Article Text
PDF
Statistics from Altmetric.com
In their recent article Kozlowski and O'Connor1 criticise a 1997 review2 on cigarette filter ventilation blocking and claim it is in error because it (1) relies on saliva based estimates, (2) does not consider degree of ventilation, (3) does not address brand-to-brand variation, and (4) omits certain tobacco industry studies. We disagree and stand by our conclusions.2,3
In their criticisms Kozlowski and O'Connor refer only to the 1997 review2 presented at a conference and not a peer reviewed article published in early 2001.3 In the latter review, Dr Baker and I considered measurement techniques, effects of vent blocking on machine smoke yields, effects of vent blocking on human smoke yields, and simultaneous determination of vent blocking and smoke yields. We concluded that vent blocking among smokers has only a relatively minor effect on human smoke yields compared to other smoking behaviour factors.3 The large effects observed with smoking machines are misleading because people do not smoke like machines.
Concerning the allegation that we erred because of our reliance on saliva based estimates, the facts are that we discussed the pros, cons, and limitations of all techniques used to estimate the extent of vent blocking.3 We reported that four studies by Kozlowski and colleagues, using the “tar”' stain technique, indicate that 50–59% of the 14 to 158 filters examined in each study showed some degree of vent blocking. Two other studies,4,5 using the same technique but each based on over 1000 filters, indicate that 21–30% of the filter vents examined were blocked, and most were only partially blocked.4 These latter studies are in reasonable agreement with large studies conducted by industry scientists using the saliva stain technique,3 which indicate that up to 24% of filters examined were blocked by lips, and again, most only partially. Direct video observation indicates finger blocking is negligible since most smokers release their fingers from the cigarette as they take a puff,2,3 but it would be virtually impossible to determine from the video whether smokers' lips had covered the vents.
We devoted a large part of our 2001 review3 to considering the degree of filter ventilation across a number of cigarette brands (cf. allegations 2 and 3). Reassuringly, some of the latest results from Kozlowski et al and industry scientists are in reasonable agreement, despite the very different experimental techniques used.
Kozlowski and O'Connor state that “one notable omission” from the 1997 review2 is a 1982 study of a 1 mg “tar” cigarette smoked under various puffing conditions6 (allegation 4). In fact, data from that study are plotted in fig 8 of the 1997 review.2 We attribute the results to RP Ferris, the project leader, rather than T Hirji, the author of the memo, but it is the same study. They quote the smoke yields from the study1 but fail to notice that the data are the same as those in our review.2
Likewise, Kozlowski and O'Connor say that we ignored pertinent Swiss7 and Canadian8 studies, but data summaries are included in our 1997 review.2 Our 2001 review3 quotes both studies as indicating a dependence of insertion depth on “tar” yield (that is, degree of ventilation). Kozlowski and O'Connor1 concentrate on the less detailed unpublished Swiss data but virtually ignore similar trends pointed out in the more comprehensive data published by Baker et al.9 (Kozlowski and O'Connor even re-plot some of the Swiss data to emphasise their point, ignoring the fact that these data were obtained using the saliva stain technique that they criticise elsewhere1).
Kozlowski and O'Connor correctly state that we did not mention a 1977 study by Creighton.10 They quote from this report that “[o]ne subject was seen to cover the ventilation holes with clear adhesive tape”. They fail to mention, however, that the “subjects” in this study were R&D scientists evaluating two competitors' filter ventilated cigarettes. Such ad libitum experimentation with the innovative (for 1977) filter design is exactly what one would expect of industry scientists. This experimentation is irrelevant to the behaviour of consumers, and there is nothing more in the report about vent blocking. We considered this report of no relevance to our reviews.
Kozlowski and O'Connor state that we have “ignored the extensive machine smoking studies by Rickert and colleagues on Canadian cigarettes”.11 We cite this study3 and discuss smoking machine data at length. Rickert et al used only one ventilation blocking condition (50%) and the studies we chose to consider used multiple vent blocking conditions.
Finally, Kozlowski and O'Connor also refer to Philip Morris reports not covered in our reviews. In fact, we did not know of their existence until recently. The topics of those memos are comprehensively covered by similar studies in our reviews, and add nothing new.
Kozlowski and O'Connor lament the fact that they cannot find on the internet some industry studies used in our reviews. Not all tobacco companies are obligated to post their internal documents on the internet. Also, as they mention, the internet databases are constantly updated and some documents may not be posted at the time of a given search.
Kozlowski and O'Connor criticise our 1997 review2 because we did not refer to certain unpublished industry studies. Yet when we sent our updated review for publication in Psychopharmacology, the manuscript was rejected on the advice of a reviewer who said it was too dependent on unpublished industry studies (and whose comments read, coincidentally, very much like the Kozlowski and O'Connor article1). It is therefore interesting that over 60% of Kozlowski's and O'Connor's references are unpublished industry documents. Many of these are short memos written for internal use, not complete research reports, and assessment by those not involved can lead to misleading conclusions, such as the discrepancy in attribution noted with Ferris and Hirji. It is very difficult to place these documents in proper context, and, in some cases, to try to do so nearly 50 years after they were written.
References
Authors' reply
Lewis takes us to task for criticising an article published in 19971 by noting that we ignore new points they made in a paper published, unknown to us, in an industry sponsored journal.2 We learned of this publication a year after our paper was accepted for publication.
Lewis implies that we had reviewed an earlier submission of their paper to Psychopharmacology. We did review this draft, but were not privy to its fate. Journal rules and professional ethics require that the information in their submitted paper be treated as confidential, and we did not mention or make use of any of this confidential draft in our articles. That Lewis and Baker publish a revised paper that was informed by our thinking and suggestions on the topic should hardly be an occasion for criticising our discussion of a work1 that had not been informed by our advice.
Our paper appeared in a special journal issue dealing with available industry documents. Ideally, review articles should derive from published, peer reviewed research. Failing that, public availability (as on the internet) of the primary reports should be expected. But when industry scientists (here from RJ Reynolds and British American Tobacco) characterise internal reports—that may not be or ever become available on the web—the opportunity for independent evaluation of findings may be lacking. Presumably, industry scientists have the ability to bring primary source internal research to peer reviewed publication. For non-industry scientists, in contrast, industry documents on the web are likely all that is available. In other words, we are limited to discuss those findings that are open to public view, while they are in a position to characterise studies to which independent scientists have no access. It would be best if all studies used to support or refute findings were available to all interested parties, preferably through peer reviewed publication.
Figure 8 in their 1997 paper,1 which they attribute to Ferris, is related to data that we attribute to Hirji.3 Compared to the Hirji version, their fig 8 contains both more data (another blocking condition) and at the same time significantly less data (for example, no mention of results from a 75 ml puff in 1 second every 25 seconds, that produces from a nominal 1 mg total particulate matter (TPM) cigarette a TPM yield of 15 mg with no blocking and 23 mg TPM with a 50% block.1 The Hirji report3 mentions by name the individuals who did the work, and Ferris is not mentioned.
Lewis writes that Creighton4 used industry scientists (as was noted in the version we have) who could be expected to conduct “ad lib experimentation” with the then innovative filter design. One of these scientists/ad hoc experimenters dropped out of the study after a day because of “an unpleasant taste in the mouth, persistent irritation and lack of satisfaction” (page 5).4 Why Creighton did not report that he received testimony from his colleagues that abuses were happening, rather than having to “observe” or write that “one subject was seen to cover up the ventilation holes” with tape, is interesting.
Lewis engages us particularly on the issue of vent blocking—a theme we think is less important overall than taste and puff volume, and probably only important for less common heavily ventilated cigarettes. (We never say the saliva based measures of blocking are worthless, just much less sensitive.) In their recent paper,2 they go into some puff volume data, but for them, interestingly, the blocked vent results (smaller puffs, fewer puffs) are caused by under-puffing on blocked cigarettes rather than over-puffing on unblocked cigarettes. Their rhetoric encourages us to see a self protecting smoker, rather a compensating smoker. Nice try!
The data in their more recent paper2 also support the position that filter ventilation is a defective and dangerous design that contributes to the misleading nature of standardised testing of cigarettes.