Brief Assessment of Cigarette Demand (BACD): Initial Development and Correlational Results in Adults and Adolescents

Abstract

Developing briefer behavioral economic measures is an important priority to ensure these measures can be used in a variety of different contexts and to reduce participant burden. We developed and sought to validate a Brief Assessment of Cigarette Demand (BACD). A 17-item Cigarette Purchase Task (CPT) and a 3-item BACD were completed concurrently in two community samples of smokers (study 1: adult smokers (n=80) with substance use disorders (SUD); study 2: adolescent smokers (n=81)). Responses on the CPT and BACD were compared on demand indices: 1) intensity (the number of cigarettes requested at no cost), 2) O_max (the maximum expenditure on cigarettes in a 24-hour period), and 3) breakpoint (the point at which consumption is totally suppressed/no cigarettes are purchased). Correlations of demand indices with cigarettes/day and nicotine dependence were calculated. Measures of cigarette demand on the CPT and BACD were significantly correlated, albeit at very different magnitudes, for all three indices in the adult sample (intensity r = .86, breakpoint r = .23, andO max r = .43) and for two of the indices in the adolescent sample (intensity r = .97, breakpoint r= .33). The CPT and BACD relationships with smoking and nicotine dependence were similar for breakpoint and intensity but not for O_max. As initial findings were mixed, additional validation work is recommended to improve psychometric properties before adoption. Valid brief measures of demand could have utility for research and treatment of addictive disorders.

Brief Assessment of Cigarette Demand (BACD): Initial Development and Correlational Validation Results in Adults and Adolescents

Purchase tasks are psychological assessments that use behavioral economic principles and methods to quantify demand (consumption of a substance as a function of its cost) (Hursh, Galuska, Winger, & Woods, 2005). Using self-report, individuals indicate how much of a substance they would purchase at increasing prices. Purchase tasks have been used to quantify demand for variety of substances including alcohol (e.g., Murphy & MacKillop, 2006; Murphy, MacKillop, Skidmore, & Pederson, 2009), cigarettes (e.g., MacKillop et al., 2008), marijuana (e.g., Aston, Metrik, & MacKillop, 2015), e-cigarettes (e.g., Cassidy, Tidey, Colby, Long, & Higgins, 2017), and even food (e.g., Epstein, Dearing, & Roba, 2010).

Several indices of demand can be generated using a purchase task including: intensity (maximum consumption when a commodity is free), elasticity (sensitivity of consumption to increases in cost), breakpoint (the price that causes consumption to be zero), and O_max (maximum expenditure). Demand assessed via purchase tasks has shown to be influenced by substance-related cues (i.e., reduced elasticity) (MacKillop, Brown, et al., 2012), stress (i.e., increased intensity, O_max, breakpoint, and reduced elasticity) (Owens, Ray, & MacKillop, 2015), and concurrent psychopathology (i.e., higher intensity and O_max) (Farris, Aston, Zvolensky, Abrantes, & Metrik, 2017; MacKillop & Tidey, 2011). In clinical research, purchase tasks have demonstrated predictive validity of substance use outcomes - elasticity, intensity and O_max predicted less abstinence during smoking treatment (MacKillop et al., 2016) and breakpoint and O max predicted more drinking 6-months post-intervention (MacKillop & Murphy, 2007). Demand has also been shown to change over time among individuals in treatment - intensity and O_max, decreased and elasticity significantly increased after a brief intervention (Dennhardt, Yurasek, & Murphy, 2015) and intensity and breakpoint decreased from baseline to quit day during smoking treatment (Murphy et al., 2017).

Nonetheless, these tasks can be burdensome for both participants and researchers, with the length (MacKillop, Few, et al., 2012) and unfamiliar analytic methods impeding more widespread adoption. Optimal assessments of demand would be brief, acceptable to participants, researchers, patients, and clinicians, and have sound psychometric properties. Accordingly, a brief measure of alcohol demand was recently developed that was sensitive to alcohol cues, paralleling purchase task findings (Owens, Murphy, & MacKillop, 2015). The development of valid briefer assessments may have utility in a variety of clinical and research contexts. Interest in using these tasks is growing; concurrent with our study, smokers completed two brief versions of a breakpoint measure online which were found to correlate with CPT indices (Athamneh, Stein, Amlung, & Bickel, 2018), showing high potential for this approach.

The objectives of this study were to develop a brief assessment of cigarette demand (BACD) and to test whether it was a valid measure of demand relative to a cigarette purchase task (CPT) and in relation to three measures of smoking involvement among adult and adolescent cigarette smokers. It was predicted that 1) there would be significant positive associations between CPT and BACD demand indices, and 2) that measures of smoking that were significantly associated with the CPT would also be significantly associated with the BACD.

Methods

Participants

Study 1 (adult smokers).

Inclusionary criteria were: (a) being 18–75 years old, (b) reporting smoking 10–42 cigarettes/day (“How many cigarettes do you smoke per day?”) for the past 6 months (“How long have you been smoking at at least this rate?”) during the telephone screening, (c) meeting DSM-5 criteria for current or past year substance use disorder for alcohol, cocaine, marijuana, and/or opiates, (d) zero breath alcohol during study visits, (e) score of 4–8 on the Contemplation Ladder (Biener and Abrams (1991); individuals interested in quitting smoking someday and/or have thought about quitting). Exclusionary criteria were: (a) active psychosis (hallucinations or delusions), (b) current smoking cessation treatment or use of any nicotine replacement, (c) use of medications that could be affected by smoking cessation (e.g., antipsychotics, warfarin, insulin) or that could affect smoking (e.g., naltrexone, buprenorphine, disulfiram), (d) use of only roll-your-own cigarettes, (e) use of illicit substances in the past month (except for marijuana), (f) changes in psychotropic medications within the past 4 weeks, (g) women: pregnant, breastfeeding, or not using effective birth control, (h) unable to understand informed consent.

Study 2 (adolescent smokers).

Inclusionary criteria were: (a) being 15–19 years old, (b) reporting smoking ≥1 cigarette/day on 28 of the past 30 days (“Out of the past 30 days, on how many days did you smoke?”) during telephone screening, (c) self-report of daily smoking for at least the last six months, (d) being able to read and write in English. Exclusionary criteria were: (a) seeking treatment for smoking or planning to quit in the next 30 days, (b) reporting daily alcohol or drug use (daily marijuana use permitted), (c) pregnancy.

Procedures

All procedures were approved by the Brown University Institutional Review Board (1303000809 - Very low-nicotine cigarettes in smokers with SUD: Smoking, substance use effects; 1404001032 - Evaluation of very low nicotine content cigarettes in adolescent smokers). Informed consent was obtained from all adult participants; parental consent and participant assent were obtained from participants under 18. The data were collected via self-report during an in-person baseline assessment session as part of two larger studies on the effects of smoking cigarettes with reduced nicotine content among adults and adolescents, respectively. Participants completed a telephone screening to determine initial eligibility.

Assessments

Descriptive information.

Demographics include age, sex, race, ethnicity, income, and education.

Cigarette demand.

The BACD was always administered after the CPT to ensure that the CPT data, collected to test other hypotheses, were not affected. The adult study utilized paper versions whereas the adolescent study utilized electronic versions that mirrored paper versions (i.e., the entire measure was presented on the screen and participants continued through all items on the tasks regardless of previous answers). On both tasks, participants were asked to “Think about HOW YOU ARE FEELING RIGHT NOW” and to indicate hypothetical smoking behavior. They were told “The available cigarettes are your usual brand. You have the same income/savings that you have now, and NO ACCESS to any cigarettes or nicotine products other than those offered at these prices. You can smoke without any restrictions and without factoring in what might occur in the next 24 hours related to your participation in the study. You would smoke the cigarettes that you request at this time, not save or stockpile cigarettes for a later date.” Both of the following two approaches began with the above instructions.

Cigarette Purchase Task (CPT) (Jacobs & Bickel, 1999; MacKillop et al., 2008) asked participants to estimate how many cigarettes they would smoke at 17 prices ranging from $0-$5/cigarette ($.00, $.02, $.05, $.10 to $1 in $.10 increments, $1 to $5 in $1 increments). Three behavioral economic indices of demand were calculated: (1) intensity - the quantity of cigarettes consumed at zero ($0) cost, (2) O_max - the maximum amount of money allocated to cigarettes, and (3) breakpoint (BP) – the first price to suppress cigarette consumption to zero.

Brief Assessment of Cigarette Demand (BACD) was developed for the present study. Open-ended questions were generated to approximate each of the three calculated CPT indices of demand. As a proxy for intensity - “How many cigarettes would you smoke if they were free?” As a proxy for O_max - “What is the most you would spend on cigarettes in total?” Two different BP questions were tested to determine which would more closely approximate CPT responses: “What is the most you would pay for each cigarette?” (BP1) and “How much would each cigarette have to cost for you not to buy any?” (BP2).

Nicotine dependence.

Fagerström Test for Nicotine Dependence (FTND; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991) was used in Study 1; higher scores (from 0–10) reflect greater nicotine dependence.

Modified Fagerström Tolerance Questionnaire (mFTQ; Prokhorov, Pallonen, Fava, Ding, & Niaura, 1996) is an adaptation of the FTND for use in adolescents that was used in Study 2; higher scores (from 0–9) reflect greater nicotine dependence.

Cigarettes per day.

The timeline follow-back interview (TLFB; Robinson, Sobell, Sobell, & Leo, 2014) was conducted in person to assess cigarettes per day during the last seven days for use in analyses.

Carbon monoxide.

Exhaled carbon monoxide (CO) was assessed via a Bedfont Micro Smokerlyzer.

Data Analysis Approach

Correlations of BACD with CPT responses and correlations of BACD and CPT responses with smoking measures were conducted.

Preliminary analyses.

CPT demand indices were determined using an observed values approach (i.e., observed responses or arithmetic calculations based on responses). This was selected because it allows for a direct comparison of BACD and CPT indices (as both are based on participant report). We tested model fit using the exponentiated model (Koffarnus, Franck, Stein, & Bickel, 2015). For overall demand, the model provided an excellent fit among adults (R² = 0.95, k = 4) and adolescents (R² = 0.98, k = 2), and for individuals, the model provided a very good fit (adult median R² = 0.86; adolescent median R² = 0.91).

Evidence of low effort on the CPT (trend, bounce, reversals from zero (Stein, Koffarnus, Snider, Quisenberry, & Bickel, 2015) or BACD (willing to spend more on one cigarette than on all cigarettes), removed 1 participant from Study 1 (CPT) and 6 from Study 2 (4 CPT, 2 BACD) from subsequent analysis, resulting in ns of 80 and 81, respectively. Variables were screened for missing data, distribution abnormalities, and outliers. Outlying values (Z >3.29) for items and demand indices were Winsorized to one unit above the next highest value (Tabachnick & Fidell, 2007). All BACD and CPT indices were normally distributed.

Among participants for which no BP was reported on the CPT (Study 1 n = 11; Study 2 n = 28), BP was coded as one unit above the highest BP value ($6). The two BP items included on the BACD were examined with regard to their relationship with CPT BP; BP1 was selected for interpretation of results based on greater correspondence in both samples (see Supplemental materials).

Results

Descriptive information.

Table 1 shows sample characteristics¹ and mean responses for BACD and CPT demand indices. The CPT took 3.5 min to complete and the BACD took 1 min to complete.

Table 1.

Mean (standard deviation) of the BACD, CPT, and other participant characteristics

	Mean (SD) or %
	Adult study		Adolescent study
Male (%)	51.3	56.8
Race (%)
White	81.0	55.6
Black/African American	8.9	8.6
Asian	0.0	17.3
American Indian/Alaskan Native	2.5	3.7
Pacific Islander	0.0	1.2
Multi-racial	7.6	13.6
Hispanic or Latino (%)	2.5	14.8
Age	43.29 (11.24)	17.95 (0.93)
Years of Education	12.10 (1.70)	11.53 (1.25)
Income	-	$140.33/wk (143.22)*
$0–9,999/yr (%)	52.5%	-
$10,000–19,999/yr (%)	22.5%	-
$20,000–29,999/yr (%)	11.3%	-
$30,000–39,999/yr (%)	7.5%	-
$50,000–59,999/yr (%)	1.3%	-
$60,000–69,999/yr (%)	2.5%	-
$70,000 and over/yr (%)	2.5%	-
CPT Intensity	26.06 (12.76)	18.21 (10.76)
BACD Intensity	27.59 (14.02)	18.49 (11.88)
CPT Breakpoint	$2.18 (1.91)	$3.40 (2.17)
BACD Breakpoint	$1.49 (1.62)	$1.16 (0.88)
CPT OMAX	$12.61 (5.84)	$10.05 (9.07)
BACD OMAX	$14.39 (7.20)	$12.55 (6.01)
Nicotine Dependence	6.08 (1.93)	3.93 (1.63)
Cigarettes/Day	19.96 (8.67)	8.12 (6.29)
Carbon Monoxide	17.08 (8.73)	10.57 (7.10)

Note. BACD = brief assessment of cigarette demand; CPT = cigarette purchase task; intensity = intensity of demand (quantity desired at zero cost); O_max = maximum amount of money allocated to cigarettes; breakpoint = the first price to suppress cigarette consumption to zero;

^*sum of money received weekly from jobs, parents/guardians, and other sources.

Associations between BACD and CPT indices.

Comparing CPT and BACD for the three indices of demand (Table 2) revealed a significant positive association in both adults and adolescents, respectively, for BP (r = .23, p = .04; r = .33, p = .003)² and Intensity (rs = .86, .97, ps < .001), whereas O_max was only significantly correlated in the adult sample (r = .43, p < .001). The magnitude of within-measure associations between indices tended to be stronger on the CPT (rs = .02–.62) than on the BACD (rs = .07–.29).

Table 2.

Correlations between BACD and CPT indices of demand and with smoking measures

Measure	Index	Adult study						Adolescent study
		CPT			BACD			CPT			BACD
		BP	INT	OMAX	BP	INT	OMAX	BP	INT	OMAX	BP	INT	OMAX
CPT	INT	0.02						0.37**
	OMAX	0.44**	0.47**					0.62**	0.55**
BACD	BP	0.23*	0.03	0.10				0.33**	0.10	0.21
	INT	−0.05	0.86**	0.34**	0.07			0.33**	0.97**	0.51**	0.09
	OMAX	0.35**	0.18	0.43**	0.22	0.11		0.23*	0.22	0.09	0.29**	0.27*
Nicotine Dependence		0.20	0.57**	0.50**	0.06	0.51**	0.16	0.26*	0.54**	0.54**	0.10	0.59**	0.08
Cigarettes/Day		0.07	0.67**	0.51**	−0.10	0.66**	0.11	0.11	0.53**	0.46**	0.09	0.51**	0.06
Carbon Monoxide		0.04	0.31**	0.16	0.04	0.37**	0.11	0.13	0.22*	0.28*	0.03	0.22*	0.04

Note. BACD = brief assessment of cigarette demand; CPT = cigarette purchase task; INT = intensity of demand (quantity desired at zero cost); O_max = maximum amount of money allocated to cigarettes; BP = breakpoint (the first price to suppress cigarette consumption to zero);

^**p < 0.01;

^*p < 0.05 level.

Associations between demand indices and smoking measures.

Associations between smoking behavior and demand (Table 2) showed similar patterns when assessed using the BACD and CPT on BP (lack of significant associations on both) and Intensity (presence of significant associations on both), however O_max tended to be significantly associated with nicotine dependence and smoking in both samples when assessed with the CPT but not the BACD.

Discussion

Initial results regarding a brief assessment of cigarette demand were generally mixed, consistently supporting some hypotheses but not supporting others. As predicted, there were significant positive associations between intensity on the CPT and BACD. The direction and magnitude of the relationships between intensity and smoking were comparable for the BACD and CPT. As predicted, there were significant positive associations between breakpoint on the CPT and BACD, however the strength of the relationships was modest. In general, measures of smoking were not significantly associated with breakpoint on the CPT or BACD. This is similar to findings that two brief breakpoint measures and CPT breakpoint were not associated with number of cigarettes smoked per day, but CPT and brief breakpoint measures were associated with each other (Athamneh et al., 2018). There was a moderately strong, significant association between O_max on the CPT and BACD in adult smokers, but not adolescents. Contrary to prediction, measures of smoking that were significantly associated with O_max on the CPT were not significantly associated with O_max on the BACD. Therefore, our data and the study of Athamneh et al. (2018) show the promising value of brief measures of breakpoint and intensity but additional work is needed, particularly for O_max.

The CPT may provide a cognitive decision making framework that generates more orderly responses than achieved with a brief assessment of demand. Similarly, these orderly responses may contribute to the associations observed between demand indices on the CPT that were of smaller magnitudes on the BACD. Associations between O_max and smoking observed on the CPT but not on the BACD may also be related to the orderly nature of the CPT task. If two individuals were equally sensitive to price (i.e., decreased purchases at the same rate as a function of price increase), and there was high correspondence between hypothetical CPT purchases and actual smoking behavior, the calculated maximum amount spent on the CPT (O_max) would, necessarily, be higher for the individual who smokes more cigarettes/day. This could account for the associations detected between nicotine dependence and daily smoking with O_max for the CPT but not the BACD. Discrepancies observed between O_max on the BACD and CPT may also be a function of BACD relying on explicit cognition (i.e., thinking about how much one is willing to spend deliberately) rather than implicit cognition when completing the CPT (i.e., reflexive response to each price without considering total resultant spending). Indeed, because many individuals are unaware of how much money they are spending on a substance, providing this calculus may be a point of intervention (Colby et al., 2005; Miller et al., 2013). Thus, asking about O_max directly does not appear to generate equivalent data, whether this is due to lack of specific knowledge as to how much is being spent, the orderly nature of the purchase task increasing the likelihood that those who smoke more heavily will report greater maximum expenditure, or other issues with question language or framing cannot be determined from the current data.

Similarly, it is unclear why there were discrepancies between samples with regard to O max. There are many differences between adolescents and adults with regard to smoking patterns, access, dependence, etc. Exploratory analyses showed that in both samples, age and motivation/readiness to quit were not associated with any of the BACD demand indices. Associations of age with the CPT varied by sample, with age associated with intensity only in adults (r = −.30, p = .007), but with breakpoint (r = −.26, p = .02) and O_max (r = −.24, p = .03) in adolescents. Motivation to quit was not associated with any CPT indices in adults and associated with O_max only in adolescents (r = −.22, p = .04). We explored whether lack of experience purchasing cigarettes among adolescents may have contributed to findings, but 88.5% of the adolescent sample reported purchasing cigarettes, making this an unlikely explanation. Thus, additional work is needed to determine whether modifying the BACD item assessing O_max will result in associations that more closely approximate the associations typically observed between smoking and O_max and whether revision may improve psychometric properties among adolescents.

This study had a number of strengths. It included three brief indices of demand rather than one, had verification of smoking using CO, and enrolled smokers from two vulnerable populations (i.e., adolescents and those with SUD), with a greater proportion of female participants, more racial diversity, and lower socioeconomic status (education and income) than recent work in this area (Athamneh et al., 2018). Accordingly, this makes the current sample more representative of smokers in the US. Nevertheless, as participants were vulnerable populations drawn from the northeastern US, they may not be representative in other ways. Another consideration is that the BACD does not allow for interpretation of the demand curve elasticity, as it is not possible to generate this. In addition, a 17-item CPT was used in both studies with a maximum price/cigarette of $5 and a considerable proportion of participants did not reach a breakpoint. This may have restricted the range of responses observed on the CPT relative to the BACD, as the BACD lacked similar constraints. Of note, the highest price of $5/cigarette [$100/pack] on the CPT was vastly higher than the average price of $9.93/pack in the state where the study was conducted (Boonn, 2018). Therefore, it is unlikely that local prices were responsible for individuals failing to reach a breakpoint. Additionally, the order of measures was not counterbalanced, completing the CPT may have influenced responses on the BACD. Additional work in this area would benefit from determining whether use of pre-set response choices rather than open-ended questions on the BACD improves psychometric properties, and whether there are differences in participant acceptability between the two measures.

In sum, although additional work is needed to improve psychometric properties before adoption, the current results suggest that for some (but not all) CPT indices, brief measures may generate similar data to more extensive purchase tasks. Development of brief measures of demand could increase the application and utility of these measures to advance the scientific understanding and treatment of addictive disorders.

Public significance statement:

Demand for cigarettes refers to an individual’s desire and willingness to pay for cigarettes across different prices. When demand was assessed using a newly developed brief measure of cigarette demand among adult and adolescents who smoke, several indices of demand were associated across methods, particularly among adults.