Thorax: first published as 10.1136/thx.35.12.925 on 1 December 1980. Downloaded from
Thorax, 1980, 35, 925-928
Inhaling habits among smokers of different types of cigarette
NICHOLAS J WALD, MARIANNE IDLE, JILLIAN BOREHAM, AND ALAN BAILEY From the ICRF Cancer Epidemiology and Clinical Trials Unit, Radcliffe Infirmary, Oxford, and BUPA Medical Centre, London
ABSTRACT Inhaling habits were studied in 1316 men who freely smoked their usual brands of cigarette. An index of inhaling was calculated for each person by dividing the estimated in- crease in carboxyhaemoglobin level from a standard number of cigarettes by the carbon mon- oxide yield of the cigarette smoked. Smokers of ventilated filter cigarettes inhaled 82% more than smokers of plain cigarettes (p
Filter cigarettes, especially those with ventilated health screening examination. The men were copyright. filters (which have perforations in the filter asked about their medical history and their usual through which air can enter and reduce the con- and recent smoking habits. All information was centration of the cigarette smoke), deliver less collected after arrival at the Medical Centre, tar and nicotine than plain cigarettes. They are and the men were not forewarned about the sur- considered less harmful to health, particularly as vey of smoking habits. The time when each far as the risk of lung cancer is concerned. How- cigarette, cigar or pipe was smoked since waking http://thorax.bmj.com/ ever, since filter cigarettes are less "strong" than that day was recorded, together with the manu- plain ones, people smoking them may inhale to a facturer's brand. Each man provided a sample greater extent or smoke more of them. A pre- of venous blood and the carboxyhaemoglobin vious study' suggested that smokers of ventilated (COHb) saturation was measured as previously filter cigarettes did inhale to a greater extent than described.2 The within-assay standard deviation other cigarette smokers, but this was not quanti- of the method was 0.05% COHb. Tar and nicotine fied and the study drew no firm conclusions yields of the cigarettes smoked were obtained regarding the inhaling habits of smokers of un- from tables published by the Health Departments on September 28, 2021 by guest. Protected ventilated filter cigarettes, the type most com- of the United Kingdom, and carbon monoxide monly smoked. We aimed to examine inhaling in (CO) yields were obtained from the Tobacco Ad- more detail in the present study, and for simplicity visory Council. Both the tar/nicotine and the used the term inhaling in a broad sense to refer CO yields were determined under the same stan- to the intake of tobacco smoke from a cigarette dard conditions. including contributions from the number of puffs, Our statistical analysis considers men seen puff volume, and the depth to which smoke is between early 1975 and September 1977, who inspired into the lungs. usually smoked only manufactured cigarettes. Men who smoked brands of cigarettes for which Methods CO yields were not available from the Tobacco Advisory Council (54) were excluded. The analysis The study population consisted of men aged 35 was restricted to the 1316 men who smoked be- to 64 years (mean 47 yr) who attended the BUPA tween one and 16 cigarettes before the blood test Medical Centre in London for a comprehensive on the day they were seen at the Medical Centre, Address for reprint requests: Dr NJ Wald, ICRF Cancer Epidemiology since only a few men smoked any given number and Clinical Trials Unit, Radcliffe fnfirmary, Oxford OX2 6HE. of cigarettes greater than 16. 925 Thorax: first published as 10.1136/thx.35.12.925 on 1 December 1980. Downloaded from
926 Nicholas J Wald, Marianne Idle, Jillian Boreham, and Alan Bailey Results lower and 21% higher, indicating that both types of filter cigarette were inhaled to a greater extent The table shows that the mean number of cigar- than plain cigarettes. ettes usually smoked each day was similar for A quantitative index of inhaling by an indi- smokers of plain, ventilated filter, and unventi- vidual smoker, which takes account of both the lated filter cigarettes. The mean COHb levels were, background and carryover COHb as well as the however, significantly different, and these differ- CO yield of the particular cigarette smoked, is ences persisted after the COHb levels were in- given by the excess COHb level divided by the CO directly standardised for the number of cigarettes yield of the cigarette. Since this index depends on smoked before the test to take account of any the number of cigarettes smoked before the differences between smokers of the three types of COHb test, the mean COHb inhaling index for cigarette, using all the smokers as the standard each group of smokers was indirectly standardised population. For smokers of ventilated and un- for the number of cigarettes smoked, and we ventilated filter cigarettes the standardised COHb called this the standardised COHb inhaling index. levels were, respectively, 23% and 35% higher than For the smokers of ventilated filter cigarettes this that for smokers of plain cigarettes (p<0001). index was 82% more than for the smokers of plain cigarettes (p<0-001), and for smokers of un- Table Mean cigarette consumption, COHb, CO ventilated filter cigarettes it was 36% more y.elds, and inhaling index according to type of cigarette (standardised values were indirectly (p Inhaling habits among smokers of different types of cigarette 927 TAR INTAKE NICOTINE INTAKE 0 Figure Tar and nicotine intake per cigarette (±standard error) as percent of a1) ct) value for plain cigarettes. The dotted lines indicate the intake of tar and nicotine which would be expected if all L- three cigarettes were smoked in the same way. P=plain, V=ventilated filter, and F=unventilated filter cigarette. The mean tar yields were 26-0, 9S5, and 18 1 mg/ cigarette respectively. The mean nicotine yields were 19, 0-8, and 13 mg/cigarette. Our results relating to unventilated filter cigar- to ventilated filter and unventilated filter cigarettes ettes are the most important because most cigar- inhaled 2% and 6% more, respectively, than those ettes sold are of this type. However, the proportion who had always smoked filter cigarettes, and these of the cigarette market occupied by ventilated differences were not statistically significant. filter cigarettes has recently been increasing. In Switching from plain to filter cigarettes was also 1974 the market share of plain, ventilated, and not associated with any material change in daily copyright. unventilated filter cigarettes in the UK was about cigarette consumption. 15%, 6%, and 79% respectively. In 1978 the The estimation of the actual intake of tar and figures were 10%, 15%, and 75% (WS Paige, nicotine shown in the figure rests on two assump- personal communication). tions. The first is that the COHb inhaling index Our results relating to ventilated cigarettes may is directly proportional to the extent of intake have been caused partly by smokers blocking the of CO from tobacco smoke. This is likely to be http://thorax.bmj.com/ perforations in the filter, say with their fingers or approximately correct since the alveolar concen- lips, as has sometimes been found to occur (LT tration of CO is likely to be proportional to the were not influenced by any such bias. extent of inhaling, and we have confirmed that The results of this study are likely to be gen- the relationship between end-tidal CO concentra- erally applicable for two reasons. Firstly, they re- tion and COHb appears approximately linear. late to men who smoked their usual brands of Secondly, it assumes that the proportions of the cigarette as they wished without being aware of different constituents of tobacco smoke inhaled the investigation until they attended the Medical remain constant, so that, for example, one smoker Centre. Secondly, the men who had changed from whose inhaling index is twice that of another on September 28, 2021 by guest. Protected plain to filter cigarettes had done so many years smoking the same brand of cigarette inhales, on before, 98% having switched at least two years average, twice as much tar as the other smoker, previously. One other long-term study yielded and this is likely to be approximately true. It does qualitatively similar results to our own6 and one not, however, necessarily follow that the deposition did not, though in the latter the change in tar of tar on the bronchial mucosa is also double. If yield was not great.7 Most other studies, usually it is less, this might explain why the risk of lung laboratory-based, have been mainly concerned with cancer associated with smoking filter cigarettes is the short-term effects of switching from high-tar less than our data would suggest. to low-tar cigarettes or vice versa. In general, it There are data to support our assumption that was concluded that switching altered inhaling nicotine delivery to the lungs is approximately habits,8"13 though where the change in tar yield proportional to the CO intake. For example, it has was not great, no material alteration in inhaling been shown that COHb levels are closely associ- habits was again noted.14 In our own study the ated with plasma nicotine levels in subjects who type of cigarette was the principal determinant of smoked cigarettes using different dilution filters.14 smoking behaviour, while switching itself was less Tar delivery is also likely to be related to CO in- important; smokers who had switched from plain take since nicotine and tar are both in the par- Thorax: first published as 10.1136/thx.35.12.925 on 1 December 1980. Downloaded from 928 Nicholas J Wald, Marianne Idle, Jillian Boreham, and Alan Bailey ticulate phase of cigarette smoke and are, Feyerabend C. Comparison of effect on tobacco therefore, likely to be delivered to the lungs in a consumption and carbon monoxide absorption of similar way. changing to high and low nicotine cigarettes. Br COHb levels in smokers can be affected by Med J 1973; 4:512-6. 11 Turner JAMcM, Sillet RW, Ball KP. Some several factors other than those considered in our effects of changing to low-tar and low-nicotine analysis, such as the time of day when blood for cigarettes. Lancet 1974; 2:737-9. the COHb test was taken, and these might have 12 Russell MAH, Wilson C, Patel UA, Feyerabend led to bias if associated with the type of cigarette C, Cole PV. Plasma nicotine levels after smoking smoked. We examined these factors as in previous cigarettes with high, medium and low nicotine studies,' 2 and again concluded that our results yields. Br Med J 1975; 2:414-6. were not influenced by any such bias. 13 Ashton H, Stepney R, Thompson JW. Self- Our data provide no indication of which con- titration by cigarette smokers. Br Med J 1979; 2:357-60. stituent of tobacco smoke may have influenced the 14 Sutton SR, Feyerabend C, Cole PV, Russell extent of inhaling. Although inhaling was con- MAH. Adjustment of smokers to dilution of sidered in relation to the tar and nicotine yield of tobacco smoke by ventilated cigarette holders. the cigarettes smoked neither need be responsible Clin Pharmacol Therap 1978; 24:395-405. for the differences in inhaling which were observed. Statistical appendix We thank Sir Richard Doll, Dr HS Cuckle, and RANDOMISATION TEST OF STATISTICAL SIGNIFICANCE Mr PN Lee for their comments, Mr R Peto for Differences in the mean value of a variable x (either x= statistical advice, the Tobacco Advisory Council COHb, or, in other analyses, x- inhaling index) measured for providing the CO yields of different brands of on men smoking two types of cigarette (A and B) were cigarette, and the Medical Research Council for for statistical significance within 16 strata defined tested copyright. part of our financial support. by the number of cigarettes smoked before the test. (Nobody in the study had smoked more than 16 cigarettes References before the test.) The underlying idea is to compare like with like, and to 1 Wald N, Idle M, Smith PG, Bailey A. Carboxy- avoid comparing, for example, people who have smoked haemoglobin levels in smokers of filter and plain only five cigarettes with people who have already smoked cigarettes. Lancet 1977; 1:110-2. 10 before the test. http://thorax.bmj.com/ 2 Wald N, Idle M, Bailey A. Carboxyhaemoglobin In a particular stratum i: levels and inhaling habits in cigarette smokers. Let Ni be the total number of men smoking i cigarettes Thorax 1978; 33:201-6. before the test, and let ni be the number smoking type A 3 Wynder EL, Mabuchi K, Beattie EJ. The epi- only. demiology of lung cancer. Recent trends. JAMA Define oi to be the sum of the ni observed values of x 1970; 213:2221-8. for men smoking type A. 4 Hammond EC, Garfinkel L, Seidman H, Lew EA. Define xi to be the overall mean value of x in stratum i "Tar" and nicotine content of cigarette smoke in and Si2 to be the variance estimator derived from the relation to death rates. Environ Res 1976; 12: x-values in stratum i by on September 28, 2021 by guest. Protected 263-74. E 5 Wald NJ. Mortality from lung cancer and Si2= (x-Ri)2/(Ni- 1). coronary heart disease in relation to changes in If the ni men smoking type A were chosen at random smoking habits. Lancet 1976; 1:136-8. from the total Ni, then by standard permutational 6 Russell MAH, Jarvis M, Iyer R, Feyerabend C. arguments the expectation of the sum of their x-values Relation of nicotine yield of cigarettes to blood would be ei and its variance vi, where ei=ni ki and nicotine concentrations in smokers. Br Med J v,=ni (I-ni/Ni) Si2. 1980; 1:972-98. Summing oi, ei andvi over all 16 strata gives: 7 Freedman S, Fletcher CM. Changes of smoking 16 16 16 habits and cough in men smoking cigarettes with 0= E oi, E= £ ei, andV=V vi. 30% NSM tobacco substitute. Br Med J 1976; 1: i=1 i=li= 1427-30. IfA smokers differ from B smokers withrespect to variable 8 Ashton H, Watson DW. Puffing frequency and x, given the number of cigarettes smoked before the test, nicotine intake in cigarette smokers. Br Med J then 0 should be systematically different from E. To test 1970; 3:679-81. whether this is so, we note that if there were no difference 9 Frith CD. The effect of varying the nicotine between A smokers and B smokers, then 0 would differ content of cigarettes on human smoking be- only randomly from E, so that (O-E) would have haviour. Psychopharmocology 1971; 19:188-92. expectation zero and variance V, and (O-E)2/V would 10 Russell MAH, Wilson C, Patel UA, Cole PV, be distributed approximately as x12.