 0.41]; six studies with 1360 participants).
This means that “open-label” research (where subjects know what they are taking) doesn’t skew the results and is just as valid as the randomized studies.
In a sense, this validates our pain by comparing a couple of indirect measurements: patient-rated pain remained the same whether the patient knew if they were getting the active drug or not.
This also disproves the common assertion that much of pain is psychological, which would show extreme differences in pain ratings based on whether subjects knew they were getting the pain medicine or placebo.
Only a minority of patients selected for opioid therapy at randomization finished the long-term open-label study.
This will always be a problem with pain treatment studies because some patients suffer intense side effects (get extremely dizzy and nauseous)
But if they also get pain relief, then it’s up to them to decide how much pain they can or want to tolerate before taking an opioid that will make them nauseous.
As always, a medication must present greater benefits than harms for it to be prescribed and used.
However, sustained effects of pain reduction could be demonstrated in these patients.
LtOT can be considered in carefully selected and monitored CNCP patients who experience clinically meaningful pain reduction with at least tolerable AE in short-term opioid therapy. The English full-text version of this article is freely available at SpringerLink (under “Supplementary Material”).
= = = end of abstract = = =
The English version of the full text is at 482_2014_1452_MOESM1_ESM.pdf
= = = excerpts from full text below = = =
The rates of long-term (>6 months) opioid therapy (LtOT) in chronic non-cancer pain (CNCP) are increasing in western countries including Germany.
Despite the several available evidence-based guidelines, the efficacy of LtoT remains controversial for the following reasons:
Firstly, although randomized controlled trials (RCTs) of up to 12 weeks demonstrated the superiority of opioids over placebo in neuropathic, low back and osteoarthritis pain, the number of published trials is hither-to limited.
One critical research gap concerning the use of opioids is the lack of effectiveness studies on the long-term benefits and harms of opioids. Most placebo-controlled RCTs in pain medicine had a study duration of 12 weeks, as required by drug agencies if the drug is submitted for approval.
How can they expect patients to tolerate the placebo arm of the study? Perhaps 12 weeks on placebo with pain that isn’t excruciating could be tolerable.
In our recent systematic search of the literature relating to RCTs with opioids in CNCP, we did not find any placebo-controlled study with a study duration >26 weeks.
26 weeks is half a year and that’s a long time to tolerate pain if you get the placebo. I imagine that most of the subjects getting a placebo would drop out to get proper medical treatment of their pain.
Secondly, the safety of opioid treatment in terms of risk of addiction and death from overdose has not been properly assessed, due to the complexity of these outcomes in RCTs.
The increasing rates of opioid prescriptions, as well as of abuse of prescribed opioids and associated deaths in the US have been highlighted in editorials. There is thus no consistent good-quality evidence available for giving a strong clinical recommendation for the long-term administration of opioids in CNCP in the form of guidelines
Clinical guidelines should reflect all knowledge—not only (meta-analyses of) placebo-controlled RCTs.
Case series, open-label and open-label extension studies of RCTs have been summarized by systematic qualitative reviews in order to address the question of long-term efficacy and safety of opioids in CNCP.
Therefore, the aims of this systematic review were as follows:
to assess the long-term efficacy and harms of opioids in any type of CNCP in open-label extension studies of RCTs with opioids.
Specifically, we studied how many patients
- remained on opioid therapy,
- reported a sustained reduction of pain and disability or
- experienced serious harms (serious adverse events, SAE; death, aberrant drug behavior)
in the long-term (=6 months).
I skipped over most of the detailed Methods section, which explains how they searched for and screened the studies they incorporated into their analysis.
However, I found a strange mistake in their assessment of the bias section. They say:
Two authors (KB, WH) independently assessed the risk of bias in each trial using eight domains recommended by the Cochrane Collaboration:
- selection bias,
- performance bias,
- detection bias,
- attrition bias,
- reporting bias,
- funding bias
Oops, their list contains 9 items instead of 8 as promised and a third of them are duplicates. When the duplicates are removed the list has only 6 items.
This is a mistake on top of a mistake, certainly not critical, just sloppy, and it shakes my confidence in the scientific community.
I sure hope this is a mistake introduced by translation because Germans are especially strict about the accuracy and precision of their statements. To me, sloppy translation is the only explanation of why multiple authors and reviewers missed it.
We slightly modified one item of the tool (selection bias) to adapt to the setting of an open-label extension trial (see supplementary table 1). The criteria were scored as “yes”, “no” or unclear”.
Any disagreements were resolved by discussion. If needed, a third review author was involved (CM).
- a high-quality study as one that fulfilled six to eight,
- a moderate-quality study as one that fulfilled three to five and
- a low-quality study as one that fulfilled zero to two
of the eight validity criteria.
The duration of the open-label extension period was
- 84 weeks in one study,
- 108 weeks in one other study,
- 52 weeks in two studies and
- 26 weeks in the remaining studies
(see . Tab.?1 for an overview and supplementary table 2 for details).
- Four studies tested oxycodone,
- two studies tramadol and
- buprenorphine, hydromorphone, morphine, oxymorphone and tapentadol were each tested in one study.
Most studies used an individually titrated flexible dosage of an extended-release formulation.
The average dosage did not exceed 120 mg/d morphine equivalents, which is considered to be a mostly sufficient and safe dosage:
The key word here is “mostly” and I’m glad they clarified this. Too many American studies assume 120MME is a hard cutoff and anything above that is just evidence of “OUD”.
- Mean oxycodone dosage was =60 mg/ day,
- mean buprenorphine transdermal dosage was <20 µg/h,
- morphine dosage was =120 mg/d,
- mean tramadol dosage was =400 mg/d and
- mean tapendatol dosage was =500 mg/d.
Three studies used dosages higher than 120 mg/d morphine equivalents:
- oxycodone up to 140 mg/d,
- oxycodone up to 80 or 140 mg/d.
- Four studies reported a slight average increase and
- three studies reported a slight decrease in opioid dosage
from the end of the randomized to the end of open-label phase (see Tab. 2).
Below are the full-text Conclusion and Discussion sections in more detail.
The evidence on long-term efficacy of opioids in CNCP is limited by the reporting of average pain scores.
Averages are not useful when measuring values that have a huge and scattered variation.
Usually, studies are designed to test the degree of change (pain) when some variable (medication) is changed, and results usually cluster around the average. For example, tramadol might alleviate 40% of some of the subjects’ pain while others might achieve 60%. Then there might be a few extreme responses like 80% efficacy in some patients balanced by only 20% in others.
This would make the average close to 50% and be applicable to most people.
But averaging is a huge problem with some of the efficacy studies on medications like Lyrica, which work extremely well for some people and not at all for others. Achieving 100% efficacy in 1/2 the patients and 0% efficacy in the other 1/2 still ends up giving an average of 50% efficacy.
But this hides the fact that test subjects rarely received partial relief; Lyrica either worked or it didn’t, even though the study had determined 50% efficacy for the “average patient” (one more reason the mythical “average patient” doesn’t exist).
With a drug like Lyrica, experience shows that most people will get either great relief or none at all, but this can’t be determined just by reading the study.
Drug approval agencies should require pain responder analyses instead of reports of average pain scores of open-label extension studies
In summary, LtOT can be considered in carefully selected and monitored CNCP patients who experience a clinically meaningful pain reduction with at least tolerable AE in short-term opioid therapy.
This is no different than the use of any other prescription medications.
Isn’t it a general truth that doctors normally prescribe medications only to patients who demonstrate some benefit and can tolerate the side effects?
And the greater the negatives, the greater the positives must be to make a drug useful for a particular patient. Every individual will have their own very specific reactions to a medication, both positive and negative, and this must be evaluated on a case by case, patient by patient basis.
This is how decisions about medication have always been made in proper medical practice.
Careful monitoring of these patients and the prescription of mean dosages of opioid equivalents less than 120 mg/d have been recommended to decrease the risk of harms caused by LtOT.
The fact that all these opioid restrictions were only recommended, not required, has been lost in the zeal to “solve” the so-called “opioid crisis” (hopefully in a way to routes money to the person having the idea).
The opioid epidemic in the US highlights the need for psychosocial services in chronic pain management
This review demonstrates—despite the limitations of open-label trials including only about one third of the primarily enrolled patients discussed later—a sustained efficacy of LtOT in chronic nociceptive (low back and osteoarthritis) and neuropathic (radicular and polyneuropathy) pain in open-label extension studies of RCTs.
The mean average reductions in pain and disability were maintained from the end of the double-blind to the end of the openlabel period up to 3 years later.
However, average pain scores are unrepresentative of patient experience and of very limited utility.
None of the open-label extension studies reported the number of patients with a clinically significant (30 or 50%) pain reduction (responders). However, we assume that only patients with any benefit were treated after the RCT.
Considering the high loss of patients (about 75%) for various reasons, there would be a long-term response rate (portion) of about 25% in CNCP. This remarkable result is in line with that of a short term RCT with strict inclusion criteria such as failure of guideline-based pretreatment
The findings from open-label extension studies also concur with the results of long-term case series with selected patients of clinical centers.
Of 121 patients at a German pain center,
- 103 (85%) still took an opioid after an average treatment time of 66 months (37–105 months;
- 87% more than 5 years).
At a US pain center, 84 patients were followed-up every 3 months for a median of 11 years. Most patients in the total sample reported 50% or greater pain relief and a moderate improvement in disability.
These case series did not report a significant increase in opioid dosage in the long-run.
This pokes a giant hole in the myth of ever-rising “opioid tolerance” that would require ever-increasing doses.
It’s only when huge amounts of illicit opioids are taken to get “high” that doses must be forever escalated to get the desired effect. The pain relief provided by opioids isn’t as affected by tolerance.
The findings are in line with the results of the open-label extension studies: the average dosage of opioids did not substantially increase in most studies.
However, these outcomes were not reported by two studies. Only one study reported the range of dosages; the other studies reported only mean values. Therefore the percentage of patients with a high-dosage therapy (>120 mg morphine equivalent) remains unknown.
This is what I’ve noticed all along: the dosages given aren’t carefully documented and controlled. They might be studying “opioids” using 2 5mg Vicodin per day, or 2 80mg oxycodone for all we know.
This seems to be another subterfuge to support the myth about the dangers of opioids.
Only 5% of the patients of the reviewed studies ended therapy due to a lack of efficacy. This finding does not support theposition of Sullivan that most patients in open-label extension studies stopped opioids due to a lack of efficacy.
Most notably, the positive effects of opioids in long-term open-label studies cannot be disentangled from those of co-therapies not controlled for, from unspecific (placebo) effects because of the lack of a placebo group or from spontaneous recovery because of the lack of a no treatment group.
This is another trick used by researchers determined to make opioids seem ineffective or dangerous: they muddle their studies with all kinds of unspecified “usual treatments”.
This review found no increase in side effects in open-label extension studies, indicating safety of LtOT in chronic nociceptive (low back and osteoarthritis) and neuropathic (radicular and polyneuropathy) pain.
The rates of abuse of prescribed opioids were higher in studies with clinical and population-based samples. In a systematic review of 67 studies, the rate of opioid abuse and addiction varied between 0.2 and 3.3%.
The overall rates of aberrant drug behavior were between 15 and 20%.
Many opioid studies like to consider “aberrant drug behavior” as equal to “opioid use disorder” and even “addiction”.
A longitudinal population-based Danish study identified at least two of the six addictive behaviors in
- 22.6% of the long-term opioid users with chronic pain compared to
- 11.5% of the nonopioid users with chronic pain and
- 8.9% of the individuals without chronic pain.
So they found “addictive behaviors” in about 10% of people without opioids and even without pain. How is it possible to find signs of opioid addiction in subjects who weren’t even taking them?
Could this mean that much of what’s labeled as “aberrant drug use” or “addictive behavior” could just be a person’s response to pain?
A US study with one national, commercially insured population and one statebased, publicly insured population reported that among users of chronic opioid therapy, 3% of both samples had a claims based opioid abuse/dependency diagnosis.
in a cohort of new opioid users in a nationwide Norwegian prescription database who started treatment with weak opioids, only 0.3 and 0.08% developed prescription patterns indicating persistent opioid use and problematic opioid use, respectively, within 3 years.
These are the real numbers, not the grossly inflated numbers being spread by PROPaganda.
In contrast to previous findings that long-term treatment facilitates osteoporosis, Li et al. recently demonstrated that the risk of fracture predominantly elevates only in the first period of opioid prescription, presumably due to titration errors, and then declines—even after years of treatment.
Another scary study about opioids is proven wrong. I’m sure all American studies inflate the downsides of opioid use, especially since they don’t seem to regard pain truly as real as any other medical malady.
The death rates during the open-label extension phases were very low.
These findings are in contrast to findings in the general US population.