Association of behavioral economic demand for kratom with DSM-5 use disorder: quantity and likelihood-based demand approaches

Highlights

• •Hypothetical purchase tasks can quantify the reinforcing efficacy of kratom.
• •Individual kratom demand is sensitive to recent, but not lifetime, kratom use disorder.
• •Quantity- and Likelihood-based purchase tasks capture different aspects of demand.

Abstract

Aims

Kratom products have increased in popularity in the United States due to their availability and their purported analgesic and stimulatory properties. We sought to examine if behavioral economic demand procedures widely used in addiction science would identify indicators of problematic kratom use and to evaluate the relative performance of differing demand approaches.

Methods

Respondents with a lifetime history of kratom use (N = 117) completed two versions of a hypothetical kratom purchase task. The first version (“Quantity”) followed typical purchase task methods; respondents were asked how much kratom they would purchase across various prices. The second version (“Likelihood”) asked how likely respondents would be to purchase a single dose of kratom across the same price range. Logistic regression was used to determine the association of demand indices with the likelihood of meeting KUD criteria.

Results

For the Quantity version, higher demand intensity (OR = 2.16, p = 0.01) and lower demand elasticity (OR = 0.45, p = 0.01) were significantly associated with past year but not lifetime KUD (p’s > 0.10). For the Likelihood version, demand intensity, demand elasticity, and Pmax were not associated with KUD (p’s > 0.10); higher breakpoint was associated with past year, but not lifetime KUD (OR = 1.44, p = 0.04).

Conclusions

This study is the first to demonstrate that purchase tasks can quantify the reinforcing efficacy of kratom and emphasizes the temporal specificity of the association between purchase tasks and problem behaviors. These findings also suggest that demand indices from Quantity- and Likelihood-based tasks differentially relate to KUD, emphasizing the importance of task and outcome selection.

1. Introduction

Kratom (Mitragyna speciosa) products have increased in popularity in the United States since 2015, with motivations for use varying from recreation to self-management of pain, anxiety, depression, and/or substance use disorder ([SUD], e.g., alcohol, opioid, stimulant use disorder) (Coe et al., 2019, Hill et al., 2024, Mun et al., 2025, Smith et al., 2022b, Smith et al., 2024a, Smith and Lawson, 2017). Alongside increasing use, concerns have emerged regarding problematic kratom use, which has been assessed by researchers and some clinicians using DSM-5 diagnostic criteria for SUD (American Psychiatric Association, 2022). Kratom use disorder (KUD) has been identified among some US adults who use kratom regularly, with symptoms most often related to tolerance, withdrawal, using more than initially intended, and craving, rather than disruptions of social or occupational functioning, although these can occur as well (Garcia-Romeu et al., 2020, Hill et al., 2024, Smith et al., 2024a).

In any SUD, the persistence and contextually determined limits of drug seeking can be examined with behavioral economic demand measures (Acuff et al., 2023, MacKillop, 2016) with these methods providing an efficient means for understanding motivational factors underlying substance use. Demand measures assess the extent to which a person will defend their consumption of a drug amidst increasing costs, capturing important variables related to relative reinforcing effects such as intensity (level of consumption when the price is free), demand elasticity (α; the rate of change in elasticity, derived from curve-fitting; Gilroy et al., 2020), Pmax (the price at where consumption shifts from inelastic to elastic), and breakpoint (the highest price before zero consumption [BP1]). Behavioral economic measures have been applied to numerous drug classes (Acuff et al., 2020, Gonzalez-Roz et al., 2019, Kaplan et al., 2018, Schwartz et al., 2021, Strickland et al., 2016, Strickland et al., 2020, Strickland et al., 2024), but not to kratom.

Here, we sought to examine if behavioral economic demand procedures widely used in addiction science would identify indicators of problematic kratom use (i.e., KUD) and to evaluate the relative performance of two different approaches—specifically, two versions of a hypothetical purchase task for kratom using either quantitative or likelihood-based assessment (see Roma et al., 2016 for introduction of these general approaches to commonly purchased commodities). Quantity-based tasks are the most commonly used approach while likelihood-based tasks have been proposed as a valuable method for assessing low-rate or one-off events or goods (e.g., demand for a bulk purchase or luxury good). Despite the popularity of these tasks, comparisons of their clinical associations in the same studies are limited. We used logistic regression to test the hypothesis that the four demand measures of interest would significantly relate to the likelihood of meeting criteria for KUD.

2. Methods

2.1. Respondents and procedures

We recontacted respondents between April-May 2021 who had reported lifetime kratom use in a larger, unrelated online survey of substance use that was conducted between September 2020-March 2021 (additional details on this kratom sample can be found in Smith et al., 2022a, Smith et al., 2022b, Smith et al., 2022c). All respondents were recruited using Amazon Mechanical Turk (mTurk), a platform for research crowdsourcing (Miller et al., 2017, Strickland and Stoops, 2019). Eligible respondents were people > 18 years of age who resided in the US, passed all data quality checks, and reported > 1 lifetime use of kratom. A total of 289 respondents were identified from the initial survey for recontact; 283 were able to be electronically notified; 134 participated; the final analytic sample was 117 after removing 17 respondents who demonstrated nonsystematic responding on both purchase tasks. Respondents were compensated $7.25. This study was approved by the National Institutes of Health Institutional Review Board (NIH IRB).

2.2. Measures

2.2.1. Hypothetical kratom purchase tasks

The two hypothetical purchase tasks are shown in the Supplemental Materials and are described briefly here. The Quantity version asked people to consider a typical quality brand of kratom powder that they had previously tried and had found to be of consistent quality across multiple purchases. Then, for a future 30-day period, they were asked how many ounces they would purchase across varying prices (e.g., $0.25/oz, $0.5/oz, $1/oz, $2/oz, $4/oz). The Likelihood version of the task asked people to consider a future 12-hour period with no obligations or responsibilities. Then, they were asked how likely they would be (0–100; would definitely not purchase − would definitely purchase) to purchase a pre-packaged dose of kratom such as a shot or extract across the same range of prices. Both versions of the task stipulated that the kratom purchased could be used whenever the participant wanted within each time period, that the kratom purchased would be the only kratom available, that any leftover would be returned without a “refund,” and that the purchased amount could not be shared with others.

2.2.2. Kratom use patterns

In addition to standard demographic information (e.g., gender, age), respondents were asked to endorse if they had engaged in or felt each of the symptoms for KUD in the past year and in their lifetime. KUD criteria were based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (American Psychiatric Association, 2022, Smith et al., 2022a).

2.3. Data analysis

Results from each version of the hypothetical purchase task were analyzed separately. For each version, those who demonstrated 0 demand across all prices were removed from only that task. There were 11 respondents who had zero demand on only one of the two versions. Respondents were also excluded if they demonstrated nonsystematic demand data as identified by an established three-point algorithm using default thresholds (Stein et al., 2015). The final sample size for each analysis was 109 for the Quantity version and 114 for the Likelihood version.

Behavioral demand measures of interest were intensity (Q0; consumption at an unconstrained price; derived), demand elasticity (α; sensitivity to changes in price; derived), Pmax (the point on the demand curve when demand became elastic, meaning the price where maximal consumption occurred; derived), and breakpoint (BP1; the highest price where any consumption occurred; observed). Demand data were analyzed using the exponentiated demand equation (Koffarnus et al., 2022, Strickland et al., 2016). Each of the behavioral demand measures were log transformed prior to analyses for normality. Logistic regression was used to determine whether demand measures were associated with meeting the criteria for KUD (any 2 or more criteria endorsed). As an exploratory analysis, demand measures were also assessed for their relationships with each individual DSM criteria for KUD (Supplemental Table 2). All analyses were conducted in R studio (version 2024.12.1 + 563).

3. Results

As reported previously (Smith et al., 2022b, Smith et al., 2022c), nearly 60 % of the initial sample had used kratom > 100 times during their lifetime and reported currently or having previously used kratom > 4 times per week (80.6%), for an average of 61.9 ± 104.3 weeks. Just under half (41.9 %) considered themselves current “regular” kratom users. The average age of this sample was 34.7 (SD = 8.70) with roughly half being male (50.42 %). The average age of kratom initiation was 29.68 (SD = 8.89).

Fig. 1 shows demand curves for each task type, separated by past-year KUD criteria endorsed. For the Quantity version of the hypothetical purchase task, intensity (OR = 2.16 [1.21, 4.03], p = 0.01) and demand elasticity (α; OR = 0.45 [0.24, 0.8], p = 0.01) were significantly associated with past-year, but not lifetime, KUD (Table 1). This remained true after adding age and gender as covariates (Supplemental Table 1). No significant associations were observed with breakpoint or Pmax across either timeframe. In the Likelihood version of the task, only breakpoint (OR = 1.44 [1.02, 2.09], p = 0.04) was significantly associated with past-year KUD, and this remained true after adding in age and gender as covariates (Table 1). Additionally, Pmax was significant when controlling for age and gender (OR = 1.54 [1.05–2.34], p = 0.03). No measures in either version of the task were significantly associated with lifetime KUD.

Fig. 1.

Demand curves generated from the Quantity (top) and Likelihood (bottom) versions of the hypothetical purchase task. Curves are separated by people who did not qualify as having KUD versus those who had 2 or more KUD criteria endorsed. R2 = 0.98-0.99. RMSE = 0.98–4.3. Q0 represents intensity derived from curve fitting the exponentiated demand model.

Table 1.

Odds ratios and confidence intervals for demand measures and meeting criteria for KUD for both the Quantity (top) and Likelihood (bottom) versions of the task. KUD = kratom use disorder.

	Predictor	Odds Ratio	95 % CI	p-value
	Quantity Task
Past Year KUD	Q0	2.16	1.21–4.03	0.01
	Alpha	0.45	0.24–0.80	0.01
	Breakpoint	1.14	0.86–1.53	0.37
	Pmax	1.18	0.90–1.56	0.23
Lifetime KUD	Q0	1.44	0.84–2.51	0.19
	Alpha	0.81	0.49–1.33	0.41
	Breakpoint	1.00	0.77–1.30	0.99
	Pmax	1.02	0.80–1.31	0.86
	Likelihood Task
Past Year KUD	Q0	1.71	0.36–14.41	0.54
	Alpha	0.63	0.31–1.20	0.18
	Breakpoint	1.44	1.02–2.09	0.04
	Pmax	1.40	0.97–2.08	0.08
Lifetime KUD	Q0	2.24	0.51–12.41	0.30
	Alpha	1.03	0.57–1.85	0.93
	Breakpoint	1.08	0.79–1.48	0.64
	Pmax	1.00	0.71–1.40	0.98

4. Discussion

These results demonstrate that—as seen with other drug classes (Gonzalez-Roz et al., 2019, Martínez-Loredo et al., 2021, Strickland et al., 2020)—behavioral economic demand measures are associated with the presence of substance use problems. In the Quantity version of the hypothetical purchase task, higher demand intensity (Q0) and lower demand elasticity (α and Pmax) were associated with being categorized as having KUD (2 + criteria endorsed). The Likelihood version showed that respondents with KUD had higher breakpoints.

These associations were only present for past-year KUD criteria endorsement and not for lifetime KUD. This is likely because the purchase task evaluates current consumption, and not during some other epoch such as during the period of heaviest kratom use. This highlights the sensitivity that hypothetical purchase tasks have for detecting motivational factors related to current substance use (see also Acuff & Murphy, 2017). Future studies could consider using alternative versions of the purchase task that manipulate the contextual conditions of the hypothetical situation (e.g., when the kratom is being used; who is present; what responsibilities someone has that day [Acuff et al., 2020, Miller et al., 2023]) to further evaluate environmental and contextual factors influencing use.

The Likelihood version of the task is unique in how it evaluates purchasing behavior. Rather than using traditional purchase task methods (what quantity of X would you purchase at price Y), these procedures ask the likelihood that a respondent would purchase kratom at a given price for a predefined quantity. The Likelihood format compressed demand intensity, particularly when comparing it to the Quantity version, as demonstrated by the fact that 85 % of respondents reported “100 %” chance of purchasing the kratom product at the first price ($0.25). However, Pmax and breakpoint were informative measures derived from the demand curve and so this version of the task still meaningfully captures motivation to purchase/consume a product and therefore has utility in measuring features of demand elasticity and relative price sensitivity. The Likelihood version of the task may be particularly valuable in contexts where a hypothetically purchased item can be best described as a one-off event (e.g., something expensive or that is purchased infrequently; for example, a kratom extract product, which is more expensive than kratom whole-leaf powder and more potent) or when the task itself would benefit from being presented in simpler terms than the Quantity version. The present findings using this task version adds to a growing literature of studies that have utilized this variation and contributes to its utility (e.g., Barnes et al., 2020, Brown et al., 2022, Morrell et al., 2021, Schwartz et al., 2021).

A limitation of the current study is that we did not have sufficient variability to characterize how the number of criteria endorsed relates to differences in demand (however, a bar graph is included as Supplemental Figs. 1 and 2 to descriptively view the trends with past-year KUD categories). This may be clinically relevant because people who are classified as having “mild” KUD (2–3 symptoms endorsed) may not demonstrate psychosocial impairments or experience net detriments from their kratom use (Smith et al., 2024b). It is therefore important to investigate whether severity of KUD or other factors like sex, duration of kratom use, or polysubstance use patterns may moderate these findings in larger, future samples. Other limitations include the study’s cross-sectional nature, potential recall bias, and the variety of kratom products that exist for purchase; at the time data were collected kratom extracts were not as widely used relative to whole-leaf kratom powder (Grundmann et al., 2024). The latter limitations are being addressed in ongoing controlled laboratory settings investigating a variety of commercial kratom products under varying conditions (i.e., pre- and post-kratom self-administration; during kratom withdrawal). We anticipate, too, that characteristics of a person’s typical kratom product or serving size may influence demand, for instance the total amount of alkaloids ingested per dose which will vary by person and product (Sharma et al., 2025). Such analyses require in-person characterization of kratom products given the variability of their formulation and sale within markets that differ across US states.

This study is the first to examine if behavioral economic demand procedures would identify indicators of problematic kratom use. We demonstrated that demand measures were significantly associated with KUD among adults with lifetime kratom use and that these associations were temporally specific. Given the rapid proliferation of kratom-derived products that differ in chemical composition, potency, and price, there is a pressing need to apply a behavioral economic framework to the study of kratom use patterns among US consumers.

CRediT authorship contribution statement

Zachary Pierce-Messick: Writing – original draft, Formal analysis. Kirsten E. Smith: Writing – review & editing, Project administration, Conceptualization. Samuel F. Acuff: Writing – review & editing, Formal analysis. Derek D. Reed: Writing – review & editing, Formal analysis. David H. Epstein: Writing – review & editing, Project administration, Funding acquisition, Conceptualization. Justin C. Strickland: Writing – review & editing, Formal analysis, Conceptualization.

Funding

Research was supported by the National Institute on Drug Abuse (NIDA) Intramural Research Program of the NIH and NIDA grant T32007209.

Declaration of competing interest

KES serves as an expert witness in legal cases involving kratom. The other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.