Dear friends,

This is in response to Mike Daube and Fenton Howell's request for background on the Covance laboratories study into non-smokers' exposure to ETS. This was reported in the Sunday Telegraph of 16 August 1998 and picked up around the world. The study used personal air quality monitors mounted on volunteers to measure their exposure to ETS as they went about their normal business. The newspaper report suggests that exposure is one thousandth of that of a smoker - equivalent to 6 cigarettes per year - 10 times lower than Government backed estimates. It is another attempt to challenge the increasingly robust consensus surrounding the health risks of passive smoking. However, it looks as though the results were misinterpreted.... read on.

Outline of study

The study was undertaken by Covance Labs of Harrogate, UK and the client was the Centre for Indoor Air Research of the US. We know the CIAR is a tobacco industry front - so this is tobacco funded research. The study covers 12 urban centres in Europe and Asia and c. 200 volunteers had personal air monitors attached for 24 hrs. Exposure to ETS was measured using a variety of indicators: ETS particles were collected in the monitors and analysed using ultra violet absorbance, flourescence and solanesol measurements. Saliva cotinine was used, but only as a means of checking that the subjects were not smokers. In Stockholm (Sweden) it is claimed that the most extreme exposure found was the equivalent of smoking 6-9 cigarettes per year. In smoky Lisbon (Portugal) a the upper figure was 20. (These figures turn out to refer to the median and 90th percentile exposures respectively, in the most exposed categories - and I think this is the most obvious weakness - see 2 below for explanation).

The main problems with this research and its reporting are the following (others may have more to say):

1. Conflicts with established evidence

The research is at sharp variance with more direct measurements of the body's exposure to ETS - which are in the form of measurements of the breakdown products of nicotine (cotinine) in the blood or saliva. There is a summary in the 1993 US-EPA report on passive smoking: 3.3.2 Biomarkers of ETS Exposure. On page 3-46 it says:

"an upper bound estimate of nicotine dose of 2.5 mg/day for passive smoke exposure has been proposed (Jarvis, 1989). This would translate to the equivalent of about one-fifth of a cigarette per day or 0.7% of the average smoker's dose of nicotine. [...] comparisons of cotinine values in ETS-exposed nonsmokers with those measured in smokers ranged from 0.1% to 2%. One analysis proposed that, on average, nonsmokers' cotinine levels are 0.5-0.7% of those found in cigarette smokers (Jarvis, 1989 *)."

Martin Jarvis adds (for this note): "studies based on cotinine measurements in non-smoking children exposed in the home continue to show nicotine intake equivalent to smoking 100-150 cigarettes per year where both parents smoke*. Although exposure levels in adults are lower than this in general, studies in particular groups, (eg. non-smoking adults working in smoky bars) show nicotine intakes as high as half a cigarette per day.**"

*Javis M., 1989 - Application of biochemical intake markers to passive smoking measurement and risk estimation. Mutat. Res. 222:101-110.
** Jarvis, Foulds, Feyeraband, 1992, Exposure to passive smoking among barstaff, British Journal of Addiction, 1992 vol 87 p.111-113)

The Covance studies made cotinine measurements and one of the researchers told me they were at variance with the air-monitor measurements. In the Scandanavian study, the authors conclude that " these findings again reinforce our previous suggestion that saliva cotinine measurements should not be used to assess exposure to ETS at low levels." It doesn't look as though they considered the alternative... "personal air monitors should not be used....". This is discrepancy is important because it means that some e xplanation need to be proposed for the difference between measurements and we can postulate that the problem lies in the air monitor measuring methodology. Had the cotinine measurements been in agreement, then we would be looking for reasons why the exposure is less than found in previous studies.

Scientists have conducted throrough examinations of the the use of cotinine as a marker at low exposures and found that it is the best available. For example: Benowitz N. Cotinine as a biomarker of environmental tobacco smoke exposure, Benowitz, Epidemiologic Reviews. 1996 vol 18 188-204. "The evidece presented in this review indicate that cotinine levels provide a valid an quantitative measure of ETS exposure over time. Cotinine is clearly the best available biomarker of ETS exposure at present."

2. Obvious questions about the studies:

2a. Reporting of results - I suspect this is the key problem. The random sample of c. 200 subjects is divided into 6 categories - with various combinations of smoking/non-smoking at work/home. Measurements are reported for the subjects in each category, but the analyst needs to find a way of characterising results for the category as a whole. One way to characterise data like this is to place all the results within each category in order of increasing exposure and then state the median (the middle result) or the 90th percentile (in a sample of 200, this would be the 180th result).

For the Stocholm study, the MEDIAN value for exposure to spouse smoking was used to generate the Sunday Telegraph headline stating that passive smoking is responsible for smoking the equivalent of 6 cigarettes per year. However, half the sample would be more exposed than this - some of them, very significantly more. I guess the Sunday Telegraph claimed this was the most exposed because this CATEGORY was the most exposed of the six. The quote from the Stockholm study abstract is:

"When the median values for nicotine and ETS particles are converted to cigarette equivalents, Stockholm housewives and househusbands living with smokers would receive 6-9 cigarette equivalents per year, working nonsmokers living with smokers would receive 0.6-0.7 cigarette equivalents at home and nonsmokers working with smokers would be exposed to 0.1-0.2 cigarette equivalents at work"

For the Lisbon study, for each category the exposure is reported as the median value and the 90th percentile for each of the four measurement techniques used. This means the 10% most highly exposed in each category are not reflected in the reported results. As an example, in the Lisbon measurements of ETS particles by using flouresence in the fifth category (people that have non-smoking home but smoking at work), the results are reported as follows:

Number in this category: 61 (of the 197 in the study) Median: 1.1 micrograms/cubic metre (this is roughly the measurement of the 31st subject ranked by measured exposure) 90th Percentile: 25 micrograms/cubic metre (this is roughly the measurement of the 55th subject).

The maximum exposure (the 61st subject) is not stated in the table or main results, but may be 'lifted' from a chart and shows the most exposed subject to have a reading of over 90 micrograms/cubic metre - almost four times the figure used to characterise maximum exposure in the study. Six of the subjects in this category will be above 25 micrograms, but the experience of these subjects is excluded from the results because of the decison to characterise the most exposed group as the 90th percentile. Obviously, the most exposed subject experienced around 90 times the exposure of the MEDIAN.

For the Lisbon study, the charts for all the measurements rise steeply after the 90th percentile - and this perhaps reflects the common sense experience that high exposure is concentrated in particular at-risk groups within the population. Is it right to characterise "the most highly exposed" as the 90th percentile value? What they have done is not exactly incorrect, it is just an inapproriate way to characterise this data. When Sunday Telegraph used this to claim that the most highly exposed passive smokers only smoke 6 cigarettes based on the Stockholm study, they weren't even using the 90th percentile figure.

Other problems.....

2b. Does the monitor accurately measure cumulative exposure to tobacco smoke?
For example, do the monitors exchange air with their surroundings in the same way the subject does when breathing? A pump is used, but does this give an accurate simulation?

2c. Does the study accurately reflect the absorption of inhaled tobacco smoke by the body?
The actual exposure to the hazardous substances in ETS depends on how readily these are absorbed in the lungs - for a given quantity of tobacco smoke taken into the lungs, is more absorbed through a 24 hour low level chronic exposure (passive smoking) or a short acute exposure to the same quantity of smoke (active smoking)? Or would it be the same? I cannot really see how this study can capture this variable, whereas it is obviously captured by measurements of the breakdown products of absorbed tobacco smoke such as cotinine. (By the way, I don't actually know if this IS a significant effect - I'm just asking the question...).

2d. Choice and size of sample.
Are the subjects chosen and the places they visit typical of at-risk groups? The studies use c. 200 subjects for 24 hrs. These are randomly selected adult non-smokers - so the study is not focussed on at-risk groups. In the Stockholm study more than half (59%) of the sample (108/182) were not exposed to smoking either at home or work. Thus the exposed group was 74 subjects of which 53 were only exposed at work and 21 at home and/or work. The definition of a smoking workplace is one where there is a smoking co-worker within 30 metres and 21% (11) of those working in a smoking workplace reported no smoke around them. This all seems to add up to be a small sample - especially if the concern is with specific at-risk groups.

3. Tobacco funding concealed

The research is funded by the tobacco industry - the Sunday Telegraph reports Dr. Keith Phillips of Covance coyly obfuscating this as follows: "They [some scientists] try to dismiss it by arguing that our research receives support from.... the tobacco industry. Our findings are completely independent of any influence from the tobacco industry." Covance has confirmed that the client is the Center for Indoor Air Quality Research - a creation of the tobacco industry and this is disclosed in the Environment International paper.

4. Balancing views absent

The Sunday Telegraph failed to report an extensive conversation between an expert in this field (Martin Jarvis) and the journalist Robert Matthews in which Jarvis was making the opposite case. We don't know if this was due to sub-editing or the journalist himself. What is clear is that the Sunday Telegraph is mounting a campaign against the scientific basis of concern about passive snoking and has become a ready conduit for tobacco industry stories. This spills out through the Conrad Black empire and via the Electronic Telegraph which is used as a source for Monday morning stories.

5. Mode of reporting

The mode of this disclosure should ring alarm bells in every newspaper office with even a fleeting concern for truthful reporting. Questions a good editor should ask before repeating tobacco stories from the Sunday Telegraph or tobacco industry:

• Has this been published in established literature? (This was, but the report was)
• If not, are copies available to journalists, critics in advance of publication? (This was a Sunday exclusive)
• Is a PR company or tobacco company promoting the research? (It is tobacco funded, albeit through a front organisation)
• Is the author available for comment or criticism? (no)
• Is the timing linked to an external political objective? (This is linked to a government White Paper and SCOTH report)
• Is the news value that it challenges a consensus then some weighing of the existing evidence is needed? (absent here)
• Has the journalist looked for and reported a dissenting view? (not reported)

Staying open-minded

Finally, we should not have closed minds and assume that all that challenges the consensus is wrong or misguided. It just may be that cotinine measurements are wrong at low ETS exposures. It may be that exposure is declining. Let's hope it is! It is just that a serious challenge to the consensus should be mounted using the established scientific approach, not through the tell-tale approach of tobacco industry PR.

Clive Bates
Director
Action on Smoking and Health
16 Fitzhardinge Street
London W1H 9PL - United Kingdom
Tel: +44 (0)171 224 0743 - Fax: +44 (0)171 224 0471
Web: http://www.ash.org.uk

GLOBALink Information and Communication in Tobacco-Control