Tobacco Documents Online

2 (ii) The main table presenting the results (Table 3) gives risk for each parent smoking adjusted for the other parent's smoking, but does not actually show the joint relationship of HPI infection to smoking by both parents. I would have liked to see an analysis comparing risk in 4 groups - neither parent smoked, father only smoked, mother only smoked and both parents smoked. (iii) No attempt has been made to relate HPI to number of cigarettes smoked by the parents. (iv) I was also concerned that the criterion used for including a factor in the logistic regression model - based on a stepwise selection procedure including a covariate only if it changed the odds ratio estimate for paternal or maternal smoking by at least 10% - may have led to relevant factors being omitted. I would have thought that if maternal history ulcer was included (as a marker of HPI status) then so too should paternal history of ulcer. It would of course have been better had parental HPI status been measured directly. While the finding is an interesting one, it clearly needs replication, preferably in a cohort study that studies onset (not presence) of HPI in children and measures actual HPI status in the parents. P N Lee 13.10.99

~ 0 og?_ REVIEW 1074 CONFIDENTIAL Subject ref 8d "Parental smoking and infection with Helicobacter pylori among preschool children in Southern Germany" H. Brenner et al Epidemiology (1998), _9, 545-549 Helicobacter pylori infection (HPI) is considered to be an important risk factor for stomach cancer and for peptic and gastric ulcers and has been under suspicion as a cause of other diseases, including heart disease. This paper describes the results of a cross-sectional study in IIIm in Germany involving 945 children aged 5-8 years with no antibiotic treatment in the last month, in which HPI status was assessed by a urea breath test (noted to be highly sensitive and specific). Parents completed a questionnaire providing data on smoking habits in the household, food consumption, demographic and socioeconomic factors, housing and living conditions, and parental history of ulcer. Information relating to the child on birth order, breastfeeding, day care attendance and lifetime number of antibiotic treatments was also collected. The main finding of the study was that after adjustment for smoking by the other parent, household crowding, birth order and maternal history of ulcer there was a highly significant positive relationship between HPI and smoking by the father (Odds ratio 3.7, 95% CI 2.3 - b.1). In contrast there was a significant negative relationship between HPI and smoking by the mother (Odds ratio 0.4, 95% CI 0.2 - 0.8). The authors note that "these striking patterns cannot be explained by current knowledge". They suggest the possibility that smoking may hinder mother- infant transmission of HPI, maternal history of ulcer (which is a strong correlate of material HPI status) being very strongly related to HPI infection in the child. I found the statistical analysis rather uninformative for a number of reasons: (i) Little information was given on the relationship of HPI to the various potential confounding variables measured.