Taper Paper Compendium Update December 16, 2022

Stefan Kertesz, MD, MSc
23 min readJun 9, 2022

Prepared by Stefan G. Kertesz, MD, MSc

Heersink UAB School of Medicine


This summary does not cover all opioid taper-related papers. It focuses primarily on observational data, which may be small single-clinic studies, single-state analyses (typically Medicaid) or national reports (typically US Department of Veterans Affairs or Medicare or private commercial payor databses)

  1. changes in dose reduction practices

2. evidence we have on benefit or adverse outcome for the patients.

Somewhat more “benefits’ are covered in two existing literature reviews (1, 2), but those reviews heavily emphasize small trials of willing volunteers. Conversely, the observational database papers describe what real-world reality.

Questions in this Summary

1. Has rapid prescription opioid taper or stoppage become more common? (yes) (pages 1–2)

2. Is there evidence either from trials or observational studies to argue that prescription opioid tapering is usually beneficial to patients. There is some (not none, but not a lot), evidence of safety gains from this practice (pages 2–3)

3. In common practice, what are the typical observed outcomes, in large databases, from prescription opioid taper or stoppage (and what are the limitations of this work)(pages 3–5)

Question 1: Did rapid taper or stoppage become more common in the US?

My take: In general, prescription opioid dose reduction and stoppage both became more common. A high percentage of opioid dose reductions exceed the rate of 10% per week. The percentage of opioid reductions that count as “rapid” varies across studies, depending not only on the population being studied, but how the authors defined dose reduction and defined “rapid”.

There are 2 papers that suggest an increase in the rate of “rapid taper” at the national level, in recent years (3, 4).

Neprash studied a 20% sample of Medicare Part D from 2012 to 2018. They focused on persons with long-term opioid therapy (LTOT)(3). “Adjusted rates of LTOT discontinuation increased from 5.7% of users in 2012 to 8.5% in 2017, a 49% relative increase (p < 0.001)…..The majority of LTOT discontinuations were stopped abruptly, and increased over time (70.1% to 81.2%, 2012–2017, p < 0.001)”

Fenton studied Commercial and Medicare Advantage database. Looking at persons on long-term opioids, 2008–2017, at doses of 50 morphine milligram equivalents or higher (4). Tapering was defined as a 15% or greater relative dose reduction. They write: “From 2008 to 2015, the age- and sex-standardized percentage of patients tapering daily opioid doses increased from 10.5% to 13.7%”. The presence of more rapid dose reduction was pronounced in 2016–17 versus prior years.

There are two other papers that shows a high prevalence of “rapid taper” in 2017–2019 but these two papers are a little different as they didn’t formally assess whether it became more common over time.

Stein: identified 810,120 new, chronic, high-dose opioid treatment episodes (Medicare) discontinued in 2017 or 2018, of which 72.0% (n=583,415) were rapidly discontinued(5)

Nataraj studied a sample of 595,078 patients on long-term opioids at doses >50 morphine milligram equivalents in the period of 2017–2019(6). About ½ had some dose reduction and ¼ had sustained (180 days or more) dose reduction. And, finally, about 9% had a discontinuation. Among persons who had dose reductions enduring 90 days or longer, 36.1% underwent “rapid taper”, ie exceeding the investigators’ designated threshold of 40% per month (which is already rapid, by my view). People receiving the highest doses (>150 morphine milligram equivalents) were the most likely to undergo rapid taper. This pattern (most rapid dose reductions for persons already at highest dose) is concerning as the highest-dose patients may be at higher risk for adverse outcomes after any dose reduction.

Question 2: Is there evidence either from trials or observational studies to argue that prescription opioid tapering can be beneficial to patients?

My take: There are indeed data that suggests a potential benefit, primarily from small clinical trials with voluntary patients, handled by experts, and from at least one large database analysis. However, there are also some large database studies that can suggest a benefit (note Hayes et al, lower in this document)

One compendium of the small trials was published in 2017 by Frank et al and offered the following tentative conclusion:

“Very low quality evidence suggests that several types of interventions may be effective to reduce or discontinue LTOT and that pain, function, and quality of life may improve with opioid dose reduction.”(2)

By 2020, more evidence had accrued, including a few more trials with benefit, but also more observational studies (not trials) hinting at harm. An updated review by Mackey, meant to expand upon the prior review by Frank et al, concluded:

“The net balance of benefits and harms of LTOT dose reduction for patients with chronic pain is unclear. Clinicians should closely monitor patients during the tapering process given the potential for harm.”(1)

Then, a 2022 review by Avery et al considered 36 studies to be “informative” on the question of whether clinical interventions could be used to bring doses down (7); this study also attempted to assess patient outcomes, but that was a secondary question. Like its predecessors, the 2022 review finds that doses can be reduced. Like its predecessor studies, this paper finds that outcomes are, at best, hazy and mixed. I’ll offer a bit more detail on Avery et al’s review:

Of the 36 studies formally reviewed, 21 were trials that the authors could describe narratively (randomized or non-randomized), with a subset of 11 being included in some of the meta-analyses offered in the same paper (some meta-analyses were for just 2 trials, others for as many as 5).

**Educational side-note: Readers may ask the difference between “review” and “meta-analysis”; “review” is a general term and means the authors read the paper and summarized the results as systematically as they could; “meta-analysis” means that the authors could feed results from a comparative trial into a statistical calculation of “average benefit” from the intervention, given multiple smaller studies in combination. Meta-analysis is a simpler undertaking with more traditional kinds of medical study (comparison of giving Drug A vs Drug B, after heart attack). It is somewhat harder to do, and harder to credit as authoritative, in prescription opioid reduction studies. This is because the reductions are not carried out in similar fashion among the various studies, and because the patients in one study are often very different from those in another**

The vast majority of the 36 studies were deemed to be at high risk of bias. For the subset included in meta-analyses , the authors found that some clinic interventions may be able to reduce dose or increase the likelihood of opioid discontinuation. The same review team commented on 126 uncontrolled studies, with over 90% being at “critical risk” of bias, mostly because they were unable to address potential confounding. Confounding, in this case, includes the challenge that patients whose dose is reduced may be people who are inherently at greater risk or lesser risk than those who are not. In this paper’s discussion, the authors declare moderate confidence that it is possible to make prescribed doses go down through clinic-based interventions. They again state that there is insufficient evidence on the harms from this practice. Several of the observational papers (i.e. not trials) that have drawn my concern are cited in Avery et al’s Discussion.

3. In practice, what are observed outcomes in large databases from prescription opioid taper or stoppage?

There were, as of November, 2022, twelve retrospective large database observational papers on opioid dose reduction or stoppage that seemed to suggest poor outcomes (and 2 more that could be argued as favorable to dose reduction or stoppage). One, by Quinn, more seemed to offer a clearly intermediate result: worse outcomes immediately before and immediately after opioid discontinuation, but not clearly worse results later.

A note of warning about this type of research: Before introducing the papers it’s crucial to declare the limitation that applies when we do backward-looking studies of outcomes in large database analyses, ie large retrospective database studies.

Retrospective studies can’t show all the factors that were going on in a patient’s life that might have contributed to the decision to taper, or to the outcomes. For example, a patient may approach their doctor, declare that they are doing really well and wish to discontinue opioids. These patients may, perhaps, have somewhat good outcomes. If there were a health system where every dose reduction was among people who were already feeling better and better, a study might find “opioid dose reduction is associated with good outcomes”. And the problem here would be that, perhaps, the “good outcomes” were driven more by the patients and their health system, and the opioid dose reduction is immaterial. And the opposite is possible: there may be patients who benefit.

Of course, this methodological problem also affects ALL prior studies that correlated prescribed dose and poor patient outcome, notably those of Dunn (8), Bohnert(9), and Gomes(10). Importantly, these three studies offered variable approaches to potential confounding, but often less robust. For example, Bohnert’s 2011 study, among the strongest in many ways, adjusted for only a few measured clinical morbidities. Its published results suggest a major problem of unmeasured factors that likely influenced both prescription decisions and outcomes (i.e. “confounding”). In that paper, the two most powerful protective factors against overdose were Black race and old age. If you think about it, neither makes sense biologically. In fact, old age should predispose to a greater, not a lesser, likelihood of overdose. This suggests that there were complex (unmeasured, uncontrolled, unaddressed) aspects guiding both prescription decisions and overdose events. Nevertheless, despite the problems of such data, they were embraced as the basis for assuming two simple cause and effect relationships, both of which were enshrined in quality metrics(11), criminal investigation thresholds (12), and the 2016 CDC Guideline(13), and slightly modified in its 2022 iteration:

· That the dose on a prescription is the primary factor governing the risk of a patient suffering drug poisoning death

· That for a given patient receiving a particular dose, reduction of that dose — with or without consent- will confer a safety benefit.

I reiterate the limitations of the prior research primarily to warn readers that the new literature, reviewed below, cannot fully escape the same basic challenge. Sometimes it is unavoidable that we will need to draw conclusions from retrospective databases. If a new blood disorder emerges with uncommon frequency only in people who receive a new psychiatric drug, we don’t need to request randomized controlled trials. There will be debate about cause-and-effect. This potential for debate about causation applies to all prior studies of opioid dose, and will applies to all future studies that peruse retrospective databases on the matter of dose reduction.

I here focus on the observational papers that are typically found in larger retrospective databases.

Among the papers with poor or concerning outcomes, there are at leaset 4 raising concerns about suicide or mental health crises, and “mental health crisis” is at least a warning for the emergence of suicidality.

a) Kennedy et al (2022) studied 14,037 persons aged 14 to 74 years on long-term opioid therapy for pain (≥90 days with ≥90% of days on therapy), median duration 3.7 years. Of this group 227 had been diagnosed with opioid use disorder and prescribed opiate agonist therapy, and 483 had diagnosis of opioid use disorder and were not prescribed that theraepy.

Among persons NOT diagnosed with Opioid Use Disorder, taper (i.e. dose reduction) was only associated with increased overdose risk in the analysis not adjusted for other characteristics (HR 1.59, 1.19–2.13). But with adjustment for other characteristics, the hazard of overdose was not increased (HR 1.14, 95% CI 0.84–1.53).

Among persons NOT diagnosed with Opioid Use Disorder, discontinuation was assocated with increased overdose risk, even after adjustment for other characteristics (HR 1.44, 1.12–1.83).

For persons diagnosed with OUD but not prescribed Opiate Agonist Therapy, taper was associated with reduced hazard of overdose in unadjusted and adjusted analysis. Conversely “discontinuation” was associated with a tripling of overdoses risk (HR 3.18, 95% CI 1.87–5.40)

Discontinuing increased “overdose risk among people without opioid use disorder (adjusted hazard ratio [AHR] = 1.44; 95% confidence interval [CI] 1.12, 1.83; p = 0.004). However, stronger associations were observed among people with opioid use disorder, including those not receiving opioid agonist therapy (AHR = 3.18; 95% CI 1.87, 5.40; p < 0.001) and those receiving opioid agonist therapy (AHR = 2.52; 95% CI 1.68, 3.78; p < 0.001).

Kennedy directly addresses the tension in their findings with Agnoli’s 2021 study, which provided a more dire picture. See the inset:

Support: Canadian Institutes of Health Research Project Grant

20% random sample of residents in the provincial health insurance client roster in British Columbia

b) LaRochelle (2022) studied a large US claims dataset of persons with commercial insurance or Medicare Advantage (2010–2018). This large database analysis was focused on tapering or stoppage in people who would NOT seem to require a diagnosis and treatment for opioid use disorder, so they excluded: substance use, abuse, or dependence, prior overdose, more than 3 prescribers or 2 or more early refills, injection-related infection, including hepatitis C.

In light of this exclusion, this study may be a better effort to study “taper” in the stable patient.

The study outcome was a composite event, encompassing a medical claim for opioid overdose or a suicide event. They controlled for a range of diagnoses and demographics.

The analysis was a kind of simulated trial approach, broadly speaking, more sophisticated than most.

The multivariable-adjusted rate of suicide OR overdose with a stable dose was: 0.96% (95% CI, 0.92%-0.99%),

With a taper strategy, it was 1.10% (95% CI, 0.99%-1.22%),

With abrupt discontinuation strategy was 1.28% (95% CI, 0.93%-1.38%)

They write two comments. First, note the summary: “In this emulated trial including more than 400 000 episodes of stable long-term opioid therapy, opioid tapering was associated with a small (0.15%) absolute increase in the risk of overdose or suicide events compared with a stable dosage during 11 months of follow-up.”

Second, note that the study has a much “milder” and less alarming result compared to prior studies: Although these data show a more neutral association of opioid tapering with opioid overdose or suicide compared with past studies, they do not show a protective association. Overall, evidence does not suggest that tapering opioid dosages achieves the goal of reducing the risk of harms associated with long-term opioid therapy.

Stefan Kertesz’s editorial response: I think this is one of the best-done studies. There is no evidence of benefit or safety from taper in this study (if you are looking for that, word-search “Hayes” below). What needs to be understood is the distinction between taper on a completely stable patient (which is what LaRochelle and team attempted to study) and taper on someone with a worsening clinical picture, who doctors are in conflict with. Doctors appear differentially to close of prescriptions or taper with the patients who present the “difficult patient” scenario.

Funding support: Centers for Disease Control and Prevention and National Institute on Drug Abuse

c) Fenton et al (2022) reported in JAMA Network Open on 19 377 commercially insured and Medicare Advantage enrollees who underwent tapering (2008–2017) after a 12-month period of stable doses

(ed note: summary language here is pulled from the editorial I published on this paper in JAMA Network Open at time of publication).

Taper was operationalized as a 15% or more reduction in mean daily dose. The analytic method (exposure-crossover method applies conditional regression models to compare periods after taper (12–24 months) with periods before in the same population, adjusting for clinical covariates such as drug- and alcohol-related diagnoses and demographic characteristics.

Compared with the period before dose reduction, the incidence of hospital or emergency department encounters for drug overdose or withdrawal was elevated by 57% (adjusted incident rate ratio, 1.57), in relative terms, and by 52% (adjusted incident rate ratio, 1.52) for mental health crisis in the 12 to 24 months following reduction. The elevations in observed risk were greater for patients whose baseline opioid dose was greater than the equivalent of 300 mg of morphine daily. Supplementary analyses comparing tapered patients with those not tapered were concordant: patients at stable dose remained at lowest risk, compared with patients whose doses were lower or higher.

d) Hallvik (2022) studying Oregon Medicaid recipients, found that discontinuation (abrupt or gradual) increased the risk of suicide four-fold among high-dose opioid recipients (14). However, opioid discontinuation was associated with reduced risk of overdose in the Oregon Medicaid recipients. As a key clarification: this study combines non-fatal events (which are more numerous) with fatal events (which are less so). The reduction in risk of overdose may or may not be convincing. A reduction in the risk of overdose was not found in the VA study by Oliva et al (which I coauthored), where overdose death was more common in Veterans whose doses were stopped (see below). In Hallvik’s Medicaid study, dose reduction without discontinuation was also associated with suicide, and with a broader category of opioid-related adverse events. However, it was also associated with lower risk of opioid overdose in particular.

Funding for the work: Centers for Disease Control and Prevention. Dr. Korthius was supported by NIH/NIDA but has other grants from Invidior and Alkermes.

e) DiPrete (2022) studied insurance claims from a large database in NC, focusing on 19,443 persons who had received at least 90 days of >90 MME in any time frame 2006–2008. Their primary risk focused on the risk of exposure to “rapid” dose reduction/discontinuation, defined as exceeding 35% relative reduction in a month. Their outcome of main interest was fatal or nonfatal overdose events (59 fatal opioid overdoses, 215 nonfatal overdoses, 268 fatal or nonfatal overdoses happened). Fatal and nonfatal overdose were more common for “rapid” compared to “gradual” reduction (year 1: HR, 1.43; 95% CI, 0.94–2.18; years 2–4: HR, 1.95; 95% CI, 1.31–2.90). The analyses controlled for many clinical and sociodemographic variables, including co-prescribed deaths. The results concerning gradual reduction proved difficult for me to understand. In the text of the article, gradual reduction is profiled as offering an initial safety “gain” but a later deterioration. Specifically, the authors write: “For the first 6 to 9 months of follow-up, patients with gradual dose reduction or discontinuation had the lowest risk of all outcomes”. The authors then declare that after 9 months there was a “dose-response relationship”. They explain that relationship as follows: the overdose risk (fatal and nonfatal) was highest for the rapidly discontinued, lowest for those kept on stable dose, and intermediate for those with gradual reduction, after 9 months. To my eyes, the graphics depicting model results are hard to reconcile with this text because they don’t suggest any safety gain for gradual taper, at all. The modeled hazard ratio for overdose risk, when shown in Figure 3F shows that in the first 12 months, there is a hint of an increased hazard in overdose risk after rapid taper/stoppage; the confidence interval does cross HR of 1.0 and thus falls shy of statistical significance. But given other findings in this paper, and in others it likely does mean that mean rapid taper/stoppage is “bad”. And based on Figure 3F, a gradual taper/stop is not associated with any reduction in risk of any fatal or nonfatal overdose in the first year (OR looks pretty close to 1.05 or so). That does somewhat contradict the text of the article. And, looking forward in time, in the 13–48 month period after taper or stoppage, the Odds Ratio for overdose looks to be above 1, although the graphic makes me thinks the lower bound of the 95% CI is 0.99 or 1.0. My reading of the graphic model results would be: gradual taper/stop is not associated with a reduction in overdose at any time, and may be associated with an increase in overdose risk after 12 months. Obviously, my reading of the graphics is not in agreement with the text of the article. As I learn more, I will update this passage.

Support: National Institute on Drug Abuse and Centers for Disease Control and Prevention.

f) Oliva (2020) found suicide and overdose deaths were both more likely in Veterans who underwent prescription opioid stoppage (15). This study is different in a few respects. First it looks at an older era of care (2013–14). Second, it includes short-term and long-term recipients. The finding for ‘stoppage’ applied regardless of short or long-term receipt. However, the longer the Veteran had received opioids, the greater the increment in risk of death by both overdose and suicide. This study controlled for a wide range of potential confounding variables, such as medical illnesses, mental illness, prior addiction, etc. I wrote the limitations section of this paper. I highlighted that this paper could not specify the reasons for reduction, or the speed of its execution. It’s a very concerning finding. And it still has significant limitations.

Support: Department of Veterans Affairs

g) Agnoli (2021), in a large national database, found that periods of dose reduction (15%) were associated with health service-seeking for overdose/withdrawal or for mental health crisis (16), in a large commercial database. The paper compared periods of time with tapering to periods of time with no tapering. The paper did attempt to control for covariates. It’s worth noting that this paper focuses on health care utilization events for overdose or mental health crisis rather than death from those events. That’s a bit different from Oliva’s paper (which collected actual death data).

Funding support: University of California–OptumLabs Research Credit and the Department of Family and Community Medicine, University of California. University of California, Davis School of Medicine Dean’s Office (Dean’s Scholarship in Women’s Health Research).

h) Coffin (2020) — in a smaller study looking at publicly insured persons with drug use issues in SF, found that discontinuation of prescribed pain relievers was associated with more frequent heroin use subsequently (17).

Support: Canadian Institutes of Health Research Project Grant (#180642); Michael Smith Foundation for Health Research and the University of British Columbia’s Steven Diamond Professorship in Addiction Care Innovation; Michael Smith Foundation for Health Research Scholar Award.

i) Binswanger (2020) — using Kaiser Colorado data- found a statistical association between heroin initiation and prior discontinuation of prescribed opioids (18). I want to note that elsewhere I’m pretty sure I have seen at least one other study that looked for such an association and didn’t find it, but this is one that did.

Support: Centers for Disease Control and Prevention

j) Glanz (with Binswanger) — again using Kaiser Colorado data- found that dose variation (upward or downward) was associated with overdose risk being greater (19), compared to dose stability. For those persons whose dose reached 0 morphine milligram equivalents, their risk of overdose was reduced.

Support: National Institute on Drug Abuse

k) James (2019)- studied a single clinic in Seattle (20). In this small sample, discontinuation of prescribed opioids was associated with a 3-fold increased likelihood of death by overdose. That said, the authors looked at the chart notes and they could see that the reasons for the doctors to stop the prescriptions often related to concerning behaviors in the patients. This does not make the discontinuation protective of the patient. However, it hints at the underlying complexity that bedevils all the research in this area.

Support: University of Washington School of Medicine’s Medical Student Research Training Program, the National Center for Advancing Translational Sciences of the National Institutesof Health

l) Mark (2019) studied high dose patients in Vermont Medicaid (over 120 MME, n=494) (21). Again this is a smaller study. They don’t know the reasons for stoppage. But they found stoppage was fast when it happened, and associated with subsequent crisis in health care, often going to the ER or hospital for new problems.

Support: RTI International (independent nonprofit research institute)

m) Demidenko et al (2017) offered observational data and reported that 11% Veterans developed suicidal ideation or self-directed action after prescription opioid stoppage. These were not followed by death and the study was small. Also, this study did not feature a comparison arm, so it is among the weaker studies cited here.

Among two papers with more reassuring outcomes

a) A Veterans Administration database study by Hayes et al looked at two somewhat complex “composite outcomes”(22). One composite was called opioid related adverse outcomes (AO). This was a composite of “accidents resulting in wounds/injuries, opioid-related accidents/overdoses, alcohol and non-opioid medication-related accidents/overdoses, self-inflicted injuries and violence-related injuries”. This AO was lower in persons who had discontinued opioids when the authors applied two different statistical methods of attempting to model the differences between the discontinuers and the non-discontinuers. A separate outcome was also complex: a new-onset diagnosis for a substance use disorder of any kind (SUD: opioid, non-opioid drug and alcohol use disorders). For the SUD outcome there was no difference between continuers and discontinuers with one type of statistical approach. But there was a reduced SUD outcome with the other. For me one challenge with this paper is the complexity of the “composite outcomes” where there are a very wide range of diverse situations that are blended together.

b) A separate Veterans Administration database study by the same group found that on statistical average, dose reduction or switching to intermittent dosing was associated with a tendency toward reduced pain intensity scores (23). This was a small subsidiary database analysis. The authors did serious work to try to address the limitation: in essence, we could guess that dose reductions are offered patients who are ready for them and feeling better. I don’t know if the authors perfectly addressed this, but they tried.

Medical Deterioration

A 2023 paper by Magnan et al looked at national data from 113,604 Medicare Advantage patients in the period of2008 to 2019, selected based on greater than 50 morphine milligram per more per day for 12 months. They assessed for evidence of opioid tapering by 15% or more for at least 7 months after the 12-month baseline period, and outcomes for at least 2 months and up to 12 months’ of follow-up. These analyses controlled for a wide range of factors that could differ between people who were tapered and people who were not tapered, including prior health care utilization, a large number of clinical conditions,

Dose reductions of 15% of the dose is associated with

19% increase in emergency department visits

16% increase in hospitalizations

5% reduction in primary care visits

40% reduction in adherence to blood pressure medications

31% reduction in adherence to diabetes medications

Small increases in both blood pressure and blood sugar.

The Paper with a Mixed Picture

Finally, one 2022 paper by Patrick Quinn offers what I might term a “mixed picture”.

These authors report some uptick in risk of substance related events right before prescribed opioids are started, right before they are stopped, and during periods of higher dose receipt ,and right after they are stopped (24). The results suggest that there are risks separate from the prescription dose change that may also be worse when prescriptions are changed. I’ll now provide a bit more detail:

Investigators identified 194,839 adolescents and adults up to age 65 (2010–2018) who received 90 or more cumulative days of opioid therapy. This paper focused on “substance-related events” (substance-related morbidity from claims for emergency visits, inpatient hospitalizations, and ambulance transportation with diagnoses of nontobacco substance use disorder (SUD) or drug or medication overdose). This paper did not look at mental health crises or suicide, however. Also, as a reminder, this paper studied “stoppage” and not “dose reduction”.

The “substance-related events” were more likely to occur during periods that patients received higher doses. However, the absolute rate of these events was, depending on your perspective, low (1 per 384 person-months of prescription, overall, rising to 1 per 137 at doses greater than 120 MME, a finding that was attenuated after adjusting for other clinical factors).

The authors found the rate of “substance-related events” after discontinuation was appreciable, but declinedfrom 1–30 days to >90 days since the discontinuation. Also, in credibly robust analyses I won’t detail, there was no sudden jump in risk of substance-related events in the 30 days before discontinuation, relative to the 30 days after. This could be seen as indicating that, on average, the discontinuation itself doesn’t suddenly add a new risk that was not their before. The authors are cautious about the findings.


In an editorial published in summer of 2022, my colleague Dr. Allyson Varley and I suggested that opioid dose reduction and opioid stoppage likely do deliver benefit for some patients while delivering harm to others. To our view, a “mixed” set of outcomes among people who differ, one from the other, is the more plausible interpretation of a mixture of studies, many showing hints of harm, but not all showing such hints of harm. The harms, such as they are, may include increasing distress, psychological crisis, or death of the patient. Where a clinical change to care could cause harm or benefit, the customary ethics of medicine call for informed consent, and we see no reason to set these customary ethics aside, for the most part.

We acknowledge that there will be some situations where a clinician believes a medication is causing direct harm, and does not have consent of the patient to reduce or stop. We do not believe a prescriber is required to continue a prescription that they assess to be imposing harm. However, in such reductions, the clinician should only proceed with the nonconsensual action after specifying (preferably in the record and with the patient in frank communication):

· Evidence of the harms

· A plan to mitigate harms and risks that could follow the prescription opioid dose reduction

· Telling the patient what criteria will be used to determine if the dose reduction has succeeded, or failed

We further state that where the dose reduction fails, as is common, then reversal of that dose reduction would be appropriate to consider. The reversal of a failed clinical plan accords with the rest of medical practice. The adage against reversing taper has been “worn thin” by the available evidence, and it should be discarded.


1. Mackey K, Anderson J, Bourne D, Chen E, Peterson K. Benefits and Harms of Long-term Opioid Dose Reduction or Discontinuation in Patients with Chronic Pain: a Rapid Review [epub 11/05/2020]. J Gen Intern Med. 2020;35(December):935–44.

2. Frank JW, Lovejoy TI, Becker WC, Morasco BJ, Koenig CJ, Hoffecker L, et al. Patient Outcomes in Dose Reduction or Discontinuation of Long-Term Opioid Therapy: A Systematic Review. Ann Intern Med. 2017;167(3):181–91.

3. Neprash HT, Gaye M, Barnett ML. Abrupt Discontinuation of Long-term Opioid Therapy Among Medicare Beneficiaries, 2012–2017. J Gen Intern Med. 2021.

4. Fenton JJ, Agnoli AL, Xing G, Hang L, Altan AE, Tancredi DJ, et al. Trends and Rapidity of Dose Tapering Among Patients Prescribed Long-term Opioid Therapy, 2008–2017. JAMA Netw Open. 2019;2(11):e1916271.

5. Stein BD, Sherry TB, O’Neill B, Taylor EA, Sorbero M. Rapid Discontinuation of Chronic, High-Dose Opioid Treatment for Pain: Prevalence and Associated Factors. Journal of General Internal Medicine. 2021.

6. Nataraj N, Strahan AE, Guy GP, Losby JL, Dowell D. Dose Tapering, Increases, and Discontinuity among Patients on Long-Term High-Dose Opioid Therapy in the United States, 2017–2019. Drug and Alcohol Dependence. 2022:109392.

7. Avery N, McNeilage AG, Stanaway F, Ashton-James CE, Blyth FM, Martin R, et al. Efficacy of interventions to reduce long term opioid treatment for chronic non-cancer pain: systematic review and meta-analysis. BMJ. 2022;377:e066375.

8. Dunn KM, Saunders KW, Rutter CM, Banta-Green CJ, Merrill JO, Sullivan MD, et al. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med. 2010;152(2):85–92.

9. Bohnert ASB, Valenstein M, Bair MJ, Ganoczy D, McCarthy JF, Ilgen MA, et al. Association between opioid prescribing patterns and opioid overdose-related deaths. JAMA. 2011;305(13):1315–21.

10. Gomes T, Mamdani MM, Dhalla IA, Paterson JM, Juurlink DN. Opioid dose and drug-related mortality in patients with nonmalignant pain. Arch Intern Med. 2011;171(7):686–91.

11. National Committee for Quality Assurance. Proposed Changes to Existing Measure for HEDIS®1 2020: Use of Opioids at High Dosage (UOD). Washington, D.C.: National Committee for Quality Assurance; 2019.

12. Kertesz SG, McCullough MB, Darnall BD, Varley AL. Promoting Patient-Centeredness in Opioid Deprescribing: a Blueprint for De-implementation Science. J Gen Intern Med. 2020;35:972–7.

13. Dowell D, Haegerich TM, Chou R. CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016. MMWR Morb Mortal Wkly Rep. 2016;65(1):1–49.

14. Hallvik SE, El Ibrahimi S, Johnston K, Geddes J, Leichtling G, Korthuis PT, et al. Patient outcomes after opioid dose reduction among patients with chronic opioid therapy. Pain. 2022;163(1):83–90.

15. Oliva EM, Bowe T, Manhapra A, Kertesz S, Hah JM, Henderson P, et al. Associations between stopping prescriptions for opioids, length of opioid treatment, and overdose or suicide deaths in US veterans: observational evaluation. BMJ. 2020;368:m283.

16. Agnoli A, Xing G, Tancredi DJ, Magnan E, Jerant A, Fenton JJ. Association of Dose Tapering With Overdose or Mental Health Crisis Among Patients Prescribed Long-term Opioids. JAMA. 2021;326(5):411–9.

17. Coffin PO, Rowe C, Oman N, Sinchek K, Santos GM, Faul M, et al. Illicit opioid use following changes in opioids prescribed for chronic non-cancer pain. PLoS One. 2020;15(5):e0232538.

18. Binswanger IA, Glanz JM, Faul M, Shoup JA, Quintana LM, Lyden J, et al. The Association between Opioid Discontinuation and Heroin Use: A Nested Case-Control Study. Drug and Alcohol Dependence. 2020;217:108248.

19. Glanz JM, Binswanger IA, Shetterly SM, Narwaney KJ, Xu S. Association Between Opioid Dose Variability and Opioid Overdose Among Adults Prescribed Long-term Opioid Therapy. JAMA Netw Open. 2019;2(4):e192613.

20. James JR, Scott JM, Klein JW, Jackson S, McKinney C, Novack M, et al. Mortality after discontinuation of primary care based chronic opioid therapy for pain: a retrospective cohort study (poster at Society of General Internal Medicine, 2018). Journal of General Internal Medicine. 2019;In Press.

21. Mark T, Parish W. Opioid Medication Discontinuation and Risk of Adverse Opioid-Related Health Care Events. Journal of Substance Abuse Treatment. 2019;103:58–63.

22. Hayes CJ, Krebs EE, Li C, Brown J, Hudson T, Martin BC. Association between discontinuing chronic opioid therapy and newly diagnosed substance use disorders, accidents, self-inflicted injuries and drug overdoses within the prescribers’ health care system: a retrospective cohort study. Addiction. 2021.

23. Hayes CJ, Krebs EE, Hudson T, Brown J, Li C, Martin BC. Impact of opioid dose escalation on pain intensity: a retrospective cohort study. Pain. 2020;161(5):979–88.

24. Quinn PD, Chang Z, Bair MJ, Rickert ME, Gibbons RD, Kroenke K, et al. Associations of opioid prescription dose and discontinuation with risk of substance-related morbidity in long-term opioid therapy. Pain. 2022;163(4):e588-e95.



Stefan Kertesz, MD, MSc

I am a primary care doctor and researcher at University of Alabama at Birmingham who focuses on how to deliver high quality care for vulnerable populations.