Starting Buprenorphine After Fentanyl Using Low-Dose Muscle Injection of Ketamine: A Preliminary Study

Buprenorphine is an effective medication for treatment of opioid use disorder (OUD), reducing mortality risk by as much as 70% [1], while producing individual and societal benefits such as reductions in illicit opioid use [2] and healthcare utilization [3, 4]. However, it is underutilized in part due to patient apprehensiveness of the treatment initiation process [5]. In individuals with physiologic opioid dependence, starting buprenorphine can result in severe acute withdrawal symptoms, known as buprenorphine-precipitated opioid withdrawal (BPOW) [6, 7], a distressing condition traditionally avoided by a prolonged delay incurring spontaneous opioid withdrawal symptoms [8]. BPOW is thought to occur because of buprenorphine’s pharmacological trait as a partial mu-opioid receptor agonist with high receptor affinity [9]. It displaces full opioid agonists from the mu-opioid receptor, producing an abrupt drop in opioid effects which triggers a physiological withdrawal syndrome.

While prescription analgesics and heroin were dominant in the illicit opioid market in the U.S., a 12 hour interval after last opioid use was often adequate to avoid BPOW [8]. However, as illicitly-manufactured fentanyl (hereafter referred to as “fentanyl”) grew to dominance in the U.S. from 2013 to 2022 [10], many patients reported severe symptoms of BPOW occurring even 48 hours after last use [6]. Spontaneous withdrawal symptoms, while largely resolved by 24 hours after abstinence from non-fentanyl opioids, can be severe even after 5 days of abstinence from fentanyl [11].

Fentanyl’s pharmacological traits, including its high mu-opioid receptor binding affinity and high potency, are thought to be responsible for the heightened risk and severity of BPOW [12]. Its lipophilicity contributes to increased distribution into adipose tissue and prolonged release into the bloodstream and results in the prolonged BPOW vulnerability [12]. Reports of BPOW during the fentanyl era have contributed to patient and provider hesitancy to initiate buprenorphine treatment [13].

New initiation strategies have been explored to avoid BPOW and/or to manage spontaneous withdrawal symptoms during the prolonged period of vulnerability [14]. The strategies include low-dose (“microdose”) initiation [15], high-dose (“macrodose”) initiation [16], early initiation of long-acting injectable (LAI) buprenorphine [17, 18] and rapid buprenorphine rescue after naloxone-induced withdrawal [19, 20]. Since none of these methods has achieved reliable success in avoiding BPOW, shared decision-making between the clinician and patient has become increasingly important to improve patient satisfaction [21].

An emerging facilitator to buprenorphine initiation is the use of ketamine. Approved by the Food and Drug Administration (FDA) in 1970 as a dissociative anesthetic agent, intravenous ketamine has entered mainstream use at a sub-anesthetic dose for a variety of off-label applications, most prominently rapid relief from treatment-resistant depression [22] and alleviation of both acute pain [23] and chronic pain [24]. Recent literature shows efficacy of ketamine or its bioactive metabolites in treating different phases of substance use disorders. Early clinical studies demonstrate the effectiveness of ketamine in treatment of alcohol, cocaine, opioid and cannabis use disorders [25]. In OUD, studies demonstrate that ketamine can reduce cravings, opioid use, and precipitated withdrawal symptoms [26]. One comprehensive study in a mouse model [27] demonstrated the prevention efficacy of (2 R,6 R)-hydroxynorketamine (HNK), a ketamine metabolite currently in Phase II clinical trials at every stage of OUD including addiction, withdrawal, and protracted abstinence - sometimes referred to as post-acute withdrawal syndrome (PAWS) [28]. Specifically, (2 R,6 R)-HNK prevented behavior changes seen in the mouse model of naloxone-precipitated opioid withdrawal. These effects were accompanied by restoration of synaptic plasticity markers, possibly through NMDA receptor-mediated and high-frequency electroencephalographic oscillation mechanisms.

Although the ketamine molecule interacts with many receptor types, its primary pharmacological mechanism of action is thought to be its antagonist effect at the N-methyl-D-aspartate-type glutamate receptor (NMDAr) [29, 30]. Blocking the NMDAr has been shown in animal models to rapidly reverse the central nervous system adaptations that mediate opioid dependence, tolerance and withdrawal, and to enhance opioid signaling [31]. In addition, ketamine rapidly reverses opioid tolerance at the receptor level by resensitizing mu-opioid receptors previously desensitized by prolonged exposure to opioids [32–34]. By increasing both the concentration and efficacy of endogenous opioids, it can be viewed as functioning as a proxy mu-opioid agonist [34–36].

In a recent inpatient study, adults with both OUD and MDD were randomized to receive intravenous ketamine or high dose buprenorphine and then compared on measures of anxiety and opioid cravings 2 hours, 24 hours and 7 days later. Individuals in the ketamine group reported a rapid and substantial reduction in anxiety symptom severity within hours of administration, accompanied by a pronounced decline in opioid craving intensity [37]. In another randomized controlled trial, an intravenous infusion of ketamine markedly blunted objective signs of precipitated opioid withdrawal over three hours while opioid-dependent subjects underwent a naloxone challenge under general anesthesia to facilitate discontinuation of opioids [38].

Recent case reports have demonstrated the use of ketamine to treat established BPOW using an intravenous infusion at a sub-anesthetic dose over several hours in a hospital emergency department [39], an inpatient ward [40] and a pain clinic [41]. A pilot study demonstrated the feasibility of lower, sub-dissociative doses of sublingual (SL) ketamine over several days to relieve spontaneous opioid withdrawal symptoms and prevent BPOW during buprenorphine initiation in the ambulatory setting [42].

Patients sometimes undergo buprenorphine initiation in non-hospital settings such as residential treatment programs and high-acuity crisis centers [43], but a role for ketamine in buprenorphine initiation has not yet been explored in these settings which are more cost-effective for withdrawal management than emergency departments or inpatient wards. However, they often lack the ready accessibility of specialized medical supplies and personnel required to manage the undesirable risks of a ketamine infusion at an anesthetic or sub-anesthetic dissociative dose, including: 1) cardiovascular complications from elevated blood pressure and heart rate [44], and 2) psychiatric side-effects that could result in psychological distress and behavioral challenges [45].

The low SL dose of ketamine used in the outpatient pilot study avoided a ketamine-induced dissociative state [42]. However, use of the SL mode of administration introduces logistical challenges: commercially available liquid ketamine is bitter tasting, but the more palatable compounded ketamine troches and syrup are difficult to obtain quickly.

The high-dose buprenorphine initiation strategy avoids an extended time-course, which has two advantages: 1) it ensures administration of a therapeutic dose, avoiding the risk of early patient drop-out, and 2) it reduces the number of expensive days in an inpatient setting. In the emergency department setting, the high-dose strategy has been shown to avoid BPOW in most patients [16, 46–48]. However, the favorable results have not yet been duplicated in the outpatient setting, where the range of strategies available for managing persistent or worsening withdrawal symptoms may be more limited.

An optimal strategy for ketamine-assisted buprenorphine initiation (KABI) would therefore: 1) avoid the undesirable risks of dissociative dose ketamine, 2) use commercially available liquid ketamine, and 3) use a high-dose buprenorphine initiation strategy.

Intramuscular (IM) injection of sub-dissociative dose ketamine has the potential to accomplish all of these goals. It employs commercially available liquid ketamine, which is practical in any medical setting. In addition, it produces a rapid increase in plasma concentration that peaks within 5–30 minutes after injection, then falls to about 50% of its peak level by about an hour, with a half-life of 2–3 hours [30, 49]. This time course closely matches the time course of BPOW symptoms which peak within 60 minutes after buprenorphine exposure in 90% of people [50]. IM ketamine could therefore be an excellent premedication for a high-dose buprenorphine initiation.

The present report addresses critical gaps in the empirical literature on the utility of ketamine in buprenorphine initiation 1) as premedication for a high-dose initiation strategy, and 2) use of the IM mode of administration. It describes a quality improvement project designed to evaluate the feasibility, dosing, safety, and effectiveness of a sub-dissociative dose of IM ketamine premedication for reducing BPOW severity during a high-dose buprenorphine initiation at a 24-hour behavioral health crisis center. By systematically capturing clinically observed withdrawal symptoms and initiation outcomes, this project aimed to inform best practices and support the development of evidence-based initiation protocols tailored to the challenges of the fentanyl era.

Over a thirteen-month period, 50 patients with recent fentanyl use agreed to buprenorphine initiation with the ketamine protocol. Their median age was 35.5 years (SD 7.3), 29 (58%) were male, 5 (10%) were Black, 1 (2%) was Hispanic, 1 (2%) was Native American, 43 (86%) were White, and 9 (18%) were unstably housed. No patients met exclusion criteria.

Full model results can be found in Table 1. Individual patients’ COWS scores at each time point and LOS are shown in Table 2.

The overall linear mixed-effects model was significant, X²(5) = 376.39, p < 0.0001, suggesting a strong effect of change over the timepoints measured. Compared to baseline scores, COWS scores 30 minutes after ketamine (time 2; b = −7.34, p < 0.0001, 95% CI [−8.37, −6.31]) and 30 minutes after buprenorphine (time 3; b = −9.50, p < 0.001, 95% CI [−10.53, −8.47]) decreased significantly. Adjusted mean COWS scores were 13.6 (range 8–21, SD 2.9, 95% CI [12.62, 14.54]) at baseline; 6.2 (range 0–17, SD 3.6, 95% CI [5.28, 7.20]) 30 minutes after ketamine; and 4.1 (range 0–18, SD 4.0, 95% CI [3.12, 5.04]) 30 minutes after buprenorphine. At 30 minutes after ketamine, the average change from baseline in COWS score was −7.34 points (SD = 3.85), Cohen’s d was −1.91, a very large effect size. At this timepoint, 48/50 (96%) had a decrease in COWS score including nine patients with COWS score 0–3; two patients had no change in COWS score. At 30 minutes after buprenorphine, the average change from baseline in COWS score was −9.50 points (SD = 4.02), Cohen’s d was −2.36, an extremely large effect size. At this timepoint, 36/50 (72%) had a further decrease in COWS score including 27/50 (54%) with COWS score 0–3; 6/50 (12%) had an increase in COWS score compared to the value after ketamine, including 2/50 (4%) who met criteria for BPOW; and a total of 49/50 (98%) had a decrease in COWS score from baseline.

Age was not significantly related to COWS score (p = 0.58). Sex (b = 1.39, p = 0.045, 95% CI [0.03–2.75]) and fentanyl verification method (b = −2.86, p < 0.001, 95% CI [11.81–19.54]) were significantly related to the COWS scores in the model, suggesting that females had, on average, slightly higher COWS scores than males and that patients whose fentanyl use was verified by a point of care urine test, rather than by self-report, had COWS scores that were on average 2.9 points lower (i.e., objective verification was associated with less severe withdrawal scores).

An examination of random-effects variance estimates indicated moderate between-patient variability (variance = 3.20, 95% CI [1.58, 6.48]) and residual variance (variance = 6.96, 95% CI [5.28, 9.18]). A likelihood ratio test confirmed the choice of using a mixed model, as it fit significantly better than a standard linear model (X² = 13.42, p = 0.0001).

Seven patients received additional medications during the first four hours: seven received additional buprenorphine (8 mg for six patients, 16 mg for one patient); four patients received additional ketamine (average 15 mg); and three received other comfort medications. Three of 50 patients (6%) stayed overnight due to a persistently elevated COWS score including the one patient whose COWS score after buprenorphine rose above its baseline value. Every other patient was stable enough for discharge within several hours after the first buprenorphine dose. No patient reported any adverse effect of ketamine.

Geometric mean LOS for patients receiving the ketamine protocol intervention was approximately 79% shorter than it was for patients who did not receive the intervention (i.e. the 24-patient pre-protocol comparison group)(exp(b) = 0.21). Neither the age nor sex covariate was significant (both p > 0.05). Because the LOS data was skewed, median LOS is most appropriate (as opposed to mean LOS) to consider descriptively. The median LOS was 7.0 hours (range 2–148) among patients receiving the intervention, and 66.0 hours (range 20–135) among patients who did not receive the intervention. The quality improvement protocol was therefore associated with a decrease in median LOS of 89%.

The multiple linear regression conducted to examine relationships between intervention status and log-transformed LOS, controlling for sex and age, was statistically significant, F(3,70) = 11.26, p < 0.001, and explained approximately 32.6% of the variance in the outcome (R² = 0.326). Adjusting for covariates, intervention status was significantly related to LOS (b = −1.57, SE = 0.28, t = −5.65, p < 0.001. This indicates that patients who received the ketamine intervention had a significantly shorter LOS than those who did not.

Within 30 days after the initial buprenorphine dose, all 50 patients (100%) were dispensed at least one prescription for buprenorphine from a pharmacy, and 26/50 patients (52%) attended a follow-up visit with a physician.

No protocol deviations or incomplete collection of COWS score data occurred. Follow-up was assessed on all 50 patients.

This report describes a novel buprenorphine initiation protocol for fentanyl-dependent patients that uses an initial dose of 8 mg of SL buprenorphine following premedication with IM ketamine at a low, sub-dissociative dose (10 mg = ~0.13 mg/kg). The reductions in COWS score, both after ketamine premedication and again after buprenorphine, were statistically significant. Half of patients had nearly complete resolution of withdrawal symptoms (COWS 0–3) by 30 minutes after buprenorphine. While two patients met criteria for BPOW, only one had an increase in COWS above the baseline level. Almost all patients were stable enough for discharge from the crisis center within hours after the first buprenorphine dose.

Previously described high-dose buprenorphine initiation strategies reliably allowed patients to reach a therapeutic dose quickly [16, 46–48]. The unique characteristic of this ketamine-assisted buprenorphine initiation protocol was the rapid reduction in withdrawal symptoms in most patients even before the initial buprenorphine dose.

Figure 2A shows the trajectory of COWS scores for all 50 patients. As shown in Fig. 2B, the average and by far most common trajectory of COWS scores was a decrease after ketamine and a further decrease after buprenorphine, similar for patients with severe (blue line) and moderate (red line) baseline withdrawal. As shown in Fig. 2C, while only one patient had a higher COWS score after buprenorphine compared to baseline (an increase from 12 to 18 for patient #12), five others (patients #3, #7, #23, #32 and #41) had higher COWS score after buprenorphine compared to the value after ketamine. These five patients are among those with the steepest average drop from baseline after ketamine: 77% vs. 54% for all patients. The distinct ‘V-shaped’ trajectory, characterized by robust initial relief from ketamine followed by a transient return of symptoms upon buprenorphine administration, is likely associated with multiple unknown individual factors that likely contributed, such as the patients’ prior opioid exposure intensity and duration, the specific opioid and the interval since last exposure, as well as non-opioid factors such as anxiety, other medications administered, and substances ingested.

Ketamine reduced fentanyl withdrawal in all but two patients (#27 and #33) whose COWS scores (15 and 10) did not drop until 30 minutes after buprenorphine. Ketamine’s reversal of fentanyl withdrawal in most patients is consistent with results in the previous report of KABI [35]. In that study, repeated doses of ketamine could sustain patient comfort over several days before the first dose of buprenorphine.

The results reported here indicate that the discomfort associated with a high COWS score was not required to avoid BPOW. Figure 2B shows that the COWS score dropped after ketamine and dropped further after buprenorphine in both high (14–21) and moderate (8–13) baseline COWS score groups. It is possible that ketamine premedication could reliably reduce the risk of BPOW even in patients with mild or no withdrawal symptoms, as long as sufficient time has elapsed since last fentanyl use to leave enough mu-opioid receptors unoccupied. Repeated doses of ketamine may be an effective way to allow time to pass while avoiding withdrawal symptoms [42].

A shorter inpatient LOS provides health system efficiencies and lower cost [53]. The median LOS was much shorter during the protocol period than the pre-protocol period (7.0 vs. 66.0 hours). The median values in both groups were somewhat inflated above the actual time needed for symptom management: each patient’s LOS included medical and administrative time before and after buprenorphine initiation, and the median value was further elevated by the delayed discharge of some patients with disposition issues such as unstable housing, no available transportation, or referral to a residential rehabilitation program.

The authors believe that the most important contributor to the protocol’s shorter LOS was the use of the rapid high-dose buprenorphine initiation strategy. In the pre-protocol comparison group, a multi-day low-dose buprenorphine taper was commonly used. It was also more common in the pre-protocol period to use buprenorphine at low doses for several days to manage acute withdrawal symptoms with the intent to start naltrexone, an option deemed reasonable at that time due to the relative rarity of fentanyl in the local drug supply. The high-dose strategy quickly became acceptable to patients after the BHCC’s positive early experience with ketamine premedication. The quality improvement team is not aware of any qualified patients who declined the protocol.

The comparison group was smaller than the protocol group because fentanyl was just entering the community drug supply in Mobile between 2021 and 2023, and fewer patients self-identified as fentanyl-dependent.

The 100% pharmacy dispense rate, while high, was comparable to that of other studies on buprenorphine prescriptions: 97.9% after audio-only telehealth buprenorphine initiation [54], and 91% after a bridge clinic telemedicine visit [55]. Likely contributors to the high rate of prescription dispensing are the presence of an AltaPointe Health pharmacy across the street from the BHCC, and an Alabama state grant that pays for buprenorphine for uninsured patients who enroll in their outpatient substance use disorders clinic.

The 52% follow-up rate is within the range of some other studies of buprenorphine initiation: 50.6% [47] and 86% [46] after an initial prescription at an emergency department. Patients in urgent and emergency settings such as the BHCC and emergency departments tend to be seeking immediate relief rather than ongoing treatment. In any case, this report assesses only the effectiveness of the buprenorphine initiation protocol. Many factors would influence follow-up rate independent of the initiation protocol, such as psychiatric, social, and transportation issues.

For clinicians considering implementing the protocol described here, use of the lowest concentration of ketamine (10 mg/mL) and a small syringe size (ideally a 1 mL insulin syringe) would provide the greatest safety. A higher concentration (20 mg/mL or 100 mg/mL) and larger syringe size would introduce risk of inadvertent administration of a dissociative or anesthetic-level dose of ketamine by miscalculating the volume to be injected. While respiratory depression does not occur with ketamine even at anesthetic-level dosing of 1–2 mg/kg (70–140 mg in a 70 kg person) [23, 24], an unexpected altered state of consciousness would pose medical, psychological, and behavioral risks as discussed in the Introduction.

This report’s protocol that used sub-dissociative dose IM ketamine as premedication for a high-dose buprenorphine initiation could be adapted for use in other settings where buprenorphine treatment is offered, including inpatient, emergency department, residential treatment, withdrawal management, crisis center, outpatient, carceral, or street medicine and warrants further study in these settings. Importantly, ketamine is a Schedule III controlled substance. In outpatient settings and street medicine, IM administration by a health care provider may be safer than oral or sublingual self-administration as it avoids the risk of misuse among patients with a substance use disorder in which impaired self-control is an integral part of the disease.

If the favorable results observed here are confirmed in other studies, sub-dissociative dose ketamine could be incorporated into the range of tools available, helping to address the significant apprehensiveness among both potential patients and providers that prevents many people with a fentanyl use disorder from initiating treatment with buprenorphine.

The limitations of this report include its observational nature at a single site in a naturalistic clinical setting. We do not have a comparison group on COWS scores and completion rates from the same time period since all patients offered ketamine consented to receive it. Data available from the pre-protocol period is limited due to lack of systematic data collection on COWS scores and completion rates during that time.

There are limitations due to the nature of this study as a quality improvement protocol rather than prospective research. The purpose of this quality improvement protocol was to explore whether adjunctive low-dose ketamine could improve reliability of buprenorphine initiation and patient comfort during the process. While we did not have a control group, the magnitude and rapidity of COWS score reduction observed in this cohort was substantially greater than what we typically see with buprenorphine alone in our clinical experience. These findings are preliminary and hypothesis-generating, warranting confirmation in a randomized controlled trial.

We did not collect blood pressure at standardized time points, so we were unable to assess the effect of ketamine on blood pressure. When designing the protocol, we anticipated that ketamine would be protective against the cardiovascular stress of precipitated withdrawal, based on vital signs measured in a previous prospective study [38] and the reduction in withdrawal symptoms among patients who used ketamine in advance of each buprenorphine dose [42]. Future prospective studies could assess the cardiovascular effect of ketamine by measuring pre-ketamine and post-ketamine blood pressure and heart rate in patients undergoing buprenorphine initiation.

Sex and fentanyl verification method were associated with baseline COWS scores, suggesting that these factors be taken into consideration in future studies and clinical applications. Participants and clinicians were non-blinded to medications given. COWS scoring was not recorded beyond 30 minutes after buprenorphine, although supplemental comfort medications were provided to 7 patients for withdrawal symptoms beyond 30 minutes. Fentanyl use by patient report for the first 40% of patients was less reliable than the urine testing available for the last 60%, though there was no obvious difference in outcome between the early and late patients. The COWS score measured by clinicians is somewhat imprecise and may not accurately reflect the subjective patient experience. Some patients may have experienced dissociative effects of ketamine that they did not report. The follow-up clinic visit rate did not include any patients who established care at clinics outside the AltaPointe Health system. Additionally, patients in the two LOS cohorts were not randomly assigned or matched, but rather the comparison cohort was a convenience sample taken from the time period before the intervention began.

A contributing factor to the higher LOS in the comparison group may have been exposure to tianeptine, a non-FDA-approved antidepressant that was previously highly prevalent on the Alabama Gulf Coast, typically obtained legally in gas stations and convenience stores. That drug’s complex pharmacology can complicate the withdrawal process in part due to agonist action at the mu-opioid receptor [56]. However, the tianeptine exposure rate in Alabama rapidly declined after it was classified as a controlled substance on March 15, 2021, two months before the start of the pre-protocol comparison group period [57]. The comparison group was drawn only from patients who reported fentanyl use. Although it is possible that some of those patients may have had exposure to tianeptine as the facility did not test for tianeptine, no one in the comparison group reported tianeptine use in addition to fentanyl.

In this observational study in a high-acuity crisis residential setting, sub-dissociative dose IM ketamine was a well-tolerated and inexpensive premedication for a high-dose buprenorphine initiation in a protocol that improved outcomes compared to prior practice. Nearly all patients undergoing this simple protocol had rapid reduction of spontaneous fentanyl withdrawal symptoms, half of patients had nearly complete resolution of withdrawal symptoms within one hour, and LOS was lower than with strategies used previously at the facility.

The protocol presented here has the characteristics needed to gain wide acceptance: it is simple to explain and implement and uses inexpensive and easily accessible medication. Widespread use of this protocol could increase uptake of buprenorphine among fentanyl-dependent patients and reduce overdose deaths. Formal research is warranted.

The authors deeply appreciate the courage and collaboration of the patients involved in this project. The authors would like to thank Nathan Sackett, MD for his guidance and input during the revision process.

JE conceptualized the protocol and drafted the initial manuscript. JE, EC, and ZO provided direct patient care and compiled quality improvement data. JE and EC conducted chart and database reviews. LG and TH contributed substantially to manuscript revision and creation of figures and table. CS performed statistical analysis. All authors read and approved the final manuscript.

No funding was received for this work.

All data generated or analyzed during this study are included in this published article.