Smoking and Celiac Disease: A Population-Based Cohort Study

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3:869 874 Smoking and Celiac Disease: A Population-Based Cohort Study JONAS F. LUDVIGSSON,*, SCOTT M. MONTGOMERY,*, and ANDERS EKBOM, *Pediatric Department, Clinical Research Centre, Örebro University Hospital, Örebro, Sweden; Clinical Epidemiology Unit, Department of Medicine, Karolinska University Hospital, Karolinska Institute, Stockholm, Sweden; and Harvard Medical School, Boston, Massachusetts Background & Aims: Earlier studies indicate a protective effect of smoking against celiac disease (CD), but have been based on small numbers and retrospective collection of smoking data. Methods: We linked the Swedish national inpatient register and the medical birth register to study the association between smoking status during pregnancy and CD (diagnosed or undiagnosed at delivery) in women who were pregnant from 1983 to 2001. We adjusted for civil status, age, and year when smoking data were collected. We identified 873 cases of CD (636 diagnosed and 237 undiagnosed). Results: Of 249,967, 67 (.27%) had undiagnosed CD (vs 170 of 794,912 non [.21%]) (odds ratio [OR], 1.25; 95% confidence interval [CI],.94 1.66; P.118). Point estimates remained unchanged when adjusting for civil status, age, and year of smoking data collection (adjusted OR [AOR], 1.25; AOR,.94 1.67). There were no associations between smoking and future (undiagnosed at delivery) CD when we adjusted for potential confounders and stratified for comorbidity or time to diagnosis (<5 vs>5 y after infant birth). In women with diagnosed CD, smoking was more common than in women who never had a diagnosis of CD (AOR, 1.36; AOR, 1.12 1.64; P.002). Conclusions: Smoking seems to have little effect on the risk for future CD in pregnant women. Celiac disease (CD) is an autoimmune disorder characterized by gluten intolerance in genetically predisposed individuals. 1 The genetic component is strong, 2 but not all monozygotic twins exposed to gluten are concordant with respect to CD. 3 This implies the existence of nongenetic risk factors in CD. In 1996, Snook et al 4 reported an increased risk for CD in non. Since then, most case reports 5,6 and observational studies 7 10 have supported an inverse relationship between smoking and CD, although exceptions exist. 11,12 Until now, most studies on smoking and risk for CD have been based on relatively small numbers of patients with CD and retrospective collection of smoking data. We used Swedish national registers to study the relationship between prospectively collected data on smoking and future CD in more than 1 million women who were pregnant from 1983 to 2001. Methods Participants Since 1973, professional health care personnel have collected health data on pregnant women and their offspring (liveborns and stillborns alike). Since 1983, the medical birth register also contains data on maternal smoking in gestational week 12. Information is collected throughout pregnancy and at birth on structured forms. These forms constitute the basis for the computerized medical birth register maintained by the Swedish National Board of Health and Welfare. In Sweden, more than 99% of all liveborns and stillborns are registered in the medical birth register. 13 Each entry contains the national civil registration number of the pregnant woman. The national civil registration number is a unique number assigned to over 99.9% of all residents in Sweden at birth or immigration. 14 It is made up of the year, month, and day of birth (eg, 19690821-) and another 4 digits (eg, 1489). The last 4 digits make it possible to distinguish between people born on the same day. Results in the current study are based on all women in Sweden giving birth from 1983 through 2001. We only included each mother once, and consistently chose the first pregnancy when smoking data were available as the source of smoking information. Definition of Smoking Within the framework of the antenatal maternal health control, pregnant women are asked about smoking in gestational week 12. Smokers are requested to grade their daily smoking into 1 9 cigarettes/day (light smoking) or 10 or more cigarettes/day (heavy smoking) (or equivalent). Data on smoking are stored in the medical birth register. Definition of Celiac Disease Celiac disease was our outcome measure. We collected data on CD through the Swedish national inpatient register. Abbreviations used in this paper: AOR, adjusted odds ratio; CD, celiac disease; CI, confidence interval; EMA, endomysial antibody; ICD, international classification of diseases; OR, odds ratio. 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(05)00414-3

870 LUDVIGSSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 9 Table 1. Characteristics of Participants Characteristics Non N (%) 1 9 cigarettes/day N (%) 10 cigarettes/day N (%) Total 795,389 159,587 90,539 Age (y) 15 19 24,445 (3.2) 12,265 (7.7) 6161 (6.8) 20 24 195,695 (24.6) 50,449 (31.6) 27,129 (30.0) 25 29 302,449 (38.1) 50,738 (31.8) 27,687 (30.6) 30 34 186,247 (23.4) 30,989 (19.4) 18,502 (20.4) 35 40 76,571 (9.6) 13,639 (8.6) 9809 (10.8) 40 44 8297 (1.0) 1406 (.9) 1195 (1.3) Data not available 685 (.1) 101 (.1) 56 (.1) Nationality Nordic country 719,292 (90.4) 151,074 (94.7) 87,003 (96.1) Other country 76,097 (9.6) 8513 (5.3) 3536 (3.9) Data not available 0 (0) 0 (0) 0 (0) Calendar period for year of infant birth 1982 1988 310,235 (39.0) 84,258 (52.8) 53,575 (59.2) 1989 1995 282,872 (35.6) 53,752 (33.7) 28,991 (32.0) 1996 2001 202,282 (25.4) 21,577 (13.5) 7973 (8.8) Data not available 0 (0) 0 (0) 0 (0) Civil status Living with child s father 730,445 (91.8) 136,639 (85.6) 73,338 (81.0) Not living with child s father 37,152 (4.7) 16,802 (10.5) 13,803 (15.2) Data not available 27,792 (3.5) 6146 (3.9) 3398 (3.8) CD Diagnosed 477 (.1) 117 (.1) 42 (.1) Undiagnosed 170 (.1) 43 (.1) 24 (.1) Never 794,742 (99.9) 159,427 (99.9) 90,473 (99.9) Data not available 0 (0) 0 (0) 0 (0) The inpatient register lists hospital-based discharge diagnoses since 1964 and is held by the Swedish National Board of Health and Welfare. A more detailed account of the inpatient register has been published elsewhere. 15 Celiac disease was defined according to the following international classification of disease (ICD) codes: ICD-7: 286.00; ICD-8: 269.00, 269.98; ICD-9: 579; ICD-10: K90.0A. Women with a hospital-based discharge diagnosis of CD between 1964 and 2001 were regarded as cases. Furthermore, cases were classified as main CD (CD was listed as the main diagnosis at the first discharge featuring CD) or secondary CD (CD not listed as the main diagnosis). Patients with secondary CD had been assigned by necessity at least 2 diagnoses (1 being CD), and the non-cd diagnosis was the reason for hospitalization. Statistics We compared proportions of undiagnosed patients with CD in and non according to smoking status using the Pearson 2 test. We also looked at the relationship to smoking according to time from infant birth to date of diagnosis ( 5 y, 5 y). Stratified analyses were performed in patients in whom CD was listed as the main discharge diagnosis vs those with secondary CD. Furthermore we compared the proportion of (light smoking, heavy smoking, or any smoking) in women with diagnosed CD and in women who never had a diagnosis of CD. Crude odds ratios (OR), adjusted odds ratios (AOR), and 95% confidence intervals () for OR were calculated using logistic regression. When calculating the AOR for CD we included smoking, maternal civil status (living with the child s father vs not living with the child s father), maternal age at infant birth (15 19, 20 24, 25 29, 30 34, 35 39, 40 44 y), and year of infant birth (calendar periods: 1983 1988; 1989 1995; 1996 2001) as independent variables in our model. Data on independent measures all were collected from the medical birth register. To study the consistency of smoking behavior we compared smoking data in the first and second pregnancy of individual women during the study period using the measure. We also tested the relationship between smoking and undiagnosed CD, restricting our dataset to mothers reporting the same smoking pattern in the first and second pregnancy during the study period (ie, nonsmoker nonsmoker, light smoker light smoker, heavy smoker heavy smoker). Statistics were calculated using the statistical software SPSS 11.0 (SPSS, Chicago, IL). 16 Ethics This study was approved by the research ethics committee of the Karolinska Institute. None of the participants were contacted. Results Background Data We had smoking data on 1,045,515 women, 250,126 of whom were (Table 1). We identified

September 2005 SMOKING AND CELIAC DISEASE 871 Table 2. Risk for Undiagnosed CD According to Smoking Status During Pregnancy Proportion of patients with CD ( ) OR; P value AOR; P value All subjects Non 170/794,912 (.21) 1.00 Light 43/159,470 (.27) 1.26;.90 1.76.175 Heavy 24/90,497 (.27) 1.24;.81 1.90.324 Any 67/249,967 (.27) 1.25;.94 1.66.118 Complete data a Non 168/766,496 (.22) 1.00 1.00 Light 42/153,232 (.27) 1.25;.89 1.75.195 1.05;.74 1.48.790 Heavy 24/87,051 (.28) 1.26;.82 1.93.293.98;.63 1.51.910 Any 66/240,283 (.27) 1.25;.94 1.67.120 1.03;.77 1.38.822 NOTE. Adjusted logistic regressions included the following independent variables: smoking, living with child s father, maternal age, year of infant birth. CD was the dependent variable. ORs 1 in nonreference categories indicate that the risk for future CD increases with smoking. a Subjects with complete data on smoking during pregnancy, CD, civil status, maternal age, and year of infant birth. 873 women with a hospital discharge diagnosis of CD (Table 1). Among women with a diagnosis of CD at infant birth, 513 of 636 had CD listed as their main diagnosis (80.7%); whereas the corresponding proportion among women with undiagnosed CD was 45.1% (107 of 237). Most women with undiagnosed CD received their diagnosis 5 years or more after infant birth (150 of 237). The majority of women were aged 20 34 years and lived with the child s father during pregnancy (Table 1). Undiagnosed Celiac Disease and Smoking Status Of 249,967, 67 (.27 ) had undiagnosed CD at delivery vs 170 of 794,912 non (.21 ) (OR, 1.25;,.94 1.66; P.118) (Table 1). ORs did not change when we restricted analyses to women with complete data on potential confounders (Table 2). Adjustment for civil status, maternal age, and year of infant birth did not affect the risk estimates for undiagnosed CD at delivery (future CD). Light and heavy did not differ with respect to the risk for undiagnosed CD (Table 2). Smoking was not associated with a diagnosis of CD within less than 5 years after infant birth (AOR,.95;,.57 1.59; P.852) or within 5 years or more after infant birth (AOR, 1.08;,.76 1.54). Neither were there any associations between light or heavy smoking during pregnancy and undiagnosed CD at infant birth, stratifying for time between infant birth and diagnosis (data not shown). In light with complete data on civil status, (maternal) age, and year of infant birth there was an increased risk for undiagnosed CD not classified as the main discharge diagnosis (OR, 1.57;, 1.03 2.41; P.036). This association disappeared when adjusting for civil status, maternal age, and year of infant birth (OR, 1.33;,.87 2.05; P.190). In women reporting any smoking, the corresponding AOR was 1.18 (,.80 1.73; P.407). There were no associations between smoking and undiagnosed CD in any other subgroup when adjusting for potential confounders and stratifying cases by comorbidity or time to diagnosis ( 5 vs 5 y after infant birth). Neither was there any association between smoking and undiagnosed CD when stratifying for time period (1983 1993 and 1994 2001; data not shown). There was no increased risk for undiagnosed CD in who reported a consistent smoking pattern (light and heavy, respectively) in the first and second pregnancy during the study period (OR, 1.18;,.67 2.06; P.574) (AOR,.93;,.52 1.68; P.810) (based on 457,300 women who never had CD and 103 women with undiagnosed CD) compared with women who consistently reported no smoking in the first and second pregnancy. Diagnosed Celiac Disease and Smoking In women with diagnosed CD, there was an increased risk for being a light but not heavy smoker when adjusting for civil status, maternal age, and year of infant birth (Table 3). The increased risk for light smoking was seen in both patients with secondary CD or main CD and there were no differences between crude ORs in the complete sample (proportion of [any smoking] according to celiac status: unexposed, 249,900 of 1,044,642 [23.9%] and diagnosed CD, 159 of 636

872 LUDVIGSSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 9 Table 3. Risk for Smoking Among Women With Diagnosed CD Light smoking (1 9 cigarettes/day) Heavy smoking ( 10 cigarettes/day) Proportion of (%) Crude OR, Adjusted OR, Proportion of (%) Crude OR, Adjusted OR, Unexposed 153,190/1,006,545 1.00 1.00 87,027/1,006,545 1.00 1.00 (15.2) (8.6) Any diagnosis 112/615 (18.2) 1.24; 1.01 1.52 a 1.44; 1.17 1.77 b 42/615 (6.8).78;.57 1.06 1.06;.77 1.46 Secondary diagnosis 27/120 (22.5) 1.62; 1.05 2.48 c 1.90; 1.23 2.95 d 7/120 (5.8).66;.30 1.40.88;.40 1.90 Main diagnosis 85/495 (17.2) 1.16;.91 1.46 1.33; 1.05 1.69 e 35/495 (7.1).80;.57 1.13 1.10;.78 1.57 NOTE. Calculations were based on subjects with complete data on smoking, CD, civil status, maternal age, and year of infant birth. Adjusted logistic regressions included the following independent variables: smoking, living with child s father, maternal age, year of infant birth. Smoking was the dependent variable. OR 1 in nonreference categories indicate that the proportion of was increased in women with diagnosed CD. Other associations were not statistically significant (ie, P.05). a P.039. b P.001. c P.028. d P.004. e P.018. [25.0%]) and in the sample of patients with complete data on civil status, maternal age, and year of infant birth (Table 3). Smoking Pattern in First and Second Pregnancy The smoking pattern changed between the first and second pregnancy ( for nonsmoking vs smoking: P.732). values were similar for both women with undiagnosed CD and women who never had CD. In mothers with undiagnosed CD and at least 2 children, 22 of 29 (76%) remained between the first and second pregnancy during the study period as compared with 72.7% of women who never had CD. The proportion of heavy remaining heavy during 2 consecutive pregnancies was 51.4% (women with undiagnosed CD, 6 of 7 [86%];.604 for first and second pregnancy comparing proportion of non, light, and heavy ). Discussion This study found that smoking status during pregnancy was not associated with undiagnosed CD at delivery. Neither could smoking be linked with undiagnosed CD in subgroups of women with CD diagnosed within 5 years after infant birth or according to the presence of comorbidity. However, in women who already had received a diagnosis of CD, smoking was more common than in women who never received a diagnosis of CD. This large cohort study comparing the risk for undiagnosed CD in vs non was based on 250,000 and almost 800,000 non. The large numbers increase the accuracy of our point estimates and allow for stratifications and adjustments. With regard to the number of patients with CD, our study is equal in size to all peer-reviewed studies on smoking and CD published so far, 4,8 11 although the literature also contains reports on endomysial antibody (EMA)-positive patients 17 and on patients with transiently increased CD serologic markers. 18 Another strength of our study is that we obtained data on CD and smoking status independently and data on smoking were collected prospectively. In our study, 1 in 4 women without CD reported smoking. This is slightly less than the 30% 33% reported in a number of case-control studies 4,8,10 but higher than in a recent British population-based cohort of EMA positivity. 17 In 3 of the case-control studies, hospital controls were used as controls 4,10,11 and 2 British studies involved controls gathered partly from orthopedic departments. 4,10 This approach risks leading to selection bias owing to a significant relationship between smoking history and risk for fractures. 19 Such controls may result in an overestimation of the protective effect of smoking on CD. It also is possible that a diagnosis in older age signals fewer symptoms and lower disease activity, which may not be sensitive to the immunomodulatory effects of smoking. 20 Most earlier studies have reported a lower risk for CD in. 4,7 10,17 Previous research, however, has been limited by relatively few cases 4,7 11,17 and a retrospective research design. 4,7 10 The Swedish antenatal care program is free of charge and, although not compulsory, involves the vast majority of pregnant women in Sweden. Other health care, in-

September 2005 SMOKING AND CELIAC DISEASE 873 cluding diagnostic measures for CD, is almost free of charge. Hence, we do not believe that recruitment bias caused by socioeconomic circumstances occurred in our study. Also previous research has failed to show a strong socioeconomic gradient in CD. 4,21 23 Smoking per se (irrespective of amount) tends to result in worse general health, greater comorbidity, and therefore secondary screening for CD. Women with diagnosed CD more often were than women who never had a diagnosis of CD. Diagnosed CD was associated with light smoking but not with heavy smoking. One explanation for the risk increase in the proportion of light among those with a prior diagnosis of CD may be that women with CD smoke as a way to cope with their disease, although they smoke moderately in response to concerns about their health. It also may be that the general health behavior of all pregnant women either has affected the actual rate of heavy smoking or resulted in a downward shift in smoking habits during pregnancy. Another possibility is that heavy with CD will report light smoking. Finally, because smoking in pregnancy has a negative public attribution, there is a risk that pregnant women in general underreport smoking. Data published by the Swedish National Institute of Public Health show that smoking rates were higher among pregnant women in the beginning of the study period than in the general population, whereas the opposite was seen in the second half of the study period. 24 The estimates generated by our study were similar in these 2 periods. Other data also suggest that women seldom start to smoke in the first 8 months after infant birth. 25 There is a risk for detection bias in the current study. Patients with more severe disease may have had a diagnosis of CD earlier in life than patients with a less severe disease. However, the positive association with smoking in diagnosed patients is not consistent with a protective effect of smoking in CD. Smoking, irrespective of the amount of cigarettes consumed, did not constitute a risk factor for subsequent CD when adjusting for civil status, age, and year of infant birth. Although the grading of smoking was somewhat insensitive, we had considerable statistical power to detect differences in risk for CD among both light and heavy. Importantly, measurement of smoking was prospective, which sets this study apart from previous research in this field. Smoking data were obtained in the 12th gestational week. Some gave up smoking after pregnancy. However, this only would have affected our point estimates had there been different smoking cessation patterns in women who subsequently would receive a diagnosis of CD and in controls. We found no such differential smoking cessation pattern. Furthermore, ORs were similar when looking at smoking and undiagnosed CD in the first and second pregnancy. Still, because we have a small sample of patients with a celiac diagnosis as the main diagnosis, we cannot exclude a protective effect of smoking in patients with only CD (OR,.88) and no comorbidity, although the OR is not statistically significant and the confidence interval is wide. The inpatient register involves more than 99% of all discharge events. However, not all patients with CD are hospitalized and a number of women with CD in this study probably were classified as false-negative controls. This only affected our results marginally because the number of true controls exceeded 1 million. We also feel certain that a vast majority of patients with a hospitalbased diagnosis of CD actually have CD. Generally, in Sweden, the diagnosis of CD is based on at least 1 small intestinal biopsy examination 26 and probably has a high specificity in the Swedish national registers. In a recent register-based study by Smedby et al, 27 the diagnosis of CD in a subset of patients with lymphoma had a specificity of 85% 90%. Our results are based exclusively on female participants. That is probably of little concern because a majority of CD patients are women 28 and there are no indications that the cause of the disease is different in women and men. With the exception of the study by Patel et al, 11 women have constituted at least 65% of the participants in the previously published studies on smoking and CD. 4,8 10,17 What is of potential concern is the risk for selection bias caused by disease activity. On the one hand, cases in this study may suffer from more severe disease than the average patient because patients with active disease and CD complications, including other autoimmune diseases, are more likely to be hospitalized. On the other hand, women appearing in the medical birth register are less likely to suffer from infertility. Women with CD probably are more prone to low fertility 29,30 and those giving birth may have less active CD than the average patient. However, hospital admission used to be common within the diagnostic framework of CD and unrelated to CD activity. Besides, ORs were similar in subsets of women receiving their diagnosis of CD within less than 5 years after infant birth and in those receiving their diagnosis within 5 years or more after infant birth. If smoking only affected the risk for CD in mothers with more active disease we would expect a lower OR for undiagnosed CD at birth among women receiving their diagnosis within less than 5

874 LUDVIGSSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 9 years, and that was not the case. In addition, prior CD was associated positively with smoking, which once again is not consistent with a protective effect of smoking. There is a risk that heavy are underrepresented in this study because of lower fertility. 31 Although heavy smoking may be less common in pregnant women, this study contained a large subset of women who smoked heavily and in these women no increased risk for undiagnosed CD was seen. Still we cannot exclude the possibility that smoking behavior could affect the risk for undiagnosed CD 10 15 years after exposure; for that our follow-up time is insufficient. Fertility decreases with age. In several studies on smoking and CD, the average age of the participants has been close to or greater than 40 years 4,8 10 ; in the EMA screening study by West et al, 17 the average participant age was 59 years. In contrast, our study was restricted to young women of childbearing age. However, the age restriction of 15 44 years only applied to age at infant birth. A woman giving birth at the age of 44 in 1983 subsequently could be diagnosed as having CD in 2001 at the age of 62 years and would be included as a case in our study. This study does not include people of old age, in whom the incidence of CD may be relatively high. 32 However, cause is likely to be the same in patients younger and older than 60 years of age. If smoking had a major protective effect we likely would be able to see it in the current subset of patients, even though the study was based on pregnant women in whom CD may be less active than in the average patient with CD. In conclusion, this study fails to show any link between smoking in pregnant women and a subsequent diagnosis of CD. References 1. Koning F. The molecular basis of celiac disease. J Mol Recognit 2003;16:333 336. 2. Greco L, Romino R, Coto I, et al. The first large population based twin study of coeliac disease. Gut 2002;50:624 628. 3. Sollid LM. Molecular basis of celiac disease. Annu Rev Immunol 2000;18:53 81. 4. Snook JA, Dwyer L, Lee-Elliott C, et al. Adult coeliac disease and cigarette smoking. Gut 1996;39:60 62. 5. Veldhuyzen van Zanten SJ. Recurrent diarrhoea and weight loss associated with cessation of smoking in a patient with undiagnosed coeliac disease. Gut 2001;49:588. 6. Jolobe O. Recurrent diarrhoea and weight loss associated with cessation of smoking in undiagnosed coeliac disease. Gut 2002; 50:582. 7. Lear JT, English JS, Jones PW. Adult coeliac disease, dermatitis herpetiformis and smoking. Gut 1997;40:289. 8. Vazquez H, Smecuol E, Flores D, et al. Relation between cigarette smoking and celiac disease: evidence from a case-control study. Am J Gastroenterol 2001;96:798 802. 9. Austin AS, Logan RF, Thomason K, et al. Cigarette smoking and adult coeliac disease. Scand J Gastroenterol 2002;37:978 982. 10. Suman S, Williams EJ, Thomas PW, et al. Is the risk of adult coeliac disease causally related to cigarette exposure? Eur J Gastroenterol Hepatol 2003;15:995 1000. 11. Patel AH, Loftus EV Jr, Murray JA, et al. Cigarette smoking and celiac sprue: a case-control study. Am J Gastroenterol 2001;96: 2388 2391. 12. Ludvigsson JF, Ludvigsson J. Coeliac disease in the father affects the newborn. Gut 2001;49:169 175. 13. Cnattingius S, Ericson A, Gunnarskog J, et al. A quality study of a medical birth registry. Scand J Soc Med 1990;18:143 148. 14. Lunde AS, Lundeborg S, Lettenstrom GS, et al. The personnumber systems of Sweden, Norway, Denmark, and Israel. Vital Health Stat 2 1980;2:1 59. 15. Askling J, Linet M, Gridley G, et al. Cancer incidence in a population-based cohort of individuals hospitalized with celiac disease or dermatitis herpetiformis. Gastroenterology 2002;123:1428 1435. 16. SPSS. 11.0 ed. Chicago, IL: SPSS Inc, 2002. 17. West J, Logan RF, Hill PG, et al. Seroprevalence, correlates, and characteristics of undetected coeliac disease in England. Gut 2003;52:960 965. 18. McMillan SA, Johnston SD, Watson RG, et al. Dietary intake, smoking, and transient anti-gliadin antibodies. Scand J Gastroenterol 1998;33:499 503. 19. Law MR, Hackshaw AK. A meta-analysis of cigarette smoking, bone mineral density and risk of hip fracture: recognition of a major effect. BMJ 1997;315:841 846. 20. Sopori M. Effects of cigarette smoke on the immune system. Nat Rev Immunol 2002;2:372 377. 21. Kokkonen J, Simila S, Vuolukka P. The incidence of coeliac disease and pyloric stenosis in children in Northern Finland. Ann Clin Res 1982;14:123 128. 22. Greco L, Tozzi AE, Mayer M, et al. Unchanging clinical picture of coeliac disease presentation in Campania, Italy. Eur J Pediatr 1989;148:610 613. 23. Ludvigsson JF. Socioeconomic characteristics in children with coeliac disease. Acta Paediatr 2005;94:107 113. 24. Sjoberg I, Persson J, Johansson S. Hälsorelaterade levnadsvanor 1980-2002: Statistics Sweden, 2004. 25. Tobaksvanor bland gravida och spädbarnsföräldrar 2002. Stockholm, Sweden: National Board of Health, 2004. 26. Danielsson L, Stenhammar L, Ascher H, et al. [Proposed criteria for diagnosis of celiac disease in children]. Lakartidningen 1998; 95:2342 2343. 27. Smedby KE, Akerman M, Hildebrand H, et al. Malignant lymphomas in coeliac disease: evidence of increased risks for lymphoma types other than enteropathy-type T cell lymphoma. Gut 2005;54:54 59. 28. Ivarsson A, Hernell O, Nystrom L, et al. Children born in the summer have increased risk for coeliac disease. J Epidemiol Community Health 2003;57:36 39. 29. Morris JS, Adjukiewicz AB, Read AE. Coeliac infertility: an indication for dietary gluten restriction? Lancet 1970;1:213 214. 30. Ferguson R, Holmes GK, Cooke WT. Coeliac disease, fertility, and pregnancy. Scand J Gastroenterol 1982;17:65 68. 31. The Practice Committee of the American Society Reproductive Medicine. Smoking and infertility. Fertil Steril 2004;81:1181 1186. 32. Murray JA, Van Dyke C, Plevak MF, et al. Trends in the identification and clinical features of celiac disease in a North American community, 1950-2001. Clin Gastroenterol Hepatol 2003;1:19 27. Address requests for reprints to: Jonas F. Ludvigsson, MD, Pediatric Department, Örebro University Hospital, Örebro, Sweden. e-mail: jonasludvigsson@yahoo.com; fax: (46)-0-19-187915. Supported in part by a grant from the Örebro University Hospital (to J.F.L.), the Swedish Society of Medicine, the Swedish Research Council, and the Majblomman Foundation.