Abstract
Background & Aims:
Fentanyl is involved in most US drug overdose deaths and its use can complicate opioid withdrawal management. Clinical applications of quantitative urine fentanyl testing have not been demonstrated previously. The aim of this study was to determine whether urine fentanyl concentration is associated with severity of opioid withdrawal.
Design:
Retrospective, cross-sectional study
Setting:
Three emergency departments in an urban, academic health system from January 1, 2020 to December 31, 2021.
Participants:
Patients with opioid use disorder, detectable urine fentanyl or norfentanyl, and Clinical Opiate Withdrawal Scale (COWS) recorded within 6 hours of urine drug testing.
Measurements:
The primary exposure was urine fentanyl concentration stratified as high (≥400 ng/mL), medium (40-399 ng/mL), or low (<40 ng/mL). The primary outcome was opioid withdrawal severity measured with COWS within six hours before or after urine specimen collection. We used a generalized linear model with gamma distribution and log-link function to estimate the adjusted association between COWS and the exposures.
Findings:
For the 1,127 patients in our sample, the mean age (SD) was 40.0 (10.7), 384 (34.1%) identified as female, 332 (29.5%) reported their race/ethnicity as non-Hispanic Black, and 658 (58.4%) reported their race/ethnicity as non-Hispanic White. For patients with high urine fentanyl concentrations, the adjusted mean COWS (95% CI) was 4.4 (3.9-4.8) compared to 5.5 (5.1-6.0) among those with medium and 7.7 (6.8-8.7) among those with low fentanyl concentrations.
Conclusions:
Lower urine fentanyl concentration was associated with more severe opioid withdrawal, suggesting potential clinical applications for quantitative urine measurements in evolving approaches to fentanyl withdrawal management.
Keywords: Opioid use disorder, opioid withdrawal, fentanyl, urine drug testing
INTRODUCTION
Amid accelerating overdose mortality,1 rising emergency department (ED) visits from opioid overdose,2 and increasing hospitalizations related to non-medical opioid use, 3 it is critical for hospital clinicians to recognize and treat opioid withdrawal. Inadequate treatment of opioid withdrawal for individuals with opioid use disorder (OUD) leads to premature hospital discharges,4,5 delays in essential care,6 and prolonged suffering.7 Conversely, evidence-based withdrawal treatment with medications such as methadone and buprenorphine can lead to sustained engagement in treatment for OUD and reduces risk of subsequent overdose.7,8
Fentanyl, a synthetic opioid significantly more potent than heroin, is now involved in the majority of drug overdose deaths in the US.1 Emerging evidence suggests that opioid withdrawal treatment can be complicated for individuals using fentanyl.9–11 Some patients require higher doses of methadone or buprenorphine for relief9,10 and preliminary studies have described a higher risk of precipitated withdrawal with buprenorphine initiation.11–15 While novel approaches to initiating treatment with buprenorphine have emerged,16–19 it remains challenging to predict the optimal strategy for individual patients.20
Fentanyl is highly lipophilic and rapidly redistributes from the brain and plasma into adipose and muscle, where it accumulates in depots that slowly release fentanyl back into systemic circulation. 21–25 This pharmacokinetic profile contributes to fentanyl’s short-acting analgesic effects and long elimination half-life. It also may explain why individuals with long-term fentanyl exposure demonstrate the prolonged presence of fentanyl and norfentanyl, its inactive metabolite, in urine.26 Persistent, low-level exposure to fentanyl may be responsible for complications with initiating buprenorphine if the fentanyl leaching from adipose and muscle stores during early abstinence induces enough mu-opioid stimulation to maintain receptor neuroadaptations, but insufficient agonism to prevent clinical withdrawal.18 This relationship is largely theoretical, however, since, to our knowledge, only a single case study has explored the clinical correlation of quantitative urine fentanyl measurements.27
To better understand the etiology of complications with fentanyl withdrawal management, we first sought to determine the correlation between quantitative urine fentanyl measurements and the severity of opioid withdrawal. Pharmacokinetic data from intravenous and transdermal fentanyl demonstrate that urine fentanyl and norfentanyl concentrations vary according to dose.28,29 This suggests that urine fentanyl concentration, adjusting for norfentanyl concentration, may correlate with clinical signs of opioid withdrawal despite patient-level variations in fentanyl metabolism, urine concentration, and route of administration. Establishing this relationship would provide the first evidence to support the theory that urine fentanyl can serve as an indicator of ongoing, clinically-relevant opioid agonism from fentanyl. Understanding this relationship could also lay the foundation for non-invasive quantitative fentanyl tests to serve as adjuncts to clinical opioid withdrawal assessments in selecting the appropriate buprenorphine induction strategy for specific patients.
The goal of this study was to describe the association of urine fentanyl with the severity of opioid withdrawal among individuals with OUD seeking care in the ED. We hypothesized that lower concentrations of fentanyl would be associated with more severe opioid withdrawal.
METHODS
A. Study design and setting
We conducted a retrospective, cross-sectional study of adult patients who presented to 3 hospital EDs in an urban academic health system. These hospitals included a tertiary referral hospital, a Level 1 trauma center, and a psychiatric crisis center, which together have more than 2000 OUD-related visits annually. We extracted data from the electronic health record (Epic) through Clarity, a reporting database which allows for detailed data queries and reports (Epic Systems Corporation, Verona, WI). The study period was January 2020 to December 2021, with data analysis performed from January to May 2022. The University of Pennsylvania Institutional Review Board approved this study with a waiver of informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cross-sectional studies.30 This analysis was not pre-registered, thus results should be considered exploratory.
B. Patient selection
We identified all ED encounters for patients suspected to have OUD during the study period using two complementary approaches. First, we included patients with OUD-related International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes (Supplement) for the ED encounter. Second, we included patients who triggered a previously-implemented automated alert to peer recovery specialists.31 The alert criteria included ED chief complaints suggestive of opioid use disorder (i.e., opioid overdose or opioid detoxification), administration of buprenorphine or naloxone in the ED, or any encounter with an OUD-related ICD-10-CM diagnosis code in the health system during the preceding year.
We included encounters in which urine drug testing (UDT) was performed and in which patients had detectable concentrations of urine fentanyl or norfentanyl on UDT. We only included patients with a Clinical Opiate Withdrawal Scale (COWS) recorded within 6 hours before or after UDT collection. We included encounters for whom UDT and COWS were performed either in the ED or after patients were admitted to inpatient or observation care in the hospital. We excluded patients with missing demographic and clinical data, including age, sex, and body mass index (BMI).
C. Measurements
Hospital laboratories processed urine specimens immediately following collection. All laboratories during the study period routinely performed the ARK Diagnostics Fentanyl I or Fentanyl II immunoassays on Olympus AU chemistry analyzers with a 1 ng/mL concentration limit of detection for fentanyl and norfentanyl. For all patients with positive immunoassays, hospital laboratories automatically measured fentanyl and norfentanyl concentrations using liquid chromatography – tandem mass spectrometry with Cerilliant fentanyl-d5 and norfentanyl-d5 internal standards (Round Rock, TX). Fentanyl and norfentanyl were separated using a Shimadzu High Performance Liquid Chromatography system followed by detection via a triple quadrupole mass spectrometer (AB Sciex 4500MD) operating with turbospray atmospheric pressure chemical ionization and multiple reaction monitoring. This method’s analytical measurement range was 2–1000 ng/mL and 5–1000 ng/mL for fentanyl and norfentanyl, respectively. Values above 1000 ng/mL were reported as “>1000 ng/mL”.
D. Outcomes
The primary study outcome was recorded COWS within six hours of urine specimen collection. The COWS is an 11-item scale of subjective symptoms and objective signs of opioid withdrawal.32 It is the standard determination for severity of opioid withdrawal in clinical settings. COWS was recorded through an automated calculator in the electronic health record available to clinicians and nurses along with an automated timestamp and record of individual COWS components. If multiple COWS were recorded within six hours of urine specimen collection, we selected the value recorded within the fewest number of absolute minutes to the time of urine collection. We chose six hours as a cut-off based on face validity and our clinical experience indicating that variations in COWS measurements after this point are likely to be influenced by emergency department and hospital treatment rather than endogenous effects of fentanyl metabolism and clearance.
The main predictor of interest was urine fentanyl concentration; however, all values over 1000 ng/mL were listed only as “≥1000”. To account for this limitation of the laboratory assay, we reviewed the distribution of values (Supplement) prior to analyses and divided fentanyl concentration into three evenly distributed terciles: low (<40 ng/mL), medium (40-399 ng/mL), and high (≥400 ng/mL).
We controlled for urine norfentanyl concentration. Approximately two-thirds of norfentanyl values were reported as “>1000 ng/mL”, thus we could not distribute norfentanyl concentrations into even terciles. Instead, we categorized norfentanyl concentration as either low (≤1000 ng/mL) or high (>1000 ng/mL). In addition, we examined patient-level characteristics that we hypothesized to be associated with urine fentanyl concentrations or COWS, including age, sex, and race/ethnicity. We included the most recently available BMI (kg/m2) to account for differing volumes for potential fentanyl storage. We included UDT results for substances other than fentanyl, including other opioids (methadone, buprenorphine, oxycodone, codeine, hydrocodone, dihydrocodeine, hydromorphone, morphine, acetylmorphine, oxymorphone, or tramadol), tetrahydrocannabinol, cocaine, amphetamine, benzodiazepine, and barbiturate. We divided opioids into three categories: methadone, buprenorphine, and other opioids. We also examined ED disposition (discharge or hospitalization), site of UDT collection (ED or hospital), and time between COWS and UDT collection.
E. Analysis
First, we described patient characteristics for the encounters included in the study, stratified by recorded COWS. For the main adjusted analysis, we used multivariable regression to examine the association between urine fentanyl and norfentanyl concentrations and the primary outcome, COWS, adjusting for patient and encounter characteristics as well as fixed effects for quarter and hospital site. We used generalized linear models with gamma distributions and log link function to account for zero-inflated, right-skewed outcome distribution (Supplement). For ease of interpretation, we report model results using predictive margins to estimate adjusted mean COWS with 95% confidence intervals 33.
As a secondary analysis, we examined the binary outcome of COWS ≥13, the threshold for moderate withdrawal and the minimum for traditional buprenorphine initiation according to recent guidelines from the Substance Abuse and Mental Health Services Administration 34. We report adjusted odds ratios and probabilities using multivariable logistic regression to examine the association between categories of urine fentanyl and norfentanyl concentrations and COWS ≥13.
Finally, we performed sensitivity analyses to confirm the robustness of our primary analysis. First, we included patients with fentanyl as the only opioid detected by UDT to exclude the effects of other opioids on withdrawal. We performed an additional analysis excluding patients who received buprenorphine or naloxone in the ED or hospital prior to COWS recording. Finally, for patients with multiple qualifying encounters during the study period, we included only the first encounter. All analyses were performed using Stata, version 16.0 (StataCorp LP).
RESULTS
The final cohort consisted of 1,127 patients who had detectable fentanyl or norfentanyl concentrations on UDT with COWS within 6 hours of UDT. Of 24,131 encounters with an OUD diagnosis, we excluded 16,017 patients without urine drug testing; 321 patients with missing age, gender, or BMI data; 5,082 patients with undetectable fentanyl and undetectable norfentanyl on UDT; and 1,584 patients without documented COWS within 6 hours before or after UDT collection. A flowchart describing development of the cohort is included in the Supplement. The mean age was 40.0 (SD 10.7), and there were 385 (34.1%) women. A total of 332 (29.5%) patients reported their race/ethnicity as non-Hispanic Black and 658 (58.4%) patients as non-Hispanic White. On UDT, 213 (18.9%) patients were positive for methadone, 37 (3.3%) were positive for buprenorphine, 517 (45.9%) were positive for other opioids, 423 (37.5%) were positive for cocaine, 302 (26.8%) were positive for tetrahydrocannabinol (THC), and 513 (45.5%) were positive for benzodiazepines or barbiturates in addition to fentanyl. Approximately 63-65% of patients across each COWS category had high urine norfentanyl concentrations (>1000 ng/mL). Of 104 patients with COWS ≥13, 21 (20.2%) had high, 39 (37.5%) had medium, and 44 (42.3%) had low urine fentanyl concentrations. Full patient characteristics are presented in Table 1.
Table 1.
Patient characteristics, by Clinical Opiate Withdrawal Scale (COWS)
| Patient Characteristics | All Patients N = 1127 n (%) |
COWS ≤ 4 N = 566 n (%) |
COWS 5 – 8 N = 301 n (%) |
COWS 9 – 12 N = 156 n (%) |
COWS ≥ 13 N = 104 n (%) |
|
|---|---|---|---|---|---|---|
| Urine fentanyl concentration | High (≥400 ng/mL) | 377 (33.5) | 212 (37.5) | 103 (34.2) | 41 (26.3) | 21 (20.2) |
| Medium (40-399 ng/mL) | 423 (37.5) | 210 (37.1) | 110 (36.5) | 64 (41.0) | 39 (37.5) | |
| Low (<40 ng/mL) | 402 (35.7) | 144 (25.4) | 88 (29.2) | 51 (33.7) | 44 (42.3) | |
| Urine norfentanyl | High (≥1000 ng/mL) | 725 (64.3) | 339 (65.2) | 189 (62.8) | 101 (64.7) | 66 (63.5) |
| Low (<1000 ng/mL) | 402 (35.7) | 197 (34.8) | 112 (37.2) | 55 (35.3) | 38 (36.5) | |
| Time between COWS and UDT, minutes (SD) | 99.6 (94.3) | 99.1 (95.1) | 100.5 (96.7) | 92.2 (86.6) | 111.1 (94.1) | |
| UDT collection Site | ED | 886 (78.6) | 357 (76.9) | 268 (79.8) | 156 (77.6) | 105 (83.3) |
| Hospital | 241 (21.4) | 107 (23.1) | 68 (20.2) | 45 (22.4) | 21 (16.7) | |
| Age, mean (SD) | 40.0 (10.7) | 40.6 (11.1) | 40.0 (10.3) | 38.1 (9.7) | 39.2 (10.8) | |
| Sex | Female | 385 (34.2) | 161 (28.4) | 116 (38.5) | 71 (45.5) | 37 (35.6) |
| Male | 742 (65.8) | 405 (71.6) | 185 (61.5) | 85 (54.5) | 67 (64.4) | |
| Race/Ethnicity | Non-Hispanic Black | 332 (29.5) | 179 (31.6) | 80 (26.6) | 47 (30.1) | 26 (25.0) |
| Non-Hispanic White | 658 (58.4) | 310 (54.8) | 188 (62.5) | 96 (61.5) | 64 (61.5) | |
| Hispanic | 79 (7.0) | 46 (8.1) | 16 (5.3) | 6 (3.9) | 11 (10.6) | |
| Other | 58 (5.2) | 31 (5.5) | 17 (5.7) | 7 (4.5) | 3 (2.9) | |
| BMI, mean (SD) | 24.8 (5.3) | 24.9 (5.0) | 24.6 (5.2) | 24.8 (6.4) | 24.4 (5.0) | |
| Methadone | 213 (18.9) | 110 (19.4) | 55 (18.3) | 35 (22.4) | 13 (12.5) | |
| Buprenorphine | 37 (3.3) | 18 (3.2) | 12 (4.0) | 5 (3.2) | 2 (1.9) | |
| Other opioids | 517 (45.9) | 260 (45.9) | 135 (44.9) | 72 (46.2) | 50 (48.1) | |
| THC | 302 (26.8) | 156 (27.6) | 76 (25.3) | 45 (28.9) | 25 (24.0) | |
| Cocaine | 423 (37.5) | 234 (41.3) | 114 (37.9) | 45 (28.9) | 30 (28.9) | |
| Amphetamine | 254 (22.5) | 137 (24.2) | 67 (22.3) | 33 (21.2) | 17 (16.4) | |
| Benzodiazepine or barbiturate | 513 (45.5) | 258 (45.6) | 139 (46.2) | 76 (48.7) | 40 (38.5) | |
In unadjusted analyses, high urine fentanyl concentration was associated with lower COWS than medium or low urine fentanyl concentration (mean COWS score for high vs. medium fentanyl concentration lower by 24% [95% CI 9.6-40.9%, p=0.001]; mean COWS score for high vs. low fentanyl concentration lower by 68% [95% CI 42.0-98.9%, p<0.001]).
In adjusted analysis, patients with high urine fentanyl concentrations had lower COWS compared to patients with medium or low fentanyl concentrations. Mean COWS score (95% CI) was 4.4 (3.9-4.8) for high fentanyl concentration, 5.5 (5.1 to 6.0) for medium fentanyl concentration, and 7.7 (6.8-8.7) for low fentanyl concentration. (Table 2).
Table 2.
Adjusted mean COWS for patient characteristics
| Patient Characteristics | Adjusted Mean COWS | Adjusted Multiplicative Change in COWS* | P | |
|---|---|---|---|---|
| Urine fentanyl, ng/mL | High (≥ 400) | 4.4 (3.9 – 4.8) | reference | -- |
| Medium (40-399) | 5.5 (5.1 – 6.0) | 1.27 (1.11 - 1.45) | 0.001 | |
| Low (< 40) | 7.7 (6.8 – 8.7) | 1.77 (1.48 - 2.13) | < 0.001 | |
| Urine norfentanyl, ng/mL | High (>= 1000) | 6.3 (5.8 – 6.8) | reference | -- |
| Low (< 1000) | 4.8 (4.3 – 5.3) | 0.78 (0.66 - 0.89) | < 0.001 | |
| Age ** | 5.7 (5.4 – 6.0) | 0.94 (0.89 - 0.99) | 0.03 | |
| Sex | Male | 5.4 (5.0 – 5.7) | reference | -- |
| Female | 6.2 (5.6 – 6.8) | 1.15 (1.03 - 1.30) | 0.01 | |
| Race/Ethnicity | Non-Hispanic White | 6.0 (5.5 – 6.4) | reference | -- |
| Non-Hispanic Black | 5.4 (4.8 – 5.9) | 0.90 (0.79 - 1.03) | 0.13 | |
| Hispanic | 5.2 (4.2 – 6.3) | 0.88 (0.71 – 1.09) | ||
| Other | 4.7 (3.6 – 5.8) | 0.79 (0.62 – 1.00) | 0.05 | |
| Body mass index ** | 5.7 (5.4 – 6.0) | 1.00 (0.99 - 1.01) | 0.89 | |
| Methadone | Negative | 5.8 (5.4 – 6.1) | reference | -- |
| Positive | 5.2 (4.5 – 5.9) | 0.90 (0.78 - 1.04) | 0.15 | |
| Buprenorphine | Negative | 5.7 (5.4 – 6.0) | reference | |
| Positive | 5.6 (3.9 – 7.2) | 0.98 (0.72 – 1.33) | 1.33 | |
| Other opioid *** | Negative | 5.5 (5.1 - 5.9) | reference | -- |
| Positive | 5.9 (5.4 – 6.4) | 1.08 (0.97 - 1.21) | 0.15 | |
| THC | Negative | 5.8 (5.4 - 6.2) | 0.91 (0.80 - 1.03) | 0.12 |
| Positive | 5.4 (4.7 – 5.8) | reference | -- | |
| Cocaine | Negative | 6.1 (5.6 – 6.5) | 0.82 (0.71 - 0.95) | 0.007 |
| Positive | 5.0 (4.5 – 5.5) | reference | -- | |
| Amphetamine | Negative | 5.8 (5.4 – 6.2) | 0.91 (0.79 - 1.04) | 0.79 |
| Positive | 5.2 (4.6 – 5.9) | reference | -- | |
| Benzodiazepine or barbiturate | Negative | 5.5 (5.0 – 5.9) | 1.09 (0.95 - 1.24) | 0.95 |
| Positive | 5.9 (5.4 – 6.5) | reference | -- | |
| Time between COWS and UDT ** | 5.7 (5.4 – 6.0) | 1.00 (1.00 - 1.00) | 0.90 | |
| UDT collection site | Hospital | 5.0 (4.3 – 5.6) | reference | -- |
| ED | 5.9 (5.5 – 6.3) | 1.18 (1.02 - 1.37) | 0.03 | |
Adjusted multiplicative change in COWS are exponentiated regression coefficients for the generalized linear model with gamma distribution and log-link function, which represent the multiplicative increase or decrease from the reference value
Adjusted mean COWS for continuous covariates are presented as the mean COWS for the mean value of the covariate in the cohort, which are 40.0 years for age, 24.8 m/kg2 for body mass index, and 99.6 minutes for time between COWS and UDT.
Other opioid category includes oxycodone, codeine, hydrocodone, dihydrocodeine, hydromorphone, morphine, acetylmorphine oxymorphone, tramadol
COWS, Clinical Opioid Withdrawal Scale; THC, tetrahydrocannabinol; UDT, urine drug test
Model also adjusted for quarter and hospital site
Low urine norfentanyl concentration was also associated with lower COWS compared to high norfentanyl concentrations (mean COWS [95% CI] for low vs. high urine norfentanyl concentration 4.8 [4.3-5.3] vs. 6.3 [5.8-6.8]). Figure 1 displays adjusted mean COWS for combinations of urine fentanyl and urine norfentanyl concentrations.
Figure 1.
Adjusted mean COWS by urine fentanyl concentration and urine norfentanyl concentration
We found cocaine positivity on UDT was significantly associated with lower COWS, with mean COWS (95% CI) of 5.0 (4.5-5.5) among cocaine positive patients compared to 6.1 (5.6-6.5) among cocaine negative patients. Male sex was associated with lower COWS, with a mean (95% CI) of 5.4 (5.0-5.7) compared to 6.2 (5.6-6.8) for female sex. Higher age was also associated with lower COWS with each one-year age increase from mean age 40.0 years associated with mean multiplicative change in COWS (95% CI) of 0.94 (0.89-0.99).
In the secondary analysis examining the association of urine fentanyl concentration with the presence of at least moderate-severity opioid withdrawal, low and medium urine fentanyl concentrations had higher odds of COWS ≥13 than high urine fentanyl concentration (low vs. high urine fentanyl concentration odds ratio 5.37 [95% CI 2.60-11.10]; medium vs. high urine fentanyl concentration odds ratio 2.14 [95% CI 1.20-3.81]). (Table 3).
Table 3.
Adjusted odds ratio of having COWS ≥13, adjusted for patient characteristics
| Patient Characteristics | Adjusted Odds Ratio of Having COWS ≥13 (95% CI) |
P | |
|---|---|---|---|
| Urine fentanyl, ng/mL | High (≥ 400) | reference | -- |
| Medium (40-399) | 2.14 (1.20 - 3.81) | 0.01 | |
| Low (≤40) | 5.37 (2.60 - 11.10) | < 0.001 | |
| Urine norfentanyl, ng/mL | High (≥1000) | reference | -- |
| Low (< 1000) | 0.45 (0.25 - 0.82) | 0.01 | |
| Age | 0.91 (0.74 - 1.13) | 0.40 | |
| Sex | Male | reference | -- |
| Female | 1.08 (0.69 - 1.69) | 0.73 | |
| Race/ethnicity | Non-Hispanic White | reference | -- |
| Non-Hispanic Black | 0.72 (0.41 - 1.25) | 0.24 | |
| Hispanic | 1.17 (0.56 - 2.47) | 0.67 | |
| Other | 0.47 (0.14 – 1.6) | 0.23 | |
| Body mass index | 1.00 (0.96 - 1.04) | 0.90 | |
| Methadone | Negative | reference | -- |
| Positive | 0.60 (0.32 - 1.13) | 0.12 | |
| Buprenorphine | Negative | reference | -- |
| Positive | 0.52 (0.12, 2.39) | 0.40 | |
| Other opioid * | Negative | reference | -- |
| Positive | 1.31 (0.85 - 2.04) | 0.22 | |
| THC | Negative | reference | -- |
| Positive | 0.78 (0.47 - 1.29) | 0.33 | |
| Cocaine | Negative | reference | -- |
| Positive | 0.60 (0.34 - 1.07) | 0.08 | |
| Amphetamine | Negative | reference | -- |
| Positive | 0.60 (0.34 - 1.07) | 0.09 | |
| Benzodiazepine or barbiturate | Negative | reference | -- |
| Positive | 1.15 (0.70 - 1.91) | 0.58 | |
| Time between COWS and UDT, minutes | 1.00 (1.00 - 1.00) | 0.41 | |
| UDT collection Site | Hospital | reference | -- |
| ED | 1.75 (0.97 - 3.15) | 0.06 | |
Other opioid category includes oxycodone, codeine, hydrocodone, dihydrocodeine, hydromorphone, morphine, acetylmorphine, oxymorphone, tramadol
COWS, Clinical Opioid Withdrawal Scale; THC, tetrahydrocannabinol; UDT, urine drug test
Model also adjusted for quarter and hospital site
None of the sensitivity analyses demonstrated differences from the main analysis (Supplement). Among the 497 (44.1%) patients with fentanyl as the only opioid present, we found a similar association between urine fentanyl and norfentanyl concentrations and COWS. When we excluded 115 (10.2%) patients who received naloxone or buprenorphine prior to the COWS used in this study, we found no significant differences from the main analysis. Finally, inclusion of only the first ED encounter during the study period for 697 (61.8%) unique patients demonstrated equivalent results.
DISCUSSION
In this study, we found that lower urine fentanyl concentrations were associated with more severe opioid withdrawal among patients with OUD. This is the first analysis, to our knowledge, to demonstrate that quantitative testing of urine fentanyl correlates with signs and symptoms of opioid withdrawal. Our findings support the hypothesis that, in general, individuals with less severe withdrawal have higher concentrations of urine fentanyl than individuals with more severe withdrawal.
If these findings are confirmed and further characterized, quantitative urine fentanyl testing may have important clinical applications. For example, quantitative fentanyl testing could be used in conjunction with clinical assessments of withdrawal to identify buprenorphine dosing strategies that lower the risk of precipitated or insufficiently treated opioid withdrawal. Current guidelines instruct patients to wait for COWS ≥8-13 as a sign that full-agonist opioids have cleared.8,34 Some patients using fentanyl develop precipitated withdrawal despite following these guidelines.12,13 One theory for why this occurs is that low levels of mu-opioid agonism from fentanyl may persist despite a period of abstinence and clinical signs of moderate opioid withdrawal. This theory is indirectly supported by our finding that one in five patients had COWS ≥13 despite high urine fentanyl concentrations. This finding, that a subset of patients with high COWS also had high urine fentanyl concentration, deserves further exploration. Future work should also assess the risk of precipitated withdrawal for various concentrations of urine, or potentially plasma, fentanyl. At lower concentrations, patients might benefit from high-dose or rapid buprenorphine titration to therapeutic maintenance doses.19 At higher concentrations, patients might require low-dose initiations with overlapping full-agonist opioids.16–18
Clinical applications of this testing are currently limited by the cost and time required for mass spectrometry. This is a crucial challenge for this approach and we are not aware of any existing efforts to develop semi-quantitative, point-of-care urine tests. These and other novel drug checking approaches, such as those involving fourier transform infrared spectroscopy or portable GC/MS, could be especially useful given the abundance of fentanyl analogues and the approximately 10% false-positivity rate of existing fentanyl lateral flow chromatographic immunoassays (known as test strips).35
We also found that low urine norfentanyl concentrations, cocaine positivity, male sex, and higher age were each associated with less severe withdrawal, holding urine fentanyl concentration constant. This study was not designed to estimate these specific effects, thus they should be considered to be hypothesis generating.36 We speculate that urine norfentanyl concentrations might serve as an indicator of quantity of fentanyl use, and therefore that lower urine norfentanyl concentrations might correlate with lower tolerance. Although cocaine is a dopamine agonist without mu-opioid receptor activity, cocaine may reduce opioid withdrawal severity.37,38 Our findings regarding cocaine may reflect this effect or varying co-use patterns of cocaine and fentanyl: some individuals using cocaine may use fentanyl sporadically and some might be exposed inadvertently to fentanyl as a contaminant. Finally, these findings align with preliminary evidence suggesting women may experience more severe opioid withdrawal than men.39,40
This study has several limitations. First, we were unable to observe many variables that influence fentanyl metabolism. These include the presence of CYP-3A modulating medications, the timing of urine specimen collection relative to most recent use, and the route of fentanyl administration. Second, hydration status affects the concentration of urine. In other studies, fentanyl and norfentanyl concentrations have been adjusted by urine creatinine26,41 or normalized to each other,28,29 however in this study, urine creatinine was not measured and more than two-thirds of urine norfentanyl concentrations were listed as “≥1000 ng/mL,” precluding us from taking a direct ratio. Instead, we adjusted for urine concentration by including a dichotomized measure of urine norfentanyl concentrations in each regression models. Future studies should consider increasing the limit of quantification to respond to the increasing concentration of fentanyl in illicit drug supplies. Third, we analyzed retrospective data collected during routine clinical care and only included patients with UDT and recorded COWS. This potentially introduced selection bias by excluding patients who left the hospital prior to UDT collection or COWS measurement. Fourth, this study was conducted in a single health system, potentially limiting generalizability. Fifth, we did not incorporate patient-reported substance use histories or clinical outcomes. Sixth, although the COWS scores by fentanyl concentration were statistically significant, the mean difference between high and low fentanyl concentration was 3.3, which may not be clinically significant. Last, given a rapidly shifting illicit drug supply, opioid withdrawal may have been complicated by exposure to undetected synthetic drugs like xylazine or fentanyl analogues with different metabolism.42
CONCLUSIONS
In summary, we found that lower urine fentanyl concentration was associated with more severe opioid withdrawal among patients seeking care in the emergency department. To our knowledge, this is the first clinically meaningful correlate of quantitative urine fentanyl testing. Non-invasive, quantitative testing for fentanyl may have important future applications to inform management of opioid withdrawal among individuals who use fentanyl, though considerable work is needed to confirm and build on these preliminary findings.
Supplementary Material
Funding:
Dr. Thakrar was supported by R25DA033211 from the National Institute on Drug Abuse. Dr. Kilaru is supported by the Agency for Healthcare Research and Quality (5K12HS026372-04)
Footnotes
Conflicts of interests: None
References
- 1.O’Donnell J, Tanz LJ, Gladden RM, Davis NL, Bitting J. Trends in and Characteristics of Drug Overdose Deaths Involving Illicitly Manufactured Fentanyls — United States, 2019–2020. MMWR Morb Mortal Wkly Rep. 2021;70(50):1740–1746. doi: 10.15585/mmwr.mm7050e3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vivolo-Kantor AM, Seth P, Gladden RM, et al. Vital Signs: Trends in Emergency Department Visits for Suspected Opioid Overdoses — United States, July 2016–September 2017. Morb Mortal Wkly Rep. 2018;67(9):279–285. doi: 10.15585/mmwr.mm6709e1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Singh JA, Cleveland JD. National U.S. time-trends in opioid use disorder hospitalizations and associated healthcare utilization and mortality. PLOS ONE. 2020;15(2):e0229174. doi: 10.1371/journal.pone.0229174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Simon R, Snow R, Wakeman S. Understanding why patients with substance use disorders leave the hospital against medical advice: A qualitative study. Subst Abuse. 2020;41(4):519–525. doi: 10.1080/08897077.2019.1671942 [DOI] [PubMed] [Google Scholar]
- 5.Ti L, Ti L. Leaving the Hospital Against Medical Advice Among People Who Use Illicit Drugs: A Systematic Review. Am J Public Health. 2015;105(12):e53–e59. doi: 10.2105/AJPH.2015.302885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Harris M. Normalised pain and severe health care delay among people who inject drugs in London: Adapting cultural safety principles to promote care. Soc Sci Med. 2020;260:113183. doi: 10.1016/j.socscimed.2020.113183 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Torres-Lockhart KE, Lu TY, Weimer MB, Stein MR, Cunningham CO. Clinical Management of Opioid Withdrawal. Addiction. Published online February 22, 2022:add.15818. doi: 10.1111/add.15818 [DOI] [PubMed] [Google Scholar]
- 8.American Society of Addiction Medicine. The ASAM National Practice Guideline for the Treatment of Opioid Use Disorder: 2020 Focused Update. J Addict Med. 2020;14(2S):1–91. doi: 10.1097/ADM.0000000000000633 [DOI] [PubMed] [Google Scholar]
- 9.Bromley L, Kahan M, Regenstreif L, Srivastava A, Wyman J, Dabam F. Methadone treatment for people who use fentanyl: Recommendations. Published online June 30, 2021. [Google Scholar]
- 10.Neimark G, Tjoa C. Treating Fentanyl Withdrawal. J Behav Health Serv Res. 2020;47(4):614–615. doi: 10.1007/s11414-020-09710-8 [DOI] [PubMed] [Google Scholar]
- 11.Sue KL, Cohen S, Tilley J, Yocheved A. A Plea From People Who Use Drugs to Clinicians: New Ways to Initiate Buprenorphine are Urgently Needed in the Fentanyl Era. J Addict Med. Published online February 10, 2022. doi: 10.1097/ADM.0000000000000952 [DOI] [PubMed] [Google Scholar]
- 12.Antoine D, Huhn AS, Strain EC, et al. Method for Successfully Inducting Individuals Who Use Illicit Fentanyl Onto Buprenorphine/Naloxone. Am J Addict. 2020;n/a(n/a). doi: 10.1111/ajad.13069 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Shearer D, Young S, Fairbairn N, Brar R. Challenges with buprenorphine inductions in the context of the fentanyl overdose crisis: A case series. Drug Alcohol Rev. Published online October 13, 2021:dar.13394. doi: 10.1111/dar.13394 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Spadaro A, Sarker A, Hogg-Bremer W, et al. Reddit discussions about buprenorphine associated precipitated withdrawal in the era of fentanyl. Clin Toxicol. Published online February 4, 2022:1–8. doi: 10.1080/15563650.2022.2032730 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Varshneya NB, Thakrar AP, Hobelmann JG, Dunn KE, Huhn AS. Evidence of Buprenorphine-precipitated Withdrawal in Persons Who Use Fentanyl. J Addict Med. 2021;Publish Ahead of Print. doi: 10.1097/ADM.0000000000000922 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ahmed S, Bhivandkar S, Lonergan BB, Suzuki J. Microinduction of Buprenorphine/Naloxone: A Review of the Literature. Am J Addict. 2020;30(4):305–315. doi: 10.1111/ajad.13135 [DOI] [PubMed] [Google Scholar]
- 17.Cohen SM, Weimer MB, Levander XA, Peckham AM, Tetrault JM, Morford KL. Low Dose Initiation of Buprenorphine: A Narrative Review and Practical Approach. J Addict Med. 2021;Publish Ahead of Print. doi: 10.1097/ADM.0000000000000945 [DOI] [PubMed] [Google Scholar]
- 18.De Aquino JP, Parida S, Sofuoglu M. The Pharmacology of Buprenorphine Microinduction for Opioid Use Disorder. Clin Drug Investig. Published online April 5, 2021. doi: 10.1007/s40261-021-01032-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Herring AA, Vosooghi AA, Luftig J, et al. High-Dose Buprenorphine Induction in the Emergency Department for Treatment of Opioid Use Disorder. JAMA Netw Open. 2021;4(7):e2117128. doi: 10.1001/jamanetworkopen.2021.17128 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Thakrar AP, Kleinman RA. Opioid withdrawal management in the fentanyl era. Addiction. Published online April 13, 2022:add.15893. doi: 10.1111/add.15893 [DOI] [PubMed] [Google Scholar]
- 21.Adams AP, Pybus DA. Delayed respiratory depression after use of fentanyl during anaesthesia. BMJ. 1978;1(6108):278–279. doi: 10.1136/bmj.1.6108.278 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Caspi J, Klausner JM, Safadi T, Amar R, Rozin RR, Merin G. Delayed respiratory depression following fentanyl anesthesia for cardiac surgery. Crit Care Med. 1988;16(3):238–240. doi: 10.1097/00003246-198803000-00006 [DOI] [PubMed] [Google Scholar]
- 23.Comer SD, Cahill CM. Fentanyl: Receptor pharmacology, abuse potential, and implications for treatment. Neurosci Biobehav Rev. 2019;106:49–57. doi: 10.1016/j.neubiorev.2018.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Klausner JM. Delayed Muscular Rigidity and Respiratory Depression Following Fentanyl Anesthesia. Arch Surg. 1988;123(1):66. doi: 10.1001/archsurg.1988.01400250076013 [DOI] [PubMed] [Google Scholar]
- 25.Palmer RB. Fentanyl in postmortem forensic toxicology. Clin Toxicol. 2010;48(8):771–784. doi: 10.3109/15563650.2010.525514 [DOI] [PubMed] [Google Scholar]
- 26.Huhn AS, Hobelmann JG, Oyler GA, Strain EC. Protracted renal clearance of fentanyl in persons with opioid use disorder. Drug Alcohol Depend. 2020;214:108147. doi: 10.1016/j.drugalcdep.2020.108147 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Jegede O, Parida S, Aquino JPD. Buprenorphine Treatment of Fentanyl-Related Opioid Use Disorder. Prim Care Companion CNS Disord. 2022;24(3):40716. doi: 10.4088/PCC.21cr03163 [DOI] [PubMed] [Google Scholar]
- 28.Schwartz JG, Garriott JC, Somerset JS, Igler EJ, Rodriguez R, Orr MD. Measurements of fentanyl and sufentanil in blood and urine after surgical application. Implication in detection of abuse. Am J Forensic Med Pathol. 1994;15(3):236–241. doi: 10.1097/00000433-199409000-00011 [DOI] [PubMed] [Google Scholar]
- 29.Tanaka N, Naito T, Yagi T, Doi M, Sato S, Kawakami J. Impact of CYP3A5*3 on plasma exposure and urinary excretion of fentanyl and norfentanyl in the early postsurgical period. Ther Drug Monit. 2014;36(3):345–352. doi: 10.1097/FTD.0000000000000029 [DOI] [PubMed] [Google Scholar]
- 30.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Ann Intern Med. 2007;147(8):573–577. doi: 10.7326/0003-4819-147-8-200710160-00010 [DOI] [PubMed] [Google Scholar]
- 31.Lowenstein M, Perrone J, Xiong RA, et al. Sustained Implementation of a Multicomponent Strategy to Increase Emergency Department-Initiated Interventions for Opioid Use Disorder. Ann Emerg Med. 2022;79(3):237–248. doi: 10.1016/j.annemergmed.2021.10.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wesson DR, Ling W. The Clinical Opiate Withdrawal Scale (COWS). J Psychoactive Drugs. 2003;35(2):253–259. doi: 10.1080/02791072.2003.10400007 [DOI] [PubMed] [Google Scholar]
- 33.Norton EC, Dowd BE, Maciejewski ML. Marginal Effects—Quantifying the Effect of Changes in Risk Factors in Logistic Regression Models. JAMA. 2019;321(13):1304. doi: 10.1001/jama.2019.1954 [DOI] [PubMed] [Google Scholar]
- 34.Substance Abuse and Mental Health Services Administration (SAMHSA). Buprenorphine Quick Start Guide. Published online 2021:6. [Google Scholar]
- 35.Park JN, Sherman SG, Sigmund V, Breaud A, Martin K, Clarke WA. Validation of a lateral flow chromatographic immunoassay for the detection of fentanyl in drug samples. Drug Alcohol Depend. 2022;240:109610. doi: 10.1016/j.drugalcdep.2022.109610 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Westreich D, Greenland S. The Table 2 Fallacy: Presenting and Interpreting Confounder and Modifier Coefficients. Am J Epidemiol. 2013;177(4):292–298. doi: 10.1093/aje/kws412 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Stine SM, Kosten TR. Reduction of opiate withdrawal-like symptoms by cocaine abuse during methadone and buprenorphine maintenance. Am J Drug Alcohol Abuse. 1994;20(4):445–458. doi: 10.3109/00952999409109183 [DOI] [PubMed] [Google Scholar]
- 38.Leri F, Flores J, Rajabi H, Stewart J. Effects of cocaine in rats exposed to heroin. Neuropsychopharmacol Off Publ Am Coll Neuropsychopharmacol. 2003;28(12):2102–2116. doi: 10.1038/sj.npp.1300284 [DOI] [PubMed] [Google Scholar]
- 39.Huhn AS, Berry MS, Dunn KE. Review: Sex-Based Differences in Treatment Outcomes for Persons With Opioid Use Disorder. Am J Addict. 2019;28(4):246–261. doi: 10.1111/ajad.12921 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Ware OD, Ellis JD, Dunn KE, Hobelmann JG, Finan P, Huhn AS. The association of chronic pain and opioid withdrawal in men and women with opioid use disorder. Drug Alcohol Depend. 2022;240:109631. doi: 10.1016/j.drugalcdep.2022.109631 [DOI] [PubMed] [Google Scholar]
- 41.Cummings OT, Enders JR, McIntire GL, Backer R, Poklis A. Fentanyl–Norfentanyl Concentrations During Transdermal Patch Application: LC–MS-MS Urine Analysis. J Anal Toxicol. 2016;40(8):595–600. doi: 10.1093/jat/bkw067 [DOI] [PubMed] [Google Scholar]
- 42.Friedman J, Montero F, Bourgois P, et al. Xylazine spreads across the US: A growing component of the increasingly synthetic and polysubstance overdose crisis. Drug Alcohol Depend. 2022;233:109380. doi: 10.1016/j.drugalcdep.2022.109380 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.