Addictive Behaviors, Vol. 15. pp. 319-322, Printed in the USA. All rights reserved.
0306-4603/90 $3.00 + .OO B 1990 Pergamon Press plc
THE COMPARABILITY OF TWO COMMONLY USED CARBON MONOXIDE ANALYSIS SYSTEMS: A TECHNICAL NOTE ROBERT C. KLESGES, MARY E. ANDERECK, EDDIE M. CLARK, LINDA H. ECK, and ANDREW W. MEYERS Center For Applied Psychological
Memphis State University
Abstract - Carbon monoxide (CO) in expired breath is a common strategy for verifying smoking abstinence. The purpose of this investigation was to determine the comparability of two carbon monoxide (CO) analyzers: the commonly used Ecolyzer Series 2000 analyzer versus the new, highly portable, and less expensive Ecolyzer Model 210. Fifty-eight CO readings were obtained from 29 subjects (17 smokers, 12 nonsmokers) and analyzed on both systems simultaneously. Results indicated that the analyzers yielded highly comparable values for both smokers and nonsmokers. Moreover, the correlation between the two units’ readings was very high (r = .92, p < ,001). There was, however, a significant analyzer by sex interaction with the Series 2000 analyzer yielding higher values for males. Overall, it was concluded that the Series 210 analyzer is a highly accurate and cost-effective instrument.
The health consequences and medical costs associated with cigarette smoking are well established. There are now more than 50,000 studies linking cigarette smoking to increased morbidity and mortality from cardiovascular diseases, various forms of cancer, and chronic obstructive lung diseases (United States Public Health Service [USPHS], 1985). It is estimated that annually in the United States smoking is causally related to 170,000 deaths from cardiovascular disease, 130,000 deaths from cancer, and 50,000 deaths from chronic obstructive lung disease (USPHS, 1979, 1988). Given the consistent demonstration of dose-dependent relationships between smoking and disease, evidence of reductions in health risks following smoking cessation, documentation of its addictive qualities, and experimental studies documenting carcinogenic effects of tobacco smoke in animals, few scientists question the causal nature of the relationship between smoking and illness (see Eysenck & Eaves, 198 1, for an exception). A critical methodological issue in both assessment and treatment outcome investigations concerns the evaluation of smoking status. Self-reports of smoking status are not considered a valid measure in treatment outcome studies (Pechacek, Fox, Murray, & Luepker, 1984), particularly investigations that utilize contingencies or incentives to increase cessation rates (Klesges, Cigrang, & Glasgow, 1987). A number of biochemical indices of smoking have been developed with the most common being carbon monoxide (CO) in expired breath, serum or saliva thiocyanate, and serum or saliva cotinine (Benowitz, 1983). Each index has its unique advantages and disadvantages. On balance, the most sensitive measure is probably cotinine, a direct metabolite of nicotine (Benowitz, 1983). Unfortunately, cotinine assays are expensive to analyze ($lS-$25 per sample), a saliva or serum sample is required of the subjects at each assessment, and immediate feedback regarding smoking status is not available.
This study was supported by two grants awarded by the National Heart, Lung, and Blood Institute: One awarded to the first author (HL No. 36553) and one awarded to both the first and fourth authors (HL No. 39332). Support was also received from Centers of Excellence grant awarded to the Department of Psychology, Memphis State University, by the state of Tennessee. Requests for reprints should be sent to Robert C. Klesges, Ph.D., Center For Applied Psychological Research, Department of Psychology, Memphis State University, Memphis, TN 38152. 319
ROBERT C. KLESGES
Given its low cost and high degree of discriminative validity (Pechacek et al.. 1984), CO assessments have become a very popular alternative to cotinine to verify self-reports of smoking and smoking abstinence. Despite its short half life (approximately 4 hours), and although its sensitivity in detecting small changes is smoking rate is questionable (see Pechacek et al., 1984, for a review), CO levels have shown to be an excellent biochemical measure of smoking/nonsmoking status. For example, Wald, Idle, Boreham, and Bailey (1981) reported the specificity (percentage of nonsmokers correctly identified) of CO was 99% while the sensitivity (percentage of smokers correctly identified) was 8 1%. Cohen and Bartsch (1980) reported specificity and sensitivity rates of 83% and 81% respectively. Despite the popularity of CO testing for verifying smoking self-reports, surprisingly little attention has been given to the different methods used to assess CO in expired air. To our knowledge, only one investigation has compared two methods of CO assessment. Irving, Clark, Crombie, and Smith (1988) compared the Ecolyzer Series 2000 (the traditional analyzer used for years) to the Bedfont EC50. Their results indicated that the two analyzers yielded highly comparable values (i.e., no reliable differences) and were highly intercorrelated (I = .96). The same company (Ecolyzer) that produced the standard laboratory analyzer has developed a lightweight, battery-powered, and highly affordable CO analyzer (viz.. Model 210). Although its cost (approximately $650) is considerably lower than the Model 2000 Ecolyzer (approximately $1995). nothing is known regarding the comparability of these two machines. Thus, the purpose of this investigation is to determine the comparability of these two devices in both smokers and nonsmokers. Additionally, it is known that both tidal and minute volume (i.e., volume of air exchange in and out of the lungs at a given point in time) vary as a function of both age and sex (Astrand & Rodahl, 1986). Because this may potentially have an effect on the volume of the air sample as well as the CO readings, the effects of age, sex, and smoking status (all important variables to the smoking research) on the accuracy of these machines will also be evaluated. METHOD
Twenty-nine subjects (17 smokers, 12 nonsmokers) were recruited to participate in an investigation that evaluated subtle behavioral and biologic changes in smokers versus nonsmokers. Sixteen of the subjects were male and 13 were female. Mean age of participants was 23.19 years (SD = 5.14; range 18-39), and smokers reported that their intake of cigarettes averaged 20.4 per day (SD = 8.79). Gender and age did not significantly differ among smokers and nonsmokers in this sample. Procedure
All subjects were seen three times a week (on Mondays, Wednesdays, and Fridays) for three weeks. On each occasion, CO readings were taken using the Ecolyzer Series 2000 Carbon Monoxide Analyzer (Energetics Science, a division of National Draeger, Park Ride, Illinois). On one of the nine visits, a second CO reading was randomly obtained using the Ecolyzer Model 210 (also manufactured by Energetics Science). That is, the subjects were randomly assigned a number from one to nine and assessed with both machines on the day that they were randomly assigned (i.e., Monday, week 1 = 1; Friday, week 3 = 9). The Series 2000, used by smoking researchers for over a decade (Hughes, Frederiksen, & Frazier, 1978) is a table-top, semi-portable, AC-powered instrument that retails for $1995. In contrast the Model 210 monitor is lightweight, battery-powered, and designed to wear on one’s belt. The Model 210 retails for approximately $650.00.
Carbon monoxide analysis systems
The use of both analyzers followed the collection procedure outlined by Hughes et al. (1978). At the beginning of each assessment session both machines were calibrated according to the manufacturer’s procedural recommendations. Subjects were asked to hold their breath for 20 seconds, then exhale half into the room and the remainder into a collection bag or balloon. This procedure is designed to better assess deep alveolar air. The air was exhaled through a small filter attached to either a plastic collection bag (Series 2000) or a balloon (Series 210). The breath sample was immediately attached to the CO monitor for analysis. Subjects rested for five minutes between assessments, and the samples were counter-balanced as to which analyzer was used first (i.e., Series 210 vs. 2000). The reading from the Ecolyzer 210 was then compared to the reading obtained that day by the Ecolyzer 2000. RESULTS Given the limited sample size in this investigation, assessment of moderator variables was limited to the assessment of two-way interactions involving a between-subjects factor (viz., age, smoking status, sex) and the within-subjects factor (viz., analyzer, Series 210 vs. 2000). Thus, separate repeated measures ANOVAs were conducted for age (using a median split), smoking status (smoker vs. nonsmoker), and sex as independent variable and analyzer as the dependent, repeated factor. Results indicated no significant differences between the Series 2000 (M = 15.79, SD = 13.46, Range 2-68) and the Series 210 (M = 14.48, SD = 14.45, Range 2-64). The correlation of the values produced by both machines was .92 @ < .OOl). Not surprisingly, a main effect for smoking status emerged, F( 1, 25) = 15.18, p < .002. Results also indicated no significant main effects for age or gender and no interactions between age and analyzer or smoking status and analyzer. However, there was a significant gender by analyzer interaction, F(1, 27) = 4.26, p < .05. Simple main effects analyses indicated that for males the Series 2000 analyzer yielded significantly higher (F(1, 27) = 5.50, p < .05, M = 19.38, SD = 15.45) readings than the Series 210 (M = 16.19, SD = 17.26). For females, however, the Series 2000 yielded similar readings (M = 11.38, SD = 9.27) as the Series 210 (M = 12.38, SD = 10.28, F(l, 27) = .44).r DISCUSSION The results of the current investigation indicate that Series 210 and the Series 2000 CO analyzers yield comparable values and significantly intercorrelate (I = .92, p< ,001). The results further indicate, however, that slightly different values may be obtained as a function of sex, depending on which analyzer is used. The highly comparable results obtained between the two units strongly suggest, for a variety of reasons, that the Series 210 analyzer be used in future smoking assessment and treatment investigations. The Series 210 is a lightweight, portable, and battery-operated analyzer, making it conductive for investigations conducted in worksite, community, school, or other nonclinic based investigations. The digital readout may increase precision over the metered display of the Series 2000 and may increase inter-rater reliability. Finally, the Series 210 is considerably less expensive (approximately $650) than the Series 2000 (approximately $1995). The two analyzers did, however, yield a slightly different pattern of results for males compared to females. That is, the Series 2000 yielded significantly higher values than the ‘The sample size was too small to enter all three independent variables (age, gender, smoking status) and the repeated dependent variable (analyzer). Nonetheless, when this analysis is conducted, an identical pattern of results is obtained except that the gender by machine interaction is only marginally different @ = .06).
ROBERT C. KLESGES
Series 210 in males, while the Series 2000 and the Series 210 yielded similar readings for females. These increased readings for the Series 2000 in males could be due to several factors. For example, the two machines use different collection procedures that may have yielded higher sample volumes for males relative to females on the Series 2000. We observed that males in our study consistently filled, and often overfilled (causing it to burst), the collection bag for the Series 2000. The Series 210, in contrast, uses a small plastic tube connected to a small balloon that has considerable resistance and overfilling did not occur with the Series 210. It is possible that the Series 2000 (and potentially both machines) may be affected by the volume of sampled air. Another possible reason for the increased values on the Series 2000 may be its metered display, which may be more difficult to read than the digital read-out of the Series 210. Future studies should evaluate these, and possibly other variables that may affect the reliability of carbon monoxide testing. In conclusion, it appears that the Series 210 analyzer displays convergent validity compared to a machine that has been shown to be valid (Hughes et al., 1978) and is a cost-effective method of assessing carbon monoxide in expired breath. However, these results must be tempered by the finding that a slightly different pattern of results is obtained for males across the two machines. The Series 210, as well as another highly promising and inexpensive CO analyzer (the Bedfont EC50; Irving et al., 1988) should make reliable CO verification of smoking status economically feasible for the vast majority of future smoking investigations.
REFERENCES Astrand, P., & Rodahl. K. (1986). Respiration. In D.B. Van Dalen. (Ed.), Textbook of work physiology (pp. 209-267). New York: McGraw-Hill. Benowitz, N.L. (1983). The use of biologic fluid samples in assaying tobacco smoke consumption. In J. Grabowski & C. Bell (Eds.). Measurement in the analysis and treatment of smoking behavior (pp. 6-26). NIDA Research Monograph No. 48. Rockville, MD: National Institute on Drug Abuse. Cohen, J.D.. & Bartsch, G.E. (1980). A comparison between carboxyhemoglobin and serum thiocyanate determinations as indicators of cigarette smoking. American Journal of Public Health, 70, 284-286. Eysenck, H.J., & Eaves. L.J. (1981). The causes and eflects ofsmoking. New York: Sage. Hughes. J.R., Frederiksen, L.W., & Frazier, M. (1978). A carbon monoxide analyzer for measurement of smoking behavior. Behavior Therapy, 9, 293-296. Irving, J.M., Clark, E.C., Crombie. I.K., & Smith, W.C. (1988). Evaluation of a portable measure of expired-air carbon monoxide. Preventive Medicine, 17, 109-I 15. Klesges. R.C., Cigrang, J., & Glasgow, R.E. (1987). Worksite smoking modification programs: A state-of-the-art review and directions for future research. Current Psychological Research & Reviews, 6, 2656. Pechacek. T.F.. Fox. B.H., Murray. D.M., & Luepker, R.V. (1984). Review of techniques for measurement of smoking behavior. In J.D. Matarazzo. S.M. Weiss, J.A. Herd, N.E. Miller, & S.M. Weiss (Eds.), Behavioral health: A handbook of health enhancement and disease prevention (pp. 729-754). New York: Wiley. United States Public Health Service. (1979). Health United States. U.S. Department of Health, Education, and Welfare, Public Health Services, Office of the Assistant Secretary for Health, Office on Smoking and Health. United States Public Health Service. (1985). The health consequences of smoking: Cancer and chronic lung disease in the workplace. A report of the Surgeon General (DHHS Pub. No. 85-50207). Rockville. MD: U.S. Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health. Wald. N.J., Idle, M.. Boreham, J.. & Bailey. A. (1981). Carbon monoxide in breath in relation to smoking and carboxyhemoglobin levels. Thorax, 36, 366-369.