Tobacco Documents Online

4 childhood actually affects high density lipoprotein levels, let alone later risk of myocardial infarction. P N Lee 12.10.99.

2 fitted models appeared to show that PS-NS differences were larger in those with a family history of cardiovascular disease (Figure 2) and that PS-NS differences were evident in boys and girls and at all weights and were larger in heavier boys and girls (Figure 3). Interactions with ETS exposure were also reported for blood pressure (Figures 4 and 5). The authors find that "Pubertal children with long-term passive cigarette smoke exposure have lower HDL-C levels. Racial differences in HDL-C levels are related to passive smoke exposure. In children with a family history of cardiovascular disease, interactions exist between passive smoking, HDL-C level, and blood pressure that differ by sex and race. White males exposed to passive smoking who have a family history of cardiovascular disease and higher weights and diastolic blood pressures may be at special risk for premature CAD." They also note that "Given the evidence that irreversible atherosclerotic changes may be caused by passive smoke exposure and that passive smoke exposure may act synergistically with other risk factors such as hypertension, avoidance of long-term passive smoke exposure during childhood is important, especially for children with known premature cardiovascular disease in their family." A number of points can be made on this paper: (i) Their claim that irreversible atherosclerosis changes may be caused by passive smoke exposure is based on a study by Howard et al (their ref 38) of intimal-medial thickness that did not actually demonstrate a statistically significant relationship in never smokers. (ii) No data on diet were collected and the authors do not even appear to have considered the possibility that the differences in HDL-C they observed between PS and NS may have arisen because of dietary differences between the two groups of children. It is well documented that diets differ in many ways between smokers, passive smokers and non ETS-exposed never smokers. This is a major weakness of the study. (iii) At visits 1 and 2, the visits with the most data available, PS was associated with reduced M

I REVIEW 1071 CONFIDENTIAL Subject ref 8f "Childhood passive smoking, race, and coronary artery disease risk. The MCV Twin Study" W B Moskowitz et al Arch Pediatr Adolese Med (1999), 153, 446-453 This paper describes the results of a study in Virginia, USA involving 408 twin pairs. When the twins were aged 11, 12%z, 14 and 15'/z their height and weight, sexual (Tanner) stage, blood pressure, lipoprotein cholesterol and serum cotinine levels were recorded, and data were collected from the parents, including those on cigarette smoking. Passive smoke exposure was determined based on whether either of the parents smoked and data from families that changed smoking status (from smoking to nonsmoking or vice versa) any time after the first visit were dropped from further analyses. Because also of loss to follow-up, numbers oftwins in the study reduced markedly by visit, from 408 initially to 295, 192 and 113 at visits 2, 3 and 4 respectively. Statistical analyses involved a randomly selected twin from each pair. As can be seen from Table 3, there was no significant difference between low-density lipoprotein levels (LDL-C) in children who were from nonsmoking families (NS) or who were from smoking families (passive smokers = PS). Indeed the direction of difference varied by visit, with levels in PS lower at visits 1 and 2 and higher at visits 3 and 4. High-density lipoprotein cholesterol levels, whether total (HDL-C) or subfracflon 2(I3DLz C), were consistently lower in PS than in NSW, although statistically significant differences were only seen at visits 1 and 3. In adults, low HDL-C and high LDL-C levels are associated with an increased risk of myocardial infarction. More detailed analyses of the HDL-C data were based on the data recorded at all four time points and involved statistical models including race, sex, passive smoking status, weight, systolic blood pressure, diastolic blood pressure and all interactions among these variables. The

3 LDL-C and reduced HDL-C. Though the only significant difference at the two visits was in HDL-C at visit 1, the magnitude of the differences between PS and NS was comparable for LDL-C and HDL-C. Can it really be right to maj or wholly on the HDL-C results in the analyses and the abstract? (iv) I found it somewhat remarkable that, though the PS/NS difference in HDL-C was only significant at some time points, the complex analyses purported to show that such a large number of interactions were statistically significant. A problem with the paper is that the authors did not actually present data showing how the PS/NS differences vary by factors such as sex, weight, blood pressure and family history of CHD, only results of fitted models. It would have been much simpler had they divided their population into subgroups by levels of other major determinants of HDL-C, presented the estimated PS/NS difference in HDL-C for each subgroup and then calculated an overall estimate of the difference adjusted for the stratifying variables. (v) I note that cotinine was measured in serum, but the only use of these data appears to be to produce the statement "Cotinine was not detected in children who were not exposed to cigarette smoke but was present in children exposed to passive smoke (8.52±17.6 nmoUL)." This raises a few questions: (a) Why did they not relate HDL-C to cotinine level? (b) Why did not children from nonsmoking families have detectable cotinine levels? This is not usually the situation. (c) Did any of the children have cotinine levels that were indicative of smoking? While not very likely at age I 1 perhaps, one would have thought that some of the children measured at 14 and 15%a would have started to smoke. (vi) The authors rejected from analysis children in families changing from NS to PS or vice versa. Would these not have provided valuable data? Change in HDL-C following change in ETS status would seem a useful thing to look at. Overall, I do not find this paper provides convincing evidence that ETS exposure in