The Electronic Nicotine Delivery System (ENDS) Purchase Task: Are Results Sensitive to Price Framing?

Abstract

Extension of the cigarette purchase task (CPT) to electronic nicotine delivery systems (ENDS) is complicated by the heterogeneous nature of this product class, as ambiguity exists regarding the appropriate price-frame (i.e., unit of the product being purchased). We explored correlations between ENDS purchase task (E-CPTs) outcomes featuring two common price-frames: 10 puffs and 1 mL of liquid. Adult exclusive ENDS users (N = 19) and dual users of ENDS and cigarettes (N = 16) completed two own-brand E-CPTs. One E-CPT used “10 puffs” as its price-frame; the other used “1 mL of liquid.” Five outcomes were generated for each E-CPT: breakpoint, intensity, O_max, P_max, and α. Exploratory Factor Analyses (EFA) considered how these outcomes captured latent structures of demand for ENDS. Spearman correlations in E-CPT outcomes assessed within-person variation between price-frames. Analyses also considered whether correlations differed by user group. E-CPT outcomes were highly correlated across price-frames (ρs > 0.57, ps < .001), and EFA revealed little difference in how outcomes from the tasks loaded onto two latent structures of demand (“Persistence” and “Amplitude”) reported in the previous literature. The magnitude of correlations for E-CPT outcomes tended to be higher for exclusive ENDS users than for dual users. Participant responses to purchase task outcomes were similar across two E-CPT price-frames. Using “10 puffs” as a price-frame may be a generalizable approach among heterogenous groups of ENDS users, but researchers should consider their target population and structure the E-CPT to reflect participants’ knowledge and purchasing behaviors.

Hypothetical purchase tasks assess the relative reinforcing value of a commodity (e.g., cigarettes) by asking participants to report how many units of the commodity they would purchase at varying prices (Jacobs & Bickel, 1999; Tidey et al., 2016). Data from purchase tasks yield multiple outcomes which, in the context of the cigarette purchase task (CPT), are validated predictors of dependence, treatment response, and abuse liability among smokers (Bidwell et al., 2012; Higgins et al., 2018; MacKillop et al., 2008; O’Connor et al., 2016; Reed et al., 2020). There is growing interest in extending the purchase task framework to study other tobacco/nicotine products like electronic nicotine delivery systems (ENDS; Cassidy et al., 2017, 2020; Pericot-Valverde et al., 2017). However, the heterogeneous nature of ENDS purchase task (E-CPT) design. Specifically, because ENDS include disposable devices, rechargeable devices with refillable tanks for liquid, and replaceable pods/cartridges—ambiguity exists regarding the appropriate “price-frame” for E-CPTs (i.e., unit of ENDS being purchased). The price-frame must capture daily consumption, variation in consumption as a function of price, and real-world purchasing behaviors (Aston & Cassidy, 2019).

Two potential price-frames for E-CPTs have emerged: puffs and mLs of ENDS liquid. One mixed methods study found that first-generation ENDS (i.e., cig-a-like) users preferred reporting consumption in units of individual devices (Cassidy et al., 2017). In contrast, advanced-generation ENDS users emphasized that mLs of liquid would better reflect their use. However, both groups’ performance in a puff-version E-CPT were well-described by an exponentiated demand equation, suggesting that puffs may be an appropriate unit for samples featuring users of different device types. In a follow-up study of dual ENDS/cigarette users and exclusive ENDS users, the investigators found that an E-CPT with “mLs of liquid” as its price-frame was more likely to be correlated with self-reported use, dependence, and cotinine levels than E-CPTs using puffs, but suggested that the latter price-frame may be more appropriate for dual users as puffs are a unit of consumption shared by ENDS and cigarettes (Cassidy et al., 2020). Moreover, several E-CPT outcomes (e.g., intensity and α) in the mL-version were uncorrelated with their counterparts in the puffs-version, suggesting E-CPT results may be sensitive to price-framing. In contrast, a clinical lab study involving 43 tank-style ENDS users concluded that indices of demand from E-CPTs were more consistent with a “mL of liquid” price-frame than a “puffs” price-frame, when comparing outcomes from a task focused on own-brand ENDS to one featuring an experimental ENDS (Pericot-Valverde et al., 2020). Finally, a crowdsourced (Amazon Mechanical Turk) sample of 132 past-week ENDS users completed purchase tasks using price-frames of individual puffs, cartridges, and mLs of liquid (Strickland et al., 2021). The authors concluded that, of the three price-frames tested, individual puffs provided the most flexibility in capturing demand within heterogenous samples of ENDS users.

These studies provide initial indications that puffs may be a superior price-frame for E-CPTs when the intended sample is a heterogenous group of ENDS users. However, continued development of the E-CPT requires new data regarding the performance of the task across alternative price-frames. In particular, the prior literature has emphasized a need to determine if price-framing might affect latent structures of demand derived from E-CPT responses (Bidwell et al., 2012; González-Roz et al., 2020; O’Connor et al., 2016). There is also a need to more thoroughly characterize whether differences in E-CPT responding across price-frames exist between different user groups (e.g., dual users vs. exclusive users) and whether yoking commodities (i.e., using multiple puffs instead of a single puff to better represent an episode of use) may provide higher quality data. We extend the literature by exploring differences in outcomes from E-CPTs featuring price-frames of “10 puffs” and “1 mL of liquid” among a sample of dual users and exclusive ENDS users. The “1 mL of liquid” price-frame was selected based on its use in the prior literature (Cassidy et al., 2017, 2020) and the “10 puffs” price-frame because it represented a rounded approximation of the number of puffs taken from a conventional cigarette as well as its use in prior E-CPTs (Barnes et al., 2017).

Method

Sample

The sample was drawn from a clinical laboratory trial (Hoetger et al., 2021) approved by the institutional review board at Virginia Commonwealth University (HM20012696) and registered on clinicaltrials.gov (NCT03830892). Data were collected between February 2019 and March 2020. Sixty-six participants consented to participate, of whom 47 enrolled and 36 completed all study activities. Of these 36 completers, 19 were exclusive ENDS users (daily ENDS use [≥3 mg/mL nicotine] with no past-month cigarette use) and 17 were dual users of tobacco cigarettes and ENDS (either [a] daily use of ENDS and someday use of cigarettes [≥3 days per week] OR [b] someday use of ENDS and daily use of cigarettes). The intended sample size for this study was 120; however, data collection were prematurely ended as a result of the COVID-19 pandemic. The analyses that follow were exploratory and the planned sample size was based on power analyses for outcomes not related to these analyses (Hoetger et al., 2021).

Eligibility required a semiquantitative urine cotinine result of “positive” at screening (≥200 ng/mL; urine cotinine cassette test). Participants were excluded for self-reported diagnoses of current medical conditions including, but not limited to, cardiovascular, pulmonary, renal, hepatic, immune, or psychiatric (including use of psychotropic medications) disorders. Participants who were pregnant or breastfeeding, reported past-month use of illicit drugs, reported >15 days out of the last 30 of marijuana use, or reported >25 days out of the past 30 for alcohol use were excluded. Those who intended to quit tobacco/nicotine products in the next 30 days were also excluded and referred to treatment resources.

Study Design and Materials

The study consisted of six laboratory sessions: in-person screening, four experimental ENDS sessions, and a final own-brand (OB) ENDS session. Demographics as well as baseline ENDS use and dependence (Patient Reported Outcome Measurement Information System Nicotine Dependence Item Bank for use with electronic cigarettes [PROMIS-E] total score) were collected at screening (Morean et al., 2019). All E-CPT data reported here were from the OB session. Sessions took place at least 48 hr apart. Participants abstained from tobacco/nicotine products for ≥12 hr before sessions, biochemically verified by exhaled CO readings (Vitalograh, Lenexa, Kansas). CO cutoffs were set at ≤10 ppm for dual users and ≤7 ppm for exclusive users because CO is a byproduct of cigarette use but not ENDS use, and CO levels have been documented as being higher among dual users than exclusive ENDS users (Carroll et al., 2018). Because of the half-life of nicotine and byproducts of ENDS—coupled with the prescribed scheduling interval for lab sessions—we could not confirm abstinence using urinary tests (e.g., changes in propylene glycol/vegetable glycerin or cotinine). However, supervised 1-hour rest periods preceding E-CPT administration ensured all participants were nicotine abstinent for this period of time prior to completing the purchase tasks.

The E-CPTs presented the following scenario:

Imagine a typical day during which you use e-cigarettes. The following questions ask how many times you would buy 10 PUFFS of your own-brand e-cigarettes if they cost various amounts of money. The only available e-cigarettes are your own-brand. Assume that you have the same income/savings that you have now and no access to any e-cigarettes or nicotine products other than those offered at these prices. In addition, assume that you would consume the puffs that you request on that day; that is, you cannot save or stockpile puffs for a later date.

Participants then reported how many times they would purchase 10 puffs of their OB ENDS at the following 16 prices (presented one at a time, in ascending order): $0.00, $0.01, $0.02, $0.04, $0.08, $0.16, $0.32, $0.64, $1.28, $2.56, $3.84, $5.12, $6.40, $7.68, $8.96, and $10.24. The “mLs of liquid” price-framed task was identical to the puffs-version, but specified mLs of e-cigarette liquid rather than 10 puffs. Counterbalancing the order the purchase tasks was impractical because of other design elements of the study reported elsewhere (Hoetger et al., 2021), thus the puffs-version of the E-CPT was always administered first followed immediately by the mL-version. Both E-CPTs were computer administered and displayed the total number of 10-puff bouts or mLs of liquid purchased, as well as the associated expenditure, for each price-point prior to participants confirming their purchases.

Outcome Measures

From both E-CPTs, five outcomes (or indices of demand) were recovered for each participant: intensity (consumption at $0.00; observed-value), breakpoint (price at which consumption was suppressed to zero; observed-value), O_max (maximum daily expenditure; observed-value), P_max (price at O_max; observed-value), and α (rate of change in elasticity [i.e., a measure of price-responsiveness]; model-derived; Gilroy et al., 2019, 2020; Hursh & Roma, 2016; Tidey et al., 2016; Zvorsky et al., 2019). From these five indices, we performed exploratory factor analyses (EFA) to identify latent structures of demand for each E-CPT. Prior investigations using similar methods have repeatedly condensed cigarette purchase task indices into two latent structures of demand—Amplitude (consumption at inexpensive prices; typically consisting of Intensity and occasionally O_max) and Persistence (insensitivity of demand to price increases; typically consisting of α, breakpoint, P_max, and occasionally O_max)—but have yet to consider demand for ENDS (Bidwell et al., 2012; González-Roz et al., 2020; Zvorsky et al., 2019).

Data Preparation and Analyses

E-CPT responses were compared to established standards for CPT data quality (Stein et al., 2015). One participant reported the same level of consumption (two 10-puff bouts; one mL) at each price-point in both E-CPTs and was excluded from these analyses, dropping sample size to 35. There were 5 (14.3%) individuals that reported purchasing one 10-puff bout, and 3 individuals (8.6%) that reported purchasing one mL of liquid, at the highest price-trial ($10.24) in the corresponding E-CPT—in such situations, breakpoint was imputed as $10.24. Most outcomes (intensity, breakpoint, O_max, and P_max) were taken directly from the observed responses on the purchase tasks. Individual demand curves were fit using an exponentiated demand equation to recover α estimates and quantify the fit of the data (R²): Q = Q₀10^k(e−“^Q0p−1) where Q is the number of units demanded at price p, Q₀ is consumption at $0.00 (set equal to observed values [i.e., “intensity”; task- and participant- specific]), α indexes the rate of change in elasticity across all modeled price-points, and k is a bounding parameter which reflects the span of the demand curve and was set equal to four for all analyses (Gilroy et al., 2020; Koffarnus et al., 2015).

We considered setting k using either individual-level or task-level (i.e., averaged across participants) values based on the span of the data (Koffarnus et al., 2015); however, both of these approaches yielded poorer average model fits (i.e., R²) compared to when k = 4 for both tasks (see Supplemental Table 1, for more information on the span of consumption across participants and tasks). Outcomes were not transformed, as no conventional transformation provided a distribution that was clearly more normal than the raw values and the primary statistical tests (i.e., factor analyses, Spearman correlations) used here do not assume normality.

EFAs were performed using the five demand indices from each E-CPT. These estimations were performed with an oblique (oblimin) rotation to permit multifactorial solutions with correlated factors (Bidwell et al., 2012; González-Roz et al., 2020; Higgins et al., 2020; O’Connor et al., 2016). Unique latent factors of demand were defined as those with eigenvalues rounded to 1.0. A factor loading of ≥0.40 determined if an index loaded onto a latent factor. Indices were allowed to cross-load onto multiple factors. Factor scores were created for each individual in both versions of the E-CPT by summing the standardized scores for individual items that loaded onto each factor. The Kaiser–Meyer–Olkin test of sample adequacy and Bartlett test of sphericity were used to determine whether the sample size and variance provided sufficient data to conduct EFA (Kaiser, 1974).

We focused on within-person variation in E-CPT outcomes across the two price-frames. We calculated Spearman correlations for each of the five indices of demand and the identified latent structures of demand, correlating each outcome from the “10 puffs” price-frame with its counterpart in the “1 mL of liquid” price-frame. We first analyzed the whole sample, then stratified by dual versus exclusive users, by participants who reported their baseline ENDS consumption in “pods/cartridges” versus “mL of liquid,” and by OB ENDS device type (pod-style devices vs. box-mod/tank-style device vs. disposable style device). The Fisher Z transformation was used to generate confidence intervals for these Spearman coefficients. We also investigated whether findings were sensitive to modeling cross-task correlations using Pearson correlations and analyzed the degree of correlation between E-CPT outcomes from the two price-frames and ENDS dependence.

Transparency and Openness

All analyses were conducted in Stata 17.0 standard edition (College Station, Texas) and GraphPad Prism 9.3.1 (San Diego, CA). α was set at 0.05 for two-sided tests. We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study, and we follow Journal Article Reporting Standards (Kazak, 2018). The study design was preregistered (NCT03830892); however, analyses reported here were exploratory and thus, not preregistered. Data, study materials, and analysis code for this study are available upon request of the corresponding author.

Results

Sample and E-CPT Outcomes

Among the 35 participants, the average age was 28.9 years, and a majority identified as male (62.9%) and as non-Hispanic White (65.7%; Supplemental Table 2). Our sample included 19 exclusive ENDS users (54.3%) and 16 dual users; most reported their baseline ENDS consumption in units of “mLs of liquid” (71.4%) rather than “pods/cartridges.” Own-brand ENDS device types included pod-(N = 17) and box-mod/tank- style (N = 17) devices, with one participant using a disposable device. Mean E-CPT outcomes did not differ significantly between the two price-frames except for intensity, which was significantly higher in the puffs-version E-CPT (Supplemental Table 3). Responses to both E-CPTs were well-described by the exponentiated demand equation, with an average R² of 0.922 in the mL-version and 0.972 in the puffs-version (Supplemental Table 1).

Exploratory Factor Analyses

Adequacy of the sample for EFA was confirmed by Kaiser–Meyer–Olkin index values and Bartlett’s sphericity test (ps < .001). Two latent factors were recovered from EFAs (Table 1). In both EFAs, Factor 1 explained roughly 60% of the variance, and Factor 2 explained roughly 20%. Loading factors revealed that, in both price-frames, Factor 1 (termed Persistence) was composed of breakpoint, P_max, O_max, and α, whereas Factor 2 (termed Amplitude) was composed of intensity. Weak correlation coefficients between Persistence and Amplitude (within-task; Table 2) provide evidence that these two factors captured distinct features of demand.

Table 1.

Results of Exploratory Factor Analyses (EFAs) Between E-CPTs Featuring Two Different Price-Frames

Sample adequacy checks
Measure		“1 mL of liquid” price-frame		“10 puffs” price-frame
Determinant of correlation matrix		0.029		0.068
Bartlett test of sphericity		χ2(111.874, 10) = p < .001		χ2(84.88, 10) = p < .001
Kaiser–Meyer–Olkin measure of sampling adequacy		0.600		0.714
Factor loading
Measure	Factor 1 (Persistence)	Factor 2 (Amplitude)	Factor 1 (Persistence)	Factor 2 (Amplitude)
Variance explained	58.15%	19.91%	57.22%	20.25%
Eigenvalue	2.90767	0.99543	2.86104	1.01249
Index
Breakpoint	0.9662	0.0102	0.9433	0.0595
Pmax	0.9008	−0.0294	0.8956	0.0083
Omax	0.8593	0.0493	0.8709	−0.0052
α	−0.6301	−0.2138	−0.6361	0.2130
Intensity	−0.1650	0.9728	0.0779	0.9816
Cronbach’s α	0.7763	—	0.7805	—

Note. E-CPT = ENDS purchase task; END = electronic nicotine delivery systems.

Factors with eigenvalues rounded to at least 1.0 were included in the final solution. A criterion of 0.40 was used to determine whether an index was significantly loaded on a factor. Oblique (oblimin) rotations were used to allow for multifactorial constructs with correlated factors. The terms “Persistence” and “Amplitude” are borrowed from previous EFAs used to characterize demand from cigarette purchase tasks (Bidwell et al., 2012; González-Roz et al., 2020; O’Connor et al., 2016).

Table 2.

Correlation in E-CPT Outcomes Featuring Two Alternative Price-Frames

Correlation in outcomes between price-frames (10 puffs vs. 1 mL of liquid)
	Intensity (1 mL)	α (1 mL)	Pmax (1 mL)	Omax (1 mL)	Breakpoint (1 mL)	Persistence (1 mL)	Amplitude (1 mL)
Intensity (10 puffs)	0.5721 (0.0003)	−0.3312 (0.0519)	0.1601 (0.3582)	0.2329 (0.1782)	0.2105 (0.2248)	0.2353 (0.1738)	0.4069 (0.0153)
α (10 puffs)	−0.0065 (0.9706)	0.7175 (<0.0001)	−0.5932 (0.0002)	−0.7269 (<0.0001)	−0.7022 (<0.0001)	−0.7304 (<0.0001)	−0.0325 (0.8530)
Pmax (10 puffs)	0.1132 (0.5172)	−0.5866 (<0.0002)	0.5274 (0.0011)	0.6403 (<0.0001)	0.6584 (<0.0001)	0.6491 (<0.0001)	0.0712 (0.6845)
Omax (10 puffs)	0.0994 (0.5699)	−0.7581 (<0.0001)	0.7003 (<0.0001)	0.7784 (<0.0001)	0.7917 (<0.0001)	0.8114 (<0.0001)	0.0960 (0.5834)
Breakpoint (10 puffs)	0.1780 (0.3064)	−0.6872 (<0.0001)	0.6147 (0.0001)	0.6708 (<0.0001)	0.7448 (<0.0001)	0.7413 (<0.0001)	0.1427 (0.4135)
Persistence (10 puffs)	0.1143 (0.5133)	−0.7360 (<0.0001)	0.6401 (<0.0001)	0.7823 (<0.0001)	0.7741 (<0.0001)	0.7858 (<0.0001)	0.0919 (0.5996)
Amplitude (10 puffs)	0.6082 (0.0001)	−0.2818 (0.1010)	0.1160 (0.5069)	0.1704 (0.3277)	0.1732 (0.3198)	0.1953 (0.2610)	0.3606 (0.0334)
Correlation in outcomes within price-frame (1 mL of liquid)
	Intensity	α	P max	O max	Breakpoint	Persistence	Amplitude
Intensity	—
α	−0.3537 (0.0371)	—
P max	−0.0304 (0.8625)	−0.6944 (<0.0001)	—
O max	0.1289 (0.4606)	−0.9323 (<0.0001)	0.7593 (<0.0001)	—
Breakpoint	0.0682 (0.6970)	−0.8495 (<0.0001)	0.9265 (<0.0001)	0.9053 (<0.0001)	—
Persistence	0.1274 (0.4658)	−0.8952 (<0.0001)	0.9193 (<0.0001)	0.9260 (<0.0001)	0.9816 (<0.0001)	—
Amplitude	0.8224 (<0.0001)	−0.2897 (0.0914)	−0.1257 (0.4753)	0.0746 (0.6703)	−0.0364 (0.8354)	0.0261 (0.8819)	—
Correlation in outcomes within price-frame (10 puffs)
	Intensity	α	P max	O max	Breakpoint	Persistence	Amplitude
Intensity	—
α	−0.1600 (0.3587)	—
P max	0.1204 (0.4908)	−0.6689 (<0.0001)	—
O max	0.2184 (0.2076)	−0.9461 (<0.0001)	0.7058 (<0.0001)	—
Breakpoint	0.1802 (0.3002)	−0.8026 (<0.0001)	0.8544 (<0.0001)	0.8397 (<0.0001)	—
Persistence	0.2303 (0.1833)	−0.8955 (<0.0001)	0.8850 (<0.0001)	0.9315 (<0.0001)	0.9097 (<0.0001)	—
Amplitude	0.9771 (<0.0001)	−0.0896 (0.6086)	0.1128 (0.5187)	0.1478 (0.3968)	0.1611 (0.3551)	0.1759 (0.3121)	—

Note. E-CPT = ENDS purchase task; END = electronic nicotine delivery systems.

Results presented above show Spearman correlation coefficients with associated p values below each in parentheses. α was set to 0.05 to determine statistical significance. Italicized values indicate the primary outcome, the between price-frame correlation of E-CPT outcomes.

Correlations Between E-CPT Outcomes Within and Across Price-Frames

The five outcomes and two factors from the puffs-version E-CPT were significantly and positively correlated with their corresponding values in the mL-version E-CPT, with ρs ranging from 0.3606 (Factor 2 [Amplitude]) to 0.7858 (Factor 1 [Persistence]; italicized diagonal line in Table 2; ps < .001). Conclusions were not sensitive to use of Pearson instead of Spearman correlations (Supplemental Table 4).

In stratified analyses (Supplemental Table 5), correlations in outcomes between the tasks tended to be higher for exclusive ENDS users than for dual users. Correlations were largely insensitive to OB ENDS device type as well as reported unit of baseline consumption. There was weak correlation between ENDS dependence (i.e., PROMIS-E total score) and E-CPT outcomes from both price-frames; however, intensity and Factor 2 were moderately and significantly correlated with ENDS dependence in the “10 puffs” price-frame only (Supplemental Table 6).

Discussion

We compared two hypothetical own-brand E-CPTs that differed by price-frame (“10 puffs” vs. “1 mL of liquid”) among a heterogenous sample of ENDS users recruited for a clinical laboratory study in a mild-to-moderate state of nicotine withdrawal. We found that outcomes were strongly correlated and mapped similarly onto latent demand structures. However, correlations in several outcomes between tasks tended to be stronger for exclusive ENDS users than dual users.

Results build on the prior literature by affirming the appropriateness of both numbers of puffs and volume of ENDS liquid as conceptually valid price-frames for E-CPTs. The strength of correlation in outcomes between the two E-CPTs suggests that respondents were able to convey their demand for ENDS under both price-frames and that results across alternative price-frames are comparable. Additionally, the EFA results presented here were highly consistent with the past literature, which has described two latent structures of demand that explain approximately 80% of the variance in CPT responses (Bidwell et al., 2012; González-Roz et al., 2020; Higgins et al., 2020), but to our knowledge had not yet been documented in E-CPT responses. The similarity to previous work suggests that both E-CPT price-frames hold high construct validity and, by extension of prior work, are likely to predict real-world tobacco use behaviors (González-Roz et al., 2020; Morean et al., 2019). Potentially because of sample size issues, however, neither price-frame demonstrated a high degree of correlation between E-CPT outcomes and ENDS dependence—with the exception of intensity/Factor 2 (Amplitude) in the “10 puffs” version of the task. Further exploration of this finding is merited, as is consideration of whether knowledge of latent demand structures can extend knowledge on ENDS demand and use behaviors (e.g., progression to sustained use, cessation, etc.). Finally, Cronbach α values and the correlation among outcomes within-task indicate internal consistency in both price-frames (Murphy et al., 2011).

Notably, our results contrast with prior findings which suggested that intensity and α were uncorrelated across “puff” and “mL” price-frames (Cassidy et al., 2020). The previous study used “1 puff’ as the price-frame, whereas the present study used “10 puffs;” perhaps ENDS users find ENDS demand easier to report in terms of blocks of puffs (i.e., “10 puffs”) or discrete use episodes, rather than individual puffs. The present study also found stronger correlations between price-frames among exclusive ENDS users than among dual users. Exclusive ENDS users may be more comfortable than dual users in reporting demand across different units of ENDS. Alongside prior work which argued that puffs may be a more interpretable price-frame for dual users (Cassidy et al., 2020; Strickland et al., 2020), these results suggest that using multiple puffs as the E-CPT price-frame could be broadly applicable to adult ENDS users and may be preferable for studies involving dual users.

A final dimension of E-CPT task performance we considered was the congruence in total purchasing volume between the “10 puffs” and “1 mL” E-CPTs. We found that participants reported relatively higher purchasing volumes at the “free” price (i.e., intensity; Supplemental Table 1) and when summed across all price-points (Supplemental Table 7) in the “1 mL” than the “10 puffs” price-frame. Across participants, on average, if total purchasing volumes were equal between the two price-frames, 1 mL of liquid would need to yield 31.23 puffs (average ratio at $0.00: 4.40 puffs/mL). These ratios are far below the marketed puff to mL ratio of popular ENDS products such as PuffBar (~278 puffs/mL; Puff Bar, 2022) and Juul (~283 puffs/mL; Truth Inititiative, 2019). Put another way, if we accepted a puffs-to-mL ratio of 280 puffs/mL, the average total purchasing in the “10 puffs” task would yield 1,378.57 puffs compared to 21,422.80 puffs in the “1 mL” task. This finding is perhaps unsurprising as the task was intended to capture consumption over a single day (when no other products were available), and we only allowed participants to purchase in discrete units. This finding highlights the need for congruence between the time- and price-frames used in E-CPTs and, we believe, further supports the utility of a “puffs” price-frame because of the greater resolution it affords in reporting intended consumption. Consideration of purchasing volumes represents an area of potential future research and development for the E-CPT.

These results should be considered alongside limitations. First, the small sample size (N = 35)—though typical of clinical laboratory studies—limits the generalizability of these findings and necessitates caution in interpreting results, particularly from subgroup analyses. Second, the findings may not generalize to novice or non-ENDS users. Third, the puffs-version E-CPT was always completed first. Responses to the puffs-version E-CPT could have anchored responses to mL-version, as could responses to puffs-framed E-CPTs used in earlier sessions.

Despite these limitations, the present analyses can inform the design of E-CPT studies and promote confidence in comparing results from E-CPTs that employ different price-frames. We found that both “10 puffs” and “1 mL of liquid” are conceptually valid price-frames for E-CPTs among adult ENDS users. Researchers should select the price-frame that most closely mirrors the knowledge and purchasing behaviors of their target population, is appropriate for the price-points and timeline specified in the task, and allows for consistency across products or conditions being tested.

Public Health Significance.

This study informs methods intended to assess the abuse liability of electronic nicotine delivery systems (ENDS). Results support the conceptual validity of using price-frames (i.e., unit of ENDS purchased) of either “10 puffs” or “1 mL of liquid” when designing hypothetical ENDS purchase tasks, but also suggest that “puffs” may be a more generalizable price-frame for studies involving heterogenous groups of ENDS users. Moreover, results suggest that the factors underlying demand for ENDS in purchase tasks are similar to those documented with other tobacco products (e.g., cigarettes).