Associations between substance use treatment and ketamine use: A hypothesis-generating analysis

Highlights

• •SUD treatment is linked to higher ketamine use across substance domains.
• •Elevated odds persist in several illicit subgroups (e.g., PCP, methamphetamine).
• •Associations are heterogeneous across drug classes and involvement levels.
• •Implications: routine ketamine screening and risk-mitigation in SUD care.
• •This study provides exploratory, hypothesis-generating insights for future research.

Abstract

Background

Ketamine is increasingly used in clinical settings for mental health and pain management, yet its misuse poses public health risks. While prior studies have examined ketamine trends, few have explored its use among individuals receiving treatment for substance use disorders (SUD).

Methods

Using 2021–2023 data from the National Survey on Drug Use and Health (NSDUH), we analyzed the association between past-year ketamine use and receipt of SUD treatment among U.S. residents aged 12 and older. Stratified analyses by substance type and use category (use, misuse, and disorder) were conducted using adjusted logistic regression models.

Results

Among 173,808 participants who reported substance use, 3.19 % received past-year treatment and 0.26 % reported past-year ketamine use. Ketamine use was more common among those in treatment (1.39 %) than not (0.22 %). Across SUD strata, treatment was associated with higher odds of ketamine use, including alcohol SUD (aOR = 2.73; 95 % CI: 1.58–4.71), marijuana SUD (2.32; 1.34–4.02), inhalant SUD (5.22; 1.96–13.94), methamphetamine SUD (5.10; 2.08–12.48), pain reliever SUD (2.62; 1.16–5.90), and opioid SUD (2.76; 1.23–6.18). Among misuse strata, associations included pain relievers (2.69; 1.40–5.16), opioids (3.13; 1.71–5.74), and psychotherapeutics (2.09; 1.21–3.62). Among use strata, treatment was associated with higher odds for cigarettes, alcohol, marijuana, heroin, PCP, DMT/AMT/FOXY, methamphetamine, pain relievers, tranquilizers, and stimulants.

Conclusions

Past-year treatment is a marker of elevated ketamine exposure across multiple substance domains. Findings are hypothesis-generating and underscore the need for clinical screening, patient education on unsupervised ketamine risks, and research clarifying timing, intent, and outcomes of ketamine use in SUD populations.

1. Introduction

Ketamine is a nonbarbiturate dissociative anesthetic that has hallucinogenic effects and can induce a state of sedation (Drug Enforcement Administration, 2020, Rosenbaum et al., 2025). This drug is approved by the United States (US) Food and Drug Administration (FDA) to use as an anesthetic (Rosenbaum et al., 2025). In addition to being used medically, ketamine has been illegally manufactured and used recreationally for decades, typically at nightclubs and festivals (National Institute on Drug Abuse [NIDA], 2024). Unsupervised ketamine poses health risks including but not limited to respiratory depression, changes in blood pressure, and gastrointestinal problems (NIDA, 2024). Despite the risk of negative health outcomes from ketamine misuse, controlled use of ketamine via prescription is suggested to have potential benefits for severe pain and mental disorders including substance use disorders (SUD), leading ketamine to be prescribed off-label (Rosenbaum et al., 2025, Walsh et al., 2022, Wilkinson and Sanacora, 2017). In 2019, the FDA approved the (S)- enantiomer of ketamine, esketamine, for treatment resistant depression (Janssen Pharmaceuticals, Inc, 2025, Rosenbaum et al., 2025). While ketamine is not FDA approved to treat SUD, multiple studies have investigated the potential benefits of the drug for SUD (Famuła et al., 2024, Jones et al., 2018, Martinotti et al., 2021, Tran and MacDougall, 2024).

The potential benefits of ketamine for SUD are important to investigate due to the overall burden of SUD. In fact, in 2022, 17.3 % of people aged 12 or older (48.7 million) had an SUD. This includes 29.5 million with alcohol use disorder, 27.2 million with a drug use disorder, and 8 million with both (Substance Abuse and Mental Health Services Administration [SAMHSA], 2023). These disorders can lead to social problems including but not limited to housing instability and unemployment (Daley, 2013). Along with social implications, SUDs pose health risks such as human immunodeficiency virus (HIV) (from intravenous drug use or risky sexual behaviors), stroke, cancer, and mental health conditions (Daley, 2013, National Institute on Drug Abuse, 2020). Moreover, the World Health Organization (WHO) reported that there were 2.6 million deaths caused by alcohol and 600,000 deaths caused by psychoactive drugs worldwide in 2019 (World Health Organization [WHO], 2024). To address the negative outcomes of SUD, treatment is available including medication and behavioral therapy (Centers for Disease Control and Prevention, 2024, Gollinge and Ploussiou, 2025). For instance, FDA approved medications to treat opioid use disorder (OUD) include methadone, buprenorphine, and naltrexone (CDC, 2024). Additionally, acamprosate, disulfiram, and naltrexone are FDA approved to treat alcohol use disorder (AUD) (Gollinge and Ploussiou, 2025, Kim et al., 2018). Treatment plans using these medications and behavioral therapy are individualized to match the needs of the individual (CDC, 2024; Hilton & Pilkonis, 2015). Due to the high prevalence and negative outcomes of SUD, it is important for researchers to continue studying potential treatment options.

Studies have addressed the need to research additional SUD treatment options by investigating the potential of ketamine as an SUD treatment (Famuła et al., 2024, Jones et al., 2018, Martinotti et al., 2021, Tran and MacDougall, 2024). A review on literature looking at ketamine for AUD, OUD, and cocaine use disorder (CoUD) found ketamine influenced abstinence, withdrawal, craving, and consumption (Tran & MacDougall, 2024). Also, a study on 58 opiate-dependent patients found patients in the ketamine treatment group needed less additional medication to manage acute opiate withdrawal at 48 h (Jovaiša et al., 2006). Moreover, Dakwar et al. (2020) found that compared to midazolam, a ketamine infusion significantly increased the likelihood of abstinence, delayed time to relapse, and reduced the likelihood of heavy drinking days in treatment-seeking alcohol-dependent adults who engage in motivational enhancement therapy. An additional study investigating CoUD found ketamine substantially reduced cocaine use, suggesting the potential benefit of ketamine infusions to improve cocaine dependence (Dakwar et al., 2014). These studies illustrate the potential of ketamine, however more research is still needed due to the safety and health risks that ketamine poses (Pai et al., 2022, Peskin et al., 2023).

With increasing use of ketamine treatment for medical indications, Ni et al. (2018) theorized that illicit use might increase. In their retrospective review of reports to US poison centers, Ni et al. (2018) found 49 % of ketamine cases reported involved multiple substances with ketamine. Furthermore, in a scoping review of clinical and pre-clinical studies, Le et al. (2022) raised the potential risk of abuse liability in using ketamine in treatment for depression, finding only four preclinical studies evaluating differential abuse liability. Moreover, two case reports observed that uncontrolled and frequent self-administration of ketamine resulted in addiction and dependence in patients with depression (Bonnet, 2015, Schak et al., 2016). Due to the findings on ketamine that show both risk and benefits, further research is needed to better understand the safety and efficacy of ketamine as treatment for SUDs.

One resource that provides data on ketamine use is the National Survey on Drug Use and Health (NSDUH), an interview-based dataset in the US that provides comprehensive data on substance use and mental health (SAMHSA, 2025). However, limited studies using NSDUH data have specifically looked at ketamine use trends and relationships to other substance use. Wu et al. (2006) looked at NSDUH data from 2002, finding ketamine users were predominantly employed youths and there was a high prevalence of lifetime heroin use among ketamine users. Moreover, Palamar et al. (2021) looked at the trend in quarterly prevalence of ketamine use between 2006 and 2019 among 12–34 year olds, finding a linear increase between 2006–2014 and a cubic increase between 2015–2019. An additional study found that lifetime use of ketamine was associated with increased odds of reporting tryptamine use in 18–25 year olds based on NSDUH data from 2007 to 2014 (Palamar & Le, 2018).

While data from NSDUH provide valuable information on substance use patterns and previous studies have looked at ketamine, there is limited research examining ketamine use specifically among individuals with recent treatment for SUDs. This study aims to fill that gap through an exploratory association analysis using NSDUH data from 2021, 2022, and 2023. Given the dual rise in ketamine misuse and its therapeutic applications, understanding its use among individuals who received SUD treatment in the past year is crucial. This study will focus on recent SUD treatment as the exposure and past-year ketamine use as the outcome, stratified by substance type, to identify ketamine use patterns in relation to substance use treatment across various substances.

While ketamine can be prescribed legally for medical purposes or administered in clinical settings, NSDUH does not distinguish between medical and non-medical ketamine use. Thus, all reported ketamine use in this study includes both possible clinical and recreational contexts. This ambiguity represents an important limitation when interpreting associations. In addition, expert consensus statements and practice guidelines generally recommend heightened caution and additional screening when considering ketamine or esketamine treatment for individuals with current or recent moderate-to-severe SUD; however, these recommendations are advisory rather than enforced requirements. Screening and exclusion practices therefore vary widely across settings, particularly as large-scale telehealth-based ketamine services have emerged. Some in-clinic programs may treat active, unstable SUD as a relative or absolute contraindication, whereas others (including some telehealth models) may apply less stringent or less consistently enforced criteria. Given this heterogeneity and limited regulatory oversight, we cannot determine how often reported ketamine use among NSDUH respondents with SUD reflects medically supervised treatment versus non-medical use, and our findings should be interpreted with this uncertainty in mind.

2. Methods

2.1. Data Source

Data are from NSDUH, which is collected annually by the Substance Abuse and Mental Health Services Administration (SAMHSA) via a cross-sectional survey (SAMHSA, 2025). Participants in the dataset are civilian, noninstitutionalized individuals, 12 years and older in the US (including all 50 states and the District of Columbia). Data are collected through face-to-face household interviews and web-based interviews. This includes households, non-institutional group quarters, and civilians living on military bases. Individuals experiencing homelessness that do not use shelters, active military personnel, and residents of institutional group quarters are excluded from interviews. Years of NSDUH data before 2021 did not use web interviewing, limiting our data access to the years 2021, 2022, and 2023, as previous years of data cannot be compared and pooled with data from 2021 and beyond (SAMHSA, 2025). To be included in our study, participants had to report substance use, yielding a total sample of 173,808 participants.

2.2. Ethical Considerations

This study was reviewed by the Loyola University Chicago Institutional Review Board (LU# 220103) and determined to meet criteria for Exempt research under 45 CFR 46.104(d)(4) (secondary research using de-identified data). Analyses relied exclusively on the publicly available NSDUH public-use files, which contain no direct identifiers, and we did not have access to any linkage keys. Consistent with guidance on ethical secondary analysis of public data, we did not attempt to identify or re-identify individual respondents or to link NSDUH records with external datasets. All data were stored on secure, password-protected institutional systems and results are reported only in aggregate form (e.g., weighted prevalences and adjusted odds ratios) for groups of respondents, thereby minimizing risks of inadvertent disclosure or stigmatizing identification of small or vulnerable subgroups.

2.3. Measures

The primary exposure, past year SUD treatment, was measured with three survey questions. Because NSDUH revised the SUD-treatment items beginning in 2022, we harmonized the exposure across years as follows: participants were categorized as “Receiving treatment” (Yes) if they reported any alcohol/drug treatment at any location in 2021 or reported inpatient/outpatient SUD treatment in 2022–2023; otherwise they were categorized as “Not receiving treatment” (No).The primary outcome, past year ketamine use, was measured by NSDUH’s revised imputation of ketamine frequency (based on the survey prompt “Time since last used ketamine”). However, the survey does not differentiate whether the use occurred under medical supervision or in non-medical (e.g., recreational) contexts. As a result, the ketamine use variable reflects any use, regardless of intent or legality.

Our study was stratified by substance type and use type (use, misuse (i.e., use without a prescription or not as prescribed), and SUD). Data were obtained from binary imputed and recoded variables in the survey and substance use indicators were restricted to past year use. For sedative use disorder, if participants were imputed as “Yes” for “Sedative use disorder, past year users but not misusers” or “Sedative use disorder, past year users but not misusers” then the participant was categorized as having a sedative use disorder. If the participant was categorized as “No” to both, they were categorized as not having a sedative disorder. This process was repeated for pain reliever use disorder, tranquilizer use disorder, and stimulant use disorder. Primary subgroup contrasts were defined using past-year status (use, misuse, SUD) to align temporally with the exposure (past-year treatment) and the outcome (past-year ketamine use), thereby avoiding heterogeneity in risk periods and reducing recall-related misclassification.

2.4. Statistical analysis

NSDUH employs a complex sampling design (stratification, clustering, and sampling weights) to produce nationally representative estimates for US residents aged ≥ 12 years. We accounted for this design using SAS survey procedures (e.g., PROC SURVEYFREQ, PROC SURVEYLOGISTIC). The analytic weight was constructed as weight = ANALWT2_C / 3 to average across the three pooled survey years (2021–2023). Survey design variables were VESTR_C (strata) and VEREP (PSU/cluster). For descriptive statistics stratified by past-year substance use treatment and past-year ketamine use, we report unweighted counts (n) and survey-weighted percentages. Associations between categorical variables and (a) past-year treatment and (b) past-year ketamine use were assessed with the design-adjusted Rao-Scott χ² test. For inferential models, we report adjusted odds ratios (aORs) with 95 % confidence intervals (CIs) from survey-weighted multivariable logistic regression. Variances were estimated using Taylor series linearization, consistent with NSDUH guidance. (See Supplemental Materials for the complete SAS syntax covering formats, recodes, descriptive analyses, and models.).

For the primary outcome (past-year ketamine use) and exposure (past-year alcohol/drug treatment), records with missing values were excluded from that specific analysis. For covariates, we conducted a complete-case (listwise) analysis; no additional imputation was performed beyond the edited/imputed variables supplied by NSDUH in the public-use files. Sample sizes (unweighted n) shown in tables reflect any exclusions due to missingness.

Primary analyses estimated aORs by fitting 39 survey-weighted multivariable logistic regression to estimate the odds of ketamine use for individuals who received substance use treatment in the past year compared to those who did not, across different subpopulations including types of SUDs, substance misuse, and substance use. The prespecified adjustment set comprised survey year, age, sex, race/ethnicity, education, family income, and urbanicity (metro/non-metro). To prevent overfitting, we applied an events-per-variable (EPV) rule (∼10 events per parameter per Vittinghoff & McCulloch (2007)). Models with adequate EPV used the full covariate set; models with limited EPV used a variation of subsets from the full covariate set (see Supplemental Materials for the entire SAS syntax for (i) used format, (ii) construction of variables, (iii) descriptive statistics, and (iv) inferential statistics). Adjusted estimates (aORs with 95 % CIs) are presented in the tables and used for statistical inference. As a robustness check for exposure harmonization, we fit a definition-restricted model aligning the 2021 item with the inpatient/outpatient construct used in 2022–2023.

To address multiplicity in subgroup comparisons, we controlled the false discovery rate (FDR) using the Benjamini-Hochberg procedure within conceptually related families: (1) Substance Use Disorders, (2) Substance Misuse, and (3) Substance Use. FDR q-values were computed from the p-values of the adjusted models and are reported alongside the corresponding p-values in Table 2. We pre-specified these three families and focused interpretation on well-powered, policy-relevant domains whereas rarer contrasts are retained for transparency but are not over-interpreted. To address sparse-data concerns, we flagged contrasts with fewer than 20 ketamine outcome events in the contrasted subgroup and interpret such estimates with caution while noting that all reported models met conventional events-per-variable guidance for a binary exposure (≥∼10 events per parameter). All tests were two-sided with α = 0.05; where exact p-values were not available (e.g., “<0.0001”), a conservative value of 0.0001 was used for FDR calculations.

Table 2.

Odds of past year ketamine use for individuals who have received alcohol or drug treatment by substance use type.

Denominator	Alcohol or Drug Treatment n (%)		Past Year Ketamine Use	Odds Ratios of Past Year Ketamine Use	p-value	FDR-adj p (BH, within family) [q-valueb]
Target population (n)			n (%)	ORa (95 % Confidence Interval)
Substance Use Disorders
Alcohol Use Disorder (18,943)	Yes No	1,784 (9.13) 17,159 (90.87)	252 (1.18) 18,691 (98.82)	2.726 (1.579–4.705) Ref = 1	0.0006	0.00390
Cannabis Use Disorder (14,427)	Yes No	1,636 (11.94) 12,791 (88.06)	248 (1.71) 14,179 (98.29)	2.321 (1.342–4.015) Ref = 1	0.0033	0.01073
Cocaine Use Disorder (827)	Yes No	281 (35.09) 546 (64.91)	84 (8.47) 743 (91.53)	0.884 (0.367–2.125) Ref = 1	0.7783	0.77830
Heroin Use Disorder (455)	Yes No	259 (58.37) 196 (41.63)	22 (3.72) 433 (96.28)	3.351 (0.764–14.696) Ref = 1	0.1066	0.15398
Hallucinogen Use Disorder (463)	Yes No	146 (36.26) 317 (63.74)	62 (15.08) 401 (84.92)	0.586 (0.153–2.248) Ref = 1	0.4279	0.48208
Inhalant Use Disorder (252)	Yes No	61 (18.67) 191 (81.33)	23 (6.46) 229 (93.54)	5.223 (1.956–13.943) Ref = 1	0.0014	0.00607
Methamphetamine Use Disorder (1,012)	Yes No	367 (37.10) 645 (62.90)	45 (3.58) 967 (96.42)	5.098 (2.083–12.478) Ref = 1	0.0006	0.00390
Pain Reliever Use Disorder (2,883)	Yes No	797 (24.00) 2,086 (76.00)	62 (1.11) 2,821 (98.89)	2.619 (1.163–5.897) Ref = 1	0.0210	0.04550
Tranquilizer Use Disorder (1,105)	Yes No	358 (32.27) 747 (67.73)	55 (3.04) 1,050 (96.96)	1.437 (0.558–3.699) Ref = 1	0.4450	0.48208
Stimulant Use Disorder (1,450)	Yes No	339 (25.41) 1,111 (74.59)	59 (3.16) 1,391 (96.84)	2.542 (1.015–6.366) Ref = 1	0.0465	0.07556
Sedative Use Disorder (490)	Yes No	146 (28.03) 344 (71.97)	23 (2.03) 467 (97.97)	2.522 (0.462–13.763) Ref = 1	0.2787	0.36231
Opioid Use Disorder (3,150)	Yes No	918 (25.63) 2,232 (74.37)	69 (1.24) 3,081 (98.76)	2.756 (1.230–6.175) Ref = 1	0.0148	0.03848
Psychotherapeutic Use Disorder (4,828)	Yes No	1,154 (22.54) 3,674 (77.46)	118 (1.47) 4,710 (98.53)	1.962 (1.061–3.630) Ref = 1	0.0324	0.06017
Substance Misuse
Pain Reliever Misuse (5,254)	Yes No	934 (15.39) 4,320 (84.61)	131 (1.96) 5,123 (98.04)	2.686 (1.398–5.160) Ref = 1	0.0038	0.01330
Tranquilizer Misuse (2,810)	Yes No	570 (18.77) 2,240 (81.23)	144 (4.71) 2,666 (95.29)	1.559 (0.762–3.192) Ref = 1	0.2190	0.30660
Stimulant Misuse (3,298)	Yes No	490 (14.79) 2,808 (85.21)	162 (4.53) 3,136 (95.47)	1.798 (0.870–3.715) Ref = 1	0.1110	0.19425
Sedative Misuse (568)	Yes No	123 (19.13) 445 (80.87)	27 (3.02) 541 (96.98)	1.164 (0.284–4.782) Ref = 1	0.8295	0.82950
Opioid Misuse (5,487)	Yes No	1,024 (16.44) 4,463 (83.56)	138 (12.06) 5,349 (97.94)	3.129 (1.707–5.737) Ref = 1	0.0004	0.00280
Psychotherapeutic Misuse (9,452)	Yes No	1,391 (13.71) 8,061 (86.29)	272 (2.47) 9,180 (97.53)	2.091 (1.210–3.616) Ref = 1	0.0093	0.02170
Oxycontin Misuse (634)	Yes No	195 (27.61) 439 (72.39)	42 (4.99) 592 (95.01)	1.416 (0.567–3.538) Ref = 1	0.4490	0.52383
Substance Use
Cigarette Use (28,311)	Yes No	2,714 (8.50) 25,597 (91.50)	317 (0.78) 27,994 (99.22)	4.290 (2.663–6.910) Ref = 1	<0.0001	0.00038
Alcohol Use (102,738)	Yes No	3,685 (3.24) 99,053 (96.76)	506 (0.39) 102,232 (99.61)	5.644 (3.499–9.104) Ref = 1	<0.0001	0.00038
Cannabis Use (41,078)	Yes No	2,908 (7.05) 38,170 (92.95)	480 (1.07) 40,598 (98.93)	2.948 (1.777––4.890) Ref = 1	<0.0001	0.00038
Cocaine Use (3,363)	Yes No	610 (19.04) 2,753 (80.96)	270 (7.42) 3,093 (92.58)	1.536 (0.908–2.600) Ref = 1	0.1076	0.13629
Crack Use (502)	Yes No	216 (40.70) 286 (59.30)	27 (4.10) 475 (95.90)	1.098 (0.203–5.942) Ref = 1	0.9121	0.91210
Heroin Use (548)	Yes No	272 (49.57) 276 (50.43)	32 (6.09) 516 (93.91)	4.412 (1.259–15.465) Ref = 1	0.0213	0.04047
Hallucinogen Use (6,528)	Yes No	738 (11.27) 5,790 (88.73)	566 (8.86) 5,962 (91.14)	1.724 (1.033–2.877) Ref = 1	0.0377	0.06444
LSD Use (1,923)	Yes No	257 (12.87) 1,666 (87.13)	200 (12.48) 1,723 (87.52)	1.512 (0.897–2.548) Ref = 1	0.1184	0.14060
PCP Use (93)	Yes No	31 (37.04) 62 (62.97)	16† (11.30) 77 (88.70)	22.991 (7.977–66.259) Ref = 1	<0.0001	0.00038
Ecstasy Use (1,614)	Yes No	205 (11.75) 1,409 (88.25)	241 (14.78) 1,373 (85.22)	1.168 (0.603–2.261) Ref = 1	0.6393	0.67482
DMT/AMT/FOXY Use (416)	Yes No	89 (24.64) 327 (75.36)	107 (21.83) 309 (78.17)	4.770 (1.588–14.325) Ref = 1	0.0063	0.01496
Salvia Use (98)	Yes No	25 (27.60) 73 (72.40)	25 (24.29) 73 (75.71)	1.442 (0.318–––6.543) Ref = 1	0.6150	0.67482
Inhalant Use (2,167)	Yes No	237 (9.76) 1,930 (90.24)	177 (8.37) 1,990 (91.63)	1.957 (0.937–4.091) Ref = 1	0.0732	0.09934
Methamphetamine Use (1,493)	Yes No	497 (32.56) 996 (67.44)	62 (3.00) 1,431 (97.00)	4.090 (1.948––8.588) Ref = 1	0.0004	0.00109
Pain Reliever Use (38,529)	Yes No	2,821 (6.22) 35,708 (93.78)	253 (0.51) 38,276 (99.49)	3.951 (2.155––7.243) Ref = 1	<0.0001	0.00038
Tranquilizer Use (17,140)	Yes No	1,617 (8.28) 15,523 (91.72)	262 (1.06) 16,878 (98.94)	2.205 (1.158–4.201) Ref = 1	0.0172	0.03631
Stimulant Use (13,847)	Yes No	1,403 (10.09) 12,444 (89.91)	276 (1.55) 13,571 (98.45)	2.529 (1.563–4.091) Ref = 1	0.0003	0.00095
Sedative Use (7,318)	Yes No	747 (7.16) 6,571 (92.84)	99 (0.67) 7,219 (99.33)	2.194 (0.975––4.933) Ref = 1	0.0572	0.08360
Oxycontin Use (4,048)	Yes No	516 (9.73) 3,532 (90.27)	86 (1.40) 3,962 (98.60)	2.104 (1.033––4.284) Ref = 1	0.0407	0.06444

^†Ketamine events < 20 in the contrasted subgroup (unweighted); interpret with caution.

^aBolded values indicated significant ORs.

^bBold indicates FDR-adjusted p (q) < 0.05 (Benjamini-Hochberg within family: SUD, Misuse, Use). Italics indicate 0.05 < q ≤ 0.10 (suggestive/borderline).

3. Results

Among the 173,808 participants, 5,951 (3.19 %) received alcohol/drug treatment in the past year and 566 (0.26 %) used ketamine in the past year (Table 1a; Table 1b). Non-Hispanic (NH) American Indian/Alaskan Native (AI/AN), Native Hawaiian (Native HI), and other NH Pacific Islanders were the race/ethnicity category with the greatest percentage (6.75 %) receiving alcohol/drug treatment, compared to 3.90 % of NH Multiracial, 3.42 % of NH White, 2.97 % of NH Black, 2.76 % of Hispanic, and 1.81 % of Asian participants (Table 1a). Among bisexual participants, 6.77 % received alcohol/drug treatment, compared to 5.18 % of gay/lesbian participants and 2.96 % of heterosexual participants. Compared to higher income categories, those with an income of less than $20,000 had the greatest percentage (5.97 %) receiving alcohol/drug treatment. Additionally, 16.95 % of those who reported past year ketamine use received alcohol/drug treatment in the past year, compared to 3.16 % of those who did not use ketamine.

Table 1a.

The prevalence of past year alcohol/drug treatment, by demographic characteristics, of U.S. residents of the age 12 + who reported substance use (NSDUH: 2021–2023).

	Past Year Alcohol/Drug Treatment
	No nb (%c)	Yes nb (%c)	Total nb (%d)	p-valuea
Total	167,857 (96.81)	5,951 (3.19)	173,808 (100)
Survey Year				<0.0001
2021	57,173 (98.44)	861 (1.56)	58,034 (33.11)
2022	56,496 (95.90)	2,573 (4.10)	59,069 (33.36)
2023	54,188 (96.10)	2,517 (3.90)	56,705 (33.53)
Sex				<0.0001
Male	76,885 (96.49)	2,698 (3.51)	79,583 (48.93)
Female	90,972 (97.11)	3,253 (2.89)	94,225 (51.07)
Race/Ethnicity				<0.0001
NH White	96,564 (96.58)	3,463 (3.42)	100,027 (60.62)
NH Black	19,923(97.03)	698 (2.97)	20,621 (12.22)
NH AI/AN or Native HI/Other Pacific Islander	3,178 (93.25)	243 (6.75)	3,421 (0.98)
NH Asian	8,916 (98.19)	142 (1.81)	9,058 (6.02)
NH Multiracial	7,420 (96.10)	332 (3.90)	7,752 (2.07)
Hispanic	31,856 (97.24)	1,073 (2.76)	32,929 (18.09)
Age				<0.0001
12–17 Years	33,100 (97.11)	1,184 (2.89)	34,284 (9.19)
18–25 Years	40,485 (96.71)	1,388 (3.29)	41,873 (12.10)
26–34 Years	27,267 (96.09)	1,137 (3.91)	28,404 (14.27)
35–49 Years	35,940 (95.93)	1,564 (4.07)	37,504 (22.34)
50 or Older	31,065 (97.48)	678 (2.52)	31,743 (42.10)
Sexual Orientation				<0.0001
Heterosexual	119,221 (97.04)	3,723 (2.96)	122,944 (89.57)
Gay/Lesbian	4,068 (94.82)	245 (5.18)	4,313 (2.56)
Bisexual	11,708 (93.23)	855 (6.77)	12,563 (5.40)
Other	4,629 (96.04)	232 (3.96)	4,861 (2.48)
Education				<0.0001
Less than High School	46,179 (95.98)	1,975 (4.02)	48,154 (18.21)
High School/GED	34,070 (95.96)	1,577 (4.04)	35,647 (24.69)
Some College/associates	39,706 (96.49)	1,598 (3.51)	41,304 (27.37)
College Graduate or Higher	47,902 (98.31)	801 (1.69)	48,703 (29.73)
Income				<0.0001
< $20,000	27,323 (94.04)	1,704 (5.97)	29,027 (15.11)
$20,000-$49,999	44,807 (96.14)	1,900 (3.86)	46,707 (26.86)
$50,000-$74,999	24,626 (97.25)	785 (2.75)	25,411 (15.18)
> $75,000	71,101 (98.05)	1,562 (1.95)	72,663 (42.86)
Employment				<0.0001
Full Time	67,635 (97.43)	1,925 (2.57)	69,560 (44.04)
Part Time	24,575 (96.47)	887 (3.53)	25,462 (12.99)
Unemployed	9,476 (94.14)	646 (5.86)	10,122 (4.93)
Other/Not in Labor Force	49,608 (96.46)	1,991 (3.55)	51,599 (38.04)
Urban/Rural				<0.0001
Large Metro	76,304 (97.15)	2,385 (2.85)	78,689 (54.80)
Small Metro	64,754 (96.52)	2,447 (3.48)	67,201 (31.30)
Nonmetro	26,799 (96.10)	1,119 (3.90)	27,918 (13.90)
Marriage				<0.0001
Married	56,538 (98.24)	1,077 (1.76)	57,615 (46.05)
Not Married	94,756 (95.52)	4,372 (4.48)	99,128 (53.95)
Past Year Ketamine Use				<0.0001
Yes	460 (83.05)	106 (16.95)	566 (0.26)
No	167,397 (96.84)	5,845 (3.16)	173,242 (99.74)

^aChisq p-values are from design-adjusted (Rao-Scott) χ² tests using NSDUH weights, strata, and PSUs.

^bFrequency (unweighted counts).

^cWeighted Row percent.

^dWeighted Column percent.

Table 1b.

The prevalence of past year ketamine use, by demographic characteristics, of U.S. residents of the age 12 + who reported substance use (NSDUH: 2021–2023).

	Past Year Ketamine Use
	No nb (%c)	Yes nb (%c)	Total nb (%d)	p-valuea
Total	173,242 (99.74)	566 (0.26)	173,808 (100)
Survey Year
2021	57,880 (99.81)	154 (0.19)	58,034 (33.11)	0.0087
2022	58,888 (99.74)	181 (0.26)	59,069 (33.36)
2023	56,474 (99.66)	231 (0.34)	56,705 (33.53)
Sex				0.0003
Male	79,268 (99.67)	315 (0.33)	79,583 (48.93)
Female	93,974 (99.80)	251 (0.20)	94,225 (51.07)
Race/Ethnicity				<0.0001
NH White	99,621 (99.68)	406 (0.32)	100,027 (60.62)
NH Black	20,598 (99.91)	23 (0.09)	20,621 (12.22)
NH AI/AN or Native HI/Other Pacific Islander	3,414 (99.95)	7 (0.05)	3,421 (0.98)
NH Asian	9,039 (99.82)	19 (0.18)	9,058 (6.02)
NH Multiracial	7,719 (99.56)	33 (0.44)	7,752 (2.07)
Hispanic	32,851 (99.80)	78 (0.20)	32,929 (18.09)
Age				<0.0001
12–17 Years	34,252 (99.94)	32 (0.06)	34,284 (9.19)
18–25 Years	41,652 (99.48)	221 (0.52)	41,873 (12.10)
26–34 Years	28,228 (99.26)	176 (0.74)	28,404 (14.27)
35–49 Years	37,391 (99.72)	113 (0.28)	37,504 (22.34)
50 or Older	31,719 (99.94)	24 (0.06)	31,743 (42.10)
Sexual Orientation				<0.0001
Heterosexual	122,646 (99.80)	298 (0.20)	122,944 (89.57)
Gay/Lesbian	4,267 (99.26)	46 (0.74)	4,313 (2.56)
Bisexual	12,428 (98.89)	135 (1.11)	12,563 (5.40)
Other	4,829 (99.18)	32 (0.82)	4,861 (2.48)
Education				<0.0001
Less than High School	48,092 (99.91)	62 (0.09)	48,154 (18.21)
High School/GED	35,540 (99.80)	107 (0.20)	35,647 (24.69)
Some College/associates	41,117 (99.66)	187 (0.34)	41,304 (27.37)
College Graduate or Higher	48,493 (99.66)	210 (0.34)	48,703 (29.73)
Income				0.1000
< $20,000	28,902 (99.68)	125 (0.32)	29,027 (15.11)
$20,000-$49,999	46,546 (99.68)	161 (0.32)	46,707 (26.86)
$50,000-$74,999	25,334 (99.78)	77 (0.22)	25,411 (15.18)
> $75,000	72,460 (99.78)	203 (0.22)	72,663 (42.86)
Employment				0.0004
Full Time	69,276 (99.68)	284 (0.32)	69,560 (44.04)
Part Time	25,349 (99.59)	113 (0.41)	25,462 (12.99)
Unemployed	10,079 (99.67)	43 (0.33)	10,122 (4.93)
Other/Not in Labor Force	51,483 (99.83)	116 (0.17)	51,599 (38.04)
Urban/Rural				<0.0001
Large Metro	78,402 (99.67)	287 (0.33)	78,689 (54.80)
Small Metro	66,976 (99.79)	225 (0.21)	67,201 (31.30)
Nonmetro	27,864 (99.88)	54 (0.12)	27,918 (13.90)
Marriage				<0.0001
Married	57,537 (99.92)	78 (0.08)	57,615 (46.05)
Not Married	98,650 (99.56)	478 (0.44)	99,128 (53.95)
Substance Use Treatment				<0.0001
Yes	5,845 (98.61)	106 (1.39)	5,951 (3.19)
No	167,397 (99.78)	460 (0.22)	167,857 (96.81)

^aChisq p-values are from design-adjusted (Rao-Scott) χ² tests using NSDUH weights, strata, and PSUs.

^bFrequency (unweighted counts).

^cWeighted Row percent.

^dWeighted Column percent.

Furthermore, 1.11 % of bisexuals reported ketamine use in the past year compared to 0.74 % of gay/lesbian participants and 0.20 % of heterosexual participants (Table 1b). NH Multiracial participants reported the greatest percentage of past year ketamine use (0.44 %) compared to all other race/ethnicity categories. Participants aged 26–34 years had the highest prevalence of ketamine use (0.74 %) compared to 0.52 % of 18–25 year olds, 0.28 % of 35–49 year olds, 0.06 % of 12–17 year olds, and 0.06 % of 50 + year olds. Among participants who received substance use treatment, 1.39 % reported past year ketamine use compared to 0.22 % of those who did not receive substance use treatment in the past year.

Among participants with specific substance use disorders such as alcohol (aOR 2.73; 95 % CI 1.58–4.71; q-value = 0.00390), cannabis (2.32; 95 % CI 1.34–4.02; q-value = 0.01073), inhalant (aOR 5.22; 95 % CI 1.96–13.94; q-value = 0.00607), methamphetamine (aOR 5.10; 95 % CI 2.08–12.48; q-value = 0.00390), pain reliever (aOR 2.62; 95 % CI 1.16–5.90; q-value = 0.04550), and opioid (aOR 2.76; 95 % CI 1.23–6.18; q-value = 0.03848) use disorder, significantly higher odds of ketamine use were found for those who received alcohol/drug treatment compared to those who did not (Table 2). Similar associations, yet on the boundary of statistical significance, were found among participants with stimulant (aOR 2.54; 95 % CI 1.02–6.37; q-value = 0.07556) and psychotherapeutic use disorder (aOR 1.96; 95 % CI 1.06–3.63; q-value = 0.06017).

Moreover, significantly higher odds of ketamine use for those who received alcohol/drug treatment, compared to those who did not, were found among people who misuse pain relievers (aOR 2.69; 95 % CI 1.40–5.16; q-value = 0.01330), opioids (aOR 3.13; 95 % 1.71–5.74; q-value = 0.00280), and psychotherapeutics (aOR 2.09; 95 % CI 1.21–3.62; q-value = 0.02170).

Additionally, significantly higher odds of ketamine use were found for those who received alcohol/drug treatment compared to those who did not in subpopulations that used cigarettes, alcohol, cannabis, heroin, PCP, DMT/AMT/FOXY, methamphetamine, pain relievers, tranquilizers, and stimulants in the past year. Specifically, those who received alcohol/drug treatment in the past year had 4.29 times the odds (95 % CI 2.66–6.91; q-value = 0.00038) among people who use cigarettes, 5.64 times the odds (95 % CI 3.50–9.10; q-value = 0.00038) among people who use alcohol, 2.95 times the odds (95 % CI 1.78–4.89; q-value = 0.00038) among people who use cannabis, 4.41 times the odds (95 % CI 1.26–15.47; q-value = 0.04047) among people who use heroin, 22.99 times the odds (95 % CI 7.98–66.26; q-value = 0.00038) among people who use PCP, 4.77 times the odds (95 % CI 1.59–14.33; q-value = 0.01496) among people who use DMT/AMT/FOXY, 4.09 times the odds (95 % CI 1.95–8.59; q-value = 0.00109) among people who use methamphetamine, 3.95 times the odds (95 % CI 2.16–7.24; q-value = 0.00038) among people who use pain relievers, 2.21 times the odds (95 % CI 1.16–4.20; q-value = 0.03631) among people who use tranquilizer, and 2.53 times the odds (95 % CI 1.56–4.09; q-value = 0.00095) among people who use stimulants, of ketamine use compared to those who did not receive treatment. Furthermore, similar associations, yet on the boundary of statistical significance, were found among people who use hallucinogen, inhalants, sedatives, and oxycontin.

4. Discussion

In this national sample of people reporting any substance use, past-year alcohol/drug treatment was consistently associated with higher odds of past-year ketamine use across numerous substance categories (SUD, misuse, and use), with several associations in the two- to six-fold range and some markedly larger (e.g., PCP users) after adjusting for multiple comparisons. This pattern suggests that engagement with treatment often co-occurs with broader polysubstance involvement and/or experimentation, rather than immediate abstinence from non-prescribed drugs. Clinically, individuals severe enough to seek treatment may also have greater exposure to emerging or club-drug markets, or may be exploring ketamine as self-medication for craving, withdrawal, or comorbid mood symptoms. Indeed, recent epidemiological analyses confirm that use of ketamine is strongly associated with use of other club drugs (e.g., MDMA, GHB) and hallucinogens in the community, underlining the tendency for polysubstance use among those in treatment (Yang et al., 2025).

These results should be interpreted alongside ketamine’s dual identity as both a potential therapy and a drug of misuse. Randomized controlled studies indicate that carefully dosed, supervised ketamine (typically combined with psychotherapy) can reduce craving and improve abstinence for some substance use disorders. For example, trials in alcohol and cocaine use disorders have shown reduced withdrawal symptoms and increased abstinence rates with ketamine treatment (Janssen-Aguilar et al., 2025). This highlights ketamine’s therapeutic promise under structured clinical protocols. Yet ketamine also carries clear misuse liability and health risks with unsupervised or repeated use. Chronic high-dose ketamine use has been linked to cognitive impairments and neuropsychiatric complications (Strous et al., 2022), as well as severe urological toxicity such as ulcerative cystitis and even kidney damage (Baetens et al., 2024). Our cross-sectional survey cannot determine the sequence of events as some individuals may have used ketamine before entering treatment, while others may try it during or after treatment. However, overall, ketamine’s role appears to be ambivalent, offering both potential benefits and notable risks.

The 2021–2023 findings indicate heterogeneous associations rather than a pattern confined to only “softer” or prescription drug categories. After controlling for FDR, several illicit subgroups (notably those reporting use of PCP, heroin, or methamphetamine) still showed elevated odds of ketamine use, whereas other contrasts were null. This underscores that the effect of treatment status on ketamine use varies by SUD vs. misuse vs. use domain and by drug class. For instance, we did not detect significant differences in ketamine use by treatment status among individuals with cocaine, heroin, hallucinogen, tranquilizer, or sedative use disorders; among those who misused tranquilizers, stimulants, sedatives, or OxyContin; or among users of cocaine, crack, LSD, ecstasy, or salvia. In contrast, elevated odds of ketamine use were concentrated in other subgroups, including individuals with alcohol, cannabis, inhalant, methamphetamine, pain reliever, or opioid use disorders; those who misused pain relievers, opioids, or psychotherapeutics; and those who used cigarettes, alcohol, cannabis, heroin, PCP, DMT/AMT/Foxy, methamphetamine, pain relievers, tranquilizers, or stimulants. A few of these significant signals rely on very small cell sizes (e.g., fewer than 20 ketamine users among people who use PCP, and sparse counts for inhalant use disorder), so these specific estimates warrant caution. Notably, the pharmacologic overlap between PCP and ketamine, both dissociative anesthetics acting on NMDA receptors, may contribute to their co-occurrence in polysubstance patterns (Journey & Bentley, 2023). This provides context for why individuals who use PCP, for example, had especially high odds of ketamine use in our data.

In addition, our results underscore that treatment uptake and ketamine use are not uniform across demographic groups. We observed higher treatment prevalence (and corresponding ketamine use) among certain subpopulations, notably bisexual individuals, AI/AN individuals, lower-income groups, and those who were unemployed. These disparities highlight the importance of tailoring policies and resources to address the unique vulnerabilities of these communities. For instance, sexual minority adults have significantly higher rates of substance use and SUD than their heterosexual peers (SAMHSA, 2023) and may face distinctive stressors that influence both treatment-seeking and substance use patterns. AI/AN populations similarly experience disproportionately high rates of SUD and often encounter barriers to care, necessitating culturally informed interventions. Ensuring equitable access to treatment for these groups while strengthening safeguards against ketamine misuse is critical. This aligns with a growing recognition of the need for comprehensive, substance-specific risk assessments in SUD treatment frameworks. By incorporating screening for emerging drugs like ketamine and addressing the social determinants of health that affect treatment engagement, clinicians can better support recovery and mitigate potential harms. Ultimately, our findings call for a balanced approach that maximizes ketamine’s therapeutic benefits in treating SUDs while vigilantly monitoring and managing its risks in real-world practice.

These findings may inform clinical guidelines and treatment planning in SUD programs. The elevated odds of ketamine use among individuals receiving treatment, especially those using licit or prescription substances, suggest that ketamine may be more accessible or perceived as safer in these contexts. As off-label ketamine use continues to grow in psychiatric practice, treatment providers and policymakers would benefit from clearer protocols to distinguish therapeutic use from potential misuse in SUD patients. For example, integrating routine screening for ketamine use in SUD treatment settings could improve safety monitoring, and educating clinicians on ketamine’s risks and benefits may support informed decision-making when treating patients who might be exposed to ketamine. Moreover, regulatory bodies may consider developing best-practice recommendations or warning systems for concurrent use of ketamine with other psychoactive substances, especially where misuse or dependence risk is heightened. Expert consensus statements and clinic-based protocols for ketamine and esketamine often recommend heightened caution or additional monitoring for patients with active SUD, and some programs treat unstable SUD as a relative or absolute contraindication (American Psychiatric Nurses Association, 2023). However, these recommendations are advisory rather than enforced requirements, and real-world screening and exclusion practices vary widely across in-clinic and telehealth-based ketamine services, where there is limited centralized oversight to ensure adherence to guidelines. As a result, we cannot determine from NSDUH data how often reported ketamine use among respondents with SUD reflects medically supervised treatment versus non-medical use. A subset of treated respondents in our sample could conceivably have received ketamine for SUD treatment itself, given emerging evidence of efficacy in that realm (Janssen-Aguilar et al., 2025), but the overall prevalence of such cases is unknown and likely heterogeneous across settings. Indeed, the NSDUH questionnaire item does not distinguish medical from nonmedical ketamine use; its wording even references street names (“Special K,” “Super K”), and prior work has assumed that most self-reported ketamine use in surveys is nonmedical (Palamar et al., 2021). Consistent with this context, our data showed no significant difference by treatment status in ketamine use for certain drug-involved groups (as noted above), whereas other subgroups, including those involving opioids, stimulants, and hallucinogens like PCP, exhibited clear elevations. This pattern reinforces that differences by treatment status may arise from factors beyond any therapeutic exposure to ketamine per se. Possible contributors include greater clinical severity or comorbidity among people in treatment, differential recall or disclosure (e.g., due to heightened surveillance or assessment in treatment), local drug availability or program culture factors, and temporal shifts in drug trends from 2021 to 2023.

To clarify mechanisms and inform practice, next steps include (i) longitudinal EHR/claims studies linking ketamine prescribing (including indication and dosing) to relapse, overdose, and service use; (ii) prospective, program-based cohorts that pair surveys with toxicology to classify clinical vs. nonmedical ketamine use during and after treatment; (iii) mixed-methods work to understand motives, decision-making, and settings of use (e.g., self-medication for withdrawal or mood vs. opportunistic/social use); and (iv) pragmatic trials of adjunctive ketamine within SUD care, under careful monitoring, with predefined co-use and safety endpoints to assess benefits and risks in real-world treatment contexts.

4.1. Limitations

Our study is not without limitations. Most notably, the NSDUH survey does not differentiate whether ketamine use was medically supervised, prescribed, or recreational. This limits our ability to interpret whether observed associations reflect clinical use, misuse, or non-medical contexts, and thus constrains our capacity to understand the nature of ketamine involvement in SUD treatment across substance use types. Furthermore, stratifying by substance use type led to small sample sizes, resulting in wide confidence intervals for some subpopulations. Additionally, because NSDUH is based on self-reported data, responses may be subject to recall bias and social desirability bias particularly given the sensitive nature of substance use. Our study is also cross-sectional, which prevents us from inferring the directionality of relationships since data were collected at a single time point. Moreover, NSDUH excludes people experiencing unsheltered homelessness and those currently incarcerated (institutional group quarters), populations with elevated SUD prevalence; thus, our estimates may understate SUD burden and may not generalize to these groups (Fovet et al., 2022, Johnson and Chamberlain, 2008). Although we adjusted for year and key sociodemographic factors, residual confounding from unmeasured variables (e.g., psychiatric comorbidities, treatment modality, or severity of substance involvement) may still influence the observed associations.

5. Conclusion

Past-year SUD treatment was associated with substantially higher odds of past-year ketamine use across multiple substance domains in US survey data. This likely reflects greater severity and polysubstance exposure among treated individuals, and possibly interest in ketamine’s perceived therapeutic effects outside clinical oversight. The field should pair vigilant screening and education with rigorous trials that clarify where ketamine may help, under supervision, without fueling new harms. While ketamine shows promise as a treatment for various addictions under controlled conditions, its presence in the recovery community outside of clinical protocols raises important concerns. These findings should stimulate further research to inform both clinical practice and policy. Prospective and qualitative studies are needed to clarify why and how people in SUD treatment are using ketamine.

CRediT authorship contribution statement

Fares Qeadan: Writing – review & editing, Validation, Supervision, Project administration, Methodology, Investigation, Conceptualization. Shanti O’Neil: Writing – original draft, Investigation, Formal analysis, Data curation.

Ethics approval and consent to participate

This study was reviewed by the Loyola University Chicago Institutional Review Board (LU# 220103) and determined to meet criteria for exempt research as secondary analysis of publicly available, fully de-identified NSDUH data. Because the datasets contain no direct identifiers and individuals cannot be personally identified, additional informed consent to participate was not required.

Consent for publication

Not applicable.

Funding

None.

Declaration of Competing Interest

The authors declare that they have no competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

The data that support the findings of this study are openly available at https://www.samhsa.gov/data/data-we-collect/nsduh-national-survey-drug-use-and-health