Abstract
Objective:
Fentanyl is a major driver of the overdose crisis in the United States, yet little is known about the intentionality of fentanyl use and its correlates among people who use drugs (PWUD). We examined the intentionality of fentanyl use and associated factors among PWUD.
Methods:
We recruited 246 PWUD in New York City in 2023. Participants completed structured interviews and provided urine samples for toxicology screening. Fentanyl use was classified as no use (negative in both self-report and urine toxicology), unintentional use (self-report negative but toxicology positive), and intentional use (positive in self-report with or without positive toxicology). We used multinomial logistic regression to identify factors associated with fentanyl use intentionality.
Results:
Fentanyl was detected in 40.7%, while 26.0% self-reported intentional fentanyl use. Intentional fentanyl use was significantly associated with chronic pain (RRR: 2.21; 95% CI: 1.02–4.80) and higher Risk Assessment Battery (RAB) scores (RRR: 1.09; 95% CI: 1.01–1.18), and inversely associated with high cannabis use severity (RRR: 0.16; 95% CI: 0.05–0.61). Unintentional fentanyl use was positively associated with older age (RRR: 1.08; 95% CI: 1.02–1.13) and negatively with moderate alcohol use severity (RRR: 0.25; 95% CI: 0.09–0.69).
Conclusions:
Unintentional fentanyl use was more common among older PWUD, highlighting the need for targeted outreach and fentanyl education. Intentional fentanyl use, while reflecting awareness of the drug supply, was associated with chronic pain and elevated HIV risk behaviors, indicating a subgroup with greater health and prevention needs. Tailored interventions integrating harm reduction, pain management, and HIV prevention are needed.
Keywords: Fentanyl, Substance-Related Disorders, HIV, Chronic Pain, Drug Overdose, People Who Use Drugs
Introduction
The United States (US) is in the fourth wave of the overdose epidemic, marked by polysubstance use and fentanyl-involved overdoses with stimulants.1 Overdose mortality accelerated in 2015, coinciding with the availability of fentanyl, with death rates doubling from 16.3 to 32.6 per 100,000 by 2022.2 Between 1999 and 2023, overdose deaths in the US approached 1 million, with synthetic opioids, including fentanyl, accounting for an increasing share.3 While data indicated a modest decline in overdose deaths in 2024,3 the latest data in 2025 show an increase of 1,400 fatal overdoses over the previous 12-month period.4 Overdose rates have increased among Black and Native American populations while declining among White populations, with impacts on older Black adults.5 Overdose deaths in New York City (NYC) follow national trends, with fentanyl-involved overdoses disproportionately affecting older Black residents.3 For example, overdose deaths in NYC declined in 2024, yet racial disparities persist as overdose rates among Black and Latino residents remain about twice as high than White residents.6
The potency of fentanyl and ease of distribution have made it a common adulterant in other drugs, sometimes without people’s knowledge.7 Fentanyl is frequently mixed with heroin, cocaine, methamphetamine, and MDMA.8 Research suggests that fentanyl often appears in stimulant supplies unintentionally, through accidental mix-ups or cross-contamination during packaging.7 The fentanyl-driven crisis is defined by an unpredictable drug supply, which make it difficult for people who use drugs (PWUD) to assess their overdose risk.5, 9 Fentanyl contamination of drugs increases unintentional use, while its rapid absorption and high concentrations increase lethality.10 Naloxone remains critical for overdose reversal, but unrecognized fentanyl exposure may limit overdose preparedness. Distinguishing intentional from unintentional fentanyl use can inform harm reduction, treatment, and prevention strategies, yet surveillance and research in this area remain limited.10, 11
Intentional and unintentional fentanyl use are influenced by distinct sociodemographic, health, and structural factors. A review found that intentional use was more common among younger White men with injection and prior overdose, whereas unintentional use was more prevalent among older PWUD and those unaware of drug contamination.8 Motivations for intentional use include the high potency of fentanyl, widespread availability, low cost, and ability to delay withdrawal.8 Structural and health-related factors, including housing instability, incarceration, and mental health conditions, may influence fentanyl exposure and intent by limiting access to safer supply or harm reduction services and increasing vulnerability to contaminated drug markets.12 Unstable housing and incarceration can disrupt treatment and harm reduction engagement, while co-occurring mental health conditions may heighten reliance on opioids to cope with stressors, influencing both intentional and unintentional fentanyl use.13, 14
Chronic pain is an underexplored driver of fentanyl use. Affecting over 1.9 billion people globally, it is a cause of disability and a barrier to daily functioning and quality of life and remains undertreated.15 Among those with opioid use disorder (OUD), 48–60% report chronic pain,16 and one study found that 76% of people who inject drugs with chronic pain used illicit opioids to manage it.17 Kim et al. (2025) identified inadequate pain relief as a factor linked to intentional fentanyl use, treatment discontinuation, and overdose.18
Evidence suggests that fentanyl use is associated with HIV and hepatitis C virus (HCV) transmission.9, 19 People with OUD face elevated risk for bloodborne infections due to unsafe injection practices, limited access to harm reduction services, and exposure to an unpredictable, adulterated drug supply.9, 19 In particular, fentanyl’s short duration of effect often leads to more frequent injections, increasing syringe sharing and unsafe injection practices, compounding overdose and infectious disease risks.9, 19 Fentanyl use has been linked to higher HCV seroconversion among PWID,9 yet the role of fentanyl use intentionality in HIV transmission risk remains limited.19
Given the persisting threat of fentanyl and the ongoing overdose epidemic, it is essential to examine correlates of both intentional and unintentional fentanyl use including both potential behavioral and structural risk factors and health-related correlates. We aimed to examine correlates of intentional and unintentional fentanyl use among PWUD in NYC using both self-report and urine toxicology data to inform more responsive harm reduction and care strategies among a volatile drug landscape. The sample included PWUD broadly and was not restricted to people with OUD.
Methods
Study design and participants
We recruited a sample of 260 PWUD in NYC between July and December 2023. Recruitment methods included flyers posted at syringe service programs (SSPs) and transitional housing clinics, referrals from other NYU Langone Health studies, and outreach in known drug use hotspots. Hotspots were determined through NYC overdose data briefs and street outreach of research staff. The study recruited from all five NYC boroughs, with outreach concentrated in Manhattan, the Bronx, and Queens. Eligible participants were aged 18 or older and reported polysubstance use on at least one day in the past 30 days as reported on the Addiction Severity Index.20 For recently incarcerated individuals (within 30 days), substance use in the 30 days prior to incarceration was assessed. Participants provided written informed consent. Urine samples were analyzed using the ECO II 19-panel Cup toxicology screening. Due to a supply chain issue, 14 participants received a different toxicology screening cup that did not include fentanyl and were excluded from the analytic sample. As data collection occurred during a condensed period from July to December, it is not expected that there would be significant changes to the drug supply and a lack of fentanyl testing for these participants was not expected to bias study results. Participants received a $40 for completing the interview and $15 for providing a urine toxicology sample.
Measures
Primary outcome
Past 30-day fentanyl use was assessed using items from the Polysubstance Assessment Tool (PAT),21 which captured intentional fentanyl use in the past 30 days. Individuals were asked about which substances they had used, even once, in the past 30-days from a select all that apply list of thirteen substance use categories. Fentanyl was included in the list as the fourth substance category, presented to participants as “Heroin or fentanyl.” For participants who endorsed use of heroin or fentanyl, they were asked, “When you used heroin or fentanyl in the past 30 days, what did you usually use? By usually, we mean most often, more than 50% or half of the time.” Answer choices included (i) heroin, does not contain fentanyl, (ii) heroin, unknown if it contains fentanyl, (iii) heroin, known or suspected to contain fentanyl, and (iv) fentanyl. The first two answer choices were recoded to indicate no self-reported fentanyl exposure, with the latter options indicative of fentanyl use. Individuals who did not indicate fentanyl among the substances used in the past 30 days were classified as having no self-reported fentanyl use. Responses were combined with urine toxicology results to classify participants into three categories. Those who were negative on both self-report and toxicology were classified as having no fentanyl use. Those who did not self-report fentanyl use but tested positive for fentanyl in urine toxicology were classified as having unintentional fentanyl use. Those who self-reported fentanyl use (including heroin known or suspected to contain fentanyl) were classified as having intentional fentanyl use, regardless of urine toxicology results, given the short detection window.
Covariates
Demographic variables included age, gender, race/ethnicity, and educational attainment, marital status, and employment status. Structural factors included housing status and incarceration history. Moreover, participants were asked whether they had experienced chronic pain defined to them as pain that persisted or recurred for three months or longer in the past 12 months. The Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5) was administered, with scores ≥31 indicating probable PTSD.22 Sensation seeking was measured with the Brief Sensation Seeking Scale (BSSS).23 Participants were asked if they had ever attempted suicide.
Substance use severity was assessed using the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST),24 which categorizes substance use involvement into none/low, moderate, and high severity based on established cut-points. These categories were used as indicators of relative substance use severity rather than DSM-5-TR diagnostic substance use disorders. Participants were asked about use, cravings, health and social consequences, and efforts to quit for alcohol, cannabis, cocaine, amphetamines, and opioids. Each response was scored per ASSIST guidelines and categorized into none/low, moderate, or high severity. The Risk Assessment Battery (RAB) was used to assess HIV risk behaviors, including sexual and injection-related behaviors.25 Injection drug use and non-fatal overdose in the past six months were assessed via self-report.
Statistical analysis
We conducted bivariate analyses comparing sociodemographic, behavioral, structural, and health-related variables across the three fentanyl use groups using chi-square tests for categorical variables and analysis of variance (ANOVA) for continuous variables. Bivariate analyses were used to screen candidate covariates for multivariable modeling using a liberal threshold (P < 0.20), consistent with purposeful covariate selection approaches described in applied epidemiologic and regression modeling texts.26 Final variable selection was based on conceptual relevance, assessment of collinearity, and model stability rather than statistical significance alone. Injection and opioid use severity were evaluated for inclusion but excluded from the primary model; injection drug use is a component of RAB and was therefore collinear, and opioid use severity had a zero cell count in the no-fentanyl-use group, precluding stable model estimation. Results are presented as relative risk ratios (RRRs), obtained by exponentiating coefficients from the multinomial logistic regression, representing the relative probability of intentional or unintentional fentanyl use compared with no fentanyl use. All analyses were conducted in Stata Version 18.
Results
Participant characteristics
The sample consisted of 246 PWUD, with a mean age of 46.2 years (SD=12.5). Most participants were men (74.8%); 44.7% identified as non-Hispanic Black, 60.6% had completed high school/GED, and 51.2% were unemployed. Unstable housing was reported by 41.1% of participants, and 55.1% reported experiencing chronic pain in the past 12 months. In terms of substance use severity, 27.6% reported high alcohol use severity, 22.0% for cannabis use severity, 39.8% for cocaine use severity, and 33.3% for opioid use severity. The mean RAB score was 6.44 (SD=4.35). Injection was reported by 17.1%, and 17.5% reported a non-fatal overdose in the past six months. Additionally, 52.4% screened positive for probable PTSD and 30.5% reported a lifetime suicide attempt (Table 1).
Table 1.
Sociodemographic, substance use, structural, and mental health characteristics by fentanyl use intentionality among people who use drugs in New York City, 2023.
| Variable | Fentanyl use |
||||
|---|---|---|---|---|---|
| Total n (%) | No use n (%) | Unintentional use n (%) | Intentional use n (%) | P value | |
|
| |||||
| Overall | 246 | 135 (54.9) | 47 (19.1) | 64 (26.0) | |
|
| |||||
| Sociodemographics | |||||
|
| |||||
| Mean age (SD) | 46.2 (12.47) | 42.80 (13.34) | 54.28 (7.66) | 47.45 (10.41) | 0.009 |
|
| |||||
| Gender | 0.440 | ||||
| Men | 184 (74.8) | 97 (71.8) | 38 (80.8) | 49 (76.6) | |
| Women | 62 (25.2) | 38 (28.2) | 9 (19.2) | 15 (23.4) | |
|
| |||||
| Race or ethnicity | 0.243 | ||||
| Non-Hispanic Black | 110 (44.7) | 66 (48.9) | 23 (48.9) | 21 (32.8) | |
| Non-Hispanic White | 44 (17.9) | 24 (17.8) | 5 (10.6) | 15 (23.5) | |
| Hispanic | 83 (33.7) | 39 (28.9) | 18 (38.3) | 26 (40.6) | |
| Multiple/Other race | 9 (3.7) | 6 (4.4) | 1 (2.1) | 2 (3.1) | |
|
| |||||
| High school diploma or GED | 0.068 | ||||
| No | 97 (39.4) | 48 (35.6) | 16 (34.0) | 33 (51.6) | |
| Yes | 149 (60.6) | 87 (64.4) | 31 (66.0) | 31 (48.4) | |
|
| |||||
| Marital status | 0.480 | ||||
| Single | 155 (63.0) | 92 (68.2) | 26 (55.3) | 37 (57.8) | |
| Married or living with partner | 48 (19.5) | 23 (17.0) | 11 (23.4) | 14 (21.9) | |
| Separated/divorced/widowed | 43 (17.5) | 20 (14.8) | 10 (21.3) | 13 (20.3) | |
|
| |||||
| Employment status | 0.014 | ||||
| Unemployed | 126 (51.2) | 69 (51.1) | 22 (46.8) | 35 (54.7) | |
| Retired/disabled, receiving SSDI | 62 (25.2) | 27 (20.0) | 20 (42.5) | 15 (23.4) | |
| Employed, full-time/part-time | 58 (23.6) | 39 (28.9) | 5 (10.6) | 14 (21.9) | |
|
| |||||
| Chronic pain, past 12 months | 0.061 | ||||
| No | 110 (44.9) | 69 (51.5) | 19 (40.4) | 22 (34.4) | |
| Yes | 135 (55.1) | 65 (48.5) | 28 (59.6) | 42 (65.6) | |
|
| |||||
| Substance use behaviors | |||||
|
| |||||
| Alcohol use severity | 0.053 | ||||
| None/low | 100 (40.7) | 44 (32.6) | 24 (51.1) | 32 (50.0) | |
| Moderate | 78 (31.7) | 47 (34.8) | 11 (23.4) | 20 (31.2) | |
| High | 68 (27.6) | 44 (32.6) | 12 (25.5) | 12 (18.8) | |
|
| |||||
| Cannabis use severity | 0.018 | ||||
| None/low | 51 (20.7) | 20 (14.8) | 12 (25.5) | 19 (29.7) | |
| Moderate | 141 (57.3) | 76 (56.3) | 28 (59.6) | 37 (57.8) | |
| High | 54 (22.0) | 39 (28.9) | 7 (14.9) | 8 (12.5) | |
|
| |||||
| Cocaine use severity | 0.004 | ||||
| None/low | 47 (19.1) | 31 (23.0) | 6 (12.8) | 10 (15.6) | |
| Moderate | 101 (41.1) | 65 (48.1) | 14 (29.8) | 22 (34.4) | |
| High | 98 (39.8) | 39 (28.9) | 27 (57.4) | 32 (50.0) | |
|
| |||||
| Amphetamine use severity | 0.268 | ||||
| None/low | 153 (62.2) | 79 (58.5) | 36 (76.6) | 38 (59.4) | |
| Moderate | 66 (26.8) | 40 (29.6) | 8 (17.0) | 18 (28.1) | |
| High | 27 (11.0) | 16 (11.9) | 3 (6.4) | 8 (12.5) | |
|
| |||||
| Opioid use severity | <0.001 | ||||
| None/low | 87 (35.4) | 78 (57.8) | 9 (19.2) | 0 (0) | |
| Moderate | 77 (31.3) | 43 (31.8) | 15 (31.9) | 19 (29.7) | |
| High | 82 (33.3) | 14 (10.4) | 23 (48.9) | 45 (70.3) | |
|
| |||||
| RAB score mean (SD) a | 6.44 (4.35) | 5.88 (3.46) | 6.28 (3.46) | 7.72 (6.08) | <0.001 |
|
| |||||
| Injection drug use, last 6 months | <0.001 | ||||
| No | 203 (82.9) | 126 (93.3) | 38 (82.6) | 39 (60.9) | |
| Yes | 42 (17.1) | 9 (6.7) | 8 (17.4) | 25 (39.1) | |
|
| |||||
| Non-fatal overdose, last 6 months | 0.071 | ||||
| No | 203 (82.5) | 117 (86.7) | 39 (83.0) | 47 (73.4) | |
| Yes | 43 (17.5) | 18 (13.3) | 8 (17.0) | 17 (26.6) | |
|
| |||||
| Structural variables | |||||
|
| |||||
| Unstable housing, current | 0.378 | ||||
| No | 145 (58.9) | 83 (61.5) | 29 (61.7) | 33 (51.6) | |
| Yes | 101 (41.1) | 52 (38.5) | 18 (38.3) | 31 (48.4) | |
|
| |||||
| Incarceration history | 0.003 | ||||
| Never | 34 (14.7) | 28 (22.4) | 1 (2.2) | 5 (8.2) | |
| More than one year ago | 168 (72.7) | 79 (63.2) | 40 (88.9) | 49 (80.3) | |
| One year ago, or less | 29 (12.6) | 18 (14.4) | 4 (8.9) | 7 (11.5) | |
|
| |||||
| Mental health variables | |||||
|
| |||||
| BSSS score mean (SD) b | 25.60 (5.95) | 26.01 (6.00) | 24.31 (6.31) | 25.67 (5.53) | 0.605 |
|
| |||||
| PCL-5 c | 0.735 | ||||
| Below threshold (<31) | 117 (47.6) | 67 (49.6) | 22 (46.8) | 28 (43.8) | |
| Probable PTSD (≥31) | 129 (52.4) | 68 (50.4) | 25 (53.2) | 36 (56.2) | |
|
| |||||
| Lifetime suicide attempt | 0.638 | ||||
| No | 171 (69.5) | 96 (71.1) | 30 (63.8) | 45 (70.3) | |
| Yes | 75 (30.5) | 39 (28.9) | 17 (36.2) | 19 (29.7) | |
The Risk Assessment Battery (RAB)
Brief measure of sensation seeking (BSSS)
The Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5)
Prevalence of fentanyl use and intentionality
Overall, 40.7% (n=100) tested positive for fentanyl by urine toxicology. Based on combined self-report and UTS results, 54.9% of participants had no fentanyl use, 19.1% were classified as having unintentional fentanyl use, and 26.0% had intentional fentanyl use.
Fentanyl use and correlates
In bivariate analyses, participants with unintentional fentanyl use were older (mean age 54.3 years) than those with no fentanyl use (42.8 years) or intentional fentanyl use (47.5 years) (P=0.009). Unemployment was more common among those with intentional use (54.7%) than among those with no use (51.1%) or unintentional use (46.8%) (P=0.014). Chronic pain was more prevalent among those with intentional use (65.6%) compared with those with no use (48.5%) (P=0.061). Regarding substance use characteristics, high cannabis use severity was less common among participants with intentional use (12.5%) than among those with no use (28.9%) (P=0.018). In contrast, high cocaine use severity was more common among participants with unintentional use (57.4%) and intentional use (50.0%) than among those with no use (28.9%) (P=0.004), and high opioid use severity was observed in 70.3% of participants with intentional use and 48.9% of participants with unintentional use, compared with 10.4% among those with no use (P<0.001). Mean RAB scores were greater among those with intentional fentanyl use than among those with no use (mean 7.72 vs. 5.88, P<0.001). Recent injection drug use was reported by 39.1% of participants with intentional use, compared with 6.7% of those with no use and 17.4% of those with unintentional use (P<0.001). Finally, incarceration more than one year ago was reported by 80.3% of participants with intentional use and 88.9% of those with unintentional use, compared with 63.2% of those with no use, while incarceration within the past year was reported by 11.5%, 8.9%, and 14.4% among participants with intentional use, unintentional use, and no use, respectively (P=0.003) (Table 1).
In multivariable multinomial logistic regression (referent category: no fentanyl use), intentional fentanyl use was significantly associated with chronic pain (RRR 2.21; 95% CI 1.02–4.80) and higher RAB scores (RRR 1.09 per unit increase; 95% CI 1.01–1.18). High severity of cannabis use was negatively associated with intentional fentanyl use (RRR 0.16; 95% CI 0.05– 0.61). Moreover, unintentional fentanyl use was significantly associated with older age (RRR 1.08; 95% CI: 1.02–1.13) and inversely associated with moderate alcohol use severity compared to none/low severity (RRR 0.25; 95% CI: 0.09–0.69) (Table 2).
Table 2.
Multivariable multinomial logistic regression of factors associated with intentionality of fentanyl use among people who use drugs in New York City, 2023.
| Variable | Unintentional fentanyl use | Intentional fentanyl use | ||
|---|---|---|---|---|
|
| ||||
| RRR (95% CI) | P value | RRR (95% CI) | P value | |
|
| ||||
| Age | 1.08 (1.02, 1.13) | 0.004 | 1.02 (0.97, 1.05) | 0.444 |
|
| ||||
| High school diploma or GED | ||||
| No | 1 | 1 | ||
| Yes | 1.24 (0.53, 2.87) | 0.621 | 0.58 (0.28, 1.21) | 0.150 |
|
| ||||
| Employment status | ||||
| Unemployed | 1 | 1 | ||
| Retired/disabled, receiving SSDI | 0.83 (0.30, 2.27) | 0.717 | 0.76 (0.28, 2.05) | 0.589 |
| Employed, full-time/part-time | 0.33 (0.09, 1.30) | 0.115 | 0.94 (0.37, 2.35) | 0.888 |
|
| ||||
| Chronic pain, past 12 months | ||||
| No | 1 | |||
| Yes | 0.93 (0.39, 2.23) | 0.885 | 2.21 (1.02, 4.80) | 0.043 |
|
| ||||
| Alcohol use severity | ||||
| None/low | 1 | 1 | ||
| Moderate | 0.25 (0.09, 0.69) | 0.008 | 0.56 (0.24, 1.29) | 0.179 |
| High | 0.46 (0.16, 1.33) | 0.154 | 0.40 (0.14, 1.09) | 0.074 |
|
| ||||
| Cannabis use severity | ||||
| None/low | 1 | 1 | ||
| Moderate | 0.72 (0.25, 2.08) | 0.552 | 0.55 (0.21, 1.40) | 0.213 |
| High | 0.31 (0.07, 1.26) | 0.103 | 0.16 (0.05, 0.61) | 0.007 |
|
| ||||
| Cocaine use severity | ||||
| None/low | 1 | 1 | ||
| Moderate | 0.49 (0.12, 1.95) | 0.315 | 1.00 (0.32, 3.11) | 0.990 |
| High | 1.72 (0.46, 6.41) | 0.416 | 2.21 (0.69, 7.04) | 0.179 |
|
| ||||
| Amphetamine use severity | ||||
| None/low | 1 | 1 | ||
| Moderate | 0.78 (0.27, 2.23) | 0.656 | 1.24 (0.52, 2.93) | 0.614 |
| High | 0.49 (0.10, 2.22) | 0.358 | 0.90 (0.29, 2.79) | 0.865 |
|
| ||||
| RAB score | 1.05 (0.94, 1.17) | 0.331 | 1.09 (1.01, 1.18) | 0.047 |
|
| ||||
| Non-fatal overdose, last 6 months | ||||
| No | 1 | 1 | ||
| Yes | 1.11 (0.37, 3.33) | 0.844 | 2.32 (0.92, 5.87) | 0.073 |
|
| ||||
| Incarceration history | ||||
| Never | 1 | 1 | ||
| More than one year ago | 7.11 (0.85, 59.45) | 0.070 | 2.05 (0.65, 6.42) | 0.216 |
| One year ago, or less | 2.83 (0.25, 32.09) | 0.401 | 1.25 (0.29, 5.41) | 0.759 |
The reference category for the multinomial model was “no fentanyl use.” Variables with a p-value < 0.2 in the bivariable analysis were included in the multivariable multinomial logistic regression model. Injection drug use and opioid use severity were excluded due to collinearity between injection and the RAB score, as well as a zero cell count for opioid use severity.
In sensitivity analyses replacing the RAB score with injection drug use, injection was strongly associated with intentional fentanyl use. Associations with chronic pain and cannabis use severity were directionally consistent with the primary model, although effect estimates were attenuated (Supplementary Table 1).
Self-reported past 30-day substance use by fentanyl use intentionality
Past 30-day substance use differed across fentanyl groups (Table 3). Unintentional fentanyl use occurred both among participants who reported using heroin believed not to contain fentanyl and among those reporting non-opioid substance use, including cocaine, cannabis, and methamphetamine, indicating multiple possible pathways of fentanyl exposure. Cannabis was reported by 80.7% of those with no fentanyl use, 57.5% with unintentional use, and 67.2% with intentional use (P=0.004). Heroin, without known fentanyl, was used by 9.6% of the no-use group, 57.5% of the unintentional group, and none of the intentional group (P<0.001). Prescription opioids were reported by 28.2%, 23.4%, and 32.8% in the no, unintentional, and intentional groups, respectively (P=0.551). Sedative use was most frequent among the intentional group (60.9%) compared with no (34.1%) or unintentional (38.3%) use (P=0.001). Prescription stimulants showed no significant difference (19.3%, 10.6%, and 15.6%; P=0.380). Methamphetamine use was similar across groups (16.3%, 8.5%, and 21.9%; P=0.169). Cocaine was common overall but highest in the intentional group (85.9% vs. 68.9% and 72.3% for no and unintentional groups; P=0.036).
Table 3.
Self-reported past 30-day substance use among people who use drugs in New York City, stratified by fentanyl use intentionality.
| Fentanyl use b | |||||
|---|---|---|---|---|---|
|
| |||||
| Drug (past 30 days) | Total N=246 | No use (n=135) | Unintentional use (n=47) | Intentional use (n=64) | P value |
| n (%) | n (%) | n (%) | n (%) | ||
|
| |||||
| Cannabis | 179 (72.8) | 109 (80.7) | 27 (57.5) | 43 (67.2) | 0.004 |
|
| |||||
| Heroin (no/unknown fentanyl) | 40 (16.3) | 13 (9.6) | 27 (57.5) | 0 (0) | <0.001 |
|
| |||||
| Prescription opioids a | 70 (28.5) | 38 (28.2) | 11 (23.4) | 21 (32.8) | 0.551 |
|
| |||||
| Sedatives (benzodiazepines, muscle relaxers, sleeping pills) a | 103 (41.9) | 46 (34.1) | 18 (38.3) | 39 (60.9) | 0.001 |
|
| |||||
| Prescription stimulants a | 41 (16.7) | 26 (19.3) | 5 (10.6) | 10 (15.6) | 0.380 |
|
| |||||
| Methamphetamine | 40 (16.3) | 22 (16.3) | 4 (8.5) | 14 (21.9) | 0.169 |
|
| |||||
| Cocaine | 182 (74.0) | 93 (68.9) | 34 (72.3) | 55 (85.9) | 0.036 |
|
| |||||
Use without prescription, more than prescribed, or for the feeling.
Substances are not mutually exclusive, and fentanyl use intentionality reflects participant-level classification.
Discussion
In this convenience sample of PWUD in NYC, more than one-quarter of participants reported intentional fentanyl use, while nearly one-fifth had unintentional use, underscoring the widespread presence of fentanyl in the drug supply and the difficulty PWUD face in detecting its presence. Our findings suggest that intentional fentanyl use is associated with chronic pain and higher HIV risk behavior burden (including injection-related behaviors), while unintentional fentanyl use is more prevalent among older individuals. High cannabis use severity was inversely associated with intentional fentanyl use. These findings have important implications for tailoring harm reduction and treatment strategies in the context of an unpredictable drug supply.
Unintentional fentanyl use was more common among older PWUD, consistent with findings among PWUD in Canada.27 This signals a critical prevention gap: older adults may be less engaged in harm reduction services and therefore less likely to receive education or tools such as fentanyl test strips, underscoring the need for broader distribution of fentanyl test strips and education on fentanyl contamination.10 Fentanyl test strips, while limited to indicating fentanyl presence/absence rather than the amount, may be especially valuable for individuals who do not anticipate fentanyl exposure but remain at risk of contamination. Expanding both access to these tools and messaging to older PWUD mitigate the disproportionate overdose burden observed in this demographic group.28 Although race was not a significant correlate in our models, national trends show disproportionate overdose burden among older Black adults in the context of a rapidly changing drug supply.29, 30 These findings underscore the heightened vulnerability of older generations of PWUD to fentanyl contamination and overdose risk.
PWUD with intentional use may require additional support to address complex patterns of use and associated medical needs.8 This subgroup seems be more informed about the contents of their drug supply, suggesting higher engagement with drug-checking services or peer support. Yet our results point to a more severe risk profile among this group. In bivariate analyses, those with intentional use were more likely to report injection, incarceration history, and recent non-fatal overdose, alongside significantly higher HIV risk behaviors. Multivariable models confirmed that intentional use was associated with higher RAB scores, suggesting that PWUD with intentional use were both more aware and more vulnerable in terms of high-risk behaviors.10, 19
The association of intentional fentanyl use and higher HIV risk suggests that individuals who intentionally use fentanyl may engage in behaviors, such as syringe sharing or unprotected sex, which increase the risk of HIV transmission. In sensitivity analyses replacing the composite RAB score with a binary measure of injection, injection was strongly associated with intentional use, indicating that injection-related risk contributes substantially to this association. However, because the RAB score incorporates both injection-related and sexual risk behaviors, the primary model captures broader HIV risk behavior burden. Several qualitative studies reinforce these findings. For example, fentanyl’s short duration of effect and high overdose risk often compel PWID to inject more frequently and in groups, which increases opportunities for sharing injection equipment.31, 32 Transitions from prescription opioids to heroin or fentanyl have also been linked to increased injection initiation and frequency, further elevating HIV risk.33 Quantitative studies have also reported the association between fentanyl use and HIV risk behaviors in the US and Estonia.19, 34, 35 These findings reinforce the need to leverage existing harm reduction engagement among intentional users to provide more comprehensive, integrated services. Individuals who already connect with SSPs or similar programs could be linked to HIV prevention strategies, such as pre-exposure prophylaxis (PrEP), and behavioral interventions.19, 34, 35 Studies also documented that PWUD report switching from injecting to smoking fentanyl to avoid vein damage or reduce stigma and infectious diseases (e.g., HIV, HCV) risks.36 This pattern may suggest that individuals with greater awareness of drug market risks are adapting their practices, but such adaptations may not fully offset risks related to frequent use, overdose, or comorbid conditions. Integrated care models that acknowledge intentional fentanyl use as both a marker of awareness and of greater severity may better address the overlapping health needs of this population.
The association between chronic pain and intentional fentanyl use suggests that pain may play a role in motivating fentanyl use among PWUD in NYC. This aligns with prior studies documenting high chronic pain prevalence among people with OUD and the use of nonmedical opioids for pain relief in the face of structural stigma due to unmet health needs.16, 18 Systematic reviews highlight opioid-related issues among those with chronic pain, including how undertreated pain can reduce OUD treatment efficacy and increase the risk of illicit opioid use.37, 38 Given its widespread availability and affordability, some individuals may intentionally seek fentanyl not just for its psychoactive effects but as a self-directed strategy to manage their ongoing, untreated pain, especially if it is the more accessible option in structurally marginalized populations.18 Future research should explore how integrating pain management support within substance use treatment could help reduce reliance on fentanyl use.
Furthermore, high cannabis use severity was inversely associated with intentional fentanyl use. Previous research suggests that some PWUD use cannabis to manage opioid withdrawal symptoms, reduce opioid cravings, and increase adherence to OUD treatment.39 Studies in Canada found that cannabis use among individuals on opioid agonist therapy was associated with reduced fentanyl exposure and overdose risk.40 However, the relationship between cannabis use and opioid outcomes remains complex, and future research should explore the mechanisms through which cannabis use might influence patterns of intentional fentanyl use among people with OUD, considering factors such as dose, frequency, context of use, and co-occurring substance use.
Our study has several limitations. The cross-sectional design limits causal inference. Despite using both self-report and toxicology data, some misclassification of fentanyl use may persist due to recall or reporting biases. UTS only detects fentanyl at a snapshot in time and may not reflect longer-term patterns of use. The PAT utilized a follow-up question to clarify if heroin with or without fentanyl was most often used (>50% of the time). Accordingly, self-reported fentanyl use reflects typical use patterns rather than any use; individuals with less frequent known fentanyl use may therefore have been classified as not reporting fentanyl use. The way fentanyl use was presented to participants from a substance use category that also mentioned heroin may have prevented them from endorsing fentanyl use as intentional. As a result, toxicology-confirmed fentanyl among participants reporting other substances (e.g., cocaine or methamphetamine) may have been classified as “unintentional,” potentially overestimating unintentional fentanyl use. Chronic pain and non-fatal overdose were assessed using self-report measures rather than validated multidimensional instruments, which may result in misclassification and limit assessment of pain-related functional impact. Lastly, the sample was recruited from community-based settings in NYC, which may limit generalizability to different geographic contexts.
Conclusion
Unintentional fentanyl use among older PWUD highlights a need for targeted outreach, fentanyl education, and broader test strip distribution. Intentional fentanyl use, while reflecting awareness of the drug supply, is associated with chronic pain, elevated HIV risk, and markers of more severe substance use. This group would benefit from integrated interventions linking harm reduction with pain management and HIV prevention. Recognizing and addressing the distinct needs of both subgroups with unintentional and intentional use is needed as public health systems adapt to an increasingly volatile drug supply.
Supplementary Material
Funding:
The project was supported through grant K01DA053435 from the US National Institute on Drug Abuse (NIDA) and the Center for Drug Use and HIV Research (CDUHR - P30 DA011041).
Footnotes
Conflicts of Interest: None declared.
Preprint Statement: This manuscript has not been posted on a preprint server.
References
- 1.Ciccarone D The rise of illicit fentanyls, stimulants and the fourth wave of the opioid overdose crisis. Current opinion in psychiatry. 2021;34(4):344–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Spencer MR, Garnett M, Miniño AM. Drug overdose deaths in the United States, 2002–2022. US Department of Health and Human Services, Centers for Disease Control and; …; 2024. [Google Scholar]
- 3.Centers for Disease Control Prevention. US overdose deaths decrease almost 27% in 2024. 2025. [December 16, 2025]. Available from: https://www.cdc.gov/nchs/pressroom/releases/20250514.html.
- 4.Mann B New Report: U.S. drug overdose deaths rise again after hopeful decline 2025. [July 31, 2025]. Available from: https://www.npr.org/2025/06/18/nx-s1-5436711/overdose-drugs-cdc.
- 5.Friedman JR, Nguemeni Tiako MJ, Hansen H. Understanding and addressing widening racial inequalities in drug overdose. American journal of psychiatry. 2024;181(5):381–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Tuazon E, Weitz A, Sun T, Gasdaska B, Clarke-Crichton G, Samaroo B, Brown L, Fulmer C, D’Anna N, Opozda K, Sun Y, Seil K, Mantha S. Unintentional Drug Poisoning (Overdose) Deaths in New York City in 2024. New York City Department of Health and Mental Hygiene: Epi Data Brief October 2025. Report No.: 150. [Google Scholar]
- 7.Kelly PJ, Vento SA, Green TC, Rich JD, Noh M, Silcox J, Hughto JM. Self-reported pathways through which illicitly manufactured fentanyl enters the stimulant supply: Novel insights from people incarcerated for drug manufacturing and distribution. SSM-Qualitative Research in Health. 2025:100568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tsang VW, Wong JS, Westenberg JN, Ramadhan NH, Fadakar H, Nikoo M, Li VW, Mathew N, Azar P, Jang KL. Systematic review on intentional non-medical fentanyl use among people who use drugs. Frontiers in Psychiatry. 2024;15:1347678. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Friedman JR, Abramovitz D, Skaathun B, Rangel G, Harvey-Vera A, Vera CF, Artamonova I, Munoz S, Martin NK, Eger W. Illicit Fentanyl Use Independently Predicts HCV Seroconversion Among a Cohort of People Who Inject Drugs in Tijuana and San Diego. medRxiv. 2024:2024.03. 22.24304609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.McKnight C, Weng CA, Reynoso M, Kimball S, Thompson LM, Des Jarlais D. Understanding intentionality of fentanyl use and drug overdose risk: findings from a mixed methods study of people who inject drugs in New York City. International Journal of Drug Policy. 2023;118:104063. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Green TC, Park JN, Gilbert M, McKenzie M, Struth E, Lucas R, Clarke W, Sherman SG. An assessment of the limits of detection, sensitivity and specificity of three devices for public health-based drug checking of fentanyl in street-acquired samples. International Journal of Drug Policy. 2020;77:102661. [DOI] [PubMed] [Google Scholar]
- 12.Frost MC, Austin EJ, Corcorran MA, Briggs ES, Behrends CN, Juarez AM, Frank ND, Healy E, Prohaska SM, LaKosky PA. Responding to a surge in overdose deaths: perspectives from US syringe services programs. Harm Reduction Journal. 2022;19(1):79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Baker O, Wellington C, Price CR, Tracey D, Powell L, Loffredo S, Moscariello S, Meyer JP. Experience delivering an integrated service model to people with criminal justice system involvement and housing insecurity. BMC Public Health. 2023;23(1):222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Downs A, Walter L, Shelton R, Li L. Co-occurring illicit fentanyl use and psychiatric disorders in emergency department patients. Scientific Reports. 2025;15(1):11192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Stretanski MF KN, Matha A, Conermann T. Chronic Pain In: StatPearls StatPearls Publishing; 2025. [July 28, 2025]. Available from: http://www.ncbi.nlm.nih.gov/books/NBK553030/. [PubMed] [Google Scholar]
- 16.Voon P, Karamouzian M, Kerr T. Chronic pain and opioid misuse: a review of reviews. Substance abuse treatment, prevention, and policy. 2017;12(1):36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Becker WC, Fiellin DA. When epidemics collide: coronavirus disease 2019 (COVID-19) and the opioid crisis. American College of Physicians; 2020. p. 59–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kim JJ, Hayati D, Zamany M, Choi F, Jang K, Ignaszewski M, Azar P, Krausz M. Chronic pain among primary fentanyl users: The concept of self-medication. European Journal of Pain. 2025;29(1):e4753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Des Jarlais DC, Weng CA, Feelemyer J, Khezri M, Reynoso M, Kimball S, Uuskula A, McKnight C. Heroin, Fentanyl, Intentional Fentanyl Use and HIV Transmission in a Community-Recruited Sample of Persons Who Inject Drugs, New York City: 2021–2023. AIDS and Behavior. 2025:1–10. [DOI] [PubMed] [Google Scholar]
- 20.McLellan AT, Kushner H, Metzger D, Peters R, Smith I, Grissom G, Pettinati H, Argeriou M. The fifth edition of the Addiction Severity Index. Journal of substance abuse treatment. 1992;9(3):199–213. [DOI] [PubMed] [Google Scholar]
- 21.Bunting A A multi-phased study to develop a novel measure of polysubstance use. Drug and Alcohol Dependence. 2024;260:110257. [Google Scholar]
- 22.Blevins CA, Weathers FW, Davis MT, Witte TK, Domino JL. The posttraumatic stress disorder checklist for DSM-5 (PCL-5): Development and initial psychometric evaluation. Journal of traumatic stress. 2015;28(6):489–98. [DOI] [PubMed] [Google Scholar]
- 23.Hoyle RH, Stephenson MT, Palmgreen P, Lorch EP, Donohew RL. Reliability and validity of a brief measure of sensation seeking. Personality and individual differences. 2002;32(3):401–14. [Google Scholar]
- 24.Humeniuk R, Henry-Edwards S, Ali R, Poznyak V, Monteiro MG. The Alcohol, Smoking and Substance Involvement Screening Test (ASSIST): manual for use in primary care 2010. [Google Scholar]
- 25.Metzger D, Woody G, Navaline H, McLellan K, Meyers T, Boney F, Mulvaney J, Williams S, Dyanick S, Jonson A. The Risk Assessment Battery (RAB): Validity and Reliability. Presented at Sixth Annual Meeting of the National Cooperative Vaccine Development Groups for AIDS. Alexandria, VA. VA: October; 1993. [Google Scholar]
- 26.Hosmer DW Jr, Lemeshow S, Sturdivant RX. Applied logistic regression: John Wiley & Sons; 2013. [Google Scholar]
- 27.Mitra S, Boyd J, Wood E, Grant C, DeBeck K, Kerr T, Hayashi K. Elevated prevalence of self-reported unintentional exposure to fentanyl among women who use drugs in a Canadian setting: a cross-sectional analysis. International Journal of Drug Policy. 2020;83:102864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Goldman JE, Waye KM, Periera KA, Krieger MS, Yedinak JL, Marshall BD. Perspectives on rapid fentanyl test strips as a harm reduction practice among young adults who use drugs: a qualitative study. Harm reduction journal. 2019;16:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Friedman JR, Hansen H. Evaluation of increases in drug overdose mortality rates in the US by race and ethnicity before and during the COVID-19 pandemic. JAMA psychiatry. 2022;79(4):379–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Mason M, Soliman R, Kim HS, Post LA. Disparities by sex and race and ethnicity in death rates due to opioid overdose among adults 55 years or older, 1999 to 2019. JAMA network open. 2022;5(1):e2142982–e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Board A, Alpren C, Hernandez B, Murray A, Dawson EL, Drumhiller K, Burrage A, Jaeger JL, Buchacz K, Klevens RM. A qualitative study of injection and sexual risk behavior among unstably housed people who inject drugs in the context of an HIV outbreak in Northeast Massachusetts, 2018. International Journal of Drug Policy. 2021;95:103368. [DOI] [PubMed] [Google Scholar]
- 32.Cranston K Notes from the field: HIV diagnoses among persons who inject drugs—Northeastern Massachusetts, 2015–2018. MMWR Morbidity and mortality weekly report. 2019;68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Dickson-Gomez J, Krechel S, Spector A, Weeks M, Ohlrich J, Green Montaque HD, Li J. The effects of opioid policy changes on transitions from prescription opioids to heroin, fentanyl and injection drug use: a qualitative analysis. Substance abuse treatment, prevention, and policy. 2022;17(1):55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Park JN, Owczarzak J, Urquhart G, Morris M, Weicker NP, Rouhani S, Sherman SG. HIV risk among urban and suburban people who inject drugs: elevated risk among fentanyl and cocaine injectors in Maryland. AIDS and Behavior. 2022:1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Talu A, Rajaleid K, Abel-Ollo K, Rüütel K, Rahu M, Rhodes T, Platt L, Bobrova N, Uusküla A. HIV infection and risk behaviour of primary fentanyl and amphetamine injectors in Tallinn, Estonia: implications for intervention. International Journal of Drug Policy. 2010;21(1):56–63. [DOI] [PubMed] [Google Scholar]
- 36.Kral AH, Lambdin BH, Browne EN, Wenger LD, Bluthenthal RN, Zibbell JE, Davidson PJ. Transition from injecting opioids to smoking fentanyl in San Francisco, California. Drug and alcohol dependence. 2021;227:109003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Delorme J, Kerckhove N, Authier N, Pereira B, Bertin C, Chenaf C. Systematic review and meta-analysis of the prevalence of chronic pain among patients with opioid use disorder and receiving opioid substitution therapy. The Journal of Pain. 2023;24(2):192–203. [DOI] [PubMed] [Google Scholar]
- 38.Vowles KE, McEntee ML, Julnes PS, Frohe T, Ney JP, Van Der Goes DN. Rates of opioid misuse, abuse, and addiction in chronic pain: a systematic review and data synthesis. Pain. 2015;156(4):569–76. [DOI] [PubMed] [Google Scholar]
- 39.Ganesh SS, Gould EE, Conner BT, Huh J, Ceasar RC, Bluthenthal RN. “Smoking weed it gets you over the hump”: Cannabis co-use as a facilitator of decreased opioid use among people who inject drugs in Los Angeles, California. Drug and Alcohol Dependence Reports. 2024;12:100257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Socías ME, Choi J, Lake S, Wood E, Valleriani J, Hayashi K, Kerr T, Milloy M-J. Cannabis use is associated with reduced risk of exposure to fentanyl among people on opioid agonist therapy during a community-wide overdose crisis. Drug and alcohol dependence. 2021;219:108420. [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.