E-liquid Purchase as a Function of Workplace Restriction in the Experimental Tobacco Marketplace

Abstract

E-cigarette use is prohibited in most smoke-free environments. The effect of this policy on tobacco consumption could be examined using the Experimental Tobacco Marketplace (ETM). The ETM allows observation of policy on smokers’ purchasing behavior under conditions that simulate “real-world” circumstances. A within-subject design was used to evaluate the effect of workplace policy (Vaping Allowed vs. Not Allowed) and nicotine concentration (24 mg/mL vs. 0 mg/mL) on tobacco product consumption. Participants (n = 31) completed one sampling and two ETM/workplace sessions per week for two weeks. During the sampling session, participants were given an e-cigarette with a 2-day supply of a commercially available e-liquid of their preferred flavor. Before purchasing, participants were informed whether e-cigarette use was permitted. During the four ETM sessions, participants purchased for the following 24 hr, including the 4-hour work shift that started immediately after buying products in the ETM. The workplace session consisted of data entry tasks in a mock office environment. Participants could use any purchased tobacco products during two 15-min breaks. Condition order was counterbalanced. The results show that permitting E-cigarette use in the workplace increased e-liquid purchase on average, but nicotine concentration had no effect on e-liquid demand. Cigarette demand was unaltered across conditions. The present study suggests that allowing e-cigarette use in the workplace increased demand for e-liquid regardless of nicotine strength. However, it would not change conventional cigarette demand.

Introduction

Cigarette smoking is a leading cause of preventable death in the United States. Worldwide, tobacco use causes more than 7 million deaths per year (World Health Organization, 2017). In 2015, less than 10% of smokers who attempted to quit were successful (Babb et al., 2017). These low rates of success and engagement suggest the need to explore alternative nicotine-delivery systems that could help the transition to less harmful sources of nicotine intake and eventually cessation. From 2004 to 2005, only about 13% of American smokers reported using nicotine replacement therapy (NRT; Hammond et al., 2008). Some evidence suggests that e-cigarettes may be more effective than traditional forms of NRT for smoking cessation (Hajek et al., 2019). However, conflicting data suggest that no clear link exists between e-cigarette use and reductions in conventional cigarette consumption (Ghosh & Drummond, 2017). Thus, further study is required.

One factor that may influence transitions between conventional cigarettes and e-cigarettes is workplace policy regarding tobacco product use. Workplace policies that allow employees to use e-cigarettes while working - either indoors (e.g., office environments) or outdoors (e.g., construction sites) - while restricting cigarette smoking to designated outdoor areas at predetermined times of day may assist in the transition from cigarettes to e-cigarettes. On the other hand, such a policy could increase total nicotine consumption and infringe on co-workers’ safety and comfort. Evaluating such a policy experimentally would indicate whether permitting e-cigarette use in smoke-free environments, such as the workplace, alters the use of conventional cigarettes and other tobacco products.

One prior observational study has shown that workplaces that allowed heated tobacco products (HTP) and e-cigarette use indoors resulted in workers who were more likely to smoke and use e-cigarettes more often (Siripongvutikorn et al., 2020). However, no study has examined the experimental effects of workplace policy, nor examined how workplace policy interacts with e-cigarette nicotine strength. Toward this end, we used the Experimental Tobacco Marketplace (ETM) in the present study to examine the effects of workplace e-cigarette policies on patterns of tobacco product purchasing.

The ETM is a simulated online store that allows rigorous, controlled examination of the effects of policies, product characteristics, price, and other factors on tobacco product purchasing (for review, see Bickel et al., 2018). The ETM displays each tobacco product offered, including price per unit, photos, and a short description. Participants are typically given a budget and asked to purchase tobacco products for a specified length of time (e.g., the following week). Using the ETM, conditions under which some commodities, such as e-cigarettes, may serve as substitutes for conventional cigarettes can be identified.

Prior work using the ETM has shown that e-cigarettes serve as a substitute for conventional cigarettes (for review, see Bickel et al., 2018), especially when used with higher nicotine concentrations (Pope et al., 2019). Although the long-term health effects of e-cigarettes are still unknown, e-cigarettes are generally accepted to have fewer adverse health side effects than combustible tobacco (CDC’s Office on Smoking & Health, 2019; Nutt et al., 2014). Accordingly, the transition of current smokers from conventional cigarettes to e-cigarettes may have a net positive effect on public health (Beaglehole et al., 2019). An analysis of the conditions under which cigarette smokers substitute conventional cigarettes for other products, including e-cigarettes, may provide new information that can be utilized in harm-reduction strategies. To this end, in the present study, we asked participants to make purchases regarding nicotine products using the ETM before working in a simulated workplace. Workplace policies regarding e-cigarette use were manipulated, as well as the available e-cigarette nicotine strength (24 or 0 mg/mL), to examine the effects of these manipulations on nicotine product purchasing.

Methods

Participants

Current cigarette smokers who were non-regular users of e-cigarettes (i.e., <12 uses in last month) were recruited from the Roanoke, Virginia area using Craigslist, flyers, online advertisements, and referrals. Participants were eligible if they: 1) were between 18 and 65 years old; 2) smoked at least an average of 10 cigarettes per day in the past month; and 3) expressed a willingness to try e-cigarettes. Of note, this study was conducted before the legal age for tobacco consumption in Virginia was raised to 21. Participants were excluded if they were trying to or had immediate plans to quit smoking, were using prescription medication that might affect smoking or nicotine metabolism (e.g., varenicline, bupropion), were pregnant or lactating, or reported regular use of nicotine replacement products (e.g., nicotine gum, lozenges, patches).

Procedure

A within-subjects design was used to examine the effects of workplace policies (i.e., Vaping Allowed and Not Allowed) and nicotine concentration (i.e., 0 mg/mL and 24 mg/mL) of e-cigarettes on tobacco-product purchasing behavior. Participants (N= 31) completed seven sessions over three weeks: one consent, two sampling periods, and four combined ETM purchasing and workplace sessions (all participants completed all four conditions), and one follow-up session (see Figure S1). Participants typically completed the sampling period and two ETM purchasing and workplace sessions within a 7-day period; this session schedule was repeated the second week. The follow-up session occurred one week after the final ETM purchasing and workplace session. The study was approved by the Biomedical Research Alliance of New York (BRANY) Institutional Review Board, contracted by Virginia Tech.

Consent/Sampling.

After obtaining informed consent, participants completed demographic questionnaires and nicotine dependence measures, such as the Fagerstrom Test for Nicotine Dependence - FTND (Heatherton et al., 1991), Questionnaire of Smoking Urges - QSU (Cox et al., 2001), Perceived Health Risk - PHR (Mooney et al., 2006) and Drug Effects/Liking Scale- DE/LS (Hatsukami et al., 1997). Using the Timeline Follow-back assessment (Brown et al., 1998), researchers calculated each participants’ estimated daily tobacco product expenditure, which determined the participant’s account balance in the ETM (see supplemental material, and Pope et al., 2019). Participants then sampled three flavors of 24 mg/mL e-liquid (VaporHQ, Oregon). The participant’s preferred flavor was exclusively available for purchase in the ETM for the study’s remainder. At the beginning of every week, participants were provided an e-cigarette (eGo ONE CT, Shenzhen Joyetech, Shenzhen, China) with a 2 mL sample of e-liquid to use at home, in their preferred flavor and nicotine concentration available for purchase in the two ETM sessions occurring that week. Participants were blind to the nicotine concentration received but were encouraged to use the e-liquid ad libitum before returning to the lab for the first purchase and workplace session, which occurred between 2–3 days after sampling. The order of the nicotine concentration (i.e., 0 mg/mL or 24 mg/mL) was counterbalanced across weeks (i.e., half of the participants received 0 mg/mL in the first week and 24 mg/mL in the second week and the other half of the participants received the opposite order).

ETM purchasing.

The ETM used in this study was a custom-built, realistic online store programmed using the Python software package (version 3; Python Software Foundation, 2008). Prior to each workplace session, participants completed a series of five price conditions per session in the ETM. Across these conditions, the price of a single cigarette varied (i.e., $0.12, $0.25, $0.50, $1.00, and $2.00), but all other products’ prices were held constant. Participants were informed of workplace policy for the session, were instructed to purchase products for the next 24 hours, including the 4-hour simulated workplace shift, and were told that they could take home any products they did not use. Participants used a computer to access the ETM. Each product displayed the price along with an image and a brief description of the product. The products available in the ETM included the participants’ usual cigarette brand, their preferred e-liquid flavor, and non-combustible nicotine and tobacco products, including Copenhagen mint dip pouches, Camel mint snus, Nicorette mint flavor 4 mg nicotine lozenges, and Nicorette white-ice mint flavor 4 mg nicotine gum. After making purchase decisions at all five price conditions, participants received the products from one condition, randomly determined; any unspent budget was returned to the participant. Participants then began the workplace portion of the session.

Workplace Session.

The 4-hour workplace sessions consisted of a 75-minute simulated work period followed by a 15-minute break, a second 75-minute work period followed by a second 15-minute break, and a final 60-minute work period. Participants could use only nicotine products purchased that day in the ETM session; cigarettes could only be used during the two 15-minute breaks. All non-combustible products could be used; e-cigarette usage during work periods was either allowed or prohibited (i.e., Vaping Allowed and Not Allowed, respectively), depending on the current workplace policy in effect. The order of the workplace policies was counterbalanced within a week. (i.e., half of the participants were exposed to the No Vaping condition in the first session of the week and to the Vaping condition in the second session of the week, and the other half of the participants received the opposite order). During the work periods, participants completed a mock data entry program that allowed users to repeatedly type roughly five-sentence paragraphs containing words and numbers, typically totaling 250 characters. After each break period, participants reported their tobacco consumption during the workplace and the break.

Statistical Analysis

Participant characteristics.

Demographic characteristics (e.g., age, education, race), cigarettes per day, and FTND were collected in Session 1. Data from the QSU/Cigarettes, QSU/E-cigarettes, and PHR were collected at the beginning of each week, and evaluated for consistency using a Pearson’s correlation coefficient. Week 1 and Week 2 assessment scores were averaged due to the high correlation (r = 0.66, 0.84, and 0.91 for QSU/Cigarettes, QSU/E-cigarettes, and PHR, respectively). Mean, standard deviation, and percentages were used to describe the sample. See Table S1 for all measures.

Own-price demand curves for cigarettes.

A total of 124 sessions (31 in each condition) were submitted to a preliminary analysis. One participant session in the 24 mg/mL no vaping condition was removed from the dataset due to an error in the implementation of the ETM. The remaining own-price demand data were first examined for systematic responding according to three criteria that indicate that purchasing behavior is not systematically affected by price (Stein et al., 2015): (1) trend (i.e., no reduction in purchasing with increasing price), as well as (2) bounce (i.e., consumption at a certain price that exceeds consumption at the lowest price by at least 25%) and (3) reversals from zero (i.e., inconsistent effects of price on purchasing). In total, 24 (19.51%) datasets failed the trend criterion, 5 (4.07%) datasets failed the bounce criterion, and zero (0.0%) failed the reversal from zero criterion. Only those sessions that failed the bounce criterion were removed from the analysis. Those with trend violations displayed stable purchasing across all prices and were therefore included in the analysis. The final ETM dataset consisted of 29 sessions in the 24 mg/mL Vaping Not Allowed condition, 30 sessions in the 0 mg/mL Vaping Not Allowed condition, 28 sessions in the 24 mg/mL Vaping Allowed condition and 31 sessions in the 0 mg/mL Vaping Allowed condition.

For individual own-price demand data, parameters were empirically determined from the observed ETM purchases (Intensity = observed consumption at 0.12 price, and Omax = highest observed expenditure) to compare across conditions. Note, these data were fitted to Koffarnus et al’s (Koffarnus et al., 2015) exponentiated demand model, using the beezdemand package in R (Kaplan et al., 2019), but failed to converge for a total of 22 sessions (18.64%). Individual intensity data were square-root transformed to allow for parametric statistical testing (Stein et al., 2018). The values of these parameters at each of the four conditions were compared using a two-factor (Workplace Policy × Nicotine Concentration) repeated measures analysis of variance (RMANOVA).

Cross-price demand curves for alternative products.

Individual demand for e-liquid was fit using linear regression and parameters were derived from ETM purchases (Y-intercept = intensity(Stein et al., 2018) and slope=substitutability). Here, the Y-intercept represents the predicted consumption when a cigarette is free 20 and the slope represents the change in the quantity of e-liquid demand divided by the change in price. In this framework, if the alternate product function’s slope is statistically greater than zero, product substitution is verified. To be able to detect differences in e-liquid demand across conditions, individually estimated Y-intercept values were shifted such that the smallest estimated Y-intercept was zero. Though these intercepts represent the purchasing in units (mL) of e-liquid when cigarettes are free, the estimated values in practice were continuous and positively skewed, and 4% zero. Therefore, these values were square root transformed to approximate a normal distribution. E-liquid derived substitution parameters (i.e., Y-intercept and slope) were tested for an interaction between workplace restrictions and nicotine strength using a factorial RMANOVA. If a significant interaction was observed, a post hoc analysis was performed using pairwise contrasts and p-value adjustment for multiple testing using Tukey’s method. These contrasts determine which groups exhibited differences in substitution parameters. Individual demand for alternative tobacco products, such as gum, snus, lozenges, and dip, were also fit using linear regression to determine individual Y-intercept and slope estimates.

Tobacco consumption in the workplace.

Consumption of e-cigarettes (binary) during work period was collected during each workplace session (two work periods per session). Consumption of cigarettes (number of cigarettes consumed) was collected during the breaks. The quantity of cigarettes consumed during a break was square root transformed to approximate a normal distribution. An RMANOVA model was used to analyze the relationship of using e-cigarettes during a work period and the number of cigarettes consumed during the subsequent break. The nicotine concentration of e-liquid was included as an independent main effect.

Drug Effects/Liking scale (DE/LS).

Individual ratings (possible range from 0 to 10) were analyzed using paired t-tests.

R software Version 3.5.1 was used for all data analysis (R Core Team (2018). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, n.d.). All statistical tests were considered significant at the < 0.05 level.

Results

Demographics and smoking-related assessments

Detailed sample characteristics regarding demographics and FTCD are available in the supplemental material (Table S1). A total of 2 out of 31 participants (6.45%) reported using e-cigarettes in the previous 30 days (Mean = 0.42 days of use and a maximum of 9 days of use). During the first sampling period, participants used an average of 0.53 (SD = 0.69) milliliters of e-liquid and during the second sampling period, they used an average of 0.59 (SD = 0.67) milliliters of e-liquid.

Substitution for cigarettes in the ETM

Figure 1A depicts group mean e-liquid purchasing across the four workplace conditions. Factorial RMANOVA revealed a significant effect of workplace policy (F(1,84) = 28.824, p < 0.001) on cross-price intensity, but no significant effect of nicotine strength (F(1,84) = 0.012, p = 0.912) or the interaction between them (F(1,84) = 0.885, p = 0.350, Figure 1B). That is, the Vaping Allowed condition showed increased intensity compared to the Vaping Not Allowed condition, regardless of nicotine strength. No significant effect of workplace restrictions were found for cross-price substitution (F(1,84) = 2.227, p = 0.139) or nicotine strength (F(1,84) = 0.273, p = 0.603) alone, but a significant effect of their interaction (F(1,84) = 5.327, p < 0.05, Figure 1C). The post hoc test indicated a significant difference between the 24 mg/mL Vaping Allowed condition and the 24 mg/mL Vaping Not Allowed condition (p < 0.05) with the latter showing a greater substitutability (i.e. greater slope) of e-liquid.

Figure 1.

(A) E-liquid substitutability as a function of the five cigarette prices, shown in log scale on the x-axis, at each condition. (B) Cross-Price intensity. (C) Group average Substitution. ***: p < 0.001; *: p < 0.05.

From the alternative products, gum was most frequently purchased and showed a substitution profile similar/comparable to e-liquid (see Table S2 for demand intensity and elasticity of alternative tobacco products). The other products were less frequently purchased.

Cigarette demand in the ETM

Group mean cigarette demand across the four workplace conditions is shown in Figure 2A. In general, demand for cigarettes decreased as cigarette unit price increased. Factorial RMANOVA revealed no significant effect of workplace restriction (F(1,84) = 0.752, p = 0.389), nicotine strength (F(1,84) = 0.001, p = 0.982) or the interaction between them (F(1,84) = 0.029, p = 0.866) on observed Intensity (Figure 2B), and no significant effect of workplace restriction (F(1,84) = 0.012, p = 0.915), nicotine strength (F(1,84) = 0.009, p = 0.924) or the interaction between them (F(1,84) = 0.0145, p = 0.704) on Omax (Figure 2C). Thus, no differences in cigarette demand were observed regardless of workplace restrictions and nicotine strength.

Figure 2.

(A) Conventional cigarette demand as a function of the five cigarette unit prices, shown in log space on the x-axis, at each condition. (B) Group average demand intensity. (C) Group average Omax. N.S.: not significant.

Tobacco consumption in the workplace

We explored the relationship between consumption of non-combustible products during the work period and cigarette consumption during breaks. Interestingly, we found that the nicotine concentration assigned to a participant was statistically significant in predicting the number of cigarettes consumed during a break (F(1,200)=5.533; p=0.020).That is, participants consumed more cigarettes during break when their work session was assigned with 0 mg/mL. Note that the quantity (i.e. number of puffs) of e-cigarette consumption during the work period exhibited an inverse relationship with the number of cigarettes consumed during the breaks, but this relationship was not statistically significant (F(1,200)=3.393; p=0.067).

Drug Effects/Liking Scale

Of the six Drug Effects/Liking Scale subscales (any effects, good effects, bad effects, liking, desire and likelihood to use again), no significant differences were found between the nicotine strengths (0 mg/mL and 24 mg/mL) assessed after the sampling periods (Table S3). This result indicates that participants rated both nicotine strengths similarly across for all subscales.

Discussion

The present study sought to investigate how workplace policy can affect tobacco purchasing. The results suggest that permitting e-cigarette use in the workplace increased e-liquid purchase on average, but nicotine concentration had no effect on e-liquid demand intensity. Consistent with their purchases, the results from the Drug Effects/Liking Scale comparing the two concentrations suggests that participants did not differentiate between the two. Based on these data, we believe that explicit workplace vaping policy exhibited stimulus control over ETM purchasing. However, the effects of consuming different e-liquid nicotine concentrations did not. Significant interaction effects were observed, with the 24 mg/mL nicotine concentration showing greater substitution (slope) when vaping was not allowed in the workplace. Despite the observed effect of workplace conditions on e-liquid demand, cigarette demand was unaltered across conditions. Moreover, data on tobacco consumption revealed that participants smoked a greater number of cigarettes during the break period when they were only allowed to use the nicotine-free e-liquid during the workplace.

These findings suggest that allowing e-cigarette use in the workplace may increase demand for e-liquid; however, it may not affect cigarette consumption. Taken together, the impact on demand for e-liquid, but the lack of impact on cigarette demand suggests an overall additive effect of e-cigarette use to cigarette consumption as opposed to a substitutive effect. Similarly, Siripongvutikorn et al.(2020) found that workplaces allowing the use of HTP and e-cigarettes indoors increased the rates of HTP, e-cigarette, but also combustible cigarette use. On the other hand, the lack of effect can be indicative that participants’ exposure to e-cigarette in the short term may have been insufficient to affect cigarette demand (Pope et al., 2019). In this regard, an effect of e-cigarette availability on cigarette demand has been demonstrated in previous studies with more experienced participants, such as dual cigarette and e-cigarette users (Johnson et al., 2017; Rass et al., 2015).

Interestingly, based on similar purchase rates, gum could also function as a substitute for cigarettes to some extent. Although this basic analysis does not control for nicotine purchased, only units of each item, this finding is consistent with previous studies (Johnson et al., 2017; Shahan et al., 2000). We advocate that further studies explore this potential as the product is familiar and is less intrusive and potentially harmful to co-workers than e-liquid. Preliminary findings indicate that e-cigarettes increase the risk to second-hand vapers (Neuberger, 2019) . Moreover, observing others vaping may increase cravings for both e-cigarettes and conventional cigarettes in those who smoke (King et al., 2015).

Even though nicotine concentration did not play a significant role in demand for tobacco products during a 24 hour-period, it significantly affected cigarette consumption during the 4-hour workplace. The fact that participants still consumed the 0 mg/ml concentration during the workplace suggests that non-nicotine aspects of e-cigarette use (or puffing behavior in general), such as taste, olfactory and respiratory tract sensations, have their own reinforcing properties (Rose et al., 2010). However, the significantly higher number of cigarettes smoked during the breaks also suggests a need to increase nicotine intake during the work shift. This study tested only a high e-liquid nicotine concentration (i.e. 24 mg/mL) in comparison to a nicotine free e-liquid (i.e. 0 mg/mL). Intermediate nicotine concentrations may exhibit an inverse gradient effect on cigarette consumption during the workplace. In this avenue, Pope et al. (2019) reported a gradient effect of nicotine concentration on e-liquid substitution for cigarettes in the ETM.

A few limitations of this study should be acknowledged. First, due to the choice in recruiting relatively naive e-cigarette users, a moderating effect of e-cigarette use prior to the study could not be explored. Second, although not limiting nicotine intake before the workplace session is consistent with the real-word, the lack of control did not allow us to elucidate how craving may play into the consumption of tobacco products during the workplace. Third, the effects of a four-hour work shift may not be externally valid with a longer, full-day work shift. Fourth, participant account balances were calculated based on 24-hours of nicotine spending. This models the case in which a user purchases tobacco products prior to a work shift and uses those products during both work and non-work hours; however, this may have limited observation of potential substitution relationships. Further constraining budget to allow purchasing of products to be used only during the 4-hour simulated workplace may have required participants to be more selective in their product purchases and increased degree of substitution. Finally, this study was conducted over a short two-week period and could not capture the effect of workplace policies on demand and substitution profiles in the long term. Future research should look at extended periods of closed economy purchasing and consumption. In addition, secondary measures such as workplace productivity should be investigated to obtain a multidimensional understanding of how these workplace restrictions may affect the employer and employee. Furthermore, this short exposure time to a novel method of nicotine administration could have influenced ETM purchasing, as some of the participants anecdotally expressed frustration with the complexity of the e-cigarette. Future research could compare the effects of closed systems, pod-style, and simple to operate e-cigarettes with tank style, refillable devices to alleviate these difficulties.

In conclusion, observed ETM purchasing suggests that allowing e-cigarettes in the workplace may lead to an increase in total nicotine consumption, as e-liquid demand was increased, but cigarette demand remained unaffected rather than a beneficial substitution and decrease substitutability between products. Public policy regarding rules for e-cigarette use should take into account a harm-reductive focus by favoring transition from cigarettes to e-cigarettes. However, the present study suggests that transition may not be achieved by allowing vaping in the workplace.

Public Significance statement.

This study uses behavioral economics methods to observe nicotine product purchasing in a simulated workplace with different e-cigarette policies and nicotine doses. We found that the e-cigarette policy exhibited stimulus control over purchasing but that this was additive as opposed to substitutive. Basic analysis suggests that gum might be a valuable focus for further studies.