Fentanyl withdrawal: Understanding symptom severity and exploring the role of body mass index on withdrawal symptoms and clearance

Abstract

Background and Aims:

Fentanyl is a highly lipophilic mu opioid receptor agonist, increasingly found in heroin and other drug supplies, that is contributing to marked increases in opioid-related overdose and may be complicating treatment of opioid use disorder (OUD). This study aimed to measure the influence of body mass index (BMI) on fentanyl withdrawal and clearance.

Design, Setting, Participants:

This secondary analysis, from a 10-day inpatient study on the safety and efficacy of sublingual dexmedetomidine for opioid withdrawal, includes participants with OUD (n = 150) recruited from three sites in New York, New Jersey and Florida, who were maintained on oral morphine (30 mg four times per day) for 5 days before starting study medication. Most participants (n = 118) tested positive for fentanyl on admission to the inpatient unit.

Measurements:

Urine toxicology and opioid withdrawal symptoms [Clinical Opioid Withdrawal Scale (COWS) and Short Opiate Withdrawal Scale (SOWS)] were assessed daily. The present analysis includes data on opioid withdrawal from days 1–5 of stabilization and urine toxicology data from days 1–10.

Findings:

Fentanyl status at admission was not significantly associated with COWS or SOWS scores after adjusting for sex, site and polysubstance use. Participants classified as overweight or obese (n = 66) had significantly higher odds of testing positive for fentanyl across days 1–10 [odds ratio (OR) = 1.65; P < 0.01] and higher SOWS maximum scores across morphine stabilization (P < 0.05) compared to those with a healthy BMI (n = 68).

Conclusions:

Among inpatients with opioid use disorder, fentanyl status does not appear to be statistically significantly associated with Clinical Opioid Withdrawal Scale and Short Opiate Withdrawal Scale mean and maximum scores. High body mass index status (overweight or obese) appears to be an important predictor of slower fentanyl clearance and higher Short Opiate Withdrawal Scale maximum scores across the inpatient period than lower body mass index status.

INTRODUCTION

Opioid use disorder (OUD) presents a major public health crisis, with 76% of deaths associated with substance use disorders (SUDs) attributed to opioids [1], and an urgent need to develop and implement effective treatments. One factor complicating treatment delivery for OUD is the proliferation of fentanyl and other potent synthetic opioids in illicit opioid supplies. Fentanyl is a lipophilic full mu-opioid receptor agonist estimated to be 50–100 times more potent than morphine [2, 3]. Fentanyl, its analogs and other novel synthetic opioids, are manufactured in clandestine laboratories and are likely to persist in opioid supplies because their high potency, ease of production and low cost confer ease of distribution and high profits for illegal drug producers [4, 5].

Recent data suggest that the presence of fentanyl and other synthetic opioids in illicit opioid supplies is rising dramatically across the United States and is closely associated with increased risk for opioid overdose [6, 7]. In the US, OD deaths increased 29% between 2020 and 2021, with 100 306 deaths attributed to OD, the highest number on record [6, 8, 9]. The majority (75%) of these deaths were attributable to fentanyl and other synthetic opioids [6, 8, 10]. The ubiquity of fentanyl is also markedly shifting drug use patterns: many individuals with OUD now express a preference for fentanyl and other synthetic opioids over heroin [11-13] and these drugs appear to be supplanting heroin in certain regions of the United States [14,15].

In addition to shifting use patterns, the proliferation of fentanyl and other synthetic opioids appears to be negatively impacting initiation of effective treatments for OUD. Clinical lore and patient reports suggest that those with OUD experience fentanyl withdrawal symptoms as more severe, more enduring and having a faster onset than heroin withdrawal, which may be driving premature exits from treatment [16-19]. Recent case reports and self-report data suggest increased likelihood of severe precipitated withdrawal from buprenorphine following recent fentanyl use [20-22]. Further, one recent study found that individuals testing positive for fentanyl at the start of treatment were 11 times less likely to be successfully inducted onto extended-release naltrexone than those testing negative for fentanyl [23]. Treatment dropout and unsuccessful induction greatly increase risks of opioid-related overdose, necessitating a greater understanding of the unique pharmacological and individual factors influencing fentanyl metabolism and associated withdrawal symptoms.

Fentanyl clearance (defined as the time it takes for an individual to test negative for fentanyl on a urine drug screen after ceasing use) also appears highly variable, and probably complicates treatment initiation and success. In one recent study, individuals admitted to a 28-day residential treatment unit continued to test positive for fentanyl an average of 7.43 days after admission, with one individual still testing positive 19 days after admission [24]. Little work has directly examined the impact of fentanyl use patterns or individual differences in fentanyl clearance on treatment outcomes. Given fentanyl’s lipophilicity, body weight or body fat percentage also warrant further exploration.

A greater understanding of the individual differences impacting fentanyl metabolism and clearance, as well as the influence of fentanyl use on withdrawal onset and severity have the capacity to improve clinical decision-making and patient outcomes. Such work is especially important considering fentanyl and other synthetic opioids are likely to persist in opioid and other drug supplies. The present secondary analysis compared withdrawal symptoms between individuals who were positive versus negative for fentanyl upon admission to an inpatient unit, examined fentanyl clearance among a sample of individuals with OUD and explored the role of body mass index on fentanyl clearance and opioid withdrawal symptoms.

METHODS

Secondary analyses were performed on data collected in a trial designed to examine the safety and preliminary efficacy of a sublingual formulation of dexmedetomidine (BXCL-501) for treating opioid withdrawal symptoms. Findings from the main trial will be published elsewhere (Jones et al., unpublished).

Individuals with OUD aged 18–65 years were recruited among three study sites in New York, New Jersey and Florida. Screening assessments included measures of drug use (e.g. time-line follow-back [25], breath alcohol test and urine drug toxicology), assessments of physical health [e.g. hematology, blood chemistry panel, liver and thyroid functioning, urinalysis, self-reported medical history, electrocardiography (ECG), physical examination and vital signs] and assessments for psychiatric diagnoses (e.g. Mini International Neuropsychiatric Interview, standard psychiatric examination and Columbia Suicide Severity Rating Scale). Participants seeking treatment for OUD, endorsing chronic pain, demonstrating current unstable medical or psychiatric conditions, endorsing use of any investigational agent in the past 30 days and women who were pregnant or breastfeeding were excluded from participation.

All study procedures were approved by a federally registered Institutional Review Board (IRB) and all study procedures were conducted in accordance with the 1964 Declaration of Helsinki. Informed consent was documented and obtained prior to screening procedures and study enrollment. The trial was pre-registered on ClinicalTrials.gov (NCT04470050; 27).

Participants who enrolled into the study were admitted to an inpatient clinical research unit for up to 10 days. All participants were maintained on oral morphine (30 mg QID at 8:00 a.m., 1:00 p.m., 6:00 p.m. and 11:00 p.m.) for the first 5 days of the inpatient stay prior to randomization onto study medication. Sublingual dexmedetomidine or placebo were administered on study days 6–10. The present analysis includes data on opioid withdrawal from study days 1–5 and data on urine drug toxicology from study days 1–10. Urine toxicology was collected at screening, at the time of admission to the inpatient unit (on day 1) and once daily on all subsequent inpatient days. The timing of urine drug collection was not standardized among participants. Qualitative dip-sticks were used at screening and each day of inpatient admission to test for amphetamines, benzodiazepines, buprenorphine, cannabis, cocaine, fentanyl, ketamine, methadone, opioids and phencyclidine (PCP). Fentanyl urine drug tests had a cut-off of 20 ng/ml. Subjective and objective assessments of opioid withdrawal were collected four times each day (8:00 a.m., 10:00 a.m., 8:00 p.m. and 10:00 p.m.).

The Clinical Opioid Withdrawal Scale (COWS [28]) is a clinician-administered assessment of opioid withdrawal that measures 11 clinician-observed signs and symptoms of opioid withdrawal. Items assess heart rate, evidence of sweating, restlessness, aching bones or joints, runny nose or lacrimation, gastrointestinal upset, yawning, anxiety or irritability, tremor, gooseflesh and pupil size. Each item is scored from 0 to 4 or 0 to 5, with a total score derived from summing all 11 items and a maximum score of 47. Generally, scores are interpreted as mild (between 5 and 12), moderate (between 13 and 24), moderately severe [25-35] and severe [35]. Although the COWS is widely used and considered well-validated and reliable, there is no established minimal clinically important difference (MCID), although some suggest that a COWS score reduction of 15% or greater for an individual is clinically significant [28,29].

The Short Opiate Withdrawal Scale (SOWS [31]) is a 10-item self-report measure that assesses opioid withdrawal severity. Participants are instructed to rate the severity of 10 opioid withdrawal symptoms as ‘none’, ‘mild’, ‘moderate’ or ‘severe’. Items include: ‘feeling sick’, ‘stomach cramps’, ‘muscle spasms/twitching’, ‘feeling of coldness’, ‘heart-pounding’, ‘muscular tension’, ‘aches and pains’, ‘yawning’, ‘runny eyes’ and ‘insomnia/problems sleeping’. Scores range from 0 to 30, with higher scores suggesting more severe opioid withdrawal symptoms. Prior work suggests that a change score of 2–4 points indicates a clinically meaningful change [31-33].

STATISTICAL ANALYSES

Baseline demographic and clinical differences between those testing positive for fentanyl on admission (n = 118) and those testing negative for fentanyl on admission (n = 32) were assessed using t-tests for continuous measures and χ² tests for categorical variables.

To examine differences in COWS and SOWS mean and maximum scores during morphine stabilization, based on fentanyl status at admission, longitudinal generalized linear mixed effect models (GLMM), with a random intercept to account for the between-subject variances and a random effect of day with an autoregressive correlation structure to account for the within-subject repeated measures was used. Each model (i.e. COWS mean, COWS maximum, SOWS mean and SOWS maximum) included the fixed effects of fentanyl status at admission (positive versus negative), study day [1-5] and the interaction of fentanyl status by day. If the interaction was not significant the term was removed, and the models were re-analyzed as main effects models. To address the potential effects from the imbalance in the fentanyl status group, the GLMM models were re-analyzed allowing for unequal group variance.

To examine the impact of BMI on COWS and SOWS scores, GLMM models as described above were used, with the fixed effects of BMI (healthy versus overweight/obese), day and the interaction of BMI and day. To examine the effect of BMI on fentanyl positivity, logistic regression models, including the effects of BMI, day and the interaction of BMI and day, were used. For all BMI analyses BMI < 25 was coded as ‘healthy BMI’ while BMI ≥ 25 was coded as ‘overweight or obese’.

All models adjusted for the covariates of sex, site and polysubstance use (the number of non-opioid drugs a participant tested positive for on admission). Five participants with missing COWS scores and 16 participants with missing BMI were excluded from analyses. All analyses were conducted using SAS version 9.4, and all hypothesis tests were two-sided using level of significance 5%.

RESULTS

Sample characteristics

Two hundred and twenty-five participants provided signed consent and were admitted to the inpatient unit. One hundred and fifty participants (67%) completed morphine stabilization (days 1–5), whereas 76 participants (34%) dropped out of the study during the morphine stabilization. Of those who discontinued participation, 18% (n = 14) withdrew on day 1, 34% (n = 26) withdrew on day 2, 30% (n = 23) withdrew on day 3 and 17% (n = 13) withdrew on day 5. There was no difference in rates of fentanyl positivity between those who discontinued participation prior to day 5 and those who completed the 5-day morphine stabilization.

Among those who completed morphine stabilization, participants had a mean age of 42 years [standard deviation (SD = 10.8]; see Table 1 for full sample characteristics. Most participants (131 of 150, 87%) tested positive for fentanyl during screening, although a slightly lower percentage of participants (118 of 150, 79%) tested positive for fentanyl on the day of admission. During screening, 64 participants (64 of 150, 43%) tested positive for fentanyl only, but not opioids. Participants who tested positive for fentanyl continued to test positive for an average of 7.2 days after admission (SD) = 2.8; range = 2–10 days). Individuals classified as overweight, who tested positive for fentanyl on admission, continued to test positive for fentanyl for 7.46 days on average (n = 61; SD = 2.82), whereas individuals with a healthy BMI continued to test positive for fentanyl for 6.40 days on average (n = 60; SD = 3.05). This difference was statistically significant (P = 0.02).

TABLE 1.

Demographic characteristics (n = 150).

Characteristics	Total sample (N = 150)		Fentanyl+ (n = 118)		Fentanyl− (n = 32)		Diff. between groups
	n	Mean (SD) or %	n	Mean (SD) or %	n	Mean (SD) or %	P-valuea
Demographics
Age (years)	150	42.0 (10.8)	118	42.6 (10.8)	32	39.8 (11)	0.188
Sex							0.644
Male	108	72.0%	86	72.9%	22	68.8%
Female	42	28.0%	32	27.1%	10	31.3%
Race							0.875
White	94	62.7%	75	63.6%	19	59.4%
Black or African American	52	34.7%	40	33.9%	12	37.5%
Multiple races	1	0.7%	1	0.8%	0	0.0%
Other	3	2.0%	2	1.7%	1	3.1%
Ethnicity							0.307
Hispanic/Latino	24	16.0%	17	14.4%	7	21.9%
Non-Hispanic/Latino	126	84.0%	101	85.6%	25	78.1%
Site							0.770
New York	78	52.0%	60	50.8%	18	56.3%
New Jersey	1	0.7%	1	0.8%	0	0.0%
Florida	71	47.3%	57	48.3%	14	43.8%
BMIb status							<0.001
Healthy	68	45.3%	59	50.0%	9	28.1%
Overweight/obese	66	44.0%	55	46.6%	11	34.4%
Missing	16	10.7%	4	3.4%	12	37.5%
Polysubstance usec at admission
Number of non-opioid substances used	150	1.1 (0.8)	118	1.1 (0.8)	32	1 (0.8)	0.555
Comorbid cocaine use							0.362
No	69	46.0%	52	44.1%	17	53.1%
Yes	81	54.0%	66	55.9%	15	46.9%
Comorbid THC use							0.573
No	100	66.7%	80	67.8%	20	62.5%
Yes	50	33.3%	38	32.2%	12	37.5%
Comorbid benzodiazepine use							0.773
No	124	82.7%	97	82.2%	27	84.4%
Yes	26	17.3%	21	17.8%	5	15.6%

^aBaseline differences are assessed using t-tests for continuous measures and χ² test for categorical measures.

^bBMI = body mass index (healthy BMI < 25; overweight or obese ≥ 25).

^cPolysubstance use derived via qualitative urine drug screen. THC = tetrahydrocannabinol; SD = standard deviation.

Examinations of polydrug use in this sample suggest notable co-use of opioids, cocaine and cannabis. At screening, in addition to opioids, 56% (84 of 150) of participants tested positive for cocaine, and 38% (57 of 150) of participants tested positive for cannabis.

Analyses of the impact of fentanyl use on opioid withdrawal symptoms

There were no significant interactions or main effects of fentanyl status at admission for COWS or SOWS mean or maximum outcomes. Allowing for unequal group variances in the models did not change the results. See Figure 1 for the model-estimated mean COWS and SOWS mean and maximum scores by day and fentanyl status.

FIGURE 1. Clinical Opioid Withdrawal Scale (COWS) and Short Opiate Withdrawal Scale (SOWS)–Gossop mean and maximum scores.

Note: pcm indicates a potentially clinically meaningful difference (≥ 15% diference in scores). COWS: Clinical Opioid Withdrawal Scale; SOWS-Gossop: Short Opiate Withdrawal Scale

Analyses of BMI

Sixty-eight participants (45%) had a BMI at 25 or above, indicating overweight or obese status. Sixty-six participants (44%) had a BMI below 25, indicating a healthy BMI. There were no significant interactions between BMI and day for COWS or SOWS mean or maximum scores. For SOWS maximum score, there was a significant main effect of BMI group (P = 0.02). SOWS maximum scores were significantly higher for the overweight/obese group compared to the healthy group; see Figure 2).

FIGURE 2. Fentanyl clearance and Clinical Opioid Withdrawal Scale (COWS) and Short Opiate Withdrawal Scale (SOWS)–Gossop scores by body mass index (BMI).

** indicates significant effect at P < .01. * indicates significant effect at P < .05. pcm Indicates a potentially clinically meaningful difference (2-4 point difference in scores). SOWS-Gossop: Short Opiate Withdrawal Scale.

There were no significant interactions between BMI and day for fentanyl positivity. After removing the interaction, there was a significant main effect of BMI (P = 0.0008), such that the odds of fentanyl positivity for overweight/obese participants was 1.65 [95% confidence interval (CI) = 1.23, 2.21; P = 0.0008] times higher than the healthy BMI group while adjusting for sex, site and polysubstance use. The expected probability of being fentanyl positive was 78.4% for overweight/obese participants compared to 68.7% for the healthy group.

DISCUSSION

This secondary analysis provides important and novel preliminary comparisons of fentanyl and heroin withdrawal symptoms during transition to oral morphine. Although clinical experience and recent survey data suggest that individuals with OUD experience fentanyl withdrawal as more severe, more enduring and having a faster onset than heroin withdrawal, few published reports have examined these differences systematically. The present data add to the literature by analyzing subjective and objective opioid withdrawal symptoms between individuals testing positive for fentanyl (n = 118) and individuals testing negative for fentanyl (n = 32) upon admission to an inpatient unit. In the present analyses, controlling for sex, site and polysubstance use, there were no statistically significant effects of fentanyl status on mean or maximum COWS or SOWS scores. Although findings were not statistically significant, it is important to consider whether observed differences are potentially clinically meaningful; and although clear guidelines for clinically meaningful differences in COWS scores remain sparse, a guideline of a 15% difference in scores has been proposed [27, 28]. In the present sample, those testing negative for fentanyl demonstrated mean and maximum COWS scores that were 15% lower than those testing positive for fentanyl on day 3 of morphine stabilization, suggesting potentially clinically meaningful differences. Using a 2–4-point difference as a threshold for clinically significant differences in SOWS scores, fentanyl status did not appear to influence SOWS scores in clinically meaningful ways. It is important to note that participants in this sample were maintained on oral morphine during stabilization, and thus may have experienced less severe symptoms of opioid withdrawal than individuals undergoing inpatient opioid detoxification, transition onto partial opioid agonists or unmanaged opioid withdrawal. However, as discussed below, the severity of opioid withdrawal symptoms in a largely fentanyl-positive sample maintained on oral morphine is notable.

The present data add to the literature on fentanyl withdrawal more broadly. Compared to a prior analysis of opioid withdrawal symptoms in a sample of participants that did not appear to be using fentanyl [33], peak SOWS and COWS scores in the present sample were several points higher. Specifically, this earlier analysis of 103 individuals with OUD was conducted between 2010 and 2015, a time when fentanyl use was relatively rare. Participants in the study were maintained on morphine and the authors noted peak COWS scores of 4.0 and peak SOWS scores of 7.4 [33]. In the present sample, peak COWS scores of 8.23 and peak SOWS scores of 11.46 were observed. Notably, Dunn and colleagues [33] utilized the 16-item SOWS scale versus the 10-item SOWS–Gossop scale in the present analysis, which complicates direct comparison, and the SOWS–Gossop has demonstrated high construct validity with the 16-item SOWS [32]. Such differences must be interpreted cautiously because the two studies were conducted at different sites using different experimental procedures, but it is interesting to note that the magnitude of differences appear to be clinically meaningful. More work is needed to systematically analyze whether current trends in fentanyl use are associated with significantly higher opioid withdrawal symptoms compared to prior eras where fentanyl use was rare. Such work has important implications for the treatment of OUD.

Specifically, more work is needed to examine the severity and successful treatment of fentanyl withdrawal in those transitioning to buprenorphine, methadone or extended-release naltrexone. Such data are critical in optimizing treatment initiation and maintenance during a time when fentanyl and other synthetic opioids continue to proliferate illicit opioid supplies—driving an opioid overdose crisis and complicating implementation of available treatments. Further systematic work is also needed to more accurately capture the extent to which the prevalence and persistence of fentanyl in opioid supplies reduces the efficacy of currently available medications for OUD and necessitates novel interventions in targeting opioid withdrawal and OUD more broadly.

Despite fentanyl’s ubiquity and known lipophilicity, to our knowledge no published studies have examined fentanyl clearance or fentanyl withdrawal symptoms in the context of body weight, body fat percentage or BMI. The present analysis provides preliminary data to address this gap and highlights the need for further inquiry. Notably, in the present sample, those who were overweight or obese were significantly more likely to be fentanyl-positive on a qualitative urine drug screen throughout days 1–10 of inpatient enrollment and demonstrated a significantly higher mean number of days of fentanyl positivity compared to those with a healthy BMI. Further, SOWS maximum scores were significantly higher for those who were overweight and obese compared to those with a healthy BMI throughout morphine stabilization, and SOWS mean and maximum scores were 2–4 points higher for those classified as overweight or obese on several days of morphine stabilization (see Figure 2), suggesting a potentially clinically significant difference. As participants began study medication on day 6, it was not possible to compare these outcomes beyond day 5 in the present sample. Given fentanyl’s ubiquity and lipophilicity, more work is clearly needed to understand the impact of body weight and body fat on its metabolism and clearance, and impact upon opioid withdrawal and treatment outcomes. Future work that builds on current findings on the role of BMI in fentanyl clearance and withdrawal symptoms may help to guide clinical decision-making and more accurately account for this potentially important biomarker in the treatment of opioid withdrawal among those who are overweight or obese, and may help to guide decisions on the timing of starting medications for opioid use disorder (MOUD). Future work should also measure body fat percentage in addition to BMI. It is notable that nearly half of all participants in the current sample were classified as overweight or obese, and more work is needed to understand the prevalence of overweight and obesity in those with OUD and its relationship to clinical outcomes. Recent work suggests that the likelihood of receiving prescription opioids increases progressively with BMI, and may relate to increased risks for joint and back pain [35]. Little work to date has examined BMI as a predictor of outcomes among those with OUD.

Despite compelling findings, the present data are not without limitations. One major limitation of the present sample is the imbalance regarding fentanyl status; the sample of individuals not testing positive for fentanyl was quite small compared to the sample of individuals testing positive for fentanyl, making interpretation of findings difficult. Secondly, the sample was predominantly male, limiting generalizability to females with OUD. A stronger design would include a larger comparative sample of individuals not testing positive for fentanyl. Unfortunately, BMI data were not available for 16 participants in the sample, reducing the sample size in analyses of BMI on fentanyl clearance withdrawal symptoms and necessitating further work with a larger sample. Finally, data on psychiatric comorbidity were not available for the current analysis, although participants were excluded during screening if they demonstrated evidence of a psychiatric illness likely to interfere with study participation. The present data offer compelling explorations of fentanyl withdrawal compared to heroin withdrawal, and the potential impacts of BMI on fentanyl withdrawal and clearance. The current findings have implications for future work exploring these questions in a larger, more gender- and racially diverse sample, with an emphasis on clinical outcomes such as successful treatment initiation and adherence. A greater understanding of differences in withdrawal severity based on fentanyl use characteristics (quantity used) and individual differences in metabolism and clearance has the potential to guide clinical decision-making, especially regarding the development of more effective treatments for opioid withdrawal, guiding clinical decision on starting medications such as buprenorphine and developing effective treatment plans.