- Email: [email protected]
Breath Carbon Monoxide as an Indication of Smoking Habit
Breath Carbon Monoxide as an Indication of Smoking Habit* Edward T. Middleton, B Med Sci; and Alyn H. Morice, MD
Study objective: To assess whether the breath carbon monoxide (CO) concentration can be used to determine a patient’s smoking habits in a respiratory outpatient clinic. Design: To provide a normal range for smokers and nonsmokers, 41 outpatients and 24 healthy subjects were questioned on their smoking habits and asked to provide two breaths into a CO monitor (EC50 Smokerlyser; Bedfont Instruments; Kent, UK). In a subsequent single-blind study, 51 different outpatients were not told of the purpose of the study and were assessed by extensive questionnaire, spirometry, and Smokerlyser estimation. Setting: The Chest Clinic and Pulmonary Medicine Department at the Northern General Hospital, Sheffield, UK. Participants: Phase 1 involved 41 outpatients attending the Chest Clinic and 24 nonoutpatient colleagues. In phase 2, an additional 51 different outpatients were studied. Measurements and results: The mean (SD) breath CO levels were 17.4 (11.6) parts per million (ppm) for smokers and 1.8 (1.3) ppm for nonsmokers (p < 0.001). A level of 6 ppm was taken as the cutoff, as this gave a selectivity of 96% and a sensitivity of 94% for outpatients. Of the 51 study patients, 5 admitted to smoking in the administered questionnaire. Eight denied smoking but had a mean breath CO > 6 ppm (7.5 to 42 ppm). Of these, three admitted to smoking after being explained the implication of the reading. Conclusions: Breath CO concentration provides an easy, noninvasive, and immediate way of assessing a patient’s smoking status. A reading > 6 ppm strongly suggests that an outpatient is a smoker. (CHEST 2000; 117:758 –763) Key words: carbon monoxide; equipment and supplies; monitoring outpatients; smoking Abbreviations: CO ⫽ carbon monoxide; COHb ⫽ carboxyhemoglobin; ppm ⫽ parts per million
a long time now, it has been known that F orsmoking is associated both etiologically and prognostically with numerous diseases of the respiratory system. However, despite this knowledge, many patients continue to smoke. Knowledge of a patient’s smoking habits is clinically important, for example as a risk factor or as a reason for failure to respond to treatment,1 and it enables appropriate antismoking advice to be given. It has been reported that active intervention, in the form of advice, significantly increased the number of patients who stopped smoking after 1 year.2 The same study also reported that *From the University of Sheffield (Mr. Middleton), Sheffield; and the University of Hull (Dr. Morice), Hull, UK. Performed at the Pulmonary Medicine Department, Northern General Hospital, Sheffield, UK. This study was supported by an educational grant from Bedfont Instruments, Kent, UK. Manuscript received April 29, 1999; revision accepted September 22, 1999. Correspondence to: Edward T. Middleton, B Med Sci, Academic Department of Medicine, Castle Hill Hospital, Castle Rd, Cottingham East Riding, HU16 5JQ, UK; e-mail: E.T.Middleton@ sheffield.ac.uk 758
the use of a carbon monoxide (CO) monitor to demonstrate an immediate and potentially harmful consequence of smoking increased further the number who complied with advice to quit.2 However, many smokers deny their habit, making appropriate counseling impossible.3 This may be because smoking is held in such poor regard by the health profession that patients are reluctant to admit to it, or they may be unwilling to admit to not doing as advised. This inaccuracy of self-reported smoking stresses the need for biochemical conformation of smoking status. For related material, see page 764 There are several methods of assessing smoking status. Nicotine, cotinine, or thiocyanate levels in the plasma or urine may be used to indicate smoking status.4 However, the blood tests are invasive and neither the blood nor the urine tests provide an immediate assessment. The measurement of breath CO level may provide an immediate, noninvasive Clinical Investigations
method of assessing smoking status. Furthermore, the development of relatively inexpensive portable CO monitors enables breath CO levels to be assessed in a wide variety of clinical settings. Following inhalation, CO displaces oxygen in the erythrocyte to form carboxyhemoglobin (COHb). In this form, CO has a half-life of about 5 to 6 h5,6 and may remain in the blood for up to 24 h depending on a number of factors, such as gender, physical activity, and ventilation rate.7–9 While some exposure to CO may occur in normal day-to-day life, due to environmental pollution, passive smoking, and occupational exposure, the most likely cause of high levels of exposure is smoking.10 Another cause of high levels of CO exposure is occult CO poisoning, for example, due to a faulty automobile exhaust system or home heating system, although this is only likely to be responsible for a minority of cases of raised breath CO levels. Finally, CO is an end product of RBC metabolism and is therefore raised in patients with acute hemolysis and following blood transfusions.11 The aims of this study were to use a portable CO monitor to compare the breath CO levels in established smokers and nonsmokers and to investigate factors that may affect breath CO levels. In a second phase of the study, we used the CO monitor to investigate the reliability of the claimed smoking status of patients attending a respiratory outpatients clinic.
on the day of the study. Informed consent was obtained, and the study was approved by the local ethics committee. The subjects were informed of the purpose of the Smokerlyser and were reassured that the results were confidential in order to encourage accurate reporting of smoking habits. Background information about their health, smoking habits, physical activity, and exposure to environmental pollution and passive smoke was obtained. Spirometry was then performed using the Escort Portable Spirometer (Vitalograph Ltd; Buckinghamshire, UK) to determine lung function. Finally, the subjects were asked to provide two breaths into the Smokerlyser as described above. Each subject was assessed using the Smokerlyser twice to check the repeatability of the test, and because a previous study has reported the second reading to be significantly higher than the first.12 In the second phase, outpatients who had not been involved in phase 1 were assessed in the same way. However, these subjects were told that we were measuring breath CO levels to monitor pollution exposure in Sheffield and were not informed that we could determine smoking status from the result. Using the values obtained in the first phase as a guide, we compared their claimed smoking status with the Smokerlyser reading. If there was a discrepancy between the subject’s claimed smoking status and their breath CO level, then the implication of the result was explained and they were asked to confirm their smoking status. Statistical Analysis All results were analyzed by the SPSS statistical package (SPSS; Chicago, IL). Descriptive statistics were used to examine the data and to assess the distribution. Depending on the distribution, a two-sample t test or a Mann-Whitney U test was used to test for a significant difference between the breath CO levels of smokers and nonsmokers and for a significant difference between outpatients and nonoutpatient smokers. A paired t test was used to test for a difference in CO levels between the two successive Smokerlyser assessments.
Materials and Methods Results
Breath CO Monitoring Breath CO monitoring was performed using the EC50 Smokerlyser (Bedfont Instruments; Kent, UK), an inexpensive, portable CO monitor that has previously been shown to be effective.12 The Smokerlyser measures breath CO levels in parts per million (ppm) based on the conversion of CO to CO2 over a catalytically active electrode. On breath holding, the CO in the blood forms an equilibrium with the CO in the alveolar air; therefore, there is a high degree of correlation between breath CO levels and COHb concentration.13 This enables the Smokerlyser to accurately estimate the blood COHb concentration from the breath CO level. The Smokerlyser was calibrated weekly using a mixture of 50 ppm CO in air. To standardize the breath being analyzed by the Smokerlyser, the subjects were asked to exhale completely, inhale fully, and then hold their breath for 15 s. If the subjects were unable to hold their breath for 15 s, they were asked to hold it for as long as possible and the length of time was recorded. Following the breath hold, the subjects were asked to exhale slowly into the Smokerlyser and were encouraged to exhale fully in order to sample the alveolar air. Study Design Outpatients and colleagues were approached without being forewarned about the study so they did not refrain from smoking
Breath CO levels were assessed in a total of 65 people: 41 outpatients attending a respiratory outpatient clinic (24 men; 16 smokers; mean [SD] age, 56.6 [13.7] years) and 24 nonoutpatient colleagues (14 men; 8 smokers; mean age, 29.1 [14.1] years). For the smokers, the mean reported daily cigarette consumption was 11.0 (7.7) cigarettes/d and 83% reported smoking on the day of testing. The second phase of the study involved an additional 51 different outpatients (28 men; mean age, 58 [12.1] years). Phase 1 The mean (SD) maximum breath CO level was 17.4 (11.6) ppm for smokers and 1.83 (1.29) ppm for nonsmokers (p ⬍ 0.001; Fig 1). The mean maximum breath CO level was significantly higher in outpatient smokers than nonoutpatient smokers, with values of 22.2 (10.7) ppm and 12.0 (11.3) ppm, respectively (p ⫽ 0.04; Fig 2). Of the 65 subjects, 48 were able to hold their breath for an equal period during CHEST / 117 / 3 / MARCH, 2000
759
Figure 1. Exhaled CO levels in smokers (n ⫽ 24) and nonsmokers (n ⫽ 41).
Figure 2. Exhaled CO levels in outpatient smokers (n ⫽ 16) and in nonoutpatient smokers (n ⫽ 8).
both assessments with the Smokerlyser. In these subjects, the CO level on the first test was significantly higher than that of the second test (mean difference between tests, 0.69 ppm; p ⬍ 0.001). The only significant correlation was between breath CO and the number of cigarettes smoked in the last 24 h (r ⫽ 0.546; p ⫽ 0.006; Fig 3). There was a trend between breath CO and the number usually smoked per day, but it was not quite statistically significant (r ⫽ 0.39; p ⫽ 0.063). Breath CO was not correlated with age, the length of time since the last cigarette, or lung function, as assessed by percentage predicted FEV1. 760
A CO level of 6 ppm was taken as the cutoff between smokers and nonsmokers, as it gave the highest sensitivity and selectivity. When smokers and nonsmokers were looked at as a whole, a cutoff of 6 ppm had a selectivity of 98% but a sensitivity of only 79%. However, as outpatient smokers had a significantly higher mean breath CO level, and as the second phase of the study only involved outpatients, the sensitivity and selectivity were recalculated only using the CO values obtained from outpatients. This gave a cutoff of 6 ppm, a sensitivity of 94%, and a selectivity of 96% for outpatients only. Clinical Investigations
Figure 3. Correlation between exhaled CO levels and the number of cigarettes smoked in the last 24 h.
Phase 2 Of the 51 outpatients involved in the second phase of the study, 5 admitted to smoking in the administered questionnaire and had maximum breath CO levels ranging from 9 to 28 ppm (mean, 17.3 ppm). An additional eight subjects denied smoking but had a maximum breath CO level ⬎ 6 ppm (range, 8 to 42 ppm; mean, 22.1 ppm). On explanation of the implication of the reading, three of these admitted to still smoking. The remaining five continued to deny smoking but could offer no other possible reason for the high Smokerlyser reading. Two of the five outpatients who continued to deny smoking despite high CO levels (15 ppm and 17 ppm, respectively) were patients pursuing claims for occupational ill health. The remaining 38 outpatients denied smoking and had maximum breath CO levels ranging from 1 to 4 ppm. Discussion This study supports a growing body of evidence12,13 that measuring breath CO levels provides an immediate, noninvasive, simple, and effective way of confirming a patient’s smoking status. A cutoff level of 6 ppm detected 94% of smokers and 96% of nonsmokers in a respiratory outpatient clinic. This is comparable with the work of Jarvis et al,4 who reported that the optimal cutoff was 8 ppm, giving 90% sensitivity and 89% selectivity. Crowley et al6 also reported that a breath CO level ⬎ 8 ppm was strongly associated with a self report of current smoking. CO levels can also be measured in the blood in the form of COHb, and Wald et al13
reported that a cutoff of 1.5% COHb had a sensitivity of 86% and a selectivity of 96.6%. A COHb level of 1.5% corresponds with a breath CO level of approximately 8 ppm13 and is thus comparable with the results of the present study. Despite all the evidence supporting the use of 6 ppm or 8 ppm as the cutoff, many studies using breath CO monitors have tended use 10 ppm (or 2% COHb) as the cutoff.14,15 This study, along with the previous studies, suggests that 10 ppm is too high to be a cutoff for a screening test, as it will reduce its sensitivity. Another problem of using a cutoff of 10 ppm arises because of the increased clearance of CO with an increased ventilation rate.9 Therefore, patients who hyperventilate or who do large amounts of physical exercise will have a lower-than-expected breath CO level and may avoid detection. This may particularly be a problem in the primary prevention of smoking related diseases, as healthy smokers may undertake regular physical exercise and thus have a relatively low breath CO level. A cutoff of 6 ppm instead of 10 ppm will be more likely to detect these smokers and enable appropriate antismoking advice to be given. The maximum breath CO values for nonoutpatient smokers were significantly lower than the values for outpatient smokers. This cannot be explained by the difference between the two groups in terms of age and lung function, as breath CO level was not correlated with either parameter in either this study or in previous studies.6 There was not a significant difference between the reported number smoked by each group (outpatients, 11.9/d; nonoutpatients, 9.2/d; p ⫽ 0.43). However, the large difference in mean breath CO levels suggests that the outpatients CHEST / 117 / 3 / MARCH, 2000
761
inaccurately reported the amount they smoke, possibly due to pressure to be seen as a light smoker. This may explain why a better correlation between the number smoked per day and breath CO was not seen. The breath CO level on the first assessment with the Smokerlyser was significantly higher than that of the second and may be due to an alteration in effort or lung function following the first test. In contrast, Jarvis et al12 reported that the second breath CO level was significantly higher than the first. In practice, any difference in repeated tests is likely to be of little clinical significance, as the difference was ⬍ 2 ppm in 96% of cases and did not alter the interpretation of the findings. We would recommend that a single Smokerlyser assessment should usually be sufficient, provided that there is adequate technique. Another difficulty with the Smokerlyser is that many patients struggle to hold their breath for 15 s, which may reduce their breath CO level. However, a short breath hold does not necessarily invalidate the test, as breath holds as short as 3 and 4 s were sufficient to produce breath CO levels of 18 ppm and 20 ppm, respectively, in two subjects tested. It is also worth briefly discussing the EC50 Smokerlyser itself. The manufacturer claims an accuracy of 2% and, over the levels of breath CO encountered in this study, this level of accuracy is more than adequate to differentiate smokers from nonsmokers. The instrument has a good reproducibility, with a change in signal of ⬍ 2%/mo. This was not a problem in this study, as the Smokerlyser was calibrated weekly; however, in clinical practice, it would be important to ensure the Smokerlyser was calibrated regularly. In most circumstances, the amount of CO in the breath is determined by the level of COHb in the blood. However, if the Smokerlyser is used in areas of high ambient levels of CO, a proportion of the breath CO may result directly from the inhaled air. The Smokerlyser will not be reliable under conditions of high ambient CO, but this situation is unlikely to be encountered in most clinical settings. Only one subject claimed to be a nonsmoker and yet had a CO level ⬎ 6 ppm. There was nothing in this subject’s history to suggest an excessive environmental exposure or an increased red cell turnover that could account for the raised breath CO, although occult CO poisoning could not definitely be excluded. There are two likely explanations for this result. Firstly, this subject may be a smoker who is denying his habit, thus demonstrating the potential benefits of the Smokerlyser. Secondly, it has recently been reported that breath CO levels may be raised 762
due to numerous inflammatory lung diseases, including bronchiectasis,16 asthma,17 and primary ciliary dyskinesia,18 with mean values of about 7 ppm being reported.16,18 Therefore, it is possible that this subject was genuinely a nonsmoker and had some degree of airway inflammation due to his COPD, which accounts for his raised breath CO levels. In light of this, for breath CO levels of between 6 ppm and 10 ppm, it may be prudent to confirm smoking status with a urinary cotinine measurement. Phase 2 of the study assessed the ability of the Smokerlyser to detect smokers who denied their habit (deceivers) in an outpatient clinic. Eight possible deceivers were detected, of which three admitted to smoking when confronted with the reading. Of the five remaining deceivers, four were almost certainly smokers, as their CO levels were ⬎ 10 ppm. Two of these subjects also had a financial incentive to conceal their habit. The subject with a CO level of 8 ppm was the only one for whom there was any doubt about smoking status, and it would have been desirable to confirm this subject’s smoking status with a urinary cotinine measurement. Conclusion The EC50 Smokerlyser provides a quick, simple, and inexpensive method of screening for smokers in the clinical setting. A breath CO level ⬎ 6 ppm indicates that an outpatient in a respiratory clinic is likely to be a smoker.
References 1 LeSon S, Gershwin ME. Risk factors for asthmatic patients requiring intubation: a comprehensive review. Allergologia et Immunopathologia 1995; 23:235–247 2 Jamrozik K, Vessey M, Fowler G, et al. Controlled trial of three different anti-smoking interventions in general practice. BMJ 1984; 288:1499 –1502 3 Daly RJ, Blann AD. Self-reported smoking in vascular disease: the need for biochemical confirmation. Br J Biomed Sci 1996; 53:204 –208 4 Jarvis MJ, Tunstall-Pedoe H, Feyerabend C, et al. Comparison of tests used to distinguish smokers from non-smokers. Am J Public Health 1987; 77:1435–1438 5 Peterson JE, Stewart RD. Absorption and elimination of carbon monoxide by inactive young men. Arch Environ Health 1970; 21:165–171 6 Crowley TJ, Andrews AE, Cheney J, et al. Carbon monoxide assessment of smoking in chronic obstructive pulmonary disease. Addict Behav 1989; 14:493–502 7 Deller A, Stenz R, Forstner K, et al. The elimination of carboxyhaemoglobin: gender specific and circadian effects. Infusther Transfusmed 1992; 19:121–126 8 Joumard R, Chiron M, Vidon R, et al. Mathematical models of the uptake of carbon monoxide on hemoglobin at low carbon monoxide levels. Environ Health Perspect 1981; 41:277–289 Clinical Investigations
9 Vesley A, Takeuchi A, Rucker J, et al. Effect of ventilation on carbon monoxide clearance in humans [abstract]. Am J Respir Crit Care Med 1999; 159:A767 10 Jarvis MJ, Tunstall-Pedoe H, Feyerabend C, et al. Biochemical markers of smoke absorption and self-reported exposure to passive smoking. J Epidemiol Community Health 1984; 38:335–339 11 Ronald F, Coburn MD. Endogenous carbon monoxide production. N Engl J Med 1970; 282:207–209 12 Jarvis MJ, Belcher M, Vesey C, et al. Low cost carbon monoxide monitors in smoking assessment. Thorax 1986; 41:886 – 887 13 Wald NJ, Idle M, Boreham J, et al. Carbon monoxide in breath in relation to smoking and carboxyhaemoglobin levels. Thorax 1981; 36:366 –369
14 Tonnesen P, Norregaard J, Mikkelsen K, et al. A double-blind trial of a nicotine inhaler for smoking cessation. JAMA 1993; 269:1268 –1271 15 Jorenby DE, Smith SS, Fiore MC, et al. Varying nicotine patch dose and type of smoking cessation counselling. JAMA 1995; 274:1347–1352 16 Hovarth I, Loukides S, Wodehouse T, et al. Increased levels of exhaled carbon monoxide in bronchiectasis: a new marker of oxidative stress. Thorax 1998; 53:867– 870 17 Zayasu K, Sekizawa K, Okinaga S, et al. Increased carbon monoxide in the exhaled air of asthmatic patients. Am J Respir Crit Care Med 1997; 156:1140 –1143 18 Hovarth I, Loukides S, Wodehouse T, et al. Nitric oxide and carbon monoxide in the exhaled air of patients with primary ciliary dyskinesia. Eur Respir J 1998; 12(suppl 28):347
CHEST / 117 / 3 / MARCH, 2000
763
Study objective: To assess whether the breath carbon monoxide (CO) concentration can be used to determine a patient’s smoking habits in a respiratory outpatient clinic. Design: To provide a normal range for smokers and nonsmokers, 41 outpatients and 24 healthy subjects were questioned on their smoking habits and asked to provide two breaths into a CO monitor (EC50 Smokerlyser; Bedfont Instruments; Kent, UK). In a subsequent single-blind study, 51 different outpatients were not told of the purpose of the study and were assessed by extensive questionnaire, spirometry, and Smokerlyser estimation. Setting: The Chest Clinic and Pulmonary Medicine Department at the Northern General Hospital, Sheffield, UK. Participants: Phase 1 involved 41 outpatients attending the Chest Clinic and 24 nonoutpatient colleagues. In phase 2, an additional 51 different outpatients were studied. Measurements and results: The mean (SD) breath CO levels were 17.4 (11.6) parts per million (ppm) for smokers and 1.8 (1.3) ppm for nonsmokers (p < 0.001). A level of 6 ppm was taken as the cutoff, as this gave a selectivity of 96% and a sensitivity of 94% for outpatients. Of the 51 study patients, 5 admitted to smoking in the administered questionnaire. Eight denied smoking but had a mean breath CO > 6 ppm (7.5 to 42 ppm). Of these, three admitted to smoking after being explained the implication of the reading. Conclusions: Breath CO concentration provides an easy, noninvasive, and immediate way of assessing a patient’s smoking status. A reading > 6 ppm strongly suggests that an outpatient is a smoker. (CHEST 2000; 117:758 –763) Key words: carbon monoxide; equipment and supplies; monitoring outpatients; smoking Abbreviations: CO ⫽ carbon monoxide; COHb ⫽ carboxyhemoglobin; ppm ⫽ parts per million
a long time now, it has been known that F orsmoking is associated both etiologically and prognostically with numerous diseases of the respiratory system. However, despite this knowledge, many patients continue to smoke. Knowledge of a patient’s smoking habits is clinically important, for example as a risk factor or as a reason for failure to respond to treatment,1 and it enables appropriate antismoking advice to be given. It has been reported that active intervention, in the form of advice, significantly increased the number of patients who stopped smoking after 1 year.2 The same study also reported that *From the University of Sheffield (Mr. Middleton), Sheffield; and the University of Hull (Dr. Morice), Hull, UK. Performed at the Pulmonary Medicine Department, Northern General Hospital, Sheffield, UK. This study was supported by an educational grant from Bedfont Instruments, Kent, UK. Manuscript received April 29, 1999; revision accepted September 22, 1999. Correspondence to: Edward T. Middleton, B Med Sci, Academic Department of Medicine, Castle Hill Hospital, Castle Rd, Cottingham East Riding, HU16 5JQ, UK; e-mail: E.T.Middleton@ sheffield.ac.uk 758
the use of a carbon monoxide (CO) monitor to demonstrate an immediate and potentially harmful consequence of smoking increased further the number who complied with advice to quit.2 However, many smokers deny their habit, making appropriate counseling impossible.3 This may be because smoking is held in such poor regard by the health profession that patients are reluctant to admit to it, or they may be unwilling to admit to not doing as advised. This inaccuracy of self-reported smoking stresses the need for biochemical conformation of smoking status. For related material, see page 764 There are several methods of assessing smoking status. Nicotine, cotinine, or thiocyanate levels in the plasma or urine may be used to indicate smoking status.4 However, the blood tests are invasive and neither the blood nor the urine tests provide an immediate assessment. The measurement of breath CO level may provide an immediate, noninvasive Clinical Investigations
method of assessing smoking status. Furthermore, the development of relatively inexpensive portable CO monitors enables breath CO levels to be assessed in a wide variety of clinical settings. Following inhalation, CO displaces oxygen in the erythrocyte to form carboxyhemoglobin (COHb). In this form, CO has a half-life of about 5 to 6 h5,6 and may remain in the blood for up to 24 h depending on a number of factors, such as gender, physical activity, and ventilation rate.7–9 While some exposure to CO may occur in normal day-to-day life, due to environmental pollution, passive smoking, and occupational exposure, the most likely cause of high levels of exposure is smoking.10 Another cause of high levels of CO exposure is occult CO poisoning, for example, due to a faulty automobile exhaust system or home heating system, although this is only likely to be responsible for a minority of cases of raised breath CO levels. Finally, CO is an end product of RBC metabolism and is therefore raised in patients with acute hemolysis and following blood transfusions.11 The aims of this study were to use a portable CO monitor to compare the breath CO levels in established smokers and nonsmokers and to investigate factors that may affect breath CO levels. In a second phase of the study, we used the CO monitor to investigate the reliability of the claimed smoking status of patients attending a respiratory outpatients clinic.
on the day of the study. Informed consent was obtained, and the study was approved by the local ethics committee. The subjects were informed of the purpose of the Smokerlyser and were reassured that the results were confidential in order to encourage accurate reporting of smoking habits. Background information about their health, smoking habits, physical activity, and exposure to environmental pollution and passive smoke was obtained. Spirometry was then performed using the Escort Portable Spirometer (Vitalograph Ltd; Buckinghamshire, UK) to determine lung function. Finally, the subjects were asked to provide two breaths into the Smokerlyser as described above. Each subject was assessed using the Smokerlyser twice to check the repeatability of the test, and because a previous study has reported the second reading to be significantly higher than the first.12 In the second phase, outpatients who had not been involved in phase 1 were assessed in the same way. However, these subjects were told that we were measuring breath CO levels to monitor pollution exposure in Sheffield and were not informed that we could determine smoking status from the result. Using the values obtained in the first phase as a guide, we compared their claimed smoking status with the Smokerlyser reading. If there was a discrepancy between the subject’s claimed smoking status and their breath CO level, then the implication of the result was explained and they were asked to confirm their smoking status. Statistical Analysis All results were analyzed by the SPSS statistical package (SPSS; Chicago, IL). Descriptive statistics were used to examine the data and to assess the distribution. Depending on the distribution, a two-sample t test or a Mann-Whitney U test was used to test for a significant difference between the breath CO levels of smokers and nonsmokers and for a significant difference between outpatients and nonoutpatient smokers. A paired t test was used to test for a difference in CO levels between the two successive Smokerlyser assessments.
Materials and Methods Results
Breath CO Monitoring Breath CO monitoring was performed using the EC50 Smokerlyser (Bedfont Instruments; Kent, UK), an inexpensive, portable CO monitor that has previously been shown to be effective.12 The Smokerlyser measures breath CO levels in parts per million (ppm) based on the conversion of CO to CO2 over a catalytically active electrode. On breath holding, the CO in the blood forms an equilibrium with the CO in the alveolar air; therefore, there is a high degree of correlation between breath CO levels and COHb concentration.13 This enables the Smokerlyser to accurately estimate the blood COHb concentration from the breath CO level. The Smokerlyser was calibrated weekly using a mixture of 50 ppm CO in air. To standardize the breath being analyzed by the Smokerlyser, the subjects were asked to exhale completely, inhale fully, and then hold their breath for 15 s. If the subjects were unable to hold their breath for 15 s, they were asked to hold it for as long as possible and the length of time was recorded. Following the breath hold, the subjects were asked to exhale slowly into the Smokerlyser and were encouraged to exhale fully in order to sample the alveolar air. Study Design Outpatients and colleagues were approached without being forewarned about the study so they did not refrain from smoking
Breath CO levels were assessed in a total of 65 people: 41 outpatients attending a respiratory outpatient clinic (24 men; 16 smokers; mean [SD] age, 56.6 [13.7] years) and 24 nonoutpatient colleagues (14 men; 8 smokers; mean age, 29.1 [14.1] years). For the smokers, the mean reported daily cigarette consumption was 11.0 (7.7) cigarettes/d and 83% reported smoking on the day of testing. The second phase of the study involved an additional 51 different outpatients (28 men; mean age, 58 [12.1] years). Phase 1 The mean (SD) maximum breath CO level was 17.4 (11.6) ppm for smokers and 1.83 (1.29) ppm for nonsmokers (p ⬍ 0.001; Fig 1). The mean maximum breath CO level was significantly higher in outpatient smokers than nonoutpatient smokers, with values of 22.2 (10.7) ppm and 12.0 (11.3) ppm, respectively (p ⫽ 0.04; Fig 2). Of the 65 subjects, 48 were able to hold their breath for an equal period during CHEST / 117 / 3 / MARCH, 2000
759
Figure 1. Exhaled CO levels in smokers (n ⫽ 24) and nonsmokers (n ⫽ 41).
Figure 2. Exhaled CO levels in outpatient smokers (n ⫽ 16) and in nonoutpatient smokers (n ⫽ 8).
both assessments with the Smokerlyser. In these subjects, the CO level on the first test was significantly higher than that of the second test (mean difference between tests, 0.69 ppm; p ⬍ 0.001). The only significant correlation was between breath CO and the number of cigarettes smoked in the last 24 h (r ⫽ 0.546; p ⫽ 0.006; Fig 3). There was a trend between breath CO and the number usually smoked per day, but it was not quite statistically significant (r ⫽ 0.39; p ⫽ 0.063). Breath CO was not correlated with age, the length of time since the last cigarette, or lung function, as assessed by percentage predicted FEV1. 760
A CO level of 6 ppm was taken as the cutoff between smokers and nonsmokers, as it gave the highest sensitivity and selectivity. When smokers and nonsmokers were looked at as a whole, a cutoff of 6 ppm had a selectivity of 98% but a sensitivity of only 79%. However, as outpatient smokers had a significantly higher mean breath CO level, and as the second phase of the study only involved outpatients, the sensitivity and selectivity were recalculated only using the CO values obtained from outpatients. This gave a cutoff of 6 ppm, a sensitivity of 94%, and a selectivity of 96% for outpatients only. Clinical Investigations
Figure 3. Correlation between exhaled CO levels and the number of cigarettes smoked in the last 24 h.
Phase 2 Of the 51 outpatients involved in the second phase of the study, 5 admitted to smoking in the administered questionnaire and had maximum breath CO levels ranging from 9 to 28 ppm (mean, 17.3 ppm). An additional eight subjects denied smoking but had a maximum breath CO level ⬎ 6 ppm (range, 8 to 42 ppm; mean, 22.1 ppm). On explanation of the implication of the reading, three of these admitted to still smoking. The remaining five continued to deny smoking but could offer no other possible reason for the high Smokerlyser reading. Two of the five outpatients who continued to deny smoking despite high CO levels (15 ppm and 17 ppm, respectively) were patients pursuing claims for occupational ill health. The remaining 38 outpatients denied smoking and had maximum breath CO levels ranging from 1 to 4 ppm. Discussion This study supports a growing body of evidence12,13 that measuring breath CO levels provides an immediate, noninvasive, simple, and effective way of confirming a patient’s smoking status. A cutoff level of 6 ppm detected 94% of smokers and 96% of nonsmokers in a respiratory outpatient clinic. This is comparable with the work of Jarvis et al,4 who reported that the optimal cutoff was 8 ppm, giving 90% sensitivity and 89% selectivity. Crowley et al6 also reported that a breath CO level ⬎ 8 ppm was strongly associated with a self report of current smoking. CO levels can also be measured in the blood in the form of COHb, and Wald et al13
reported that a cutoff of 1.5% COHb had a sensitivity of 86% and a selectivity of 96.6%. A COHb level of 1.5% corresponds with a breath CO level of approximately 8 ppm13 and is thus comparable with the results of the present study. Despite all the evidence supporting the use of 6 ppm or 8 ppm as the cutoff, many studies using breath CO monitors have tended use 10 ppm (or 2% COHb) as the cutoff.14,15 This study, along with the previous studies, suggests that 10 ppm is too high to be a cutoff for a screening test, as it will reduce its sensitivity. Another problem of using a cutoff of 10 ppm arises because of the increased clearance of CO with an increased ventilation rate.9 Therefore, patients who hyperventilate or who do large amounts of physical exercise will have a lower-than-expected breath CO level and may avoid detection. This may particularly be a problem in the primary prevention of smoking related diseases, as healthy smokers may undertake regular physical exercise and thus have a relatively low breath CO level. A cutoff of 6 ppm instead of 10 ppm will be more likely to detect these smokers and enable appropriate antismoking advice to be given. The maximum breath CO values for nonoutpatient smokers were significantly lower than the values for outpatient smokers. This cannot be explained by the difference between the two groups in terms of age and lung function, as breath CO level was not correlated with either parameter in either this study or in previous studies.6 There was not a significant difference between the reported number smoked by each group (outpatients, 11.9/d; nonoutpatients, 9.2/d; p ⫽ 0.43). However, the large difference in mean breath CO levels suggests that the outpatients CHEST / 117 / 3 / MARCH, 2000
761
inaccurately reported the amount they smoke, possibly due to pressure to be seen as a light smoker. This may explain why a better correlation between the number smoked per day and breath CO was not seen. The breath CO level on the first assessment with the Smokerlyser was significantly higher than that of the second and may be due to an alteration in effort or lung function following the first test. In contrast, Jarvis et al12 reported that the second breath CO level was significantly higher than the first. In practice, any difference in repeated tests is likely to be of little clinical significance, as the difference was ⬍ 2 ppm in 96% of cases and did not alter the interpretation of the findings. We would recommend that a single Smokerlyser assessment should usually be sufficient, provided that there is adequate technique. Another difficulty with the Smokerlyser is that many patients struggle to hold their breath for 15 s, which may reduce their breath CO level. However, a short breath hold does not necessarily invalidate the test, as breath holds as short as 3 and 4 s were sufficient to produce breath CO levels of 18 ppm and 20 ppm, respectively, in two subjects tested. It is also worth briefly discussing the EC50 Smokerlyser itself. The manufacturer claims an accuracy of 2% and, over the levels of breath CO encountered in this study, this level of accuracy is more than adequate to differentiate smokers from nonsmokers. The instrument has a good reproducibility, with a change in signal of ⬍ 2%/mo. This was not a problem in this study, as the Smokerlyser was calibrated weekly; however, in clinical practice, it would be important to ensure the Smokerlyser was calibrated regularly. In most circumstances, the amount of CO in the breath is determined by the level of COHb in the blood. However, if the Smokerlyser is used in areas of high ambient levels of CO, a proportion of the breath CO may result directly from the inhaled air. The Smokerlyser will not be reliable under conditions of high ambient CO, but this situation is unlikely to be encountered in most clinical settings. Only one subject claimed to be a nonsmoker and yet had a CO level ⬎ 6 ppm. There was nothing in this subject’s history to suggest an excessive environmental exposure or an increased red cell turnover that could account for the raised breath CO, although occult CO poisoning could not definitely be excluded. There are two likely explanations for this result. Firstly, this subject may be a smoker who is denying his habit, thus demonstrating the potential benefits of the Smokerlyser. Secondly, it has recently been reported that breath CO levels may be raised 762
due to numerous inflammatory lung diseases, including bronchiectasis,16 asthma,17 and primary ciliary dyskinesia,18 with mean values of about 7 ppm being reported.16,18 Therefore, it is possible that this subject was genuinely a nonsmoker and had some degree of airway inflammation due to his COPD, which accounts for his raised breath CO levels. In light of this, for breath CO levels of between 6 ppm and 10 ppm, it may be prudent to confirm smoking status with a urinary cotinine measurement. Phase 2 of the study assessed the ability of the Smokerlyser to detect smokers who denied their habit (deceivers) in an outpatient clinic. Eight possible deceivers were detected, of which three admitted to smoking when confronted with the reading. Of the five remaining deceivers, four were almost certainly smokers, as their CO levels were ⬎ 10 ppm. Two of these subjects also had a financial incentive to conceal their habit. The subject with a CO level of 8 ppm was the only one for whom there was any doubt about smoking status, and it would have been desirable to confirm this subject’s smoking status with a urinary cotinine measurement. Conclusion The EC50 Smokerlyser provides a quick, simple, and inexpensive method of screening for smokers in the clinical setting. A breath CO level ⬎ 6 ppm indicates that an outpatient in a respiratory clinic is likely to be a smoker.
References 1 LeSon S, Gershwin ME. Risk factors for asthmatic patients requiring intubation: a comprehensive review. Allergologia et Immunopathologia 1995; 23:235–247 2 Jamrozik K, Vessey M, Fowler G, et al. Controlled trial of three different anti-smoking interventions in general practice. BMJ 1984; 288:1499 –1502 3 Daly RJ, Blann AD. Self-reported smoking in vascular disease: the need for biochemical confirmation. Br J Biomed Sci 1996; 53:204 –208 4 Jarvis MJ, Tunstall-Pedoe H, Feyerabend C, et al. Comparison of tests used to distinguish smokers from non-smokers. Am J Public Health 1987; 77:1435–1438 5 Peterson JE, Stewart RD. Absorption and elimination of carbon monoxide by inactive young men. Arch Environ Health 1970; 21:165–171 6 Crowley TJ, Andrews AE, Cheney J, et al. Carbon monoxide assessment of smoking in chronic obstructive pulmonary disease. Addict Behav 1989; 14:493–502 7 Deller A, Stenz R, Forstner K, et al. The elimination of carboxyhaemoglobin: gender specific and circadian effects. Infusther Transfusmed 1992; 19:121–126 8 Joumard R, Chiron M, Vidon R, et al. Mathematical models of the uptake of carbon monoxide on hemoglobin at low carbon monoxide levels. Environ Health Perspect 1981; 41:277–289 Clinical Investigations
9 Vesley A, Takeuchi A, Rucker J, et al. Effect of ventilation on carbon monoxide clearance in humans [abstract]. Am J Respir Crit Care Med 1999; 159:A767 10 Jarvis MJ, Tunstall-Pedoe H, Feyerabend C, et al. Biochemical markers of smoke absorption and self-reported exposure to passive smoking. J Epidemiol Community Health 1984; 38:335–339 11 Ronald F, Coburn MD. Endogenous carbon monoxide production. N Engl J Med 1970; 282:207–209 12 Jarvis MJ, Belcher M, Vesey C, et al. Low cost carbon monoxide monitors in smoking assessment. Thorax 1986; 41:886 – 887 13 Wald NJ, Idle M, Boreham J, et al. Carbon monoxide in breath in relation to smoking and carboxyhaemoglobin levels. Thorax 1981; 36:366 –369
14 Tonnesen P, Norregaard J, Mikkelsen K, et al. A double-blind trial of a nicotine inhaler for smoking cessation. JAMA 1993; 269:1268 –1271 15 Jorenby DE, Smith SS, Fiore MC, et al. Varying nicotine patch dose and type of smoking cessation counselling. JAMA 1995; 274:1347–1352 16 Hovarth I, Loukides S, Wodehouse T, et al. Increased levels of exhaled carbon monoxide in bronchiectasis: a new marker of oxidative stress. Thorax 1998; 53:867– 870 17 Zayasu K, Sekizawa K, Okinaga S, et al. Increased carbon monoxide in the exhaled air of asthmatic patients. Am J Respir Crit Care Med 1997; 156:1140 –1143 18 Hovarth I, Loukides S, Wodehouse T, et al. Nitric oxide and carbon monoxide in the exhaled air of patients with primary ciliary dyskinesia. Eur Respir J 1998; 12(suppl 28):347
CHEST / 117 / 3 / MARCH, 2000
763