Behavioral Economic Predictors of Brief Alcohol Intervention Outcomes

Abstract

Objective

The present study attempted to determine if behavioral economic indices of elevated alcohol reward value, measured before and immediately after a brief alcohol intervention, predict treatment response.

Method

Participants were 133 heavy drinking college students (49.6% female, 51.4% male; 64.3% Caucasian, 29.5% African American) who were randomized to one of three conditions: motivational interviewing plus personalized feedback (BMI), computerized personalized feedback intervention (e-CHUG), and assessment only.

Results

Baseline levels of alcohol demand significantly predicted drinks per week and alcohol problems at 1-month (demand intensity= maximum expenditure) and 6-month (relative discretionary expenditures on alcohol) follow-up. BMI and e-CHUG were associated with an immediate post-session reduction in alcohol demand (p < .001, ηρ² = .29) that persisted at the 1-month follow-up, with greater post-session reductions in the BMI condition (p = .02, ηρ² = .06). Reductions in demand intensity and O_max (maximum expenditure) immediately post-intervention significantly predicted drinking reductions at one-month follow up (p = .04, ΔR² = .02 & p = .01, ΔR² = .03, respectively). Reductions in relative discretionary expenditures on alcohol at 1-month significantly predicted drinking (p = .002, ΔR² = .06,) and alcohol problem (p < .001, ΔR² = .13) reductions at the 6-month follow-up.

Conclusions

These results suggest that behavioral economic reward value indices may function as risk factors for poor intervention response and as clinically-relevant markers of change in heavy drinkers.

Brief motivational interventions (BMIs) have been shown to be efficacious for reducing college drinking (Cronce & Larimer 2011). BMIs typically utilize a motivational interviewing style to deliver personalized feedback on an individual’s drinking or drug use. Motivational interviewing (MI) is a nonjudgmental, person-centered therapeutic approach that is specifically designed to address the ambivalence that a person may have about changing a behavior (Miller & Rollnick, 2013). Systematic reviews of this literature have concluded that although BMIs appear to be effective in reducing college student substance abuse, effect sizes are generally small to moderate (d = .11 −.40) and effects dissipate over time (Cronce & Larimer, 2011; Scott-Sheldon et al., 2014). More research is necessary to examine participant characteristics associated with response to BMIs (Carey, Scott-Sheldon, Carey, & DeMartini, 2007; Ewing, LaChance, Bryan, & Hutchison, 2009). This would allow for efficient matching of students to more vs. less intensive intervention approaches (Borsari, O’Leary Tevyaw, Barnett, Kahler, & Monti, 2007; Murphy et al., 2012), and might facilitate the development of additional intervention components that are designed to address identified risk factors (Feldstein Ewing et al., 2012; Schuckit, Kalmijn, Smith, Saunders, & Fromme, 2012; Turrisi et al., 2009). Another important next step in this literature is to identify dynamic, proximal markers that signal the likelihood of a successful intervention. Such markers might indicate the need for additional intervention elements. One putatively central indicator of change that has not been investigated is whether BMI outcomes can be predicted by the extent to which the intervention reduces the value of alcohol as a reinforcer.

Behavioral Economic Predictors of Treatment Change

Behavioral economic (BE) research has focused on studying patterns of drug and alcohol consumption as they develop and evolve over time in the context of changes in access to substances and other activities (Bickel, Johnson, Koffarnus, MacKillop, & Murphy, 2014; Hursh & Silberburg, 2008; Tucker, Roth, Vignolo, & Westfall, 2009). Behavioral economic theory predicts that the potential for substance use and related problems is highest when an individual values substance-related rewards more than the rewards associated with available substance-free activities. Considerable experimental evidence supports the importance of alternative reinforcers in the understanding of drug and alcohol use (Higgins, Heil, & Plebani Lussier, 2004). In preclinical animal models, the highest rates of substance use are present in contexts with the least availability of substance-free reinforcement, such as palatable food, exercise, enriched housing, and social access (Carroll, Ankera, & Perry, 2009) and similar results have been found in human laboratory and clinical studies (e.g., Bickel et al., 2014; Higgins et al., 2004). Individuals with problematic substance use report less reinforcement from nondrug activities compared to control participants (Audrain-McGovern, Rodriguez, Rodgers, & Cuevas, 2011; Correia, Carey, Simons, & Borsari, 2003; Correia, Simons, Carey, & Borsari, 1998; Van Etten, Higgins, Budney, & Badger, 1998), and there is evidence that both successful treatment and “natural” recovery (that occurs without treatment) are associated with increased engagement in substance-free activities (King & Tucker, 1998; Murphy et al., 2005, 2012; Rogers et al., 2008). In addition to considering the availability of alternative reinforcers, research has also examined participants’ response to these alternatives. A recent functional magnetic resonance imaging (fMRI) study found that individuals with drug dependence exhibited diminished neural activation to nondrug rewards (Lubman et al., 2009).

BE theory uses the term reward value or reinforcing efficacy to describe the relative level of motivation for a drug (Bentzley, Jhou, & Aston-Jones, 2014; Hursh & Silberberg, 2008). Following acquisition of use, the reinforcing efficacy of a given drug is theorized to both be a relatively stable product of the direct reinforcing effects of the drug and individual differences in decision making (e.g., delay discounting), but also to vary as a function of experiential and environmental states (e.g., craving, negative affect, the availability of alternative reinforcers; Amlung & MacKillop, 2014; Bickel et al., 2014; Rousseau, Irons, & Correia, 2011). There is some evidence that elevated alcohol reward value is associated with indices of substance use problems and dependence (Heinz, Lilje, Kassel, & de Witt, 2012; Skidmore, Murphy, & Martens, 2014) and predicts increasing drinking trajectories over time (Tucker et al., 2009; Tucker, Roth, Huang, Crawford, & Simpson, 2012), as well as poor response to intervention (MacKillop & Murphy, 2007; Murphy et al., 2005; Worley, Shoptaw, Bickel, & Ling, 2015), but to date it has not been measured as a dynamic marker of intervention response.

Behavioral economic indices of reward value

There are three general self-report approaches to measuring the reward value of alcohol that are all derived from behavioral economic laboratory drug self-administration models: alcohol purchase tasks, reinforcement survey instruments, and measures of discretionary monetary expenditures towards alcohol (Heinz et al., 2012; Murphy, MacKillop, Skidmore, & Pederson, 2009; Tucker et al., 2009). Alcohol purchase tasks (APTs) estimate alcohol reward value by generating demand curves that plot consumption as a function of price and identify how much someone would drink given unrestricted (free) access to alcohol, how much money they would spend on alcohol, and the extent to which their consumption level is price sensitive. These hypothetical tasks yield reliable and valid individual difference measures of reward value that are correlated with lab-based alcohol consumption and a variety of collateral indices of alcohol problem severity, including alcohol use disorder symptoms and craving (Amlung, Acker, Stojek, Murphy, & MacKillop, 2012; Kiselica & Borders, 2013). Alcohol demand has also been shown to increase acutely in response to experimentally induced elevations in craving (MacKillop et al., 2010) and stress (Amlung & MacKillop, 2014) and to decrease acutely following administration of the anti-craving medication Naltrexone (Bujarski, MacKillop, & Ray, 2012). In one study elevated demand at baseline predicted greater levels of typical weekly drinking six months following a brief intervention, even after controlling for baseline drinking and gender (MacKillop & Murphy, 2007). Thus, even within samples of regular drinkers there are individual differences in demand that may reflect the propensity for alcohol problems and drinking patterns that are relatively impervious to standard brief intervention approaches. Although laboratory research indicates that demand is sensitive to experimental manipulations, no study has examined the impact of a psychosocial intervention such a BMI on demand to date.

Reinforcement survey instruments define reinforcement as the product of activity frequency and enjoyment ratings, and addiction researchers have modified these measures to differentiate and quantify substance-related and substance-free reinforcement (Correia et al., 1998, 2003). Correia and colleagues (1998) found that a measure of the ratio of substance-related to total reinforcement was significantly predictive of substance use days. The brief alcohol intervention study described previously found that, similar to the demand curve measures of alcohol reward value, this relative substance-related reinforcement-ratio predicted drinking at six-month follow-up, even after controlling for baseline drinking (Murphy et al., 2005). Thus, participants who, prior to treatment, had a number of enjoyable alternatives to drinking were more likely to reduce their drinking following an intervention than participants with similar drinking levels but lower proportional reinforcement related to alcohol. Additionally, men and women who reduced their drinking post-intervention were examined and they showed increased proportional reinforcement from substance-free activities.

The Alcohol-Savings Discretionary Expenditure Index (ASDE) measures the difference between relative discretionary expenditures towards alcohol and relative discretionary expenditures towards savings. Savings is used in order to capture relative preference for alcohol (an immediate reinforcer) versus the delayed reinforcement associated with monetary savings. Larger ASDE values reflect greater valuation of alcohol relative to savings and have been shown to predict relapse following alcohol problem resolution, even after controlling for other significant measures of problem severity (Tucker et al., 2009, 2012). The ASDE, like the demand curve and reinforcement-ratio indices, may provide a clinically relevant index of alcohol value that is associated with greater problem severity and poor treatment response. Because young adults typically do not allocate significant amounts of money to savings, this measure has been modified in two studies with college students to reflect Relative total Discretionary Expenditures on Alcohol rather than discretionary expenditures on alcohol relative to savings (RDEA; Murphy et al., 2009), and this metric has shown significant relations with other alcohol reward metrics and with alcohol problems (Skidmore et al. 2014). Another recent study found that proportional baseline expenditures on drugs predicted response to opioid dependence treatment (Worley et al., 2014).

Taken together, these findings suggest that these three behavioral economic indices of alcohol reward value may capture clinically relevant individual differences in strength of desire for alcohol that are not redundant with standard measures of drinking or alcohol problems. The current study aimed to extend previous research indicating that elevated alcohol reward value at baseline is associated with poor BMI treatment response by including multiple reward value measures (demand, proportionate activity participation and enjoyment related to substance use, and relative discretionary expenditures related to alcohol), by evaluating the hypotheses that alcohol interventions would reduce alcohol valuation, and that the amount of reduction in alcohol valuation would predict subsequent drinking reductions (i.e., function as an dynamic marker of intervention response; Bentzley et al., 2014).

Method

Participants

This study is a secondary analysis of a previously published brief alcohol intervention trial (Murphy et al., 2010). A diagram of participant flow through the stages of this RCT is available as supplemental material. Participants were 133 first-year undergraduate students (49.6% women, 50.4% men) from a large public university in the southern United States. Eligible students were at least 18 years old and reported one or more heavy drinking episodes (5/4 drinks on one occasion for a man/woman) in the past month. The sample was ethnically diverse: 64.3% of participants identified as European American, 29.5% as African American, 2.3% as Hispanic/Latino, 2.3% as Native American, 0.8% as Asian, and 0.8% as Hawaiian. Participants reported drinking an average of 15.84 standard drinks during a typical week in the past month (SD = 13.57). The average age was 18.6 years (SD = 1.2).

Procedure

All procedures were approved by the university’s institutional review board. Data for this study were derived from a randomized trial that examined the effects of a BMI, Electronic Check-Up to Go (e-CHUG), and an assessment only condition (AO) on alcohol use and problems in heavy-drinking college students (see Murphy et al., 2010 for additional details). Participants were recruited from a required university-wide course where they completed a screening evaluation. Eligible participants completed baseline measures during an individual laboratory-based assessment appointment and were then randomly assigned to a condition using a random number table that was stratified by gender and ethnicity. Interventions were completed in the laboratory immediately after the baseline assessment. The BMI intervention consisted of a decisional balance exercise, personalized feedback delivered in MI style and goal setting, along with a discussion of harm reduction strategies if the student indicated interest. e-CHUG is an interactive web-based program that requires students to complete a brief assessment (6–7 minutes) that is used to generate personalized feedback about the student’s drinking. Students reviewed this feedback for at least 30 minutes. Follow-up assessments were conducted one and six months following the intervention. Additionally, indices of alcohol demand were collected immediately following the intervention for participants in the BMI or e-CHUG condition.

Measures

Alcohol consumption

The Daily Drinking Questionnaire (DDQ; Collins, Parks, & Marlatt, 1985) was used to measure alcohol consumption. Participants were asked to provide an estimate of the total number of standard drinks they consumed on each day during a typical week in the past month. The DDQ has been used frequently with college students and is a reliable measure that is highly correlated with self-monitored drinking reports (Kivlahan, Marlatt, Fromme, Coppel, & Williams, 1990).

Alcohol-related problems

Alcohol-related problems were measured using the Young Adult Alcohol Consequences Questionnaire (YAACQ; Read, Kahler, Strong, & Colder, 2006). The YAACQ is a 48-item self-report measure that dichotomously assesses alcohol-related consequences over the past 6 months. The YAACQ has demonstrated good test-retest reliability (r = .86 for the total YAACQ scale) as well as good validity with college students (Read, Merrill, Kahler, & Strong, 2007). Internal consistency for the YAACQ total score was .92 in our sample.

Relative discretionary expenditures on alcohol (RDEA)

RDEA was measured by calculating the proportion of discretionary money participants spent on alcohol relative to all discretionary money available. Participants were asked to respond to the following questions, “How much money did you have available to spend for non-essential items (e.g., clothing, music, entertainment, alcohol, eating in restaurants, going to the movies, etc.) during the past month? (Do not include money budgeted for essentials, such as rent, school books, gasoline, utility bills, groceries, etc.)” and, “How much money did you spend on alcohol in the past month (this includes any alcohol that you purchased, regardless of whether or not you consumed the alcohol)?” Past month alcohol expenditures were then divided by past-month discretionary funds available (scores ranging from 0 to 1). Larger values suggest greater relative valuation of alcohol and have been uniquely associated with alcohol problems (Skidmore et al., 2014).

Alcohol demand

An alcohol purchase task (APT; Murphy & MacKillop, 2006) was used to measure alcohol demand. APTs are simulation measures that assess self-reported alcohol consumption and financial expenditures across a range of drink prices. Participants report the number of standard drinks they would purchase and consume during a specified time frame (5 hours) at 17 price increments ranging from free ($0) to $20 per drink. The APT is used to generate both consumption and expenditure curves from which several measures of alcohol reward value can be derived. Because we were interested in examining changes in demand resulting from treatment, we focused on the two demand indices that have demonstrated the highest test-retest reliability and that we anticipated would be most sensitive to treatment: intensity and O_max (rs = 89 - .90 over two week interval, Murphy et al., 2009). Intensity is defined as the maximum level of consumption when drinks are free and O_max is the maximum expenditure generated after participants’ purchase responses are multiplied by drink price. These two APT metrics have demonstrated the most consistent pattern of associations with alcohol-related outcomes across studies (Kiselica & Borders, 2013; Skidmore et al., 2014).

Proportionate alcohol-related reinforcement (reinforcement ratio)

The Adolescent Reinforcement Survey-Substance Use Version (ARSS-SUV; Murphy, et al. 2005) is a measure of past-month reinforcement from substance-related and substance-free activities. Participants were provided with a list of 45 activities and asked to rate the frequency of participation and enjoyment associated with each of the activities during the previous 30 days. Each activity is rated twice; once reflecting times the participant was drinking and using drugs (substance-related frequency and enjoyment) and again for times the participant engaged in the activity without drinking or using drugs (substance-free frequency and enjoyment). Frequency ratings range from 0 (zero times per week) to 4 (more than once per day), and enjoyment ratings range from 0 (unpleasant or neutral) to 4 (extremely pleasant). The frequency and enjoyment ratings are multiplied to obtain a cross-product score that reflects reinforcement derived from the activity (Correia et al., 2003). Previous research has demonstrated that these methods provide reliable estimates of participation in activities that are consistent with observer reports of behavior (MacPhillamy & Lewinsohn, 1982). The reinforcement ratio (R-ratio) was calculated as substance-related reinforcement / (substance-free reinforcement + substance-related reinforcement), and this index has been shown to be uniquely associated with alcohol problems and change in drinking in multivariate models (Murphy et al., 2005; Skidmore et al., 2014).

Data analysis plan

To minimize the impact of outliers, values greater than 3.29 SDs above the mean on a given variable were changed to one unit greater than the greatest non-outlier value (Tabachnick & Fidell, 2006). There were one or fewer outlier values for all variables except for weekly drinking at 6-months for which there were 3 outlier values (2.7% of those who completed the 6-month follow-up). Variables that were skewed or kurtotic were transformed prior to analyses; typical weekly drinking, alcohol problems, intensity, O_max, and RDEA were transformed using a square root transformation and R-ratio was corrected using a log transformation.

A previous paper (Murphy et al., 2010) described one-month drinking outcomes and indicated a treatment advantage for BMI, but subsequent analyses showed no significant differences between the three treatment conditions at the 6-month follow-up (see Table 1). Changes in intensity and O_max were measured immediately post-session and changes in RDEA and R-ratio were measured at 1-month (behavioral and money allocation changes could not have taken place immediately post-session). The goal of the current paper is to evaluate three hypotheses. First, to examine the hypothesis that elevated baseline alcohol reward values would predict change in drinking and alcohol problems, we conducted a series of hierarchical regression analyses that controlled for gender, and baseline drinking and problems. Second, to assess the hypothesis that treatment would reduce alcohol reward value, we conducted repeated measures ANOVAs comparing baseline to post-session (for intensity and O_max only), and baseline to one-month alcohol reward values, while controlling for gender. Third we conducted a similar series of regression analyses that examined the hypothesis that the amount of change in reward value following treatment would predict change in drinking and problems. Analyses that examined baseline to post-session changes in demand included only those assigned to MI or eCHUG because participants assigned to the assessment-only condition did not complete a post-session assessment. All other analyses included participants from all three groups.

Table 1.

Associations among baseline drinking and behavioral economic variables

	1	2	3	4	5	6
1. Drinks per week	--
2. Alcohol Problems	.49***	--
3. R-ratio	.42***	.35***	--
4. Intensity	.56***	.48***	.40***	--
5. Omax	.51***	.37***	.44***	.67***	--
6. RDEA	.34***	.09	.26***	.39***	.41***	--

Note.

^*p< .05.

^**p< .01.

^***p< .001

R-ratio = Reinforcement Ratio, RDEA = Relative total Discretionary Expenditures on Alcohol, Intensity = maximum consumption level on the alcohol purchase task demand curve measure, O_max = maximum alcohol expenditure on the alcohol purchase task demand curve measure

Results

There were no significant treatment group differences on demographic, drinking, or BE variables at baseline. Follow-up rates were 89% (n = 118) at one-month follow-up and 83% (n = 111) at six-month follow-up, with no between-group differences in rates. There were no demographic or baseline drinking differences between follow-up completers vs. non- completers.

Associations among Alcohol Use and Behavioral Economic Variables

Typical weekly drinking and alcohol-related problems were significantly positively correlated with intensity, O_max, and R-ratio. RDEA was positively correlated with drinks per week, but not alcohol problems (See Table 1).

Changes in Behavioral Economic Variables by Condition

Means, standard deviations, and within-subjects effect sizes for BE variables are presented in Table 2. There was a significant immediate, post-intervention session reduction in intensity values (F(1, 87) = 34.97, p < .001, ηp² = .29), which was greater for individuals in BMI vs. e-CHUG (F(1,87) = 5.29, p = .02, ηp² = .06). There was also a significant reduction for O_max (F (1,88) = 36.15, p < .001, ηp² = .29), but no condition effect (See Figure 1).

Table 2.

Means, SDs, and within-subjects effect sizes for weekly drinking, alcohol problems and behavioral economic variables at baseline, post-session, one month follow-up and six month follow-up by condition.^†

	Baseline		Post-session		Baseline to post-session	1 month		Baseline to 1 month	6 month		Baseline to 6 month
Variable	Mean	SD	Mean	SD	d	Mean	SD	d	Mean	SD	d
Drinks per week
MI (n = 37)	14.61	14.62	-	-	-	9.43	11.84	.57	11.14	14.56	.29
e-CHUG (n = 35)	16.57	16.30	-	-	-	11.93	11.33	.42	15.20	15.14	.11
AO (n = 39)	15.44	11.02	-	-	-	14.99	11.34	.05	14.04	16.01	.13
Alcohol problems
MI (n = 37)	12.34	9.28	-	-	-	12.17	8.77	.03	9.70	8.81	.29
e-CHUG (n = 35)	11.74	7.62	-	-	-	11.29	8.40	.08	10.91	8.51	.10
AO (n = 39)	10.90	8.11	-	-	-	11.67	9.89	−.09	11.12	10.38	−.02
Intensity
BMI (n = 39)	9.74	7.17	7.08	5.56	.60	7.06	5.50	.62	-	-	-
e-CHUG (n = 38)	9.34	4.07	8.06	3.92	.46	8.68	5.32	.21	-	-	-
AO (n = 38)	10.08	6.24	-	-	-	9.26	4.63	.22	-	-	-
Omax
BMI (n = 38)	15.95	11.64	11.90	10.10	.59	11.73	11.76	.47	-	-	-
e-CHUG (n = 38)	17.60	10.66	14.44	10.19	.48	14.09	11.40	.41	-	-	-
AO (n = 37)	16.62	8.62	-	-	-	15.62	10.56	.14	-	-	-
R-ratio
BMI (n = 39)	.25	.18	-	-	-	.25	.15	.00	-	-	-
e-CHUG (n = 39)	.30	.18	-	-	-	.33	.19	−.11	-	-	-
AO (n = 34)	.30	.19	-	-	-	.31	.22	−.03	-	-	-
RDEA
BMI (n = 36)	.17	.22	-	-	-	.14	.17	.13	-	-	-
e-CHUG (n = 36)	.18	.24	-	-	-	.14	.17	.17	-	-	-
AO (n = 34)	.27	.30	-	-	-	.24	.25	.09	-	-	-

MI = motivational intervention, e-CHUG = electronic check-up to go, AO = assessment only; sample size for the one-month comparisons differed slightly than those for the six-month comparisons for MI, e-CHUG, and AO ns = 41, 38, and 39, respectively. Intensity = maximum consumption level on the alcohol purchase task demand curve measure, Omax = maximum alcohol expenditure on the alcohol purchase task demand curve measure, R-ratio = Reinforcement Ratio, RDEA = Relative total Discretionary Expenditures on Alcohol.

^†One-month drinking outcomes were originally published in Murphy et al. (2010).

Figure 1.

Change in Intensity and Omax from baseline to post-session by intervention condition. MI = motivational intervention, e-CHUG = electronic check-up to go, AO = assessment only. Intensity = maximum consumption level on the alcohol purchase task demand curve measure, Omax = maximum alcohol expenditure on the alcohol purchase task demand curve measure

Changes in hypothetical consumption on the APT from baseline to one-month follow-up, across the three treatment conditions, are presented in Figure 2. A series of paired-sample t-tests indicated that MI was associated with significant reductions in demand across 9 hypothetical price-points, e-CHUG participants decreased at 5 prices, and control participants had no significant changes. ANOVAs indicated reductions in intensity (F(1, 112) = 12.41, p = .001, ηp² = .10) and O_max values (F(1, 110) = 19.27, p < .001, ηp² = .15) across all groups, but no significant difference in these changes between groups. There were no significant changes in RDEA or R-ratio from baseline to one-month.

Figure 2.

Demand curves for hypothetical consumption on the APT for baseline and one-month by condition. Paired-samples t-tests revealed MI participants decreased at 9 prices, e-CHUG participants decreased at 5 prices, and control participants had no significant changes. *p > .05.

Predictive Utility of Behavioral Economic Variables

Baseline alcohol reward value variables were entered in the first block of these regression equations. Results for both baseline and change values are included in Table 3. Baseline intensity significantly predicted drinks per week and alcohol problems at one-month follow-up. Baseline RDEA predicted drinks per week and alcohol problems at six-months. Baseline values of O_max and R-ratio did not predict outcomes.

Table 3.

Regression results for predictive utility of changes in behavioral economic variables for typical weekly drinking and alcohol related problems at one and six month follow-ups.

	B	SEB	β	t	p-value	ΔR2
Change in intensity predicting substance use
1-month typical drinking
Baseline Intensity	.559	.201	.298	2.785	.007*	.020
Δ Intensity	−.085	.041	−.156	−2.092	.040*	.020
1-month alcohol problems
Baseline Intensity	.258	.158	.178	1.632	.107	.005
Δ Intensity	−.065	.036	−.154	−1.791	.078	.019
6-month typical drinking
Baseline Intensity	.561	.339	.222	1.655	.103	.007
Δ Intensity	−.113	.069	−.157	−1.621	.110	.019
6-month alcohol problems
Baseline Intensity	.437	.211	.270	2.069	.043*	.024
ΔIntensity	−.076	.051	−.166	−1.489	.141	.022
Change in Omax predicting substance use
1-month typical drinking
Baseline Omax	.198	.106	.163	1.871	.065	.007
Δ Omax	−.041	.016	−.180	−2.626	.011*	.030
1-month alcohol problems
Baseline Omax	.088	.078	.093	1.121	.266	.002
Δ Omax	−.036	.014	−.199	−2.585	.012*	.037
6-month typical drinking
Baseline Omax	.199	.176	.123	1.134	.261	.006
Δ Omax	−.038	.029	−.113	−1.304	.197	.012
6-month alcohol problems
Baseline Omax	.109	.107	.105	1.016	.313	.007
Δ Omax	−.024	.022	−.113	−1.102	.274	.012
Change in R-ratio predicting substance use
6-month typical drinking
Baseline R-ratio	.592	4.178	.017	.142	.888	.001
Δ R-ratio	.296	.904	.037	.327	.744	.001
6-month alcohol problems
Baseline R-ratio	4.593	3.098	.191	1.482	.142	.012
Δ R-ratio	−5.79	.687	−.107	−.843	.402	.006
Change in RDEA predicting substance use
6-month typical drinking
Baseline RDEA	9.979	2.582	.516	3.865	.000*	.026
Δ RDEA	−3.087	.954	−.396	−3.235	.002*	.060
6-month alcohol problems
Baseline RDEA	5.929	1.459	.491	4.063	.000*	.003
Δ ADE	−2.813	.598	−.577	−4.701	.000*	.134

PS = change in relevant variable from baseline to post-session, 1 month = change in relevant variable from baseline to one-month follow-up

Baseline to post-session changes in demand predicting drinking

Greater pre-post decreases in intensity and O_max were associated with fewer drinks per week at one-month follow up. For alcohol problems at one-month, there was a trend-level finding in the same direction for change in intensity, and change in O_max was a significant predictor in the expected direction. Post-session change in demand did not predict six-month outcomes.

Baseline to one-month changes in RDEA and R-ratio predicting drinking

Greater baseline to one-month decreases in RDEA were associated with fewer drinks per week and alcohol problems at the six-month follow-up, but change in R-ratio did not predict outcomes.

Discussion

The goal of this study was to evaluate several behavioral economic hypotheses related to treatment response following a brief alcohol intervention. We administered three measures of the relative valuation of alcohol that were derived from basic laboratory approaches to measuring drug reward value (Bickel et al., 2014). Although they are empirically distinct, all can be viewed as indexing the relative strength of alcohol as a reinforcer (Hursh & Silberberg, 2008; Skidmore et al., 2014). We hypothesized that elevated alcohol reward value at baseline would predict poor response to intervention, that a BMI would decrease alcohol reward value, and that this change would predict subsequent drinking reductions. Our findings provided partial support for these hypotheses and suggest that there is differential utility among the three measurement approaches in indexing change in alcohol reward value and predicting drinking outcomes.

First, consistent with previous research (Mackillop & Murphy, 2007; Murphy et al., 2005; Tucker et al. 2009; Worley et al., 2014), elevated baseline levels of alcohol demand and relative discretionary expenditures related to alcohol (RDEA) predicted poor response to BMI in multivariate models. Although effect sizes were small and there were inconsistencies in the prediction of one versus six-month drinking change, in general these results suggest that elevated alcohol reward value is a risk factor for persistent drinking following intervention. Second, there were measurable reductions in alcohol demand immediately following the intervention, and a clinician-delivered BMI was associated with greater and more enduring (out to one month) reductions than a computerized intervention. This suggests that, consistent with behavioral economic theory and laboratory research (Amlung & MacKillop, 2014; MacKillop et al., 2011; Rousseau et al., 2011), alcohol reward value is a dynamic construct that is sensitive to environmental contingencies (i.e., intervention), and may function as a marker of intervention response. Participants also reduced the relative amount of money they allocated to alcohol in the month after treatment, and although the amount of reduction predicted subsequent six-month levels of alcohol problems, there was no effect for treatment condition on this variable (control group participants reduced at the same level). In general, the amount of change in demand and RDEA following treatment was a more robust predictor of future drinking than baseline level of these variables, suggesting that researchers and clinicians who are interested in examining this construct as a predictor of future drinking should consider repeated measurement before and after an intervention or another event that is expected to change alcohol valuation.

The hypothetical demand curve approach to measuring alcohol reward value may be an especially practical and clinically useful index because of its brevity and acute sensitivity to proximal state changes in alcohol reward related to factors such as stress (Amlung & MacKillop, 2014), craving (MacKillop et al., 2010), naltrexone (Bujarski et al., 2012), and, as the current results suggest, brief alcohol interventions. The current results provide preliminary evidence that reductions in alcohol demand may provide a marker for successful treatment, and that lack of reductions may indicate the need for additional intervention elements. Clinicians can easily administer and score an alcohol purchase task prior to and after (or over the course of) treatment to gauge dynamic changes in a patient’s strength of desire for alcohol. Similarly, researchers who are evaluating new intervention elements could use the alcohol purchase task to obtain an immediate proxy of initial treatment efficacy. The utility of the demand curve approach in predicting future drinking may be related to the fact that it a “clean” measure of an individual’s desired level of consumption, unconstrained by the myriad contextual features (access to alcohol, fear of sanctions, etc.) that might serve to influence actual young adult drinking during the relatively brief time-frame covered by most alcohol consumption measures.

Another advantage to the demand curve approach is its amenability to translational research on alcohol-related decision-making (Feldstein Ewing & Chung, 2013; Morgenstern, Naqvi, Debellis, & Breiter, 2013). Purchase tasks can be administered in fMRI scanners and a recent study indicates that there are unique brain regions responsible for decision making related to the initial peak consumption region of the demand curve relative to the higher price region where responses are more price sensitive (MacKillop et al., 2014). A recent translational study that examined change in demand for cocaine among rats injected with oxytocin found results that were highly similar to the current study; demand was a predictor of cocaine severity, and post-treatment changes in demand provided a dynamic marker of treatment response that predicted relapse (Bentzley et al., 2014). Thus, excessive demand may be a cross-species biomarker of addiction propensity and response to treatment. However, the current results suggest that although demand may be an ideal proximal (post-session) indicator of the efficacy of the intervention, change in actual alcohol-related expenditures (relative to other discretionary expenditures) during the month after treatment may provide a more enduring and larger effect-size predictor of longer-term outcomes (Tucker et al., 2012). In both cases resource allocation (whether real or hypothetical) appears to be a unique index of strength of desire for alcohol.

Surprisingly, neither baseline level nor change in relative behavioral allocation and enjoyment related to drug and alcohol use (R-ratio) predicted outcomes, nor did this variable decrease following treatment. This is inconsistent with previous research with college drinkers completing a similar BMI (Murphy et al., 2005) and with more recent results indicating that participants in weight loss treatment who reported greater increases in relative levels of participation in and enjoyment related to food-free compared to food-related activities lost more weight (Buscemi, Murphy, Berlin, & Raynor, 2014). More research is needed to delineate the relative utility of the three approaches to quantifying reward value. It is possible that non-treatment seeking young adult heavy drinkers with mild to moderate levels of alcohol problems may generally have reasonably high levels of substance-free activity participation and that change in their drinking may occur in a manner that is relatively independent of more global change in activity participation. The reinforcement ratio may be more useful as a predictor and marker of change in more severe substance abusing populations, for whom a shift in activity participation away from alcohol use and towards reinforcing social and leisure substitutes may be a more salient mechanism of change (King & Tucker, 1998; Rogers et al., 2008). It is important to note that the null results with the reinforcement ratio variable may also be related to imprecise measurement rather than to the lack of relevance of the construct to change in drinking. Whereas it is relatively easy to quantify hypothetical demand and recent monetary expenditures, there is presumably a large degree of measurement error in the retrospective measure of activity participation and enjoyment. Accurate measurement of activity participation and enjoyment may require a more intensive prospective measurement approach.

Strengths of the study include the inclusion of multiple theoretically-based and translational measures of alcohol reward value administered in the context of a randomized controlled trial with high risk young adult drinkers, as well as the measurement of change in demand immediately post-intervention session. The application of a novel theoretical approach (behavioral economics) to study the change process related to brief alcohol interventions is also a strength. In addition to the aforementioned measurement issues, other limitations include the relatively small sample size, the use of self-report measures of alcohol use, and the relatively brief follow-up period. Although our results indicated that an in-person MI reduced demand slightly more than a computerized intervention, and that change in demand predicted outcomes, we were not able to test formal mediation analyses due to the small sample size. Nontheless, these results suggests that changes in alcohol-related demand may occur as a result of a variety of intervention methods, including assessment. These results also suggest that reductions in alcohol reward value may be an important mechanism of change that is not specific to treatment yet may portend more enduring drinking reductions. Future research should replicate and extend these results using larger samples, more potent interventions, and a measurement approach that might allow for more sensitive measurement of changes in alcohol reward value and drinking. Because all participants in the current study reported recent heavy drinking, Future studies should examine the predictive utility of alcohol reward value for those with a greater range of drinking levels. There is also a need for additional laboratory research that investigates potential treatment-related mechanisms that might reduce demand, particularly in the context of stressors or other factors that might increase demand and potential risk for relapse or risky drinking (Amlung & MacKillop, 2014; Bujarski et al., 2012; Rousseau et al., 2011). In particular, research that studies intervention response using novel behavioral and neuroimaging based approaches to measuring dynamic changes in alcohol reward valuation have the potential to advance our understanding of mechanisms of treatment response (Morgenstern et al., 2013).

In conclusion, the present results add to a growing literature investigating processes that are relevant to changes in alcohol use following brief interventions (Magill et al., 2014). Previous research indicates that successful BMIs increase use of protective behavioral strategies (Kulesza, McVay, Larimer, & Copeland, 2012), reduce estimates of how much other students drink (Merrill, Carey, Reid, & Carey, 2014), increase alternatives to alcohol use (Murphy et al., 2005, 2012), enhance self-efficacy to moderate drinking (Black et al., 2012), and elicit client language associated with change (Magill et al., 2014). The current results suggest that brief alcohol interventions also reduce alcohol reward value, and that change in alcohol reward value predicts subsequent drinking reductions.

Public Health Significance.

This study suggests that brief alcohol interventions reduce the reward value of alcohol. Measures of alcohol reward value can be administered following an intervention to determine the likelihood of subsequent drinking reductions and the need for further treatment.