Scholarly article on topic 'Mouth level smoke exposure using analysis of filters from smoked cigarettes: A study of eight countries'

Mouth level smoke exposure using analysis of filters from smoked cigarettes: A study of eight countries Academic research paper on "Health sciences"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{Cigarette / "Smoke exposure" / "Mouth level exposure" / "Filter analysis" / Nicotine / Tar / Australia / Brazil / Canada / Germany / Japan / "New Zealand" / "South Africa" / Switzerland}

Abstract of research paper on Health sciences, author of scientific article — D.C. Mariner, M. Ashley, C.J. Shepperd, G. Mullard, M. Dixon

Abstract The analysis of spent cigarette filters enables the estimation of the nicotine and tar (nicotine-free dry particulate matter) yields obtained by smokers in their everyday environment and has been shown to correlate well with biomarkers of exposure. Leading products across the range of ISO tar yields were selected from Australia, Brazil, Canada, Germany, Japan, New Zealand, South Africa and Switzerland. At least fifty demographically representative smokers were recruited per product. Subjects, ⩾21years of age and smoking ⩾5 cigarettes per day, were asked to collect ⩾15 filters from cigarettes they had smoked. The collected filters were analysed for nicotine and UV absorbance to enable the smokers’ mouth level exposure to nicotine and tar to be estimated and a comparison of countries and tobacco blend styles to be made. Smoking history data were also collected. More than 80,000 filters were collected from 5703 smokers of 106 products from eight countries. Mean±SD estimated nicotine exposures per cigarette and per day ranged from 0.93±0.34mg/cigarette (Brazil) to 1.77±0.69mg/cigarette (South Africa) and from 16.4±11.1mg/day (Germany) to 31.5±14.8mg/day (South Africa), respectively. Male smokers obtained higher mean estimated tar and nicotine exposures than female smokers. These gender differences were statistically significant for six countries. Significant correlations were found between estimated nicotine exposure and ISO nicotine yield, and between estimated tar exposure and ISO tar yield (p <0.001).

Academic research paper on topic "Mouth level smoke exposure using analysis of filters from smoked cigarettes: A study of eight countries"

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology

journal homepage: www.elsevier.com/locate/yrtph

Mouth level smoke exposure using analysis of filters from smoked cigarettes: A study of eight countries

D.C. Mariner**, M. Ashleya, C.J. Shepperda, G. Mullarda, M. Dixon b

a Group Research and Development, British American Tobacco, Regents Park Road, Southampton SO15 8TL, UK b Dixon Consultancy, Hazeldene Road, Liphook GU30 7PH, UK

ARTICLE INFO

Article history:

Available online 25 May 2010

Keywords:

Cigarette

Smoke exposure

Mouth level exposure

Filter analysis

Nicotine

Australia

Brazil

Canada

Germany

New Zealand South Africa Switzerland

ABSTRACT

The analysis of spent cigarette filters enables the estimation of the nicotine and tar (nicotine-free dry particulate matter) yields obtained by smokers in their everyday environment and has been shown to correlate well with biomarkers of exposure.

Leading products across the range of ISO tar yields were selected from Australia, Brazil, Canada, Germany, Japan, New Zealand, South Africa and Switzerland. At least fifty demographically representative smokers were recruited per product. Subjects, P 21 years of age and smoking P 5 cigarettes per day, were asked to collect P15 filters from cigarettes they had smoked. The collected filters were analysed for nicotine and UV absorbance to enable the smokers' mouth level exposure to nicotine and tar to be estimated and a comparison of countries and tobacco blend styles to be made. Smoking history data were also collected.

More than 80,000 filters were collected from 5703 smokers of 106 products from eight countries. Mean ± SD estimated nicotine exposures per cigarette and per day ranged from 0.93 ± 0.34 mg/cigarette (Brazil) to 1.77 ± 0.69 mg/cigarette (South Africa) and from 16.4 ± 11.1 mg/day (Germany) to 31.5 ± 14.8 mg/day (South Africa), respectively. Male smokers obtained higher mean estimated tar and nicotine exposures than female smokers. These gender differences were statistically significant for six countries. Significant correlations were found between estimated nicotine exposure and ISO nicotine yield, and between estimated tar exposure and ISO tar yield (p < 0.001).

© 2010 Elsevier Inc. All rights reserved.

1. Introduction

For the past four decades regulatory bodies have required the yields of nicotine-free dry particulate matter (NFDPM or 'tar'), nicotine and carbon monoxide from cigarettes to be determined using machine smoking methodologies such as those specified by the Federal Trade Commission (FTC) and International Organization for Standardization (ISO). In both cases, the yields are determined when cigarettes are smoked on a machine, taking 35 mL puffs of 2 s duration once a minute, smoking until a specified length of cigarette remains. These methods were designed to provide a means of ranking cigarettes in terms of smoke yield. In 1967 the FTC recognised the variability in human smoking behaviour and stated that the purpose of their test was not to determine the amount of tar and nicotine inhaled by any human smoker, but rather to determine the amount of tar and nicotine generated when a cigarette is smoked by a machine in accordance with the prescribed method

* Corresponding author. Fax: +44 (0)2380 779715. E-mail addresses: derek_mariner@bat.com (D.C. Mariner), madeleine_ashley@ bat.com (M. Ashley), jim_shepperd@bat.com (C.J. Shepperd), gavin_mullard@bat. com (G. Mullard), dixon.consultancy@virgin.net (M. Dixon).

0273-2300/$ - see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2010.05.006

(Federal Trade Commission, 1967). In addition, the 1988 report of the UK Independent Scientific Committee on Smoking and Health (ISCSH) recognised that machine derived yields were generally less than those obtained by smokers (Independent Scientific Committee on Smoking and Health, 1988).

Despite the differences between the standardised machine smoking puff parameters and human puffing topography, a number of regulatory authorities have used machine derived yields of tar, nicotine and carbon monoxide (CO) in the regulation of cigarette smoke mainstream emissions, e.g., the European Union has an upper limit of 10 mg tar, 1 mg nicotine and 10 mg CO as measured under ISO machine smoking conditions (European Union, 2001). In some countries, e.g., USA, Japan, Australia and New Zealand, there are currently no mandatory upper limits on cigarette smoke emissions.

For several decades many public health and regulatory authorities viewed the reduction of ISO/FTC tar yields as a possible means of reducing the adverse health effects of cigarette smoking (e.g., US Department Health and Human Services, 1966, 1981; Royal College of Physicians, 1971, 1977; Independent Scientific Committee on Smoking and Health, 1988; Swan and Froggatt, 1996). More recently, doubts have been expressed about the

health benefits produced by reducing ISO/FTC tar yields and therefore the usefulness of the FTC/ISO method in terms of providing a basis for tobacco control regulations and consumer information has been questioned (National Institutes of Health, 1996; Bates et al., 1999; Wilkenfeld et al., 2000; National Institutes of Health, 2001; Jarvis et al., 2001; World Health Organization (WHO) Study Group on Tobacco Product Regulation (TobReg), 2004, 2008) with many concluding that machine yields do not offer smokers meaningful information about relative exposure differences between cigarettes. In November 2008, the FTC rescinded their guidance issued in 1966 that generally permitted statements concerning tar and nicotine yields measured using what was termed the FTC method.

A number of studies have been published over the last 30 years that attempt to estimate smokers' exposure to mainstream cigarette smoke from cigarettes. The methodology used falls into three broad categories:

(1) The analysis of biomarkers in human body fluids or expired breath. These have focussed primarily on nicotine and CO exposure (see reviews by Stephen et al. (1989), Scherer (1999), and National Institutes of Health (2001) for references). However, in recent years studies have been conducted in which other biomarkers for exposure to tobacco smoke constituents such as tobacco specific nitrosamines, and polycyclic aromatic hydrocarbons have been conducted (e.g., Hecht et al., 2005; Scherer et al., 2007; Shepperd et al., 2009; Mendes et al., 2009).

(2) The measurement of human smoking behaviour (puff volume, duration, and frequency) followed by machine duplication. In this method tar and nicotine yields are determined using the smoking machine set to duplicate human puffing conditions rather than standardised machine smoking conditions (e.g., Creighton et al., 1978; Djordjevic et al., 2000; Hammond et al., 2006).

(3) The analysis of nicotine and tar deposited in spent cigarette filters and the derivation of human smoke yields from the tar and nicotine retention characteristics of the filters (e.g., Rawbone, 1984; Baker et al., 1998; Pauly et al., 2009; Shepperd et al., 2006, 2009; St.Charles et al., 2006, 2009).

These methods have different levels of invasiveness, either associated with the sample collection or the environment with the consequence that smoking behaviour may be modified (Comer and Creighton, 1978; Ossip-Klein et al., 1983). In addition, smoking behaviour/duplication studies are based on individuals smoking just 1 or 2 cigarettes hence they only allow a snapshot estimate of the amount of smoke that a smoker receives.

Studies involving the collection of urine, saliva, and spent cigarette filters can be conducted outside the laboratory and are more representative of the smoker's behaviour in their everyday environment than laboratory based studies. However, estimates of smoke exposure using urine and saliva analyses involve the measurements of metabolites of smoke components, e.g., nicotine. As individuals may differ markedly in their metabolism of smoke components such as nicotine, problems can arise in attempting to assess exposure to cigarette smoke from the levels of metabolites in urine or saliva.

Filter analysis methodology is used to estimate human smoke yields from the nicotine and tar content of spent cigarette filters. This is possible since the yield of nicotine or tar in the smoke and that remaining in the filter are related by the filtration efficiency (FE) of the filter. The main advantage of filter analysis is that, unlike methods based on biomarker analyses, it can provide a direct estimate of the nicotine and tar yields produced by cigarettes when smoked by humans.

The use of filter analysis for estimating nicotine yields to smokers was originally used in the late 1960s and early 1970s (Ashton and Watson, 1970; Schulz and Seehofer, 1978; Forbes et al., 1976) Cigarettes marketed at that time were predominantly unventilated and their FEs were relatively constant over the range of puff flow-rates produced by smokers. However, smoke yields have been markedly reduced since the early 1970s and many products have incorporated filter ventilation as one means of reducing smoke yields. FE is dependant on smoke velocity (or flow). As velocity increases the FE falls, becoming stable above approximately 30 cm/s (Dwyer and Abel, 1986). With ventilated cigarettes, the smoke velocities in the portion of the filter upstream of the ventilation holes generally fall within the range that influences FE and therefore any changes in human puff flow-rates may alter FE. Some earlier methods analysed the whole filter (Rawbone, 1984; Baker et al., 1998) and attempted to take into account the flow dependency of FE. In 2001, St.Charles (2001) developed a simple but effective means of improving the accuracy of filter analysis methodologies. The velocity through the mouth-end portion of the filter, downstream of the ventilation holes, is generally above the FE sensitive range so here the FE is generally constant, hence the analysis of a 10 mm mouth end section of the filter improves estimates. This 'part-filter' technique provides a good estimate of the tar and nicotine exiting the cigarette during human smoking (Shepperd et al., 2006) and thus modern filter analysis techniques have the capability of producing robust estimates of mouth level exposures to cigarette smoke in large-scale studies of smokers (Pauly et al., 2009; Polzin et al., 2009; Shepperd et al., 2009; St.Charles et al., 2009). However, potential disadvantages of filter analysis techniques include the need to provide the subjects with cigarettes from a common batch, calibrations for each cigarette type and the possibility of subjects returning filters from cigarettes smoked by other smokers.

Filter analysis techniques provide a measure of mouth level exposure (MLE) to cigarette smoke constituents but they do not take account of smoke spilled by the smoker prior to inhalation, or smoke exhaled. Thus filter analysis derived data represent the maximum available to the smoker rather than absolute smoke amounts retained in, and absorbed, from the respiratory system. However, St.Charles et al. (2006), Morin et al. (this issue), and Shepperd et al. (2009) used filter analysis techniques and reported strong correlations between mouth level nicotine exposure and nicotine metabolite levels in urine and saliva in US, Canadian and German smokers, suggesting that measures of smoke exposure using filter analysis are indeed robust.

There have been a number of cross-sectional studies involving the measurement of nicotine biomarkers in groups of smokers of cigarettes differing in ISO/FTC tar and nicotine yields. Although many cross-sectional studies show statistically significant correlations between machine-derived nicotine yields and biomarkers of nicotine uptake in individual smokers, these relationships tend to be relatively weak with r values in the range 0.2-0.5 (e.g., Russell et al., 1980; Ebert et al., 1983; Gori and Lynch, 1985; Russell et al., 1986; Rosa et al., 1992; Bridges et al., 1990; Byrd et al., 1998; Jarvis et al., 2001; Ueda et al., 2002). These studies also use a variety of biomarkers for nicotine uptake. These include plasma nicotine (Russell et al., 1980, 1986; Ebert et al., 1983; Gori and Lynch, 1985), plasma cotinine (Gori and Lynch, 1985; Bridges et al., 1990; Rosaetal., 1992), saliva cotinine (Jarvis et al., 2001) and 24 h urinary nicotine metabolites (Byrd et al., 1998; Ueda et al., 2002). The use of different biomarkers makes it difficult to make direct comparisons between studies.

The majority of studies measuring nicotine uptake are from North America and Europe. We now report a cross-sectional study using subjects from a wider range of countries to estimate the nicotine and tar MLEs obtained for smokers in their everyday environ-

ment. Our aim was to determine whether or not there was a relationship between ISO smoke yields and tar and nicotine MLEs for smokers from a range of countries smoking cigarettes differing in yields and blend styles. The countries were Australia, Brazil, Canada, Germany, Japan, New Zealand, South Africa and Switzerland. This is the first time such data have been collected on such a large scale and the first published smoke exposure data of any description in some of these countries.

2. Materials and methods

2.1. Product selection

The protocol was essentially the same in all countries, with minor variations to suit local needs. The aim was to cover the range of cigarettes marketed in each of the countries. Products were selected according to the following criteria:

(1) The full range of ISO tar yields within that country should be represented.

(2) Leading legitimate products (by market share) from each ISO tar sector/band.

(3) Depending on the number of products selected using the first two criteria, other products would be selected to provide a better overall coverage of the cigarette market in that country.

(4) Filtered cigarettes, 24-25 mm circumference, 72, 84 or 100 mm length.

(5) A maximum of 15 products in total per country.

Market share data were obtained from The Nielsen Company (New York, USA) as part of a range of marketing information services provided to British American Tobacco (BAT).

2.2. Subject selection

Local market research agencies (MRAs) conducted telephone interviews to screen potential subjects from multiple locations/cities/regions. Details of smoking history, including time with current cigarette brand and current daily consumption rates, were taken and subjects were recruited using the following criteria:

(1) Fifty smokers per product to complete the study (in practise about 60 were recruited).

(2) At least 21 years of age.

(3) Regular smokers of at least 5 cigarettes per day of one of the products being assessed.

(4) The product being assessed had been their preferred product for at least 6 months.

(5) Representative of the demographics (age and gender) of smokers of the products being assessed.

2.3. Study procedure

Those subjects who met the recruitment criteria attended two appointments held at the MRA's central location offices. At the first appointment, subjects were fully briefed on the survey protocol before giving their written informed consent to participate in the study. The subjects were shown how to use a specially designed filter collector to cut, collect and protect the cut mouth-end portions of filter tips as soon as possible after smoking. The subjects were given a supply of their 'own-brand' of cigarettes (typically one or two packs) and were asked to collect a minimum of 15 filter tips from cigarettes they had smoked according to their normal prac-

tise. Once this was done, the subject informed the agency and a second appointment was arranged to return the filter collector and complete an exit questionnaire. Payments were made to the subjects only in countries where such payment is normal practise in consumer research.

Returned filter collectors were checked by the MRA staff to confirm that the anti-tamper seals on the storage box had not been broken or removed. If the seals had been damaged, the subject/ sample was rejected from the study. All filter collectors were logged by subject number, product and region (where appropriate) and then stored in a cool/dry environment for no more than one week before being shipped to the BAT laboratories. Samples from Brazil were sent to the BAT laboratories in Rio de Janeiro. All other samples were sent to the BAT laboratories in Southampton, UK. Each laboratory is accredited by national accreditation bodies, and a comprehensive program of method validation and ongoing cross-checks between laboratories was undertaken to ensure data consistency (St.Charles et al., 2009).

Once at the BAT laboratories, samples were inspected and rejected if there were <12 filter tips, whole filter tips, poorly cut or damaged part-filters, or tobacco or ash was collected in the sample. If samples were rejected, the MRA contacted those subjects again and they were asked to repeat the sample collection. If the subject was not contactable, or declined to repeat the sample collection, a new subject was recruited if the target of 50 smokers per product was not going to be achieved.

2.4. Filter analysis method

The part-filter analysis technique used a slightly modified version of the methods described previously (Shepperd et al., 2006; St.Charles et al., 2006). Calibration curves for tar yield vs UV absor-bance per part-filter tip, and nicotine yield vs nicotine content per tip were prepared for each of the cigarette types used by the subjects. The six machine smoking regimes described in Shepperd et al. (2009) were used to construct each calibration curve. Five cigarettes were smoked onto each Cambridge pad and this procedure was repeated for each cigarette type and each machine smoking regime. The mainstream yields of total particulate matter (TPM), nicotine, water and NFDPM were determined according to the ISO (ISO 10362-1, 1999; ISO 3308, 2000; ISO 4387, 2000; ISO 10315, 2000). The average values from the repeat smoking procedures were used for the calibration curves.

The part-filters from the calibration procedures and subjects were extracted using methanol incorporating n-heptadecane as an internal standard and were analysed for nicotine content by GC with FID detection. Full details of this procedure were reported by Shepperd et al. (2009). The tar content of the part-filters was also determined from the extracts using an UV absorbance method (HPLC with UV detection (310 nm)). Quinoline in methanol solutions (20, 50, 100, 200, 400 and 1000 ppm) was used as an instrument check for the UV absorbance method.

Linear regression equations were produced from the calibration curves for mainstream tar vs UV absorbance per part-filter, and mainstream nicotine vs nicotine content per part-filter tip. Separate regression equations were produced for each brand of cigarette used in the study.

The part-filters from each smoker were extracted and analysed in three batches of five tips. The UV absorbance and nicotine values obtained from these three batches of part-filters were averaged for each smoker and converted into estimated mouth level nicotine and tar exposures per cigarette by use of the appropriate regression equations.

2.5. Daily cigarette consumption rate assessment

Two records of daily cigarette consumption rates were obtained from the subjects. The first was the response to the question ''How many manufactured cigarettes do you normally smoke a day?" asked during the recruitment phase of the study. The second was the response to the question ''About how many cigarettes would you say you smoked in the past 24 h (please be as accurate as possible)" asked when the filters and collectors were retrieved from the smokers. This latter question was not asked in the Swiss survey. The consumption figures covering the filter collection periods were used to convert estimated mouth level nicotine and tar exposures per cigarette into estimated daily exposures for each subject for seven of the countries. However, the consumption data obtained from the recruitment question were used for the Swiss study.

2.6. Measurement of ¡SO smoke yields

All of the cigarettes used in the study were smoked under the ISO machine smoking regime (ISO 3308, 2000) and mainstream smoke yields of tar and nicotine were determined (ISO 4387, 2000; ISO 10315, 2000; ISO 10362-1, 1999).

2.7. Data analysis

Minitab (version 15, Minitab Ltd., UK) was used for the statistical analyses. The ISO tar and nicotine yields of the cigarettes smoked by the subjects were averaged for each country. The statistical significance of the differences in the smokers' estimated mean mouth level exposures between the different countries was assessed using one-way analysis of variance (ANOVA) and Tukey's tests.

The relationships between ISO yields, mouth level tar and nicotine exposures and daily cigarette consumption rates were assessed using linear regression analyses.

Two sample t tests were also conducted to assess the effects of gender, ISO yields above and below 6 mg and blend.

3. Results

3.1. Products and subjects

Table 1 summarizes the countries, products and subjects. The highest market coverage was in South Africa with products representing 65% of the market, and lowest was in Canada with 31%,

Table 1

Country, product and subject data.

Country Number ISO Combined Number Total Survey

of tar market of subjects date

products range share of subjects

assessed (mg) products per

(%) product

Australia 15 1-13 41.2 24-60 705 Q1/Q2

Brazil 11 1-9 56.0 50-61 607 Q4 2005

Canada 12 1-15 30.6 54-60 690 Q2/Q3

Germany 14 1-10 45.3 51-60 767 Q4 2005

Japan 15 1-12 57.2 55-60 861 Q2 2006

New 9 5-13 52.1 7-58 391 Q1/Q2

Zealand 2005

South 15 2-14 64.6 54-62 860 Q2/Q3

Africa 2005

Switzerland 15 1-10 48.4 51-60 822 Q2 2006

Note: the Canadian survey was conducted prior to the introduction of low ignition propensity regulations.

with the remainder between 41% and 57%, based on contemporary market research data.

In most cases the target of 50 subjects per product was reached. The exceptions were: 1 and 2 mg pack tar products in Australia (combined market share 4.2%) for which 24 (1 mg) and 40 (2 mg) subjects were recruited, and 1 and 8 mg pack tar products in New Zealand (market shares 0.03% and 1.9%, respectively) for which only 1 and 7 subjects were recruited, respectively. Data from the single smoker of the 1 mg product from New Zealand have not been included. In addition, 227 sets of samples were rejected on receipt at the laboratory and were replaced by new samples from the same subjects or new subjects.

The overall data-set therefore comprises eight countries, 106 products and 5703 smokers, and is based on the analysis of filters from more than 80,000 cigarettes smoked by the subjects.

3.2. Cigarette consumption rates

Two estimates of cigarette consumption rates were obtained from the subjects in six of the eight countries. The two exceptions were Switzerland and Japan, where daily consumption rates from the recruitment questionnaire were only available for the former, and consumption rates 'per last 24 h' from the exit questionnaire were only available for the latter.

The correlations (r values) between both the consumption rates obtained from the recruitment questionnaire ('consumption/day') and consumption rates from the exit questionnaire (consumption/last 24 h) and the ISO nicotine yields of the cigarettes smoked by the subjects are shown in Table 2. The correlations between the two estimates of cigarette consumptions rates are also shown in Table 2.

There were significant positive correlations between cigarette consumption/last 24 h and ISO nicotine yield for Australia (r = 0.178), Canada (r = 0.135), Germany (r =0.121) and New Zealand (r =0.221) and a negative significant correlation for South Africa (r = -0.200). Significant positive correlations between cigarette consumption per day and ISO nicotine yield were obtained for Germany (r = 0.099) and New Zealand (r = 0.106) and a negative correlation was obtained for South Africa (r = -0.116). Although statistically significant these correlations were very low indicating very weak relationships between cigarette consumption rates and ISO nicotine yield.

Significant correlations were seen between the two estimates of cigarette consumption. However, with the possible exception of Brazil, these correlations were low indicating weak relationships between the two measures. Fig. 1 shows the relationship between these two measures of cigarette consumption.

Table 2

Least squares linear regression correlations (r values) between cigarette consumption/last 24 h, cigarette/day and ISO nicotine yield.

Country Cons/last 24 h vs Cons/day vs ISO Cons/last 24 h vs

ISO nicotine nicotine Cons/day

r value p value r value p value r value p value

Australia 0.178 <0.001 0.018 0.634 0.479 <0.001

Brazil 0.065 0.112 0.037 0.366 0.733 <0.001

Canada 0.135 <0.001 0.045 0.239 0.344 <0.001

Germany 0.121 0.002 0.099 0.007 0.512 <0.001

Japan 0.032 0.344 NA NA

New Zealand 0.221 <0.001 0.106 0.036 0.130 0.010

South Africa -0.200 <0.001 -0.116 0.001 0.367 <0.001

Switzerland NA -0.019 0.578 NA

Note: 'Cons/last 24 h' refers to cigarette consumption data obtained from the exit

questionnaire given at the end of the filter collection period.

'Cons/day' refers to cigarette consumption data obtained from the recruitment

questionnaire.

Cigarettes/day (recruitment questionnaire)

Fig. 1. Relationship between self-reported daily cigarette consumption at recruitment and self-reported cigarette consumption during the 24 h filter collection period (N = 3756; linear regression y = 0.613x + 6.39; r =0.518, p <0.001).

3.3. Analysis by country

Mean estimated tar and nicotine MLEs are presented by country in Table 3, on a per cigarette and per day basis. The per day values for seven of the eight countries have been calculated by multiplying each subject's per cigarette MLEs by their self-reported cigarette consumption in the last 24 h, reported when the filter collectors were returned to the research agency staff, since they are more relevant to the period during which the filters were collected than the daily cigarette consumption reported on recruitment. However, as these data were not available for the Swiss study the cigarettes per day data collected at recruitment were used to calculate MLEs per day.

Subjects from New Zealand and South Africa obtained the highest estimated tar and nicotine MLEs per cigarette and those from Brazil and Japan the lowest. This reflects the differences in the mean ISO tar and nicotine yields of the cigarettes from those countries used in the study. However, the mean MLEs were higher than the corresponding mean ISO yields in all cases. The relationships between mean MLEs and mean ISO yields for the eight countries are shown in Fig. 2 (tar) and Fig. 3 (nicotine). There were statistically significant correlations between mean MLE and mean ISO tar (r = 0.708, p < 0.05) and mean ISO nicotine and mean nicotine MLE (r = 0.865, p <0.01).

Table 3

Mean tar and nicotine MLE data, per cigarette and per 24 h, by country.

The rank order for per cigarette MLEs across the eight countries was to a large extent maintained for MLEs calculated on a per day basis. However, there was a notable change in rank order for Germany. As a result of a very low mean cigarette consumption rate 'in the last 24 h' German subjects obtained the lowest estimated mean daily MLEs.

Linear regression analyses were conducted on the relationships between the estimated tar and nicotine MLEs (per cigarette and per day) and the corresponding ISO tar and nicotine yields of the cigarettes smoked by the subjects (Table 4). Statistically significant positive correlations were obtained in all countries for tar/cigarette (r = 0.280-0.643), tar/day (r = 0.339-0.524), nicotine/cigarette (r = 0.401-0.616) and nicotine/day (r = 0.261-0.461). In seven of the countries the correlations for the MLEs per day were weaker than those for MLEs per cigarette.

The nicotine MLE per cigarette data for each country were grouped into four bands based on the ISO nicotine yields of the cigarettes. These bands were based on those used by Benowitz et al. (1986) and were <0.20 mg, 0.21-0.60 mg, 0.61-1.00 mg and >1.00 mg. The mean nicotine MLEs are plotted against the mean ISO nicotine yields for cigarettes in these four bands in Fig. 4. Data were not obtained for the <0.20 mg band in New Zealand, nor for the >1.00 mg band in Brazil, Germany, Switzerland and Japan.

Country ISO tar (mg) ISO nicotine (mg) Est. tar MLE (mg/ Est. nicotine MLE (mg/ Cigarettes in last Est. tar MLE (mg/ Est. nicotine MLE (mg/

cigarette) cigarette) 24 hA/C day) day)

Mean Mean Mean SD B Mean SD B N Mean SD B Mean SD B Mean SD B NA

Australia 6.6 0.61 12.8 4.6 c 1.36 0.50 c 705 21.5 11.1 bc 281 192 b 29.6 20.0 c 701

Brazil 5.8 0.54 11.0 4.3 a 0.93 0.34 a 607 18.9 9.9 b 208 135 a 17.5 11.3 a 597

Canada 8.3 0.79 12.0 4.4 b 1.32 0.48 c 690 22.4 8.2 c 271 151 b 30.1 17.0 c 685

Germany 6.3 0.55 14.8 5.6 d 1.25 0.42 b 767 12.2 6.5 a 196 138 a 16.4 11.1 a 673

Japan 5.1 0.43 12.1 5.4 bc 0.99 0.39 a 861 18.2 5.9 b 222 125 a 18.1 9.5 a 861

New Zealand 10.2 0.87 17.4 4.8 e 1.62 0.49 d 391 18.8 9.1 b 321 160 d 30.0 15.6 c 391

South Africa 9.1 0.89 17.2 6.7 e 1.77 0.69 e 860 18.4 6.2 b 307 146 cd 31.6 14.8 c 851

SwitzerlandC 6.2 0.55 14.9 5.5 d 1.32 0.44 c 822 19.3 9.1 b 288 174 bc 25.5 14.8 b 822

A Some cigarette consumption data from the preceding 24 h were not recorded: Australia 4, Brazil 10, Canada 5, Germany 94, Japan 0, New Zealand 0, South Africa 9. B The MLE values for countries having the same letter are not significantly different (p > 0.05; Tukey's test). C Cigarettes/day from recruitment questionnaire used for Switzerland estimated MLE per 24 h calculations.

£ 16 h £

-J 14 H

J AUS H-S-H

& BR CDN

ISO Tar yield (mg/cig)

Fig. 2. Relationship between mean (±SEM) estimated tar MLE and mean ISO tar yield per country (linear regression y = 0.971x + 7.039; r = 0.708, p = 0.049).

ig /ci

ê 1.2-1

ZA i-JH

CH m AUS ' Ï '

D i-iH CDN

J BR 1-fH

0.6 0.7 0.8

ISO Nicotine yield (mg/cig)

Fig. 3. Relationship between mean (±SEM) nicotine MLE vs mean ISO nicotine yield per country (linear regression y = 1.421x + 0.391; r =0.865, p = 0.006).

Table 4

Correlations (r values) between tar and nicotine MLE (per cigarette and per day) vs ISO tar and nicotine yields (all p < 0.001).

Country Tar/ Tar/day Nicotine/ Nicotine/day

cigarette vs vs ISO tar cigarette vs ISO vs ISO nicotine

ISO tar nicotine

Australia 0.481 0.433 0.456 0.394

Brazil 0.481 0.339 0.409 0.291

Canada 0.560 0.431 0.616 0.461

Germany 0.523 0.346 0.401 0.273

Japan 0.643 0.524 0.563 0.441

New 0.280 0.359 0.415 0.421

Zealand

South 0.557 0.360 0.561 0.317

Africa

Switzerland 0.573 0.365 0.457 0.261

3.4. Gender differences

Male smokers tended to obtain higher mean tar and nicotine MLEs per cigarette than female smokers in all eight countries (Table 5). These differences were statistically significant in all

countries except Japan and New Zealand. The mean ISO tar and nicotine yields of the cigarettes smoked by the males were also significantly higher than those smoked by females in all countries except Japan and New Zealand. Data averaged for all countries showed male smokers obtained 8.8% more nicotine MLE and 11.1% more tar MLE than female smokers but smoked cigarettes that were on average 9.3% higher in ISO nicotine and 11.4% higher in ISO tar yields than those smoked by the females.

3.5. MLEs and ¡SO tar and nicotine yields

The data from all countries have been combined in Figs. 5 and 6 to show the relationships between the estimated tar and nicotine MLEs for each subject plotted against the corresponding ISO tar and nicotine yields for the products. Linear regressions were calculated and significant correlations were found between estimated tar MLE and ISO tar yield (r = 0.547) and between estimated nicotine MLE and ISO nicotine yield (r = 0.591) (p < 0.001 for both).

There was somewhat higher variation in MLEs between subjects for the lowest yield products than for the highest yield products

Japan ♦ Aus

NZ ■ Brazil

S.Africa Canada

Switz X Germany

0.2 0.4 0.6 0.8 1

ISO nicotine yield (mg/cig)

Fig. 4. Mean nicotine MLE for each country grouped into four ranges of ISO nicotine yield.

(Table 6). The variation in MLEs was greater for the per day data than the per cigarette data.

MLEs for smokers of 66 mg ISO tar products were less than those for smokers of >6 mg ISO tar products, on a per cigarette and per day basis, for male and female subjects separately and combined (Table 7). MLEs for females were significantly less than males in all cases except nicotine MLE per cigarette for 66 mg ISO tar products (Table 8).

3.6. MLEs and cigarette blend style

The products were separated into flue-cured (Virginia tobacco) and American blend (Virginia, Burley and Oriental) styles and the MLE data analysed (Table 9). Overall, MLEs for smokers of flue-cured products were significantly greater than those of air cured product smokers, reflecting differences in ISO yields. After adjusting the MLE data for the differences in ISO yields, the nicotine and tar MLEs per cigarette were higher for the USB products than the flue-cured (both for all subjects and when separated by gender). However, the ISO yield adjusted nicotine MLE per day data were higher for Virginia products for all subjects combined and female subjects, but not males, and the ISO yield adjusted tar MLEs

Table 5

Mean tar and nicotine MLE data per cigarette, by country and gender.

Country Mean tar MLE (mg/cigarette) Mean nicotine MLE (mg/cigarette)

Male Female P Male Female P

Mean SD N Mean SD N Mean SD N Mean SD N

Australia 14.1 4.9 302 11.9 4.1 402 <0.001 1.46 0.53 302 1.27 0.46 402 <0.001

Brazil 11.5 4.2 331 10.5 4.5 266 0.009 0.96 0.33 331 0.89 0.34 266 0.007

Canada 12.7 4.4 386 11.0 4.1 300 <0.001 1.40 0.49 386 1.22 0.46 299 <0.001

Germany 15.8 5.9 364 14.0 5.2 399 <0.001 1.31 0.45 364 1.19 0.39 399 <0.001

Japan 12.3 5.3 665 11.8 5.7 196 0.321 1.00 0.38 665 0.94 0.41 196 0.063

New Zealand 17.9 5.1 162 17.2 4.6 229 0.160 1.65 0.52 162 1.60 0.46 229 0.323

South Africa 18.4 6.8 579 14.6 5.6 272 <0.001 1.85 0.70 579 1.59 0.63 272 <0.001

Switzerland 15.7 5.7 400 14.2 5.2 420 <0.001 1.38 0.46 400 1.27 0.42 420 0.001

Gender not recorded: Australia 1, Brazil 10, Canada 5, Germany 4, Japan 0, New Zealand 0, South Africa 9, Switzerland 2. * Two sample t test assuming unequal variances.

Fig. 5. Tar MLE data, by subject, from all markets. Each point represents the mean MLE for one smoker (N = 5703; linear regression y = 0.829x + 8.154; r = 0.547, p < 0.001).

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

ISO Nicotine yield (mg/cig)

Fig. 6. Nicotine MLE data, by subject, from all markets. Each point represents the mean MLE for one smoker (N =5703; linear regression y = 1.006x + 0.667; r =0.591, p <0.001).

Table 6

Coefficients of variation of tar and nicotine MLE data at different ISO tar yields.

ISO tar group 1 mg 2-9 mg P10mg

No products 12 63 31

Est. MLE data Mean SD CV(%) Subjects (n) Mean SD CV (%) Subjects (n) Mean SD CV (%) Subjects (n)

Tar (mg/cigarette) 7.8 3.5 44.1 612 13.2 4.7 35.9 3352 17.7 5.7 32.2 1739

Nicotine (mg/cigarette) 0.77 0.34 44.3 612 1.22 0.45 36.9 3352 1.7 0.56 33.4 1739

Tar (mg/24 h) 138 90.9 65.9 611 238 136 57.1 3329 343 177 51.6 1729

Nicotine (mg/24 h) 13.5 8.9 66.2 611 22.3 13.5 60.4 3329 32.9 17.5 53.2 1729

Table 7

MLEs for all subjects and by gender, for subjects smoking 66 mg ISO tar products and >6 mg ISO tar products.

All subjects Female Male

66 mg ISO tar >6 mg ISO tar p value 66 mg ISO tar >6 mg ISO tar p value 66 mg ISO tar >6 mg ISO tar p value

MLE/cigarette (mg) n 2676 3027 1278 1205 1381 1808

Nicotine Mean (SD) 1.06 (0.43) 1.53 (0.56) <0.001 1.05 (0.42) 1.46 (0.50) <0.001 1.08 (0.44) 1.58 (0.59) <0.001

Tar Mean (SD) 11.2 (4.6) 16.4 (5.6) <0.001 10.9 (4.4) 15.5 (5.0) <0.001 11.5 (4.7) 17.0 (5.9) <0.001

MLE/day (mg) n 2659 3010 1277 1203 1381 1807

Nicotine Mean (SD) 19.2 (12.4) 29.4 (16.7) <0.001 17.9 (11.8) 27.2 (16.2) <0.001 20.3 (12.8) 30.8 (16.8) <0.001

Tar Mean (SD) 200 (124) 312 (168) <0.001 183 (114) 284 (159) <0.001 216 (131) 330(172) <0.001

Table 8

MLEs from all products, 66 mg ISO tar and >6 mg ISO tar split by gender.

All products 66 mg ISO tar >6 mg ISO tar

Female Male p value Female Male p value Female Male p value

MLE/cigarette (mg) n 2483 3189 1278 1381 1205 1808

Nicotine Mean (SD) 1.25 (0.50) 1.36 (0.59) <0.001 1.05 (0.42) 1.08 (0.44) 0.175 1.46 (0.50) 1.58 (0.59) <0.001

Tar Mean (SD) 13.1 (5.2) 14.6 (6.0) <0.001 10.9 (4.4) 11.5 (4.7) <0.001 15.5 (5.0) 17.0 (5.9) <0.001

MLE/day (mg) n 2480 3188 1277 1381 1203 1807

Nicotine Mean (SD) 22.4 (14.9) 26.3 (16.1) <0.001 17.9 (11.8) 20.3 (12.8) <0.001 27.2 (16.2) 30.8 (16.8) <0.001

Tar Mean (SD) 232 (147) 281 (165) <0.001 183 (114) 216 (131) <0.001 284 (159) 330(172) <0.001

per day were higher for USB products for all subjects combined and male subjects, but not females.

4. Discussion

The technique of filter analysis provides estimates of the amounts of tar and nicotine generated per cigarette in a smoker's

normal smoking environment. This can also be described as a measure of a smoker's mouth level exposure to tar and nicotine resulting from each cigarette that is smoked. Daily mouth level exposure to tar and nicotine can also be estimated from filter analysis provided accurate measures of daily cigarette consumption rates are obtained during the filter collection period.

Table 9

MLEs from all products split by tobacco blend style.

Flue-cured American blend p value

ISO yields n 2251 3452

(mg/cigarette)

ISO yields n 2251 3452

(mg/cigarette)

Nicotine Mean (SD) 0.80 (0.37) 0.54 (0.25) <0.001

Tar Mean (sd) 8.29 (4.18) 6.18 (3.24) <0.001

MLE n 2254 3452

(mg/cigarette)

Nicotine Mean (SD) 1.53 (0.61) 1.17 (0.46) <0.001

Tar Mean (SD) 14.2 (5.64) 13.8 (5.79) 0.011

MLE (mg/day) n 2235 3434

Nicotine Mean (SD) 31.1 (17.4) 20.3 (12.8) <0.001

Tar Mean (SD) 291 (164) 239 (152) <0.001

To our knowledge this is the first time tar and nicotine exposure data obtained by the same method has been presented for smokers across a wide range of countries. Other studies have estimated nicotine intakes per cigarette for groups of smokers in the US (e.g., Gori and Lynch, 1985; Benowitz et al., 1986; Byrd et al., 1998), UK (e.g., Jarvis et al., 2001) and Japan (Ueda et al., 2002). These studies have measured indices of nicotine absorption, e.g., plasma nicotine (Gori and Lynch, 1985), plasma cotinine (Benowitz et al., 1986), salivary cotinine (Jarvis et al., 2001) and used estimates of daily cigarette consumption rates to convert these indices into nicotine intakes per cigarette. The use of different indices of nicotine exposure and potential problems with the accuracy of self-reported cigarette consumption makes it difficult to compare nicotine intakes per cigarette across different studies and countries. However, the filter analysis method provides per cigarette tar and nicotine data without the complications of the accuracy of cigarette consumption data or metabolic issues surrounding the index of nicotine exposure. The part-filter analysis technique used in the study has been shown to produce good estimates of the nicotine and tar yields for different cigarette types smoked under a wide range of human puffing behaviours (Shepperd et al., 2006).

In recent years two alternatives to the ISO/FTC method have been introduced to characterise mainstream cigarette smoke yields. One of these, the 'Canadian intense method' has been nominated by WHO TobReg as the smoking regime for the implementation of regulations designed to lower exposure to cigarette smoke toxicants (WHO, 2008; Burns et al., 2008). It is very likely that in future the ISO/FTC method will be replaced or supplemented by an alternative regime, and the data presented in our current paper may provide an opportunity for researchers to compare the performance of cigarettes under actual human smoking conditions with the performance under nominated smoking regimes.

Although the filter analysis method does not require daily cigarette consumption data in order to provide estimates of tar and nicotine exposures per cigarette, it does require such a measure to provide estimates of daily exposure. We obtained two estimates of daily consumption. One was the self-reported daily consumption figure given by the smokers during the recruitment process. The other was the self-reported consumption over the previous 24 h given in response to the question asked at the end of the filter collection period. As shown in Fig. 1 there was a considerable degree of variability between the two consumption measures and the correlations between the two measures were not strong, especially for New Zealand (r = 0.130), Canada (r = 0.344) and South Africa (r = 0.367). We used the 'consumption over the past 24 h' data to calculate daily tar and nicotine exposures as we believed

this to be more reliable than the data given at recruitment. As this was also a self-reported value there was no check on its accuracy, therefore the daily exposure data may still be subject to errors caused by the use of self-reported consumption rates.

The discrepancies between the two sets of consumption data obtained from the same smokers highlight the difficulties in obtaining accurate self-reported consumption data. This can cause major problems when estimates of daily nicotine exposure (from nicotine metabolites in body fluids) are divided by self-reported cigarettes per day to produce estimates of nicotine intake per cigarette. These derived per cigarette estimates are frequently used to compare the performance of cigarettes under human smoking conditions with those obtained under machine smoking regimes. Arguably, data obtained from filter analysis can provide a more reliable indicator of cigarette performance than nicotine per cigarette exposures derived from biomarker data as it does not require an accurate assessment of daily cigarette consumption rates.

Burns et al. (2001) analysed self-reported cigarette consumption data from the 1990 and 1996 California Tobacco Surveys and reported an inverse relationship between average daily cigarette consumption data and FTC nicotine yield of the cigarettes smoked by participants in the survey. These authors claimed daily cigarette consumption increased as the FTC nicotine yield fell below approximately 0.95 mg/cigarette.

A weak but statistically significant negative correlation between either measure of consumption and ISO nicotine yield was found for only the South African smokers in our study (Table 2). Either a weak, positive, significant correlation or no significant correlation between either measure of cigarette consumption and ISO nicotine yield was observed for seven of the eight countries (Table 2). These results are consistent with the results from a number of cross-sectional studies of groups of smokers of cigarettes differing in ISO/ FTC tar and nicotine yields. The majority of these studies did not find a negative relationship between ISO/FTC tar or nicotine yield and cigarette consumption rates. These include studies in the US (Folsom et al., 1984; Gori and Lynch 1985; Bridges et al., 1990; Djordjevic et al., 2000; Bowman et al., 2002; Hecht et al., 2005; St.Charles et al., 2006, 2009; Mendes et al., 2009), UK (Russell et al., 1980; Wald et al., 1981; Rawbone, 1984; Woodward and Tunstall-Pedoe, 1992; Jarvis et al., 2001), Germany (Sepkovic et al., 1990; Shepperd et al., 2009), Switzerland (Hofer et al., 1991), Italy (Rosa et al., 1992), France (Hee et al., 1995) and Japan (Ueda et al., 2002; Nakazawa et al., 2004). It should be stressed for most, but not all, of these studies cigarette consumption rates were self-reported and therefore may be subject to errors. The exceptions were St.Charles et al. (2006) and Shepperd et al. (2009) where exact accounting for daily consumption rates was assured.

Filter analysis provides measures of the amounts of tar and nicotine exiting the cigarette when a smoker takes puffs on the cigarette. This gives an indication of a smoker's mouth level exposure to tar and nicotine because each puff is achieved entirely as a result of a mouth action against a closed soft palate, and the smoke generated is initially contained within the mouth (Rodenstein and Stanescu, 1985; Fairweather, 1989; Dixon and Baker, 2003; Bernstein, 2004). Shortly after the end of the puff period most smokers relax their soft palates and draw the smoke remaining in the mouth into the lower respiratory tract during the inhalation process. Some of the smoke may be expelled from the mouth prior to inhalation (mouth spill), and some of the smoke may be exhaled. Additionally, for some smoke components, the site and degree of smoke component retention in the respiratory tract and the rate of absorption into the systemic circulation will be dependent on factors such as the depth of inhalation, smoke residence time in the respiratory tract and the chemical nature of the smoke component (Armitage et al., 2004; Baker and Dixon, 2006; Moldoveanu and St.Charles, 2007; Feng et al., 2007). Thus differences between

cigarettes in mouth level exposure of smoke components may not necessarily be reflected in differences in respiratory tract exposures or systemic absorptions.

Three recent studies have investigated the relationships between mouth level exposure to nicotine estimated by filter analysis and the systemic absorption of nicotine assessed by the measurement of nicotine metabolites in 24-h urine samples (St.Charles et al., 2006; Shepperd et al., 2009; Morin et al., this issue). Accurate records of the numbers and types of cigarettes smoked during the monitoring periods were kept by the researchers in these studies. St.Charles et al. (2006) compared estimated mouth level nicotine exposure (from part-filter analysis) with urinary total nicotine equivalents in 74 smokers of US cigarettes (FTC tar yields 119 mg) participating in an in-clinic study. They reported a r value of 0.91 for the correlation between the two measures of nicotine exposure. In a similar study using 140 German smokers of cigarettes ranging from 1 to 10 mg ISO tar yield, Shepperd et al. (2009) reported a r value of 0.83 for the correlation between mouth level exposure and total nicotine equivalents in 24-h urine samples. A further study using 142 Canadian smokers, smoking cigarettes ranging from 4 to 15 mg ISO tar yield found a correlation of 0.81 for the two measures (Morin et al., this issue).

These studies indicate that mouth level exposure to nicotine is a good predictor of nicotine uptake from cigarette smoke. This is because changes in puffing parameters (e.g., puff volume) are important determiners of nicotine uptake, but changes in inhalation depth and breath-hold time are not (Zacny et al., 1987). Nicotine retention in the respiratory tract is typically greater than 90% irrespective of differences in post-puff respiratory behaviour provided the smoker inhales (Armitage et al., 2004; Bernstein, 2004; Baker and Dixon, 2006; Feng et al., 2007). Consequently any change in puff volume which is sufficient to influence the amounts of nicotine absorbed from cigarette smoke will be detected as a change in nicotine MLE.

Cigarette smoke 'tar' is a complex mixture of chemical compounds and the deposition and retention characteristics of these compounds in the respiratory tract are dependent upon factors such as molecular weight and water solubility (Moldoveanu and St.Charles, 2007). Additionally, the deposition and retention of many of the compounds in 'tar' is influenced by differences in post-puff respiratory characteristics (Baker and Dixon, 2006; Feng et al., 2007). Consequently, unlike the situation with nicotine, it is not possible to directly relate differences between cigarettes in mouth level tar exposure to potential differences in the lower respiratory tract exposures to specific components of tar. Nevertheless, information on mouth level tar exposure from filter analysis could provide a useful indication of the performance of different cigarette types when smoked by individuals in their normal, everyday smoking environments.

The smokers in our study obtained a wide range of estimated tar and nicotine mouth level exposures for cigarettes with the same ISO tar and nicotine yield (Tables 2 and 3). This is due, mainly, to the well-documented wide range of puffing topographies, e.g., puff numbers, puff volumes, etc. for smokers of cigarettes of all types and yields. Consequently, ISO/FTC yields cannot be used to predict an individual smoker's exposure to tar and nicotine. This point was stressed by the FTC when their machine smoking method was introduced (FTC Press Release, 1967) and re-iterated by the UK Independent Scientific Committee on Smoking and Health in 1988 (ISCSH 4th Report, 1988).

Despite the wide inter-smoker variability in tar and nicotine exposures, for each country there were statistically significant correlations between estimated mouth level exposures per cigarette or per day and ISO tar and nicotine yields of the cigarettes smoked by the subjects (Table 4). Our correlations for mouth level nicotine exposure per cigarette vs ISO nicotine yield (r = 0.401-0.616) are

consistent with an r value of 0.50 reported by St.Charles et al.(2009) for the correlation between nicotine MLE per cigarette and FTC nicotine yield in a group of 784 US smokers. St.Charles et al. (2009) reported that the published correlations between the levels of nicotine uptake biomarkers and ISO/FTC nicotine yields tended to be lower than the MLE correlation. The majority of the nicotine biomarker vs nicotine yield correlations are based on the analysis of a single metabolite of nicotine and are statistically significant but in the range of 0.15-0.30 (see Table 3.1 National Institutes of Health, 2001). St.Charles et al. (2009) postulated that the lower correlation from those biomarker studies relying on a single metabolite of nicotine, e.g., cotinine in biofluids, may have resulted from increased inter-subject variability due to individual differences in the metabolism of nicotine. Inter-subject variability in nicotine metabolism is not an issue with the filter analysis method.

The relationships between MLEs and ISO nicotine and tar yields are evident in other aspects of our data analysis. First, as shown in Table 3 and Figs. 2 and 3, average tar and nicotine MLEs per cigarette for each country appeared to be related to the mean ISO yields of the products in the countries surveyed. Secondly, when the nicotine MLE data were grouped into four ISO nicotine yield bands (Fig. 4) there was a progressive increase in mean nicotine exposure across the ascending bands in most of the countries surveyed. The only exceptions being Australia where the mean nicotine MLEs were similar for 0.61-1.00 mg and the >1.00 mg bands, and Brazil where the 0.21-0.60 mg and the 0.61-1.00 mg bands were similar. Thirdly, the male smokers obtained higher mean tar and nicotine MLEs than the female smokers. The average ISO tar and nicotine yields of the cigarettes smoked by the males were also higher than those for the cigarettes smoked by the females.

Although mean MLEs tended to be higher in those countries where the average ISO tar and nicotine yields were higher than in those countries having lower average ISO yields there were also inter-country differences in MLEs that cannot be explained by differences in ISO yields. For example, as shown in Fig. 4, Brazilian smokers tended to obtain the lowest mean nicotine MLE per cigarette in the <0.2 mg, 0.21-0.60 mg and 0.61-1.00 mg ISO nicotine bands whereas South African smokers tended to obtain the highest mean nicotine MLE in the 0.21-0.60 mg, 0.61-1.00 mg and >1.00 mg bands. The reasons for these differences are unclear. It is possible that design differences between cigarettes from different countries are responsible for differences in mean MLE and/or that cultural differences in smoker behaviour may be responsible. Future studies involving smokers from one country smoking cigarettes from another country, e.g., Brazilians smoking South African cigarettes and vice versa, and the measurement of puffing behaviour patterns could provide an insight into the reasons for these between-country differences in MLE.

We observed significantly greater mean tar and nicotine MLEs per cigarette in male smokers than in female smokers in all countries except Japan and New Zealand (Table 5). The mean ISO tar and nicotine yields of the cigarettes smoked by the males were also significantly higher than those smoked by females in all countries except Japan and New Zealand. Consequently, differences in the yields of cigarettes may be a factor responsible for the gender difference in mouth exposures. However, smoking behavioural factors may have been involved, for example, in the few studies that have separately analysed male and female puffing topographies, males tend to take larger average and total puff volumes than females (Battig et al., 1982; Hofer et al., 1991; Hee et al., 1995; Eissenberg et al., 1999).

It has been suggested that some ingredients typically used in American blended style cigarettes, e.g., casings containing sugar, and flavourings, may enhance a smoker's exposure to cigarette smoke constituents by masking the irritant properties of cigarette

smoke and thereby making the cigarette more palatable (Bates et al., 1999; Henningfield et al., 2004). Casing and flavouring ingredients are not typically used in flue-cured cigarettes. The mean tar and nicotine MLEs tended to be lower for the American blend cigarettes (containing casings, etc.) than for the flue-cured cigarettes used in the study but this was also true for ISO yields. On the other hand the MLEs/cigarette expressed as ratios to ISO yields were higher for the American blend cigarettes. It is not possible to determine whether this was due to a blend effect on smoking behaviour or an effect caused by the lower ISO yields of the American blend cigarettes. With regard to this second point, both tar and nicotine MLEs/cigarette expressed as ratios to ISO yields rose steeply as ISO yields were reduced. For example the mean nicotine MLE/cig-arette to ISO yield ratios for the four nicotine yield bands used in Fig. 4 were 1.6 for the >1.0 mg band, 1.8 for the 0.6-1.0 mg band, 2.6 for the 0.21-0.6 mg band, and 7.3 for the <0.2 mg band. In order to avoid complications due to differences in ISO yields of the US blended and flue-cured cigarettes one would need to compare the exposures obtained from similar ISO yield cigarettes from the two blend styles, ideally in smokers from the same country. Unfortunately, we could only make two of these comparisons from our data-set. In Australia, an American blended cigarette (0.5 mg ISO nicotine yield) gave a mean nicotine MLE/cigarette to ISO ratio of 2.27, and the corresponding flue-cured cigarette (0.51 mg ISO yield) produced a mean ratio of 2.44. In New Zealand the mean nicotine MLE/cigarette ratio for an American blend cigarette (0.45 mg ISO yield) was 2.62 and the corresponding value for a flue-cured cigarette (0.47 mg ISO yield) was 2.53. The differences between the ratios were statistically insignificant in both instances. As our results on blend differences are not conclusive we would recommend that further work be conducted to determine the effect, if any, on ingredients on smoke exposure.

5. Conclusions

This is the largest study of its type reported to date, encompassing eight countries, 106 different cigarette products and 5703 smokers in total. It enables comparison with published data and provides the first estimates of smoke exposure for some countries. The data confirm the wide range of actual smoke exposures from a given product. Despite this variation, there were significant correlations between the estimated mouth level exposures per cigarette or per day and the ISO tar and nicotine yields of the cigarettes smoked. Generally, the mouth level exposures were higher in countries where the ISO yields of tar and nicotine were higher. Also, mouth level exposures were higher in males than in females, but this reflects the tendency of the males in this study to smoke higher ISO yield products.

Conflict of interest statement

The authors declare that there are no conflicts of interest. Acknowledgments

The authors would like to express their thanks to the MRAs who were the interface with the subjects, the laboratory staff at British American Tobacco Group Research and Development who undertook the filter analysis, particularly Pam Saunders, and Jon Shepp-ard and Graham Errington who undertook the preliminary data analysis.

References

Armitage, A.K., Dixon, M., Frost, B., Mariner, D.C., Sinclair, N.M., 2004. The effect of

inhalation volume and breath-hold on the retention of nicotine and solanesol in

the human respiratory tract and subsequent plasma nicotine concentration during cigarette smoking. Beitrage zur Tabakforschung International 21 (4), 240-249.

Ashton, H., Watson, D.W., 1970. Puffing frequency and nicotine intake in cigarette smokers. BMJ 3, 679-681.

Baker, R.R., Dixon, M., 2006. The retention of tobacco smoke constituents in the human respiratory tract. Inhalation Toxicology 18 (4), 255-294.

Baker, R.R., Dixon, M., Hill, C., 1998. The incidence and consequences of filter vent blocking amongst British smokers. Beitrage Zur Tabakforschung International 18 (2), 71-83.

Bates, C., McNeil, A., Jarvis, M., Gray, N., 1999a. The future of tobacco product regulation and labelling in Europe: implication for the forthcoming European Union Directive. Tobacco Control 8, 225-235.

Bates, C., Jarvis, M., Connolly, G., 1999. Tobacco Additives Cigarette Engineering and Nicotine Addiction. <http://www.ash.org.uk/papers/additives.html>.

Battig, K., Buzzi, R., Nil, R., 1982. Smoke yield of cigarettes and puffing behaviour in men and women. Psychopharmacology 76,139-148.

Benowitz, N.L., Jacob III, P., Yu, L., Talcott, R., Hall, S., Jones, R.T., 1986. Reduced tar, nicotine and carbon monoxide exposure while smoking ultralow but not low yield cigarettes. JAMA 256, 241-248.

Bernstein, D.M., 2004. A review of the influence of particle size, puff volume, and inhalation pattern on the deposition of cigarette smoke particles in the respiratory tract. Inhalation Toxicology 16, 675-689.

Bowman, D.L., Smith, C.J., Bombick, B.R., Avalos, J.T., Davis, R.A., Morgan, W.T., Doolittle, D.J., 2002. Relationship between FTC tar and urine mutagenicity in smokers of tobacco burning or eclipse cigarettes. Mutation Research 521,137149.

Bridges, R.B., Combs, J.G., Humble, J.W., Turbeck, J.A., Rehms, R., Haley, N.J., 1990. Population characteristics and cigarette yield as determinants of smoke exposure. Pharmacology, Biochemistry, and Behavior 37,17-28.

Burns, D.M., Major, J.M., Shanks, T.G., Thun, M.J., Samet, J.M., 2001. Smoking lower yield cigarettes and disease risks. In: National Cancer Institute. Risks Associated with Smoking Cigarettes with Low Machine-measured Yields of Tar and Nicotine. Smoking and Tobacco Control Monograph 13. US Dept. of Health and Human Sciences, National Institutes of Health, National Cancer Institute, NIH Pub. No. 02-5074, October 2001, Bethesda, MD.

Burns, D.M., Dybing, E., Gray, N., Hecht, S.S., Anderson, C., Sanner, T., O'Connor, R., Djordjevic, M., Dresler, C., Hainaut, P., Jarvis, M., Opperhuizen, A., Straif, K., 2008. Mandated lowering of toxicants in cigarette smoke: a description of the World Health Organization TobReg proposal. Tobacco Control 17,132-141.

Byrd, G.D., Davis, R.A., Caldwell, W.S., Robinson, J.H., deBethizy, D., 1998. A further study of FTC yield and nicotine absorption in smokers. Psychopharmacology 134, 291-299.

Comer, A.K., Creighton, D.E., 1978. The effect of experimental conditions on smoking behaviour. In: Thornton, R.E. (Ed.), Smoker Behaviour. Churchill Livingstone, Edinburgh, pp. 76-86.

Creighton, D.E., Noble, M.J., Whewell, R.T., 1978. Instruments to measure, record and duplicate human smoking patterns. In: Thornton, R.E. (Ed.), Smoking-behaviour: Physiological and Psychological Influences. Churchill Livingstone, Edinburgh, London and New York, pp. 277-288.

Dixon, M., Baker, R.R., 2003. Influence of cigarette design on human smoking-behaviour and sensory response. Recent Advances in Tobacco Science 29, 77-98.

Djordjevic, M.V., Stellman, S.D., Zang, E., 2000. Doses of nicotine and lung carcinogens delivered to cigarette smokers. Journal of the National Cancer Institute 92, 106-111.

Dwyer, R.W., Abel, S.G., 1986. The efficiencies of cellulose acetate filters. Beitrage zur Tabakforschung International 13, 243-253.

Ebert, R.V., McNabb, M.E., McCusker, K.T., Snow, S.L., 1983. Amount of nicotine and carbon monoxide inhaled by smokers of low tar, low nicotine cigarettes. JAMA 250, 2840-2842.

Eissenberg, T., Adams, C., Riggins III, E.C., Likness, M., 1999. Smokers' sex and the effects of tobacco cigarettes: subject rated and physiological measures. Nicotine and Tobacco Research 1, 317-324.

European Union Directive, 2001. 37/EC concerning the manufacture, presentation and sale of tobacco products. Official Journal of the European Communities.

Fairweather, F.A., 1989. The possible role of factors other than nicotine in compensatory-smoking. In: Wald, N., Froggatt, P. (Eds.), Nicotine and the Low Tar Programme. Oxford University Press, Oxford, pp. 212-219.

Federal Trade Commission Press Release, 1967. FTC to begin Cigarette Testing. Federal Trade Commission News Release. <www.philipmorrisinternational. com/global/downloads/SH/1967_FTC_press.pdf>.

Feng, S., Plunkett, S.E., Lam, K., Kapur, S., Muhammed, R., Jin, Y., Zimmermann, M., Mendes, P., Kinser, R., Roethig, H., 2007. A new method for estimating the retention of selected smoke constituents in the respiratory tract of smokers during cigarette smoking. Inhalation Toxicology 19,169-179.

Folsom, A.R., Pechacek, T.F., De Gaudemaris, R., Luepker, R.V., Jacobs, D.R., Gillum, R.F., 1984. Consumption of low yield cigarettes. Its frequency and relationship to serum thiocyanate. American Journal of Public Health 74, 564-568.

Forbes, W.F., Robinson, J.C., Hanley, J.A., Colburn, H.N., 1976. Studies on the nicotine exposure of individual smokers. The International Journal of Addiction 11, 933950.

Gori, G.B., Lynch, C., 1985. Analytic cigarette yields as predictors of smoke bioavailability. Regulatory Toxicology and Pharmacology 5, 314-326.

Hammond, D., Fong, G.T., Cummings, K.M., O'Connor, R.J., Giovino, G.A., McNeill, A., 2006. Cigarette yields and human exposure: a comparison of alternative testing regimes. Cancer Epidemiology, Biomarkers and Prevention 15, 1495-1501.

Hecht, S.S., Murphy, S.E., Carmella, S.G., Li, S., Jensen, J., Le, C., Joseph, A.M., Hatsukami, D.K., 2005. Similar uptake of lung carcinogens by smokers of regular, light, and ultralight cigarettes. Cancer Epidemiology, Biomarkers and Prevention 14, 693-698.

Hee, J., Callais, F., Mormas, I., Laurent, A.M., Min, S., Molinier, P., Chastagnier, M., Claude, J.R., Festy, B., 1995. Smokers behaviour and exposure according to cigarette yield and smoking experience. Pharmacology, Biochemistry, and Behavior 52, 195-203.

Henningfield, J.E., Benowitz, N.L., Connolly, G.N., Davis, R.M., Gray, N., Myers, M.L., Zeller, M., 2004. Reducing tobacco addiction through tobacco product regulation. Tobacco Control 13, 132-135.

Hofer, I., Nil, R., Battig, K., 1991. Nicotine yield as a determinant of smoke exposure indicators and puffing behaviour. Pharmacology, Biochemistry, and Behavior 40,139-149.

International Organization for Standardization, 1999. Cigarettes - Determination of Water in Smoke Condensates Part 1. Gas-chromatographic Method. ISO 10362:1999, Geneva, Switzerland.

International Organization for Standardization, 2000. Routine Analytical Smoking Machine - Definition and Standard Conditions. ISO 3308:2000, Geneva, Switzerland.

International Organization for Standardization, 2000. Cigarettes - Determination of Total and Nicotine Free Dry Particulate Matter using a Routine Analytical Smoking Machine. ISO 4387:2000, Geneva, Switzerland.

International Organization for Standardization, 2000. Cigarettes - Determination of Nicotine in Smoke Condensates. Gas-chromatographic Method. ISO 10315:2000, Geneva, Switzerland.

ISCSH, 1988. Fourth Report of the Independent Scientific Committee on Smoking and Health, Her Majesty's Stationary Office, London..

Jarvis, M., Boreham, P.P., Feyerabend, C., Bryant, A., 2001. Nicotine yield from machine-smoked cigarettes and nicotine intake from smokers: evidence from a representative population survey. Journal of the National Cancer Institute 93 (2), 134-138.

Mendes, P., Liang, Q., Frost-Pineda, K., Munjal, S., Walk, R.A., Roethig, H., 2009. The relationships between smoking machine derived tar yields and biomarkers of exposure in adult cigarette smokers in the US. Regulatory Toxicology and Pharmacology 55,17-27.

Moldoveanu, S.C., St.Charles, F.K., 2007. Differences in the chemical composition of the particulate phase of inhaled and exhaled cigarette mainstream smoke. Beitrage zur Tabakforschung International 22, 290-302.

Morin, A., Shepperd, C.J., Eldridge, A.C., Poirier, N., Voisine, R., this issue. Estimation and Correlation of Cigarette Smoke Exposure in Canadian Smokers as Determined by Filter Analysis and Biomarkers of Exposure.

Nakazawa, A., Shigela, M., Ozasa, K., 2004. Smoking cigarettes of low nicotine yield does not reduce nicotine intake as expected: a study of nicotine dependency in Japanese males. BMC Public Health 4, 28.

National Cancer Institute, 1996. Smoking and Tobacco Control. Monograph 7. The FTC Cigarette Test Method for Determining Tar, Nicotine, and Carbon Monoxide Yields of US Cigarettes. Report of the NCI Expert Committee. US Department of Health and Human Services. Public Health Service. National Institutes of Health. Bethesda, MD.

Ossip-Klein, D.J., Martin, J.E., Lomax, B.D., Prue, D.M., Davis, C.J., 1983. Assessment of smoking topography generalisation across laboratory, clinical and naturalistic settings. Addictive Behaviours 8,11-17.

Pauly, J.L., O'Connor, R.J., Paszkiewicz, M., Cummings, K.M., Djordjevic, M.V., Shields, P.G., 2009. Cigarette filter-based assays as proxies for toxicant exposure and smoking behavior - a literature review. Cancer Epidemiology, Biomarkers and Prevention 18 (12), 3321-3333.

Polzin, G.M., Wu, W., Yan, X., McCraw, J.M., Abdul-Salaam, S., Tavakoli, A.D., Zhang, L., Ashley, D.L., Watson, C.H., 2009. Estimating smokers' mouth level exposures to select mainstream smoke constituents from discarded cigarette filter butts. Nicotine and Tobacco Research 11, 868-874.

Rawbone, R.G., 1984. Switching to low tar cigarettes: are the tar league tables relevant? Thorax 39 (9), 657-662.

Rodenstein, D.O., Stanescu, D.C., 1985. Pattern of inhalation of tobacco smoke in pipe, cigarette, and never smokers. American Review of Respiratory Disease 132, 628-632.

Rosa, M., Pacifici, R., Altieri, I., Pichini, S., Ottaviani, G., Zuccaro, P., 1992. How the steady state cotinine concentration in cigarette smokers is directly related to nicotine intake. Clinical Pharmacology and Therapeutics 52, 324-329.

Royal College of Physicians, 1971. Smoking and Health Now. Pitman Medical and Scientific Publishing Co., London.

Royal College of Physicians, 1977. The Third Report from the Royal College of Physicians of London, Smoking or Health. Pitman Medical and Scientific Publishing Co..

Russell, M.A.H., Jarvis, M.J., Iyer, R., Feyerabend, C., 1980. Relation of nicotine yield of cigarettes to blood nicotine concentrations in smokers. BMJ 280, 972-976.

Russell, M.A.H., Jarvis, M.J., Feyerabend, C., Saloojee, Y., 1986. Reduction of tar, nicotine and carbon monoxide intake in low tar smokers. Journal of Epidemiology and Community Health 40, 80-85.

Scherer, G., 1999. Smoking behavior and compensation: a review of the literature. Psychopharmacology 145,1-20.

Scherer, G., Engl, J., Urban, M., Gilch, G., Janket, D., Riedel, K., 2007. Relationship between machine-derived smoke yields and biomarkers in cigarette smokers in Germany. Regulatory Toxicology and Pharmacology 47,171-183.

Schulz, W., Seehofer, F., 1978. Smoking-behaviour in Germany - the analysis of cigarette butts (KIPA). In: Thornton, R.E. (Ed.), Smoking-behaviour: Physiological and Psychological Influences. Churchill Livingstone, Edinburgh, London and New York, pp. 259-276.

Sepkovic, D.W., Colosimo, S.G., Axelrod, C.M., Haley, N.J., 1990. Long term use of low nicotine yield cigarettes. Poster Presentation at 'Effects of Nicotine on Biological Systems' Symposium, Hamburg, Germany.

Shepperd, C.J., St.Charles, F.K., Lien, M., Dixon, M., 2006. Validation of methods for determining consumer smoked cigarette yields from cigarette filter analysis. Beitrage Zur Tabakforschung International 22 (3), 176-184.

Shepperd, C.J., Aldridge, A.C., Mariner, D.C., McEwan, M., Errington, G., Dixon, M., 2009. A study to estimate and correlate cigarette smoke exposure in smokers in Germany as determined by filter analysis and biomarkers of exposure. Regulatory Toxicology and Pharmacology 55, 97-109.

St.Charles, F.K., 2001. A robust method for determining consumer smoked cigarette yields from filter analytical data. In: 55th Tobacco Science Research Conference, Program Booklet and Abstracts, vol. 55, paper 92, pp. 73-74.

St.Charles, F.K., Krautter, G.R., Dixon, M., Mariner, D.C., 2006. A comparison of nicotine dose estimates in smokers between filter analysis, salivary cotinine, and urinary excretion of nicotine metabolites. Psychopharmacology 189 (3), 345-354.

St.Charles, F.K., Kabbani, A.A., Borgerding, M.F., 2009. Estimating tar and nicotine exposure: human smoking versus machine smoking. Regulatory Toxicology and Pharmacology. doi:10.1016/j.yrtph.2009.08.11.

Stephen, A., Frost, C., Thompson, S., Wald, N., 1989. Estimating the extent of compensatory smoking. In: Wald, N., Froggatt, P. (Eds.), Nicotine, Smoking and the Low Tar Programme. Oxford Medical Publications, Oxford, pp. 100-115.

Swan, C., Froggatt, P., 1996. The Tobacco Products Research Trust 1982-1996. Royal Society of Medicine Press.

Ueda, K., Kawach, I., Nakamura, M., Nogami, H., Shirokawa, N., Masui, S., Okayama, A., Oshima, A., 2002. Cigarette nicotine yields and nicotine intake among Japanese male workers. Tobacco Control 11, 55-60.

US Department of Health and Human Services, 1981. The Health Consequences of Smoking. The Changing Cigarette: A Report of the Surgeon General.

US Department of Health and Human Services, National Institutes of Health, 2001. Smoking and Tobacco Control, Monograph 13, Risks Associated with Smoking Cigarettes with low Machine-measured Yields of Tar and Nicotine.

US Public Health Services, 1966. Technical Report on Tar and Nicotine. November 25.

Wald, N., Idle, M., Boreham, J., Bailey, A., 1981. Carbon monoxide in breath in relation to smoking and carboxyhaemoglobin levels. Thorax 36, 366-369.

WHO Study Group on Tobacco Product Regulation, 2004. Recommendation 1: Guiding Principles for the Development of Tobacco Product Research and Testing Capacity and Proposed Protocols for the Initiation of Tobacco Product Testing. World Health Organization, Geneva, Switzerland. <http:// www.who.int/tobacco/globaLinteraction/tobreg/goa_2003_principles/en/>.

WHO Study Group on Tobacco Product Regulation, 2008. The Scientific Basis of Tobacco Product Regulation; Second Report of a WHO Study Group; WHO Technical Report Series 951, ISBN:978 92 4 120951 9. <http://www.who.int/ tobacco/global_interaction/tobreg/publications/tsr_951/en/index.html> (accessed September 2009).

Wilkenfeld, J., Henningfield, J., Slade, J., Burns, D., Pinney, J., 2000. It's time for a change: cigarette smokers deserve meaningful information about their cigarettes. Journal of the National Cancer Institute 92, 90-92.

Woodward, M., Tunstall-Pedoe, H., 1992. Do smokers of lower tar cigarettes consume lower amounts of smoke components? Results from the Scottish Heart Health Study. British Journal of Addiction 87, 921-928.

Zacny, J.P., Stitzer, M.L., Brown, F.J., Yingling, J.E., Griffiths, R.R., 1987. Human cigarette smoking: effects of puff and inhalation parameters on smoke exposure. The Journal of Pharmacology and Experimental Therapeutics 240 (2), 554-564.