Social, Psychological, and Substance Use Characteristics of U.S. Adults Who Use Kratom: Initial Findings From an Online, Crowdsourced Study

Abstract

Kratom, a plant that produces opioid-like effects, has gained popularity in the U.S. for self-treating symptoms of chronic pain, mood disorders, and substance-use disorders (SUDs). Most data on kratom are from surveys into which current kratom-using adults could self-select; such surveys may underrepresent people who have used kratom and chosen to stop. Available data also do not adequately assess important psychosocial factors surrounding kratom use. In this study, U.S. adults who reported past 6-month alcohol, opioid, and/or stimulant use (N = 1,670) were recruited via Amazon Mechanical Turk between September and December 2020. Of the 1,510 evaluable respondents, 202 (13.4%) reported lifetime kratom use. Kratom-using adults, relative to others, were typically younger, male, unpartnered, without children, and had lower income. They had higher rates of chronic pain (31.7% vs. 21.9%, p = .003), childhood adversity, anxiety, and depression (p < .001), and lower perceived social rank (d = .19, .02–.22) and socioeconomic status (d = .37 .16–.26). They also reported higher use rates for most substances (except alcohol); this included medically supervised and unsupervised use of prescription opioids and diverted opioid agonist therapy (OAT) medications. Most (83.2%) met diagnostic criteria for any past-year SUD. Those reporting kratom use were less likely to reside in an urban/suburban area. The strongest predictors of kratom use were use of other drugs: cannabidiol (OR = 3.73), psychedelics (OR = 3.39), and nonmedical prescription opioids (OR = 1.72). Another strong predictor was lifetime OAT utilization (OR = 2.31). Despite seemingly poorer psychosocial functioning and health among respondents reporting lifetime kratom use, use of other substances may be the strongest indicators of kratom use.

What Is Kratom?

Mitragyna speciosa (“kratom”) is a botanical indigenous to Southeast Asia. Kratom leaves contain over 40 known alkaloids that interact to produce wide-ranging, dose-dependent stimulatory, analgesic, and anxiolytic effects (Kruegel et al., 2019; Kruegel & Grundmann, 2018; Todd et al., 2020; Vicknasingam et al., 2020). Two of these alkaloids, mitragynine (MG) and 7-hydroxymitragynine (7-HMG), act as partial, putatively “biased” agonists at mu opioid receptors (MORs) and comprise part of the complex matrix of alkaloids responsible for kratom’s analgesia, with 7-HMG showing higher affinity than MG at MORs (Kruegel et al., 2019; Obeng et al., 2021; Todd, et al., 2020). The alkaloids’ relative seeming functional selectivity for the MOR G-protein signaling pathway, rather than the β-arrestin pathway, suggests kratom may have less risk of adverse effects (e.g., respiratory depression, constipation) compared to traditional opioids (Basiliere & Kerrigan, 2020; Behnood-Rod et al., 2020; Henningfield et al., 2018, 2019; Singh, Müller, et al., 2018; Todd, et al., 2020). However, kratom’s pharmacology suggests that additional, non-opioid mechanisms of action are responsible for many effects, underscoring the complexity of this plant and its derivatives (Fowble & Musah, 2019; Hiranita et al., 2019; Kamble et al., 2021).

Who Is Using Kratom?

Human kratom research consists mainly of self-report surveys, case studies, and observational investigations from Southeast Asia and the U.S. (Agapoff & Kilaru, 2019; Coe et al., 2019; Garcia-Romeu et al., 2020; Grundmann, 2017; Overbeek et al., 2019; Phillip, 2019; Saingam et al., 2013; Schimmel & Dart, 2020; Singh, et al., 2015; Singh, Müller, et al., 2018; Singh, Murugaiyah, et al., 2018; Vicknasingam et al., 2020). Kratom has a long-established history in Southeast Asia, though it is difficult to ascertain the prevalence of kratom use in the U.S. Anecdotal reports suggest kratom was introduced to the U.S. during the Vietnam War (Legislative Analysis and Public Policy Association, 2019), and evidence suggests its use began to increase around 2015 amidst the opioid crisis (Boyer et al., 2007, 2008; Grundmann, 2017). While data from imports, sales, and advocacy groups estimate 5–15 million persons in the U.S. engage in regular kratom use (American Kratom Association, 2019), epidemiological surveys from, 2018 to 2019, including the National Survey on Drug Use and Health, estimate past-month rates at approximately 0.3% and past-year use at 0.8% (approximately 2.6 million). The former may be an underestimate because of the lack of sampling of possibly high-kratòm use groups, such as persons who are homeless or incarcerated (Substance Abuse and Mental Health Services Administration, 2020). More recent estimates suggest that 6.1% (approximately 20 million) U.S. adults use kratom (Covvey et al., 2020).

Many report trying kratom for a variety of purposes. Although this includes social and recreational use, widely endorsed motivations involve self-treatment of symptoms associated with chronic pain, psychological disorders, and substance-use disorders (SUDs), including iatrogenic opioid dependence (Bath et al., 2020; Coe et al., 2019; Grundmann, 2017; Smith & Lawson, 2017; Smith, Rogers, Schriefer, et al., 2021; Swogger et al., 2015). For some, this reflects using kratom as a means for mitigating symptoms of withdrawal from opioids, as a short- or long-acting substitute for prescription and illicit opioids, as a self-treatment for opioid-use disorder (OUD), and possibly as a self-treatment for alcohol-use disorder (AUD) or stimulant-use disorders (Assanangkornchai et al., 2007; Boyer et al., 2008; Coe et al., 2019; Garcia-Romeu et al., 2020; Gutridge et al., 2020; Saref et al., 2019; Singh et al., 2021; Smid et al., 2018; Smith & Lawson, 2017; Smith, Rogers, Strickland, et al., 2021; Swogger et al., 2015; Swogger & Walsh, 2018; Tanguay, 2011; Vicknasingam et al., 2010).

In U.S. surveys, people have routinely reported multiple motivations for use (Bath et al., 2020; Coe et al., 2019; Grundmann et al., 2017; Smith, Rogers, Schriefer, et al., 2021; Swogger & Walsh, 2018). Important within-group heterogeneity is reflected by the fact that some, but not all, people who use kratom also report prior or contemporaneous use of illicit psychoactive substances (Bath et al., 2020; Covvey et al., 2020; Garcia-Romeu et al., 2020; Smith & Lawson, 2017; Smith, Rogers, Schriefer, et al., 2021). While recent kratom surveys have investigated patterns of use among regular kratom users, this is also a limitation, because it underrepresents people who have tried kratom and discontinued use. Yet, in one small study among adults in SUD treatment, 20.8% reported lifetime kratom use, but only 10.2% reported past-year use (Smith & Lawson, 2017). This study, along with examination of social media data (Grundmann et al., 2021; Smith, Rogers, Schriefer, et al., 2021; Smith, Rogers, Strickland, et al., 2021), shows that not all people who try kratom continue using. However, data from these sources did or could not capture motivations for use (e.g., self-treating anxiety) with validated assessments for conditions likely to be found among people who have used kratom.

Thus, while the preponderance of kratom survey data likely reflects experiences of many kratom-using adults, they are not reflective of all users, nor do they reflect experiences of people who used kratom intermittently or quit. Most also lack a comparison group (of non-users) that may be similar in some important respects (e.g., chronic pain, psychological disorders, substance use) but who have not tried kratom. The use of a comparison group, particularly in a survey where the sampling strategy was not focused on kratom, can address self-selection bias and provide points of cross-drug comparison.

Aims

Present study aims were: (a) to refine conceptualizations of U.S. adults with kratom-use histories by obtaining data from people who did not self-select into a kratom-specific survey, but rather who participated in a larger online survey about substance use and social conditions which sampled for inclusion based on licit and illicit substance use, meaning that people reporting kratom use here may be more representative of the U.S. kratom-using population.

Respondents with lifetime kratom use histories would not have self-selected to the survey based on regular use of or favorable attitudes toward kratom. They were also not recruited by kratom advocacy groups (a useful but potentially biasing source of enrollees), but instead from a popular crowdsourcing platform.

(b) To assess whether survey respondents who reported lifetime kratom use were distinguished from persons with no kratom use by: demographic characteristics; social, psychological, physical health indicators, and substance use history. These questions have received minimal research attention to date.

(c) To use validated psychometric instruments, rather than single-item symptom indicators, as assessments for conditions commonly cited as reasons for kratom “self-treatment,” thereby increasing the reliability and validity of assessment.

Doing this, we hoped to lay the groundwork for targeted follow-up studies.

Method

Below, we describe how we determined our sample size, handled data, and study measures. As this was an exploratory investigation without hypotheses, this study and the analyses presented here examining characteristics of kratom-using adults were not preregistered.

Recruiting Platform

Amazon Mechanical Turk (mTurk), an online crowdsourcing platform, was used for study recruitment, screening, and compensation. Use of crowdsourcing for obtaining convenience samples in research has grown rapidly with mTurk, unique in its ability to reach a diverse group and rigorously screen for eligibility prior to enrollment (Chandler & Shapiro, 2016; Miller et al., 2017; Mortensen & Hughes, 2018; Peer et al., 2014; Shank, 2016; Sheehan, 2018; Strickland & Stoops, 2018, 2020). On mTurk, crowdsourced survey respondents are referred to as “workers.”

Inclusion Criteria

Study eligibility required respondents to be ≥18 years or older, U.S. residents, English language proficient at an approximately 8th grade level or greater, and have ≥100 completed mTurk human intelligence tasks (HITs), indicating that they were experienced mTurk workers and would potentially provide more reliable data (Peer et al., 2014). Prospective respondents also had to endorse past 6-month substance use (at least 1 day of use during the 6 months prior to screening) for one of the following: (a) alcohol use only (nicotine and caffeine permitted); (b) licit opioid use (prescription opioid analgesics, prescribed methadone, and/or prescribed buprenorphine); (c) illicit opioid use (heroin, fentanyl, nonmedical/diverted prescription opioids, nonmedical/diverted methadone, and/or buprenorphine); (d) kratom use (which we considered an opioid); and (e) illicit stimulant use (powder or crack cocaine, synthetic cathinones, “street” methamphetamine, 3,4-methylenedioxy-methamphetamine, nonmedical/diverted amphetamine medications). For Groups b through e, additional drug use during the 6 months was acceptable.

Data Collection and Validity

Between September 3 and December 16, 2020, a total of 10,169 m Turk workers were screened. Of these, 4,027 were eligible and passed inclusion and validity checks. The full survey was hosted on Qualtrics. Participants were compensated $0.08 for completing the 8-item screener and $7.25 for completing the survey. To ensure data validity, four quality “attention checks” were programmed into the screener and 26 into the survey. Failing ≥3 checks, or exceeding the 4-hr completion window, resulted in automatic study unenrollment. Participants were tracked by mTurk worker ID, rather than a completed HIT, so as to prohibit “ballot stuffing.” IP addresses were analyzed via IPHub, a service that evaluates IP addresses, in order to detect proxy or VPN addresses. As no personally identifiable information was collected (except IP addresses, which were used to check for “ballot stuffing,” then deleted), this study was given exempt status by the NIH Institutional Review Board.

Measure

Demographic Characteristics

These were collected using a locally developed questionnaire to examine the following variables: age, sex/gender, race/ethnicity, past-year employment status, education status, incarceration history, past-year annual income, relationship and parental status, and past-year U.S. region of residence. Past-year urban/suburban (vs. rural) residence was also measured by converting zip codes into counties and categorizing residence using the U.S. Department of Agriculture’s, (2013) Rural-Urban Continuum, where participants who resided in or adjacent to urban/metropolitan areas of ≥250,000 people were coded “urban/suburban” (vs. rural; U.S. Department of Agriculture, 2013).

Social, Psychological, and Health Indicators

Unless otherwise noted, primary outcomes for the below items was total score or subscale scores.

Social Comparison Scale (SCS; Allan & Gilbert, 1995): Eleven items, each rated 1–10 scale, with lower scores representing lower self-perceived social status and greater inferiority (range 11–110).

Perceived Socioeconomic Status (SES; Adler et al., 2000; Singh-Manoux et al., 2003): A three-item questionnaire that asked respondents to think of themselves on a 10-rung ladder wherein 10 is the most well off and 1 is the worst off in U.S. society, and to indicate their ladder rung (range 1–10) during their childhood, their current life, and their expected future. Primary outcome was ladder rung for each time period.

Adverse Childhood Experiences Questionnaire (ACE; Dong et al., 2004): A 10-item measure of social dysfunction, abuse, neglect, and other stressors during the first 18 years of life. Results are rated as “yes” or “no” and summed, with higher values representing greater adversity (range 0–10).

Perceived Stress Scale (PSS; Cohen et al., 1983): A 14-item measure of self-rated stress levels and coping ability over the past month rated on a 4-point Likert scale, with higher values representing greater stress (range 0–56).

Center for Epidemiologic Studies Short Depression Scale (CES-D-R-10; Björgvinsson et al., 2013; Miller et al., 2008; Radloff, 1977), a shortened version of the 20-itemCES-D, consisting of 10 items measuring past-week depression symptom severity across domains corresponding to DSM-5 criteria. Items are measured using 4-point Likert scales (range 0–30), with higher values representing greater depressive symptomatology.

Generalized Anxiety Disorder Scale (GAD-7; Spitzer et al., 2006): A 7-item assessment of past-month GAD symptoms based on DSM-IV diagnostic criteria. Ratings are made on 4-point Likert scales, with higher values representing greater severity (range 0–21).

Brief Pain Inventory Short Form (BPI; Cleeland & Ryan, 1994): A 9-item measure of chronic pain. Primary outcome was endorsement of experiencing pain today and for the past ≥3 months (yes/no).

World Health Organization Quality of Life (QOL)–BREF (WHO-QOL-BREF; WHO, 2004; WHOQOL Group, 1998): A 26-item measure that yields ratings of physical health, psychological health, social relationships, environment, and total QOL score. Items are rated on 5-point scales, with higher values representing greater QOL total score (range 26–120).

Substance-Use Indicators

Lifetime Substance Experience:

Participants were asked whether they had used any of the following substances at least once in their lifetime: kratom, nicotine, alcohol, cannabidiol, medications taken as prescribed (e.g., opioid analgesics, benzodiazepines, medicinal cannabis), medications taken not as prescribed (e.g., nonmedical/diverted opioid analgesics, benzodiazepines, psychostimulants, antidepressants), and illicit drugs (e.g., cannabis, cocaine, crack cocaine, heroin). Respondents also indicated whether they had experienced a drug overdose (excluding alcohol poisoning); been diagnosed or told by a medical professional they had an SUD/AUD; received treatment for an SUD/AUD; or considered themselves to be in recovery. Respondents next indicated frequency of their polydrug use during the past 30 days (Never = 0 to Every time = 5) coded “Any past-month polydrug use” versus “Never.”

SUDs:

Respondents were asked to complete the DSM-5 (American Psychiatric Association, 2013) checklist for the substance they identified as the biggest problem for them in the past year; respondents indicating they had no problems were asked to complete the DSM-5 checklist for the substance they used the most frequently in the past year. Values were summed into a total score and categorized: mild (2–3 items), moderate (4–5 items), or severe (≥6 items).

Perceived Stigma of Addiction Scale (PSAS; Luoma et al., 2007, 2013): An 8-item assessment of the degree to which respondents believe that a person with a history of substance use would be stigmatized, rated on a four-point Likert-scale. The PSAS was adapted here, with permission, to include history of treatment for an SUD. Higher scores indicate greater perceived stigma. Primary outcome was total score (range 8–32).

Analytic Plan

Analyses focused on between-group comparisons of respondents who did and did not report lifetime kratom use. Means and proportions were generated for descriptive purposes. Between-group differences were examined using 2 × 2 Pearson chi-squares and independent-sample t-tests. Effect sizes for comparisons of means were expressed as Cohen’s d. Variables statistically significant at p < .05 in bivariate analyses were used to build a logistic-regression model examining their relationship with lifetime kratom use, excluding items examined only for descriptive purposes. Regression analysis focused on variables of interest based on prior kratom research. Collinearity was assessed via variance inflation factor (VIF); an approximate VIF < 1.5 or 1/VIF > 0.85 was considered acceptable (Hair et al., 1995). Collinearity was found for lifetime smoked tobacco use and e-cigarette use; we chose to retain the latter. Collinearity was also found for past-month WHO-QOL total and past-month perceived stress, which is unsurprising given the additional measures considered. We chose to retain more precise measures, anxiety, and depression specifically, as kratom has been used for the purposes of reducing mood symptoms. Analyses were conducted using IBM SPSS version 26.

Raw study data and a copy of the full survey instrument may be obtained by contacting the corresponding author (Kirsten E. Smith) contingent on authorization from the Principal Investigator (David H. Epstein).

Results

Sample:

Of the 2,354 survey HITs assigned, 1,670 were completed. Of these, 1,510 (90.6%) were included in the final sample. The 160 excluded cases were removed for one of the following reasons: duplicate IP addresses; IP addresses outside the U.S. or of indeterminate location; discrepant screener and full-survey responses for drug-use inclusion criteria; and unrealistically short survey completion times.

Sample Demographic Characteristics:

Table 1 displays means and proportions for the entire sample. On average, the age of the sample was 36.4 years old, female (50.9%), White (74.1%), employed full-time (57.7%) or part-time (20.7%), and held a college degree (65.6%). Approximately 12% had ever been incarcerated and one-third reported an annual household income in the range $55,000–$64,999. Less than half reported being in a partnered relationship or having children. Modal regions of residence were in the South (27.0%) or Mid-Atlantic, Northeast, or New England (21.4%); 79.0% resided in or adjacent to an urban/metro location.

Table 1.

Sample Demographic Characteristics and Between-Group Differences Among Participants Endorsing Lifetime Kratom Use and Those Reporting No Use (N = 1,510)

Survey Variable	Total N = 1,510	Lifetime kratom use N = 202 (13.4%)	No kratom use N = 1,308 (86.6%)	Chi-square statistic or Cohen’s d	P
Parent study group (%)
Past 6-month alcohol-only use	51.0% (n = 770)	3.0%	58.4%	χ2 = 213.0	.001
Past 6-month opioid/stimulant use	49.0% (n = 740)	97.0%	41.6%
Age (M, SD)	36.4 (10.7)	33.6 (8.4)	36.8 (10.9)	d = 0.30	.001
Malea (%)	40.1%	52.8%	38.5%	χ2 = 13.8	.001
Race/ethnicityb (%)
White	74.1%	72.7%	74.2%		N/A
Asian	7.0%	7.9%	6.9%
African American	6.6%	6.5%	6.6%
Hispanic	4.4%	2.9%	4.6%
Biracial	3.7%	5.8%	3.4%
Indian	2.5%	2.2%	2.5%
Native American	1.2%	0.7%	1.3%
Middle Eastern	0.6%	1.4%	0.5%
Past-year employment status (%)
Full-time	57.7%	52.5%	58.6%		.181
Part-time/student	20.7%	23.3%	20.3%
Unemployed, looking for work	9.7%	13.4%	9.1%
Unemployed, not looking for work	7.2%	5.9%	7.3%
Disabled	2.1%	2.5%	2.0%
Retired	2.0%	1.0%	2.1%
“Hustling,” selling drugs, incarcerated	0.7%	1.5%	0.6%
Employed (%), past year	78.4%	75.7%	78.8%	χ2 = 0.81	.369
Education status (%)
No high school degree (HSD)/GED	1.1%	0.5%	1.2%	χ2 = 48.4	.363
HSD/GED only	9.6%	15.9%	8.6%
Some college	23.6%	35.3%	21.8%
Associates degree/vocational certificate	12.9%	15.4%	12.5%
Bachelor’s degree	36.6%	27.9%	37.9%
Master’s degree/PhD	16.3%	5.0%	18.0%
College degree (%)	65.6%	48.3%	68.4%	χ2 = 30.6	.001
Has ever been incarcerated (%)	12.1%	24.8%	10.2%	χ2 = 33.6	.001
Past-year annual income (%)
$0–$4,999	10.1%	10.4%	10.1%	χ2 = 28.2	.002
$5,000–$9,999	5.4%	7.4%	5.1%
$10,000–$14,999	6.3%	11.4%	5.5%
$15,000–$19,999	5.4%	5.9%	5.3%
$20,000–$24,999	6.6%	8.4%	6.3%
$25,000–$34,999	11.7%	12.9%	11.5%
$35,000–$44,999	10.7%	12.9%	10.3%
$45,000–$54,999	11.3%	11.4%	11.3%
$55,000–$64,999	7.2%	5.9%	7.3%
$65,000–$74,999	6.2%	4.0%	6.6%
$75,000–$99,999	9.5%	4.5%	10.3%
≥$100,000	9.7%	5.0%	10.4%
Annual income above $55,000–$64,999 (%)	32.6%	19.3%	34.6%	χ2 = 18.0	.001
(approximate U.S. median cut-off)
Currently in partnered relationship (%)	37.0%	23.8%	39.0%	χ2 = 16.8	.001
Has children (%)	49.1%	42.1%	50.2%	χ2 = 4.3	.037
Past-year region of primary residence (%)					.093
West Pacific	12.7%	11.9%	12.9%	χ2 = 10.9
West Mountain	7.0%	9.4%	6.7%
Midwest West North Central	6.2%	8.4%	5.9%
Midwest East North Central	14.1%	11.9%	14.4%
West South Central	11.5%	12.9%	11.3%
South (East South and South Atlantic)	27.0%	30.7%	26.4%
Mid-Atlantic, Northeast, New England	21.4%	14.9%	22.5%
Past-year urban/suburban residence (vs. rural)	79.0%	72.8%	80.0%	χ2 = 5.09	.024

Note. N/A Due to small cell counts, tests of significance could not be performed.

^aThe variable for gender included 1,498 total cases that were analyzed; 12 cases for nonbinary were excluded due to the fact that gender identity was not established.

^bDue to a data collection error, the variable for race/ethnicity included 1,203 total cases that were analyzed.

Sample Social, Psychological, and Health Indicators:

Table 2 displays means, standard deviations, and proportions for the entire sample. The mean current perceived SES was 5.2/10, though most respondents believed that their SES would increase in the future (to a mean of 6.5). The mean ACE score was 2.7/10. Mean scores on the CES-D-R-10 and GAD-7 were 11.6/30 and 7.8/21, respectively, indicating moderate symptoms of recent depression and anxiety. Approximately 23.0% of the sample reported chronic pain. Total WHOQOL scores indicated overall moderate QOL, with physical and psychological health rated slightly higher than the other QOL subdomains.

Table 2.

Social, Psychological, and Physical Health Indicators for Entire Sample and Between Participants Endorsing Lifetime Kratom Use and Those Reporting No Use (N = 1,510)

Survey Variable	Total N = 1,510	Lifetime kratom use N = 202 (13.4%)	No kratom use N = 1,308 (86.6%)	Chi-square statistic or Cohen’s d	p
Social comparison, SCS Total Score (11–110)	62.3 (17.5)	59.4 (15.9)	62.8 (17.7)	d = 0.19	.015
Perceived SES total scores (1–10)
Where you started out as child	5.0 (2.0)	4.7 (2.1)	5.0 (2.0)	d = 0.16	.031
Where you stand now	5.2 (1.8)	4.5 (1.8)	5.2 (1.8)	d = 0.37	.001
Where you think you might be in future	6.5 (1.9)	6.2 (1.9)	6.5 (1.9)	d = 0.15	.043
Childhood adversity, ACE total score (0–10)	2.7 (2.6)	3.5 (2.7)	2.6 (2.6)	d = −0.34	.001
Past month perceived stress, PSS total score (0–56)	25.8 (10.1)	28.7 (9.7)	25.4 (10.0)	d = −0.33	.001
Past-month depression, CES-D-R-10 (0–30)	11.6 (6.9)	13.9 (6.7)	11.3 (6.9)	d = −0.40	.001
Past-month anxiety, GAD-7 total score (0–21)	7.8 (5.8)	9.7 (5.6)	7.5 (5.7)	d = −0.39	.001
Chronic pain (%), BPI	23.2%	31.7%	21.9%	χ2 = 8.8	.003
Past-month QOL (WHOQOL)
Total (26–120)	83.8 (17.2)	76.7 (16.7)	84.7 (17.0)	d = 0.47	.001
Physical health (7–35)	22.2 (4.9)	24.3 (5.3)	26.3 (5.3)	d = 0.37	.001
Psychological health (6–30)	18.0 (4.9)	17.5 (5.0)	19.4 (5.4)	d = 0.36	.001
Social relationships (3–15)	6.6 (3.2)	9.0 (3.2)	10.0 (3.0)	d = 0.32	.001
Environment (8–40)	24.7 (6.5)	25.9 (6.1)	29.1 (6.0)	d = 0.52	.001

Note. M (SD). SCS = Social Comparison Scale; SES = socioeconomic status; ACE = Adverse Childhood Experiences Questionnaire; PSS = Perceived Stress Scale; CES-D-R = Center for Epidemiologic Studies Short Depression Scale; GAD-7 = Generalized Anxiety Disorder Scale-7 item; BPI = Brief Pain Inventory.

Sample Substance-Use Indicators:

Table 3 displays means and proportions for the entire sample. Nearly all participants reported lifetime use of alcohol or nicotine. The next most commonly used substances were diverted and/or prescribed benzodiazepines (49.5%), medically prescribed opioids (43.8%), amphetamines (40.9%), non-medical/diverted psychiatric medication (38.9%), cannabidiol (36.1%), nonmedical/diverted prescription opioids (36.0%), and cocaine (31.4%). Less than one-third reported lifetime use of psychedelics, heroin, synthetic drugs, or nonmedical/diverted methadone and/or buprenorphine. Past-month polydrug use was reported by 21.9%. Nearly 60.0% of the sample met past-year SUD criteria for at least one substance. Only 5.0% reported ever having received opioid agonist treatment (OAT).

Table 3.

Lifetime Substance Use for Entire Sample and Between Participants Endorsing Lifetime Kratom Use and Those Reporting No Use

Survey Variable	Total N = 1,510	Lifetime kratom use N = 202 (13.4%)	No Kratom use N = 1,308 (86.6%)	Chi-square statistic or Cohen’s d	P
Lifetime substance use (%)
Alcohol	99.5%	99.0%	99.5%	χ2 = 0.20	N/A
Smoked tobacco	78.7%	94.1%	76.4%	χ2 = 31.6	<.001
Cannabis	72.1%	96.5%	68.3%	χ2 = 68.0	<.001
Electronic cigarettes	52.4%	86.1%	47.2%	χ2 = 105.0	<.001
Nonmedical/diverted benzodiazepines	49.5%	82.7%	44.4%	χ2 = 100.9	<.001
Medically prescribed opioids	43.8%	68.3%	40.0%	χ2 = 55.9	<.001
Amphetamine, any type	40.9%	80.2%	34.9%	χ2 = 146.9	<.001
Nonmedical/diverted psychiatric medication	38.9%	78.7%	32.8%	χ2 = 153.2	<.001
Cannabidiol/CBD	36.1%	78.7%	29.5%	χ2 = 181.6	<.001
Nonmedical/diverted prescription opioids	36.0%	75.7%	29.9%	χ2 = 157.6	<.001
Cocaine, any type	31.4%	63.9%	26.4%	χ2 = 112.4	<.001
Psychedelics	29.0%	72.8%	22.2%	χ2 = 214.5	<.001
Synthetic cannabinoids	15.2%	42.1%	11.1%	χ2 = 127.8	<.001
Nonmedical/diverted methadone/buprenorphine	10.8%	32.7%	7.4%	χ2 = 113.3	<.001
Medicinal cannabis	9.3%	18.8%	7.8%	χ2 = 23.9	<.001
Heroin	8.5%	27.7%	5.6%	χ2 = 107.0	<.001
Synthetic cathinones	2.8%	8.4%	2.0%	χ2 = 23.9	<.001
Ever experienced drug overdose (%)	8.2%	23.3%	5.9%	χ2 = 67.8	<.001
Suspected SUD, lifetime (%)	40.3%	72.3%	35.4%	χ2 = 97.4	<.001
Diagnosed SUD, lifetime (%)	13.2%	28.2%	10.9%	χ2 = 44.0	<.001
Treatment for any AUD/SUD, lifetime (%)	14.0%	33.2%	11.1%	χ2 = 68.9	<.001
Lifetime OAT utilization (methadone, buprenorphine) (%)	5.0%	16.8%	3.1%	χ2 = 66.7	<.001
Past-year any SUD (any substance) (%)	59.5%	83.2	55.9%	χ2 = 52.9	<.001
Past-year SUD severitya (%)
None	40.5%	16.8%	44.1%	χ2 = 78.3	<.001
Mild	17.4%	13.9%	18.0%
Moderate	13.0%	18.8%	12.2%
Severe	29.1%	50.5%	25.8%
Past-month polydrug use (%)	21.9%	30.7%	20.6%	χ2 = 9.90	.002
Currently in recovery (%)	14.4%	23.3%	13.0%	χ2 = 14.2	<.001
PSAS total score (8–32) (M, SD)	23.7 (4.0)	23.7 (4.2)	23.7 (3.9)	d = −.015	.847

Note. N/A Due to small cell counts, tests of significance could not be performed. OAT = opioid agonist treatment; SUD = substance use disorder; PSAS = Perceived Stigma of Addiction Scale.

^aSUD determined for substance identified as largest problem or used at highest frequency if no problem substance was identified.

Kratom Use

Lifetime kratom use was reported by 202/1510 respondents (13.4%). Past-year kratom use was reported by 123 (8.1%) of the sample, all of whom had used within the past 6-month period based on screener responses.

Kratom Versus Non-Kratom-Using Respondents

Demographic Characteristics:

Table 1 displays differences for demographic characteristics between respondents reporting lifetime kratom use and those reporting no use. Those who reported lifetime kratom use were slightly younger (33.6 vs. 36.8; d = 0.30, p < .001) and more often male (52.8% vs. 38.5%; χ² = 13.8, p < .001). Race/ethnicity did not reliably differ between groups, though, with all categories considered, cell sizes were too small to permit a confident conclusion about the absence of such differences. When race/ethnicity was dichotomized as “White” (vs. “Non-white”), groups did not differ (p = .966). Fewer in the kratom use versus nonuse group reported being in a partnered relationship (23.8% vs. 39.0%; χ² = 16.8, p < .001) or having children (42.1% vs. 50.2%; χ² = 4.3, p = .037). No between-group differences were found for being employed at least part-time (75.7% vs. 78.8%; χ² = 0.81, p = .369). When examined categorically, no group differences were found for education status, but when we dichotomized (by college degree), the proportion of respondents with a college degree was lower in the kratom-use group (48.3% vs. 68.4%; χ² = 30.6, p < .001).

Annual income differed between groups, including when income was dichotomized to reflect earnings above the approximate U.S. median income of $55,000–$64,999: the kratom-use group having a lower proportion of people with incomes above $55,000–$64,999 (19.3% vs. 34.6%; χ² = 18.0, p < .001). Groups differed by residence, with a slightly lower proportion of users residing in urban/ suburban (vs. rural) areas compared to people who reported no kratom use (72.8% vs. 80.0%, χ² = 5.09, p = .024).

Social, Psychological, and Health Indicators:

Table 2 shows between-group differences for these indicators. Participants who reported lifetime kratom use scored slightly lower than those who did not use kratom on the SCS (59.4/110 vs. 62.8/110, d = .019, p = .015), meaning that, on average, they ranked their social status lower than did those who had never used kratom. A larger difference in the same direction was found for perceived SES, across multiple time points, especially current perceived SES (4.5/10 vs. 5.2/10, d = 0.37, p < .001). ACE scores were also higher for respondents who reported lifetime kratom use (3.5/10 vs. 2.6/10, d = −0.34, p < .001).

For psychological and physical health factors, PSS scores were higher for the kratom-use group (28.7/56 vs. 25.4/56, d = −0.33, p < .001) as were CES-D-R-10 (13.9/20 vs. 11.3/20, d = −0.40, p < .001) and GAD-7 scores (9.7/21 vs. 7.5/21, d = 0.39, p < .001), indicating moderate-severe symptomatology for these domains. Over one-third of the kratom-use group, compared to 21.9% of the nonuse group, reported chronic pain (χ² = 8.8, p = .003). Total WHOQOL scores were lower for the kratom-use group (76.7/120 vs. 84.7/120, d = 0.47, p < .001), as were all WHOQOL subdomains.

Substance Use Indicators:

Rates of lifetime use for all substances (presented in Table 3) were higher in the kratom-use group, with the exception of alcohol. Differences were particularly large for electronic cigarettes (86.1% vs. 47.2%; χ² = 105.0, p < .001), medicinal cannabis (18.8% vs. 7.8%; χ² = 23.9, p < .001), cannabidiol (78.7% vs. 29.5%; χ² = 181.6, p < .001), nonmedical/diverted prescription opioids (75.7% vs. 29.9%; χ² = 157.6, p < .001), medically prescribed opioids (68.3% vs. 40.0%; χ² = 55.9, p < .001), nonmedical/diverted methadone and/or buprenorphine (32.7% vs. 7.4%; χ² = 113.3, p < .001), heroin (27.7% vs. 5.6%; χ² = 107.0, p < .001), cocaine (63.9% vs. 26.4%; χ² = 112.4, p < .001), amphetamines (80.2% vs. 34.9%; χ² = 146.9, p < .001), benzodiazepines (82.7% vs. 44.4%; χ² = 100.9, p < .001), nonmedical/diverted psychiatric medication (78.7% vs. 32.8%; χ² = 153.2, p < .001), and psychedelics (72.8% vs. 22.2%; χ² = 214.5, p < .001).

Multivariate Analysis of Associations with Kratom Use

Table 4 displays results from the regression model. The strongest statistical predictors of lifetime kratom use were having ever used cannabidiol (OR = 3.73, p < .001) or psychedelics (OR = 3.39, p < .001). Other strong predictors included nonmedical/diverted use of prescription opioids (OR = 1.72, p = .035) (though not use of heroin [OR = 1.55, p = .139]) and lifetime OAT utilization (OR = 2.31, p = .018). Lifetime nonmedical/diverted use of methadone and/or buprenorphine also tended (OR = 1.69, p = .067) to be associated with lifetime kratom use. For demographic predictors, lifetime kratom use was higher for males (OR = 1.67, p = .039) and slightly lower for people who resided in a urban/suburban areas (OR = .063, p = .029).

Table 4.

Results From Binary Logistic Regression Model Examining Associations for Lifetime Kratom Use (N = 1,507^a)

Survey Variable	SE	OR	95% confidence interval	p
Age	.012	0.98	[0.96–1.01]	.130
Male	.200	1.67	[1.02–2.24]	.039
College degree	.204	0.91	[0.61–1.36]	.651
Ever incarcerated	.262	0.66	[0.39–1.10]	.110
Income above U.S. median ($55,000)	.240	0.87	[0.54–1.39]	.565
In relationship	.236	1.01	[0.69–1.74]	.692
Has children	.221	0.75	[0.48–1.15]	.185
Past-year urban proximity	.217	0.63	[0.46–0.95]	.029
Social comparison, SCS	.006	1.00	[0.99–1.02]	.227
Childhood adversity, ACE	.038	0.94	[0.89–1.03]	.325
Past-month depression, CES-D-R	.026	1.04	[0.98–1.09]	.181
Past-month anxiety, GAD-7	.029	0.99	[0.93–1.04]	.639
Chronic pain	.226	1.50	[0.96–2.33]	.073
E-cigarette/vape	.274	1.35	[0.79–2.31]	.270
Cannabis	.453	1.98	[0.81–4.77]	.134
Medicinal cannabis	.269	0.58	[0.35–0.99]	.045
Cannabidiol/CBD	.224	3.73	[2.40–5.72]	<.001
Synthetic cannabinoids	.226	1.21	[0.78–1.88]	.400
Nonmedical/diverted prescription opioids	.257	1.72	[1.04–2.84]	.035
Medically prescribed opioids	.220	0.94	[0.61–1.44]	.787
Nonmedical/diverted methadone/buprenorphine	.284	1.69	[0.96–2.94]	.067
Heroin	.297	1.55	[0.87–2.76]	.139
Cocaine, any type	.248	0.82	[0.50–1.32]	.409
Amphetamine, any type	.299	0.88	[0.49–1.58]	.666
Synthetic cathinones	.415	1.12	[0.54–2.70]	.647
Nonmedical/diverted benzodiazepines	.339	1.24	[0.64–2.41]	.525
Nonmedical/diverted psychiatric medication	.338	1.20	[0.64–2.41]	.518
Psychedelics	.238	3.39	[1.92–4.94]	<.001
Ever drug overdose	.279	1.56	[0.87–2.60]	.140
Lifetime medical OAT	.348	2.31	[1.15–4.51]	.018
Past year SUD	.246	1.18	[0.73–1.90]	.426
Past month polydrug use	.217	0.68	[0.44–1.02]	.063
In recovery	.261	0.93	[0.56–1.56]	.805

Note. SCS = Social Comparison Scale ACE = Adverse Childhood Experiences Questionnaire; PSS = Perceived Stress Scale; CES-D-R = Center for Epidemiologic Studies Short Depression Scale; GAD-7 = Generalized Anxiety Disorder Scale-7 item; BPI = Brief Pain Inventory. χ² = 369.389 (df = 33) Cox-Snell/ML = 0.218; Cragg-Uhler/Nagelkerke = 0.400; Homer & Lemeshow, p = .917.

^aDue to missingness for gender, three cases were excluded from analyses.

Discussion

This investigation helped refine conceptualizations of U.S. adults with kratom-use histories, in part by determining the prevalence and associated features of lifetime kratom use among a sample of U.S. adults who were recruited for a larger study examining social adversity and substance use, not a kratom-specific study (Aim 1). Our study built upon the results of prior U.S.-based kratom surveys that specifically enrolled people with current or regular kratom use and thereby may have overrepresented people who had biases about kratom. Our enrollment procedures helped ensure that the sample was more diverse and not skewed toward people with strong attitudes about kratom. Here, we assessed differences between people who reported even one lifetime use of kratom and those who did not, in terms of: demographic characteristics; indicators of social, psychological, and physical health; and substance use history (Aim 2). We were also able to explore data gathered using standardized instruments, which have not been used in most prior kratom surveys (Aim 3).

A Step Toward Refining Conceptualizations of the American “Kratom User”

Demographic Characteristics

Because we compared people with kratom-use histories to those without, we were able to detect several subtle demographic differences, some of which support prior characterizations of U.S. kratom-using adults. The lifetime-kratom-use group was, on average, over the age of 30 (skewing slightly younger than in some prior surveys) with some college education. However, kratom-using adults in this sample were less likely to have a college degree, be partnered, or have children, and had overall lower incomes compared to the nonuse group; these findings diverge from those of other population-level characterizations (Coe et al., 2019; Garcia-Romeu et al., 2020; Grundmann, 2017), particularly for education and income, which was sometimes higher than the general population. While our sample was majority white (74.1%) kratom use did not vary by race/ethnicity.

Similar to one finding from Garcia-Romeu et al. (2020), people in this study who had tried kratom were slightly more likely to be located in the Southern U.S. Ours was the first study to examine urban/suburban proximity among people reporting kratom use. Our finding that kratom use was associated with a slightly decreased likelihood of residence in an urban/suburban (vs. rural) area warrants focused attention. It may be that persistent barriers to SUD treatment in rural regions hard hit by the opioid epidemic (Andrilla et al., 2017; Jones et al., 2018; Keyes et al., 2014; Luu et al., 2018; Prunuske et al., 2014) have led some rural residents to try kratom as a form of self-treatment, or that the characteristics of rural drug markets, compared to urban ones, make kratom an appealing form of withdrawal mitigation when preferred drugs cannot be readily obtained due to cost, accessibility (e.g., driving long distances), or supply disruptions (e.g., increased opioid prescribing guidelines and monitoring), which have changed in rural drug markets over the past decade (Cicero et al., 2007; Habecker et al., 2018; Havens et al., 2007; Lebin et al., 2019; Monnat et al., 2019; Moody et al., 2017; Patrick et al., 2016, 2019; Sexton et al., 2008; Surratt et al., 2014). Heroin, despite its proliferation, still cannot be obtained as cheaply and readily in most rural compared to urban areas; whereas kratom can be purchased online and at a variety of in-person retail shops. Ultimately, that both groups had a majority of participants residing in an urban/suburban area and that participants were distributed fairly evenly across U.S. regions across makes geographical use patterns still unclear and of interest for future study.

Social, Psychological, and Health Indicators

Among the most interesting takeaways from our survey is that social, psychological, and health indicators were consistently poorer in the kratom-use group than the non-use group. Although effect sizes were small-moderate, the finding was consistent across measures. This finding argues against a narrative in which people who use kratom are primarily members of a socioeconomically elite psychonaut subculture (Rolando and Beccaria, 2019), though there may be some overlap between the two. Instead, our lifetime-kratomuse group reported more use of highly stigmatized illicit substances that are themselves associated with poor health outcomes and highly comorbid with conditions such as chronic pain or mood disorders (Boscarino et al., 2011; Edlund et al., 2010; Grant et al., 2004; Martins et al., 2012; Swendsen & Merikangas, 2000). However, it was interesting that PSAS scores did not differ significantly between groups. Although it is possible that kratom use contributes to these or other social problems in some instances, there is no clear evidence for it. Given that these social adversity and health problems include greater reported childhood adversity and lower perceived childhood SES, it is likely that these indicators, many of which arise from social determinants, predated kratom use (Hatzenbuehler et al., 2013; Link & Phelan, 2006; McEwen, 2012; McEwen & Gianaros, 2010; Mersky et al., 2013; ROOM, 2005; Wilkinson & Marmot, 2003). Higher rates of lifetime incarceration (spending ≥1 night in jail) among respondents who used kratom may also be explained by the higher rates of illicit polydrug use among the kratom-use group and the fact that such use is often associated with criminal justice system involvement (Smith et al., 2017; Smith & Lawson, 2017). It is also unsurprising that psychiatric health, QOL, and chronic pain were worse among people who reported lifetime kratom use, given that these are established motivations for initiating kratom use, rather than resulting from it (Bath et al., 2020; Coe et al., 2019; Garcia-Romeu et al., 2020; Grundmann, 2017; Singh et al., 2015; Swogger & Walsh, 2018). What is surprising is that these were ultimately not strong predictors of lifetime kratom use. Rather, other substance use and lifetime OAT utilization were.

Kratom in the Context of Other Drug Use

In our heterogeneous sample, 13.4% reported lifetime kratom use, and a subset (8.1%) reported past-year use, all of which occurred during the past 6 months. This is the first U.S.-based online study to assess rates of kratom use in such a sample. Despite not prospectively selecting for lifetime kratom use, we found that rates of lifetime kratom use were higher than those of nonmedical/diverted buprenorphine and/or methadone, medicinal cannabis, heroin, and synthetic cathinones. Nonetheless, other “novel” substances (e.g., cannabidiol, synthetic cannabinoids) had been tried by a greater proportion of respondents than kratom. Accordingly, lifetime kratom use was not endemic, though it was reported by a non-trivial number of respondents.

In some ways, findings here are similar to those from (Smith & Lawson, 2017) in which respondents who endorsed lifetime kratom use had higher use rates for nearly all substances compared to respondents who had never tried kratom. We also found greater diversity in the types of drugs used by those who had tried kratom, which is in keeping with some surveys and analyses of social media data (Garcia-Romeu et al., 2020; Smith, Rogers, Schriefer, et al., 2021; Smith, Rogers, Strickland, et al., 2021). This may be unsurprising, given that the willingness or propensity to try one relatively novel substance (e.g., cannabidiol, electronic cigarettes, synthetics) may carry over to others (Smith et al., 2019; Smith & Staton, 2019; Smith & Stoops, 2019). It is unclear to what extent kratom-using respondents could be classified as “psychonauts” or as part of the subculture of people seeking cognitive- and mood-enhancing polydrug regimens (Corazza et al., 2014; Deluca et al., 2012; Napoletano et al., 2020; O’Brien et al., 2015; Smith, Rogers, Strickland, et al., 2021). That lifetime use of cannabidiol and psychedelics showed the strongest associations with lifetime kratom use underscores drug-use versatility as a characteristic of some who have tried kratom.

Respondents who used kratom also showed high rates of lifetime use of nonmedical/diverted prescription opioids or enrollment in OAT (along with high rates of identifying as “in recovery” from an SUD). It is uncertain here, but elsewhere kratom is used for nonmedical OUD self-treatment (Coe et al., 2019; Garcia-Romeu et al., 2020; Grundmann, 2017; Smith, Rogers, Schriefer, et al., 2021; Smith, Rogers, Strickland, et al., 2021). Kratom use may thus be part of a larger approach to medical and nonmedical treatment of OUD, including OAT engagement as a result of kratom use (Bowe & Kerr, 2020; Buresh, 2018; Khazaeli et al., 2018; Smith et al., 2019; Weiss & Douglas, 2021). We found that 16.8% of our kratom-use group had been enrolled in OAT, compared to 3.1% of the nonuse group. Follow-up work should clarify temporal and causal relations between kratom use and OAT enrollment.

Our finding of greater lifetime substance use in the kratom-use group extended to many of the negative consequences of substance use: lifetime drug overdoses, SUD prevalence, and SUD severity. There have been limited and mixed findings in U.S. adults regarding the prevalence of kratom-use disorder (a diagnostic category that is not in the DSM-5 but can be operationalized with SUD criteria for other opioids). For instance, using an SUD checklist, Garcia-Romeu et al. (2020) found that nearly 90.0% of kratom-using adults did not meet DSM-5 diagnostic criteria for use disorder. However, when using the Drug Abuse Screening Test, Schimmel et al. (2021) found scores were higher among people who reported kratom use, a conclusion that remains contested (Grundmann et al., 2021). Although we did not assess criteria for kratom use disorder, our findings do show that this lifetime kratom-use group had higher past-year rates of DSM-5 SUD for any substance. However, based on published accounts of reasons for kratom use, it is reasonable to speculate that some respondents were using kratom to mitigate SUD symptoms for other substances.

Limitations

This study has notable limitations, particularly, the cross-sectional design and the lack of generalizability of data crowdsourced from one platform. While these limitations prevent us from definitively addressing our aim to better conceptualize U.S. adults with a kratom-use history, it does not leave it unaddressed, as data collected here were done by recruiting via mTurk into a survey not specific to kratom. As such, we were able to also contrast people with lifetime kratom use to those without. Still, it is impossible to know whether respondents initiated kratom use prior to other drug use or subsequent to use for purposes of SUD symptom mitigation. These hindrances limit generalization of results to the broader kratom-using population. The generalizability of findings is also limited to an online sample of U.S. adults who, in the past 6 months, used either a normative, relatively nonstigmatized drug (alcohol) or one of several more stigmatized drugs (opioids, psychostimulants). Extrapolation to the broader population is probably impossible, but our survey is still a stride forward from surveys that enrolled only people currently using kratom. The study is also limited by lack of kratom-specific follow-ups, such as reasons for use. For instance, while rates of chronic pain were higher among people who reported kratom use, we cannot infer that they were using kratom to self-treat pain, though this has been documented elsewhere. These limitations are actively being addressed by our group in a follow-up survey. By recontacting kratom-using adults identified here for further investigation, we may better understand not only multiple motivations for use, but also changes in use, dose, and kratom products over time, including reducing or discontinuing use.

Conclusion

Overall, these data suggest that a nontrivial number of people in the U.S. have ever tried kratom. Though our understanding of U.S. adults with a history of kratom use remains incomplete, these findings allow for the provisional characterization that heterogeneity of people who use kratom may best be reflected in terms of other drug use histories, specifically those reporting polysubstance use or SUD histories (particularly illicit substances). Here, we found versatile drug-use histories and high SUD rates among kratom-using adults compared to those in prior work (Garcia-Romeu et al., 2020). Indeed, the broader literature is beginning to suggest a distinct subgroup of regular kratom users who report using specifically to self-treat illicit SUDs, compared to regular users reporting other motivations, including supplementing licit opioids (Bath et al., 2020; Garcia-Romeu et al., 2020; Grundmann et al., 2017). As we could not assess motivations for kratom use here, future work should discern primary and secondary motivations for initiating and continuing kratom use and determine if motivations are dynamic when measured proximally to kratom use events. Our findings also show that psychosocial factors differ between people with kratom use histories and those without, with anxiety and depression symptoms far greater among the former. Future self-report studies should include comparison groups and work to reduce self-selection bias by sampling people who have discontinued kratom use or who have had adverse kratom experiences. More broadly, addiction researchers and clinicians who assess substance use histories, particularly for diagnosing SUDs, need to assess kratom use and disordered use, especially given that it is not readily detected via urine analyses. Doing this will enable multiple types of data collection which can help establish kratom-use prevalence within subpopulations and initiate more accurate understanding of differences among kratom-using adults and differences between those who have and have not used kratom. Such work will inform both clinical practice and public health policies surrounding kratom.

Public Health Significance.

Self-report data collected via kratom-specific surveys have lacked standardized psychosocial measures and may not be representative of the broader U.S. kratom-using population. Using different sampling methods, we found that adults with kratom use histories had poorer psychological health, greater chronic pain, and higher rates of substance use disorder and polysubstance use compared to those with no kratom use history. As use of other drugs had been the strongest predictors of kratom use, clinicians should incorporate kratom into clinical assessments and be aware of the physical and psychiatric symptoms associated with its use.