Monday, January 12, 2009

Original Study on Thirdhand Smoke Does Not Support Conclusion that Health Threat is Significant

It is important to evaluate the original study which is being used to support the idea that thirdhand smoke poses a significant risk to nonsmokers, especially to children and infants. That study appeared in Tobacco Control in 2004 (see: Matt GE, et al. Households contaminated by environmental tobacco smoke: sources of infant exposures. Tobacco Control 2004; 13:29-37).

The study compares airborne nicotine levels and urine cotinine levels in rooms and infants, respectively, in three settings:

1. Homes in which there are no smokers (referred to in the study as "no exposure").
2. Homes in which there are smokers, but they report making great efforts not to expose the infant (referred to in the study as "indirect exposure").
3. Homes in which there are smokers who do not report making great efforts not to expose the infant (referred to in the study as "direct exposure").

The major findings of the study are that air nicotine levels were 2-3 times higher in homes under the indirect exposure conditions as compared to the no exposure condition and that urine cotinine levels of infants living in indirect exposure conditions were about 8 times higher than among infants living in no exposure conditions. Based on these findings, the article (and subsequent articles which use this as a basis) concludes that thirdhand smoke is a significant health hazard.

The Rest of the Story

There are so many aspects of the rest of the story that it's difficult to know where to begin. To make it easier to follow, let me number each point.

1. Twice As High As a Miniscule Number is Also a Miniscule Number

While at first glance it may sound dangerous to have airborne nicotine levels that are twice as high as in a nonsmoking home, the level of nicotine in nonexposure homes is so low that even at twice this level, it does not appear to represent any substantial health threat. The average airborne nicotine level in nonexposure homes was about 0.1 ug/m3 (micrograms per cubic meter). In indirect exposure homes, the average nicotine level was between 0.2 and 0.3 ug/m3. These are extremely low levels; in fact, they are lower than levels which have been measured in many nonsmoking environments. So the simple fact that nicotine levels in indirect exposure conditions was twice as high as in direct exposure conditions does not necessarily mean that thirdhand smoke is a substantial health risk. In fact, based on the nicotine levels reported in the paper, it does not appear that thirdhand smoke results in airborne exposure to nicotine that is high enough to represent a significant health risk.

2. Some of the Indirect Exposure Homes in the Study Were Almost Certainly Direct Exposure Homes

The study relies upon self-report to determine whether parents smoke in the home or not. Almost certainly, parents are going to under-report smoking in the home. Given the societal scorn that is placed upon parents who expose their children to secondhand smoke (believe it or not, some anti-smoking groups and advocates go so far as to call them child abusers), it is not surprising that in a survey, parents are going to be hesitant to admit that they do smoke in the home around the child.

Thus, the indirect exposure condition almost certainly contains a number of homes in which the parents do smoke around the child. This misclassification severely biases the results of the study and makes it impossible to conclude that the levels of exposure of infants in "indirect" exposure homes is actually due to thirdhand smoke, rather than to secondhand smoke.

Ironically, one way to verify the presence of absence of smoking in the household would be to measure ambient nicotine levels, which would indicate whether smoking is taking place or not. So ironically, this study uses as an outcome measure a variable that actually should have been used as a verification of the parents' non-smoking status in the home.

To make matters worse, most of the homes had multiple smokers, so even if the parent interviewed was telling the truth, it may have been the case that another household member smoked inside the home. Even worse, a large proportion of these infants were exposed directly to smokers outside the home, which could well explain their elevated cotinine levels. In fact, in the indirect exposure group, one-third of the respondents reported that the infant had visited someone who smokes in the past 30 days. Thus, these subjects should not have been used in the study, especially for the measurement of cotinine levels, because it messes up the results.

While many parents in the indirect exposure group reported not smoking in the home in the past 30 days, 8% admitted to smoking in the home occasionally. These cannot be used as true indirect exposure homes, since the parent is admitting to smoking inside the home.

The bottom line is that this study is almost certainly not measuring the true exposure of infants who live in conditions in which parents never smoke in the home. These results therefore cannot be used reliably to evaluate the risk of thirdhand smoke.

3. Examining the Minimum Exposure Levels Appears to Confirm that if Stringently Adhered to, Smoking Outside the Home Results in No Significant Airborne Nicotine

Given the likelihood that many of the indirect exposure homes were actually direct exposure homes, it now becomes important to examine the minimum airborne nicotine levels in these homes. Importantly, the minimum airborne nicotine level in an indirect exposure home was about 0.1 ug/m3. This is no different from the airborne nicotine level in a nonexposure home. It therefore suggests that it is possible to smoke outside the home and to have the infant exposed to no significant airborne nicotine inside the home.

More provocatively, perhaps what the range of airborne nicotine levels indicates is the degree to which smoking is taking place inside the home, rather than the differences between homes, all of which represent the complete absence of any smoking.

The bottom line here is that it is possible for a smoker to protect the infant from any significant airborne nicotine exposure by smoking outside the home and perhaps taking other measures (such as ensuring that doors and windows are shut when smoking).

It is also interesting to note that the minimum level of nicotine in dust from the indirect exposure homes was 0. Again, this confirms that it is possible to have no infant airborne or dust nicotine exposure, even though the parent is a regular smoker.

This same pattern of results held for urine cotinine measurements as well.

4. The Study Does Not Acknowledge Any of These Limitations

Most concerning and worrisome to me is that the paper does not acknowledge as a limitation the possibility that parents are not being accurate in their reporting of smoking inside the home. We know that there is significant under-reporting even of just a person's smoking status. In other words, people are reluctant to even admit that they smoke, much less to admit that they smoke in the presence of infants. If there is significant under-reporting of smoking status, then the under-reporting of smoking in the presence of infants has to be very substantial.

That this is not even mentioned in the paper is concerning and has two important implications. First, it renders the conclusions of the study invalid. If researchers cannot defend their conclusions against the most plausible and likely limitation such as this, then there are plausible alternative explanations of the study findings, and the study conclusion cannot be accepted.

Second, it raises the spectre of investigator bias. One would not expect to see an omission of this magnitude in a situation where the researchers were free of substantial bias. Unfortunately, it raises the possibility that the researchers were trying to show that thirdhand smoke is harmful, rather than to objectively discover whether there is significant child exposure to tobacco smoke toxins when a parent is careful to never smoke in the home or around the child.

For further commentary on this study, see the insightful discussion of the study's conclusions and limitations over at Velvet Glove Iron Fist.

See also Dr. Alan Blum's take on the thirdhand smoke issue at The Birmingham News blog.

And here is Sandy Swarc's commentary on the issue over at her Junkfood Science blog.

No comments: