Tobacco Documents Online

Parental Smoking and Infection with Helicobacter pylori among Preschool Children in Southern Germany Hermann Brenner,' Dietrich Rothenbacher,t Gunter Bode,z Regina Gommel,l Gabriele Berg, I and Guido Adlerz Exposure to parental smoking is known to increase children's susceptibility to a variety of infections. We investigated the relation of parental smoking to infection with Helicobaccer pylori in a population-based study among preschool children who were screened for school fitness in the city of Ulm, Germany, in 1996. Current infection with H. pylori was mea- sured by a "C-labeled urea breath test Information on factors potentially related to H. pylori infection, including parental smoking in the household, was obtained from children's par- ents through a self-administered standardized questionnaire. Among 1,201 eligible children, 945 participated in the study (response rate = 79%). Overall prevalence of active infection was 13.7%. After adjustment for confounding factors, we found a strong positive relation between smoking by the father in the household and H. pylmi infection (odds ratio = 3.7; 95% confidence interval = 2.3-6.1). By contrast, there was a strong negative relation between smoking by the mother and H- pylon infection (odds ratio = 0.4; 95% confidence interval = 0.2- 0.8) that was most pronounced among children who had been breastfed. These striking patterns cannot be explained by cur- rent knowledge. (Epidemiology 1998;9:545-549)' Keywords: child, Helicobucter ¢ylori, infection, smoking. Infection with Heticobacter pylori is now considered the major cause of chronic atrophic gastritis and gastric and peptic ulcerst and an important risk factor for the de- velopment of gastric cancer.' Childhood is the critical period during which most H. pylori infections are ac- quired. Direct (oral-oral or fecal-oral) person-to-person spread is the most likely form of transmission of infec- tion,3 bur there is also evidence of waterborne transmis- sion in developing countries.4 In sharp contrast to de- veloping countries, childhood prevalence is now relatively low in many developed countries, which may reflect major improvements in housing conditions and hygiene.; Previous studies have identified age, low so- cioeconomic status, and poor living conditions during childhood as major risk factors of infection within both developing and developed countries.4•b•2 The nature of this association is not fully understood. Socioeconomic LJ factors and living.conditions during childhood may be related both to exposure to the infectious agent and to susceptibility to colonization of gastric or duodenal mu- cosa by H. pylori. Exposure to environmental tobacco smoke (ETS) has been shown to be associated with susceptibility to a variety of infectious diseases in childhood, mainly those of the respiratory tract 8•9 To our knowledge, the relation between exposure to ETS and infection with H. pyiori in childhood has not been studied. The main sources of exposure to ETS in childhood are the children's parents. The aim of this paper was to assess the relation between parental smoking and H. pylori infection in a large population sample of preschool children in Germany. Subjects and Methods STUDY DE9ICN AND STUDY POPULATION In Germany, school attendance is mandatory, and all children who are to attend school for the first time are examined for school fitness by physicians of the public health service- We conducted a cross-sectional study among 1,201 preschool children living in Ulm,.a city of about 100,000 inhabitants located in the south of Ger- many, who were examined for school fitness by the public health service in 1996. Details of the study design have been reported elsewhere.10 Prom ehe Depariments of'Eptdemfotogy and rlmemal Medicine I, Unwersiry of Ulm, Ulm, Gennany. Address conesponden<e tn: Hem,ann Bmnneq Deparrment of Epidemiology, Unlversiry of Ulm, Alberr-EinsteimAlkc 43. Db908I UIm, Getmany. Submitted luly 21, 1997; Soel venion accepted Febmary 13, 1998. ® 1998 by Epidemiology Resources Inc. DATA COLLECTION Active infection status was determined by a°C-labeied urea breath test. First, an initial breath sample was collected in a plastic bag. The children then were given 200 ml of apple juice (pH 2.2-2.4), which contained 60 mg of nonradioactive labeled 13Gurea (Mass Trace, Wobum, MA). Thirty minutes later, a second breath 545

546 BRENNER ET AL sample was collected. The breath samples were analyzed with an isotope-selective nondispersive infrared spec- trometer (Wagner Analytical Systems, Worpswede, Germany). A change of the °C-value over baseline of more than 5%o was considered to indicate active infec- tion. Both sensitivity and specificity of this test have been found to be close to 100%.11-I3 To reduce the possibility of false-negative test results, children who had received antibiotic treatment within the last month were excluded from the analysis. Information on housing and living conditions, includ- ing exposure to smoking by parents in the household, frequency of consumption of various foods, family demo- graphics, and socioeconomic factors, was obtained by a self-administered questionnaire, which was sent to the children's parents 1-2 weeks before the examinations and which was collected and checked for completeness and plausibility during the examinations. Because the study population included a large proportion of Turkish children, the questionnaire was also offered in Turkish. . Parents were asked whether either the mother or the father smoked in the household. In addition, informa- tion on exposure to ETS was quantified by the total number of cigarettes smoked in the household. STATISRCAL ANALYSIS In the assessment of the relation between parental smok- ing and H. pyloi-i infection, potential confounding by other risk factors has to be taken into account. There- fore, we first described the relation with parental smok- ing of the following known or suspected risk factors for the infecrion'o,t4,ts: nationality (German or other), school education of mother and of father (:~;9 years, 10-12 years, or ?13 years, reflecting standard categories of the German school system), housing density (mz per person living in the household), maternal and paternal history of gastric or duodenal ulcer, birth order (1st, 2nd, or ?3rd), breastfeeding (none, s3 months, or >3 months), day care attendance (yes or no), and lifetime number of antibiotic treatments (0, 1-3, or ?4). Then, we assessed the prevalence of H. pylori infec- tion in the children according to exposure to paternal and matemal smoking in the household. We used mul- tiple logistic regression with infection status as the de- pendent variable to allow for simultaneous adjustment for the covartates listed above. We derived estimates of odds ratios with 95% confidence intervals to quantify the crude and'adjusted relation of exposure to patemal and matemal smoking with H. pylon infection. To limit the number of covariates in the model while ensuring efficient control for confounding, we used a stepwise forward selection procedure in which covariates were included if and only if they changed the odds ratio for either maternal or paternal smoking by more than 10%. All analyses were carried our with the SAS statistical software package.16 Results Overall, 945 of 1,201 eligible children participated in the study (response rate = 79%). We excluded 82 chil- Epidemiology September 1998, Volume 9 Number. 5 TABLE 1. Smoking by Parents in the Household, According to Levels of Covariates Mother Father Smokes Smokes Covariate N (% (%) Nationality of child German 637 17 19 Foreign " 224 22 38 School education of mother (years) <_9 356 28 35 10-12 262 13 18 ?13 235 1i 15 School education of father (years) t9 333 22 34 10-12 205 21 23 a13 296 13 16 mr per person living in household <19.5 276 24 40 19.5- 269 19 21 26.9 a27.0 289 14 13 Matemal history of ulcer No 841 18 24 ~ Yes 22 45 32 Patcmal history of ulccr No 835 19 24 Yes 28 21 36 Birth order lst 416 20 22 2nd 244 18 24 a3rd 180 18 29 Breestfeeding (months) No 175 24 23 s3 280 25 30 >3 402 12 20 Day care attendance No 785 17 24 Yes 72 35 24 Antibiotic neatmen,s 0 172 20 33 1-3 385 17 20 z4 225 19 21 Total 863 19 24 dren from the analysis because of antibiotic treatment within the last month, leading to a final sample size of 863 children. The children were 5 (24%), 6 (68%), 7 (7%), or 8 (1%) years of age at the time of the exami- nation. About one-fourth (24%) of the children were living in a household in which the father smoked. Smok- ing by the mother in the household was slightly less common (19%). - Parental smoking by levels of covariates is shown in Table 1. Both maternal and paternal smoking showed an inverse relation with levels of parental school education and were less common among children who lived in less crowded (?27.0 m2 per person) households. Smoking and history of gastric,or duodenal ulcer were positively related among both children's mothers and children's fathers. In addition, patemal smoking was more com- mon among foreign children (38%) than among Ger- 2505586052

Epidemiology September [998, Volume 9 Numberj TABLE 2. Smoking by Father in the Household, Accord- ing to Smoking by Mother in the Household Mother Smokes N - Father Smokes (%) No 701 16 Yes 161 59 man children (19%) and among children who had never received antibiotic treatment than among other chil- dren. Maternal smoking was less common among chil- dren who had been breastfed for more than 3 months and more common among the minority of children who had attended day care than among other children. Although smoking by the father and the mother showed partly divergent relations with the covariates considered in this analysis, both factors were strongly related (Table 2). More than half (59%) of the children whose mother smoked in the household were also ex- posed to smoking by the father in the household, whereas only one-sixth (16%) of the children whose mother did not smoke at home were exposed to smoking by the father. Overall prevalence of infection was 13.7%. As re- ported elsecvhere,10•is foreign (in particular, Turkish) na- tionality, residential crowding, and maternal history of peptic ulcer were major risk factors of H. pylort infection in this population, whereas use of antibiotics was asso- ciated with a lower prevalence of infection. The relation between exposure to parental smoking in the household and H. pylon infection is shown in Table 3. Prevalence of H. pylori infection was much higher among children whose father smoked in the household (26.9%) than among other children. (9.5%), resulting in a crude odds ratio of 3.5 (95% confidence interval = 2.4-5.3 ). No such relation was seen between smoking by the mother and prevalence of infection in the crude analysis (crude odds ratio = 1.0; 95% confidence inter- val = 0.6-1.7). H. PYLORI INFECTION IN CHILDREN 547 When the odds ratio for paternal smoking was ad- justed for maternal smoking, it increased to 4.5 (95% confidence intecval = 2.9-7.1). Conversely, the odds ratio for materrral smoking was reduced to 0.5 (95% confidence interval = 0.3-0.9) by adjustment for parer- nal smoking. This pattern suggests that both exposures are related to H. pylon infection in a divergent manner, and that adjustment for one another is essential because the two exposures are positively correlated with one another (Table 2). Additional adjustment forr the covariates identified as relevant confounders in the variable selection procedure (residential crowding, birth order, and maternal history of ulcer) reduced both the odds ratio for paternal smok- ing and that for maternal smoking to some extent. After control for the covariates, smoking by the father was still associated with an almost fourfold increase of H. pylon infection odds (odds ratio = 3.7; 95% confidence inter- val = 2.3-6.1), however, whereas smoking by the mother was associated with a clear reduction of the prevalence odds of H- pylori infection (odds ratio = 0.4; 95% confidence interval = 0.2-0.8). Additional control for the other variables listed in Table 1 led to only negligible changes in the estimates of odds ratios; these variables were therefore not included in the final model. To address potential differences in dosage of exposure to ETS from paternal and maternal smoking, we com- pared the average number of cigarettes smoked in the household by maternal and paternal smoking. Although the mean amount of smoking was clearly highest in households in which both parents smoked (17.0 ciga- rettes per day), we found no, difference in the average number of cigarettes smoked per day between house- holds in which only the mother smoked (10.6 cigarettes per day) and households in which only the father smoked (10.3 cigarettes per day). Discussion To our knowledge, this is the first study that addresses the impact of parental smoking on the risk of H. pylori ~~ TABLE 3. Number of Children (N), Prevalence of H. pylori Infection (HP+), and Crude and Adjusted Odds Ratios with 95% Confidence Interval According to Smoking by the Father and Smoking by the Mother in the Household Odds Ratio (95% Confidence Interval) Smoking by Parents N HP+ (%) Cmde Adjusted for Smoking by Other Parent Only Additionally Adjusted for Multiple Other Covariates• Father smokes Not 655 9.5 1.0 1.0 1.0 Yes 208 26.9 3.5 (2.4-5.3) 4.5 (2.9-7.1) 3.7 (23-6.1) Mother smokes Not 701 13.6 1.0 1.0 1.0 Yes 161 13.7 1.0 (0.6-1.?) 0.5 (0.3-0.9) 0.4 (02-0.8) • Adpated for mt per persor. liv5ng in the household, binh ordex, maternal history of gastric ur duodenal ulcer, and smoking by other parene Additional control for the other variables listed In Table I led to only negligible changev in the esrimares ofodds 2tios; these variables v.ere therefore not included in the final model. f Reference category. infection in childhood. We found strongly divergent relations of paternal and maternal smoking with H-. pylari infection. These striking patterns were unexpected and may even appear im- plausible on firse view. They therefore require particularly careful discussion. Given the size ofthe study popula- tion and the strength of the associa- tions, chance is an unlikely explana- tion for our findings. Our study was limited, however, by the fact that in- . formation on parental smoking was obtained at the time of the examina- tion only, which hindered a more de- tailed study of specific effects of paren- cal smoking during various periods in infancy. There is now increasing evi- dence that infection is acqnired in early infancy9•'7.'s Therefore, changes

548 BRENNER ET AL of smoking habits in the household over time may have led to some misclassification of the relevant exposure. In the assessment of the role of parental smoking on major respiratory diseases, such changes may lead to severe bias, because parents often tend to reduce exposure subsequent to the development of such diseases.19 Be- cause H. pylorf infection is typically asymptomatic among children" (which was also true in our study), potential changes of exposure would be expected to be unrelated to H. pylori infection. Such changes would have led to underestimation of the potential effects of parental smoking.r` There are indications, however, that parental smoking habits might have been rather stable over time in this study population: the mean birthweight of children whose mother smoked in the household was 171 gin lower than the mean birthweight of other chiL dren. This difference is very similar to the difference in mean birthweight that has been found between children whose mothers did and did not smoke during preg- nancY 22,23 - Another major concern in the assessment of parental smoking effects is confounding by socioeconomic factors that are related to both smoking habits and susceptibility to infection. Such factors were carefully controlled in our study. Control for additional factors, including nu- tritional factors and body mass index, did nor materially alter the results. Therefore, confounding also seems to be an unlikely explanation of the observed patterns. Although we cannot offer an explanation for the intriguing findings of our study on the basis of existing knowledge, we suggest that the following potential mechanisms deserve careful consideration. Several stud- ies have indicated that mother-infant transmission may play a very important role in acquisition of the infection. Although parental infection status was not determined in our study, we were able to show that maternal (but not patemal) history of peptic ulcer, which is known to be strongly related to maternal infection, was associated with a more than 10-fold prevalence of infection among their children.15 Our results might thus be consistent with the hypothesis that matemal smoking may hinder mother-infant transmission of the infection. Because oral-oral transmission is the most likely form of spread of the infection, colonization of the oral cavity may be important in this context. H. pylori has been detected in saliva and dental plaques.2q Although most studies have not found a relation between smoking and gastric'infec- tion with H. pyiori; 5 data are tacking on the coloniza- tion of the oral cavity among smokers and nonsmokers. Another mechanism by which smoking by the mother could have different effects than smoking by the father is macernal smoking during breastfeeding. It is well known that maternal smoking affects both the volume and the composition of breast milk.Z6,27 In particular, nicotine (which is a weak base) is concentrated two- to threefold in breast milk, compared with matemal serum.28•29 An- imal studies have shown that topical effects of nicotine on the gastric mucosa are complex and strongly dose dependent. Observed effects include effects on secretion Epidemiology September 1998, Volume 9 Number 5 of acid and mucus (which contains bactericidal compo- nents, such as lysozyme) and on gastric motility.3o To assess a potential role of nicotine in breast milk, we carried out additional analyses in which the sample was stratified by history of breastfeeding. These analyses indicated that the inverse relation of maternal smoking with children's infection might be stronger among the majority of children who were breastfed (adjusted odds ratio = 0.3; 95% confidence interval = 0.2-0.7) than among other children (adjusted odds ratio = 0.7; 95% confidence interval - 0.2-2.4). Acknowledgments We thank Diedind Wehrhahn and Ingrid Wamer and che staff of rhe Ulm Public Healch Scrvice for their support in conducring this study. References 1. NIH Consensus Coaference. He4cobacrer bylon in pepaic ulcer di:.-•. JAMA 1994;272:65-69. ,~ j 2. International Agency for Research on Cancer. IARC Monographs on'Se Evaluation of Carcinogenic Risks to Humans- vol. 61. Infection v.,irh Hdi- cobocter pylan. In: Schisrosomes, Liver Flukes and Hei+cobacrer pybri. Lyon: Inremational Agency for Research an Cancer, 1994;177-241. 3. Goodman K], Correa P. The transmission of Helicobauer pylmi: a crirical ew of the evidence. Int J Epidemiol 1995;24:875-887. 4. Goodman KJ. Cortea P. Tenganl Aux HJ, RamGes H, DcLany JP, Pepinosa OG, Quinones ML,_Cnllaaos Parra T. Helicnbocter pyb.i infection in the Colombian Andes; a populanon-based study of transmission parhways. Am J Epidemiol 1996; I443 90-2 99. 5. Taylor DN, Blaser MJ. The epidemiology of He4coboner pytrrd lnfecnon. EpideminlRev 1991;13ri2-59. 6. MendaLl MA, Goggin PM. Molineaux N, Levy J, Toosy T, Scrachan D, Northfield TC. Childhood liv(ng-condicions and Helicobacrer pybd seropos- itiviry in adulc life. Lanacr'1992;339:896-897. 7. Webb PM, Knight T. Greaves S, Wilson A. Newell DO. Elder J, Forman D. Relation between infection wich Helfcobacrerpylori and living corsdirions in childhood: evidence for person-to-person tmnsmiuion in carly life. BMJ 1994;308:750-753. 8. Graham NMH. The epidemiology of acure respiramry infections in children and adults: a global perspective. Epidemiol Rev 199Q12:149-178. 9. Poswiilo D, Alberman E. Effects of Smoking on the Fetus. Neonate and Child. Oxford, England: Oxford Univeraity Press, 1992. 10. Rothenbaaher D, Bode 0. Betg 0, Gommel R, Gonser T. Adler G, Brenner H. Prevalence and detenninants ofHelicn6ocrer pylon infection in pre-school childrere a population based study from Germany. Inr J Epidemiol 1998;27: 135-141. 11. Braden B, Duan LP, Caspary WF, Le,nbcke B. More convenient oC breath r«t modif,catfons still ,neet the cdreda for valid dianosis of colwcrn pylmi infection. Z Gastrcenrerol 1994:32:198-202. 12. Thijs JC, van Zwet AA, 7hijs WJ, Oey HB. Kanenbeld A, Scellaard F, Luljr DS, Meyer BC, Kleibeuker JH. Diagrostic rears for Helicobacrer pylori: a proapective evaluation of cheir accuracy, without selecting a single test as the goLd standard. Am J Gastrcenterol I996;9I:2125-2129. =. 13. Braden B, Schafer F. Caspary WE, Lembeke B. Nondispersivc iwrope, selective infrared specttoscopy: z new analytical method for uC-yrea breach rests. Scand 3 Gasacen¢rol 1996;31:442-445. I4. Vincent P. Transmission and acquislcion of He&obacrcr prluri infecflon: evidences and hypothes6. Biomed Pharmacother 1995;49:I1-18. 15. Brenner H, Rorhenbachec D, Bode 0, Adler G. Parental history of gastric or duodenal ulcer and HeG<obacrer pjlari Infeceion among predchool children; evidence for morher-infanc transmission. BMJ 1998;316:665. 16. SAS Inscieuaq Inc. SAS Language: Refetence, Version 6, 1sc ed. Cary, NC: SAS Institute, Inc., 1990. 17. Klein PD, Giltn2n RH, Leon-Bama R. Diaz F, O'Brian Smith E, Graham DY. The epidemiology of Hdirola<rer pylori in Pemvian children between 6 and 30 months of age. Am J Gastrcenterol 1994:89:2I96-2200. 18. Mahalanabis D, Rahman MM, Sarker SA, Bardhan PK, Hilde6rand P, Beglfnger C, Gyr K. Hdirducrerpyb.i infection in the young in Bangladesh: prevalence, socioeconomic and nunrtional aspects. Inr J Epidemiol 1996;25: 894-898. 19. Brenner H, Bletrner M. Effects of disease-dependenr changes of ecposure in <ruzvsecttonal studies. Epidemiology I996;7:502-506. 2505586054

I Epidemiology September 1998, Volume 9 Number 5 20. Sinatra FR, Pienak MM. HeG'colveter py(mi infection in children. Curr Opin [nfect Dis 1996:9:187-190. 21. Copeland KT, Checkoway H, M<Michael AJ, Holbrook RH. Bixs due to isclassificzHOn in the estimation of rdaaive risk(Am J Epidemiol 1977; 105:488-495. - 22. Stein Z, Kline J. Smoking, alcohol znd reproduction. Am J Public Health 1983;73:I154-1156. 23. Abell TD, Baker LC, Ramsey CN Jr. The effects of maretnzl smoking an infanr, bbirth weighc Fara Med 1991;23:103-I07. 24. PyrkoaPoloneryk J, Konturek S), Karo.ewska E, Bielanski W, Kaeemarcryck- Srachovrska A. Oral cavity zs permaneat reservoir of He6mbacrer pjion and potential soume of infection. J Physiol Phannacol t996;47:121-129. 25. EUROGAST Study Group. Epidemiology of. and rbk factors for, HeGcobac- ,.H. PYLORI INFECTION IN CHILDREN 549 ter pyfori infection among 3,194 asympmmatic subjecrs in 17 populations. Gut 1993d4a672-1676. 26. VfoF.Balaur0.InfanreC.Smokingduringpregnancyandla<tarionandirs effects on breasr-milk~ volume. Am J Clin Nutr 1991;54:101 t-fOl6. 27. Byrd RS, Hovrord CR. Children's passive and prenatal exposure to cigaretre smoke. Pediaa Ann 1995;24:640-645. 28. Luck W,Nau H. Nicotine and eorinine concentrations in serum and milk of nursing smokers. Br J Clin Pharnucol I984;I8:9-15. 29. DahLstrrim A. Lundell B, Curvall M. Thapper L. Nicotine and cotinine concenrrarions in the nursing inother and her infant. Acna Paediatr Scand 1990;79:142-147. 30. Endoh K, Letmg FW. Efkca of smokutg and nicotine en the gavic mucom: a review ddinid and expcrimenral evidcnce. Gancenterology 1994;107:864-878.