Tobacco Documents Online

3 affect non-whites but not whites, and it seems more probable that this is either a chance finding or it is due to uncontrolled confounding by a factor to which non-whites are more commonly exposed than whites. Limitations of the study include failure to measure ETS exposure objectively, relatively small number of ETS exposed individuals with the endpoints of interest and the limited number of potential confounders studied. The authors note that they "had no data on history of other diseases, particularly sexually transmitted diseases, which may play a role in preterm delivery, or on the specific causes of preterm birth." Lack of data on diet is another possible limitation. Overall this can be regarded as yet another study relating ETS to birthweight and prematurity which fails clearly to demonstrate an effect of ETS exposure. P N Lee 22.1.2001

2 Of more interest were the results for ETS exposure, which are summarized in the table below: No Low ETS HiQh ETS Any ETS* BW . N Mean (g) Adjusted difference (g) (95% CI) exposure 2887 3514.1 Baseline 625 3495.8 0.68 (-46.7 to 48.1) _ 134 3516.6 8.2 (-86.1 lo 102.5) 759 3499.5 2.0 (Not significant) LBW : N 85 20 7 27 Adjusted odds ratio 1.0 1.0 1.8 1.1 (95% CI) (0.61-1.7) (0.82-4.1) (0.71-1.17) SGA : N 173 42 5 47 Adjusted odds ratio 1.0 1.1 0.62 -1.0 (95% CI) (0.74-1.5) (0.25-1.5) (Not significant) PTB : N 174 46 13 59 Adjusted odds ratio 1.0 1.1 1.Fi 1.2 (95% Cl) (0.79-1.6) (0.87-2.9) (0.90-1.17) VPTB : N 58 23 7 30 Adjusted odds ratio 1.0 1.5 2.4 -1.7 (95% Cl) (0.90-2.5) (1.0-5.3) (?significant) * Results only given in the paper for LB W and PTB; other data presented are approximate estimates Generally, the results show no real indication of a relationship with B W(where ETS was associated with a slight increase , LB W, SGA or total PTB. A possible association is seen with VPTB, but this is of marginal significance and based on relatively few cases. Additional analysis (see Table 3 of the paper) suggested that high ETS exposure was associated with some increase in LBW in whites and some decrease in LBW in non-whites. However, as none of the differences are significantly different from zero, this may well be chance variation. The subgroup analyses in Table 5 suggest an association of ETS with increased LB W, PTB and VPTB in non-whites, but not in whites. However, this would seem to be based on very few cases, bearing in mind that there are only 7,13 and 7 total cases in the high ETS group before they are subdivided by race. In any case, it would seem implausible that ETS might actually

REVIEW 1140 CONFIDENTIAL Subiect ref 8d "Prenatal active or passive tobacco smoke exposure and the risk of preterm delivery or low birth weight" G C Windham et al Epidemiology (2000), 11, 427-433 In 1999 (see Review 1048), Windham et al reported results relating to spontaneous abortion from a prospective study of over 5000 women conducted in California in 1990-1991. This paper describes results from the same study relating to low birth weight and the risk of preterm delivery. The study compares outcomes in six groups of women, classified based on smoking and ETS data collected by telephone in the first trimester of pregnancy and relating to the previous week and to the week before the last menstrual period before pregnancy: 2887 nonsmokers with no (or <0.5 hours a day) ETS exposure, 625 nonsmokers with 1-6 hours a day ETS exposure, 134 nonsmokers with 7 or more hours a day ETS exposure, 180 smokers of less than 5 cigarettes a day, 186 smokers of 5-10 cigarettes a day and 87 smokers of more than 10 cigarettes a day. Endpoints considered were mean birthweight (BW), incidence of low, i.e. <2500 g, birthweight (LBW), small for gestational age, i.e. less than the 10th percentile for the week of gestation (SGA), preterm birth, i.e. less than 37 weeks gestation (PTB) and very preterm birth, i.e. less than 35 weeks gestation (VPTB). LB W analyses were adjusted for age, race, body mass index (BMI), parity and prior pregnancy loss, education, marital status, stressful life events, caffeine and alcohol consumption, while other analyses were adjusted for a reduced set of these variables (prior pregnancy history, race, BMI, stressful life events and education) which caused a defined degree of confounding. The analyses showed that smoking was associated with a clear decrease in B W(greater in non-whites than whites and in older than younger mothers) and an increased risk of LB W and SGA, but not so clearly with an increase in PTB and VTPB.