Contingent reinforcement for reduced breath carbon monoxide levels: Target-specific effects on cigarette smoking

Vol. 10. pp. 345-349.1985 Printed in the USA. All rights reserved.

0306-4603/85 53.00 + .OO

AddictiveBehaviors.

Copyright * 1985 Pergamon Press Ltd

CONTINGENT REINFORCEMENT FOR REDUCED BREATH CARBON MONOXIDE LEVELS: TARGET-SPECIFIC EFFECTS ON CIGARETTE SMOKING MAXINE

L. STITZER

and GEORGE

E. BIGELOW

The Johns Hopkins University School of Medicine and Francis Scott Key Medical Center Abstract-This study determined the effects on smoking behavior of providing contingent reinforcement for nonsmoking versusreduced smoking afternoon breath carbon monoxide (CO) target levels. Twenty-eight hired chronic smoker volunteers were randomly assigned to one of three experimental conditions during a 1Gday intervention: (a) 8 ppm target CO, $5 per day incentive (n = 11); (b) 16 ppm target CO, S5 per day incentive (n = 8); or (c) 8 ppm target CO, no incentive (n = 9). Both payment groups showed significantly lower CO levels and greater amounts of daytime smoking reduction than the no-pay group. A specific effect of CO target was also seen; 45% of subjects in the 8 ppm group compared with 0% of subjects in the 16 ppm target and nepay groups produced average afternoon CO levels of 8.5 ppm or lower during the intervention. Average levels of CO and smoking reduction did not differ for the two paid groups, however, because some subjects in the 8 ppm group failed to reduce CO sufficiently to contact the reinforcer. Contingent reinforcement based on expired air CO levels can exercise powerful and precise (target-specific) control over smoking behavior. but there may be individual differences in ability to meet reinforcement contingencies if difficult targets are introduced abruptly.

Carbon monoxide (CO) concentration in expired breath provides a convenient objective measure of cigarette smoke exposure. Previous studies have shown that study volunteers will decrease their daytime smoking when offered monetary incentives for reduced afternoon carbon monoxide levels (Stitzer & Bigelow, 1982, 1983). In previous studies, where 50% reductions in afternoon CO levels have been reinforced, 50% reductions in CO levels have been obtained on the average, suggesting that subjects can regulate their daytime smoking with precision in order to produce the CO levels necessary for contingent reinforcement. The purpose of the present study was to compare the effects on smoking behavior of providing contingent reinforcement for reduced smoking versus nonsmoking afternoon CO levels. The study provides an opportunity to further explore the precision of smoking regulation when reinforcement is offered for different specific carbon monoxide reduction targets. METHODS

Subjects Twenty eight cigarette smokers, 25 females and 3 males, participated in two successive groups. These were employees of a large municipal hospital, primarily clerical and nursing personnel. Average age was 36 years (range = 19 to 61 years). Subjects reported an average smoking history of 20.2 years (range = 4 to 46 years) and reported using about 30 cigarettes per day (range = 10 to 46 cigarettes per day). Volunteers with CO levels below 18 ppm at initial screening were excluded. This research was supported by USPHS grant DA-03893 and Research Scientist Development Award K02-DA00050 from the National Institute on Drug Abuse. Requests for reprints should be sent to Maxine L. Stitzer, Behavioral Pharmacology Research: D-j-West. Francis Scott Key Medical Center. 4940 Eastern Avenue, Baltimore, MD 212X 345

346

MAXINE

L. STITZER

and GEORGE

E. BIGELOW

Prior to participation, subjects were told that this was not a program to help them quit smoking, but that optional changes in smoking might be encouraged during the study. Incentives for participation were free cigarettes and money earned for study completion. Study procedures

Subjects reported on Monday through Friday to a convenient location in the hospital between 3 and 5 p.m. to give a breath sample (following 20 seconds of breath holding) which was analyzed immediately for CO using an Ecolyzer (Energetic Science model 2180). At these study contacts, subjects received a supply of their own brand of cigarettes and turned in cards on which they had recorded the time of day each cigarette was smoked during the previous 24 hours. During the first and last 2 weeks of the 6 week study, subjects were instructed to smoke in their normal fashion, and baseline data were collected. On Friday of Week 2, all subjects received a notice that encouraged them to cut down their smoking as much as possible during the next 2 weeks and to quit smoking entirely if they could. At this time subjects were stratified on average baseline CO level (20-29 ppm, 30-39 ppm, 40-49 ppm, and ~50 ppm) and randomly assigned to one of three experimental interventions: (a) No pay (n = 9); subjects were told that the target CO reading that they should try to achieve during the next 2 weeks was 8 ppm or lower but no contingent incentive was offered. (b) Nonsmoking target (n = 11); subjects were told that they could earn $5 per day for providing afternoon CO readings at or below 8 ppm. (c) Reduced smoking target (n = 8); subjects were told that they could earn $5 per day for providing afternoon CO readings at or below 16 ppm. During the IO-day intervention, payment for samples at (within 0.5 ppm) or below the specified target was made immediately after analysis of the afternoon breath sample. Data analysis

Four measures were used: (a) carbon monoxide levels obtained at the afternoon study contact; (b) daytime cigarettes recorded from the time of arising in the morning to the time of the afternoon study contact; (c) evening cigarettes recorded from the afternoon study contact until the next morning; and (d) time since the last cigarette, which was the time elapsed between collection of the breath sample and recorded time of the most recent cigarette. Data analysis utilized the average values obtained for each subject on each smoking variable during each 2-week phase of the study. RESULTS

Figure 1 shows the distribution of average CO values by study group during the IO-day experimental intervention. In the no-pay group, 7 of the 9 subjects (78%) had average CO values above 24 ppm, two (22%) had an average CO value between 17 and 24 ppm, and none had average CO values below 17 ppm. In the 16 ppm target group, one subject .had an average CO level above 17 ppm, whereas 7 of 8 subjects (87.5%) had average CO values between 9 and 16.5 ppm; these latter subjects earned monetary reinforcement on 7 or more of the 10 opportunities. in the 8 ppm group, five of 11 subjects (45.5%) had average afternoon CO values of 8.5 ppm or below and earned the available reinforcers on 6 or more occasions; one of these subjects quit smoking entirely at the start of the intervention and remained abstinent for the rest of the study, Two subjects (18%) cut down their smoking appreciably, produced average CO levels between 9 and 16.5 ppm during the intervention, and earned payment on 5 of the 10 opportunities. Two other subjects produced average CO read-

Reinforcement for reduced carbon monoxide

TARGET:

TARGET:

NO

AVERAGE

347

8 PPm

16

PPm

PAY

CO

(ppm)

Fig. 1. Distribution of average CO values during a IO-day contingent reinforcement intervention is shown for three study groups.

ings of 17-24 ppm; one received payment on 3 occasions, the other on 0 occasions. The remaining two subjects in the 8 ppm group did not cut down their smoking at all, had average CO levels above 24 ppm, and never received reinforcement. The percent of subjects with average CO levels at or below 8.5 ppm was significantly greater in the 8 ppm target group (45.5%) than in the 16 ppm (0%) or no-pay (0%) groups (2 = 3.0, p < .003). Average baseline CO values were about 40 ppm for each group. During the intervention, average CO values were 32.9, 14.4, and 20.2 ppm for the no-pay, 16 ppm, and 8 ppm target groups, respectively (analysis of covariance F = 8.52, df = 2,24p < .002). Average CO values returned toward baseline for all groups following withdrawal of the intervention. During baseline, subjects in each group reported smoking about 15 daytime cigarettes on average. During the intervention, average reported daytime cigarettes were 10.7, 6.0, and 5.8 for the no-pay, 16 ppm, and 8 ppm target groups, respectively (F = 4.49, df = 2,24, p < .03). During baseline, average time since the last cigarette

348

MAXINE

L. STITZER

and GEORGE

E. BIGELOW

was about 40 minutes for each group. During the intervention, average time since the last cigarette was 83, 369, and 432 minutes for the no-pay, 16 ppm, and 8 ppm target groups, respectively (F = 7.18, df = 2,24, p < .004). Post hoc tests revealed significant differences between the no-pay group and each of the contingent reinforcement groups on average CO and daytime smoking measures, but the 8 ppm and 16 ppm target groups did not differ from each other. Average reported number of evening cigarettes during the intervention was 13.3, 15.7, and 10.5 for the no-pay, 16 ppm, and 8 ppm groups respectively (F = 2.77, df = 2,24, p < .09). These values represent a reduction from baseline of 0.5, 1.3 and 4.0 cigarettes per day in the no-pay, 16 ppm, and 8 ppm target groups respectively. Pearson correlation coefficients between average baseline CO and intervention CO values were .82 and .89 for the 8 ppm and 16 ppm groups, respectively; both were significant at p < .Ol. In contrast, correlations between baseline total cigarettes per day and average intervention CO values were .32 and .25 for the 8 ppm and 16 ppm groups, respectively; neither of these was significant. DISCUSSION

Contingent payment for CO reduction clearly had a substantial impact on afternoon carbon monoxide levels of chronic smokers. Thus, for example, 74% of subjects in the two contingent pay groups produced average CO readings of 16.5 ppm or below during the intervention compared to 0% of subjects in the no-pay group. The magnitude of this differential response in subjects exposed to pay and no-pay conditions attests to the potency of contingent payment procedures for motivating hired research volunteers to change their smoking behavior. Effects of the specific CO target reinforced were seen primarily in the distribution of average CO levels during the intervention (Figure 1). Specifically, 45% of subjects paid to achieve an 8 ppm target had average afternoon CO levels of 8.5 ppm or lower during the intervention whereas none of the subjects paid to reach a I6 ppm criterion had average CO levels this low. These group differences attest to the precision with which subjects can regulate their smoking in order to achieve carbon monoxide targets specified for contingent reinforcement. Although substantial CO reduction was seen with the 8 ppm CO target in many subjects, the nonsmoking target was insufficient to motivate total smoking cessation in any but one subject. Nor did payment for an 8 ppm target result in a greater average decrease in CO levels than did payment for a 16 ppm target. This was because of the diverse individual responses seen in the 8 ppm group. Specifically, two subjects showed no smoking reduction and two others failed to reduce their average CO levels sufficiently to consistently earn reinforcers. In contrast, all subjects in the 16 ppm group reduced their daytime smoking from baseline levels and all but one consistently earned the available reinforcers. Results for the 8 ppm group suggest that this target required too great an initial behavior change. It is possible that more subjects could be influenced to make large reductions in smoking if difficult CO targets were introduced gradually. As in previous studies of this type (Stitzer & Bigelow, 1982, 1983), reductions in afternoon CO levels were associated with decreases in the reported number of daytime cigarettes used and increases in cigarette abstinence time prior to the afternoon study contact. Also consistent with previous studies was the lack of compensation in evening smoking during portions of the study when subjects were making dramatic reductions in their daytime smoking. On the contrary, subjects who cut down their daytime smoking in the present study tended to make reductions in their evening smoking as well. A final point of consistency is that subjects returned to regular smoking when contingent reinforcers were withdrawn.

Reinforcement

for reduced carbon monoxide

349

A range of smoking rates and patterns was associated with afternoon CO levels of 8 ppm or lower in the seven subjects who achieved these levels on five or more occasions. One subject quit smoking altogether, three reportedly abstained from all daytime smoking but continued to smoke in the evenings, and three others reported using up to five cigarettes in the morning then abstaining for 5 to 8 hours prior to the afternoon study contact. This latter smoking pattern is consistent with results from a previous controlled study (Henningfield, Stitzer, & Griffiths, 1980) in which periodic carbon monoxide readings were taken after subjects smoked five consecutive cigarettes in the morning: Carbon monoxide levels fell below 8 ppm for most (five of eight) subjects after 5 hours of abstinence. These results suggest that it would be necessary to obtain multiple CO samples during the day and evening for verification and/or reinforcement of total smoking abstinence. An interesting relationship that emerged in this study was the high correlation between average baseline CO levels and average CO levels during motivated smoking reduction. This experimental correlation is consistent with previous observations that baseline CO levels are related to smoking cessation success (Vogt, Selvin, & Billings, 1979) such that light smokers are more likely to quit than are heavier smokers (Hughes, Hymowitz, Ockene, Simon, & Vogt, 1981). Number of cigarettes smoked, however, was a very poor predictor of smoking reduction outcomes in the present study. This argues for the potential utility of expired air carbon monoxide rather than reported number of cigarettes smoked as a screening measure to identify smokers who will have difficulty cutting down or quitting. Additional clinical implications of this study are limited by its focus on smoking reduction rather than cessation. Contingent reinforcement in the form of security deposit contracting has been effectively utilized to promote initial cessation among smokers enrolled in treatment programs (Paxton, 1980; Winett, 1973). Contingent reinforcement may also be useful for promoting cessation among smokers not seeking treatment or for maintaining long-term smoking abstinence, provided that appropriate objective measures of recent smoking behavior are available. REFERENCES Henningfield, J.E., Stitzer, M.L., & Grifflths. R.R. (1980). Expired air carbon monoxide accumulation and elimination as a function of number of cigarettes smoked. Addictive Behaviors, 5, 265-272. Hughes, G.H., Hymowitr, N.. Ockene, J.K., Simon, N., & Vogt, T.M. (1981). The multiple risk factor intervention trial (MRFIT). Preventive Medicine, 10,476-500. Paxton, R. (1980). The effects of a deposit contract as a component in a behavioral programme for stopping smoking. Behaviour Research and Therapy, 18,45-50. Stitzer, M.L., & Bigelow, GE. (1982). Contingent reinforcement for reduced carbon monoxide levels in cigarette smokers. Addictive Behaviors, 7, 403-412. Stitzer, M.L., & Bigelow, GE. (1983). Contingent payment for carbon monoxide reduction: Effects of pay amount. Behavior Therapy, 14, 647-656. Vogt, T.M., Selvin, S.. & Billings, J.H. (1979). Smoking cessation program: Baseline carbon monoxide and serum thiocyanate levels as predictors of outcome. American Jouma/ of Public Health, 69, 1156-l 159. Winett, R.A. (1973). Parameters of deposit contracts in the modification of smoking. Psychofogical Record, 23,49-m