sign in sign up
<<
Home , Type 1 diabetes

Epidemiology/Health Services/Psychosocial Research ORIGINAL ARTICLE

A Meta-Analysis of the Association Between Childhood Type 1 and Atopic

1 2 CHRIS R. CARDWELL, MSC DENNIS J. CARSON, MB size and consequent lack of power. In the 2 1 MIKE D. SHIELDS, MD CHRIS C. PATTERSON, PHD latter situation a meta-analysis is valuable in synthesizing the available evidence (4). We therefore assessed the evidence for an association between and atopic disorders in children by conduct- OBJECTIVE — To review the published literature and perform a meta-analysis summarizing ing a systematic review of published epi- the evidence in support of an inverse association between type 1 diabetes and the atopic disor- demiological studies. ders: , eczema, and in children.

RESEARCH DESIGN AND METHODS — MEDLINE, Web of Science, and PubMed RESEARCH DESIGN AND were searched to identify relevant studies. These were assessed on quality criteria, and odds METHODS ratios (ORs) and 95% CIs were calculated for each study from the reported prevalences of in children with diabetes and in control children. Meta-analysis was then used to derive a combined OR and test for heterogeneity in findings between studies. Search The search used MEDLINE through RESULTS — Twenty-five studies were identified. Heterogeneity in the findings from different OVID ONLINE using this strategy: (“ex- studies was evident but was considerably reduced when the asthma and rhinitis analyses were plode ‘Diabetes Mellitus, Depen- restricted to those studies judged to be of adequate design. The meta-analysis revealed an inverse dent’/all subheadings”) and ((“explode association between asthma and type 1 diabetes, but the finding only attained significance when ‘Asthma’ or ‘asthma’ textword”) or (“ex- analysis was restricted to the studies of adequate design (OR 0.82, 95% CI 0.68–0.99). In this plode ‘Eczema’ or ‘eczema’ textword”) or subset an association of similar magnitude was observed between eczema and type 1 diabetes (“explode ‘Rhinitis’ or ‘rhinitis’ textword”) (0.82, 0.62–1.10) although this failed to attain statistical significance, and heterogeneity between studies was still present. There was little evidence of an association between rhinitis and type 1 or (“‘atopy’ textword”)). A similar strategy diabetes (0.97, 0.82–1.16) in this subset of studies. was used in searches on Web of Science and PubMed. Finally, to identify epidemi- CONCLUSIONS — Our analysis suggests that there is a small but significant reduction in the ological studies that potentially could prevalence of asthma in children with type 1 diabetes, but the findings for the other atopic have investigated atopy as one of a num- are less conclusive. ber of risk factors, a more general search was conducted on MEDLINE using this Diabetes Care 26:2568–2574, 2003 broader strategy: (“explode ‘Diabetes Mel- litus, Insulin Dependent’/all subhead- ings”) and ((“explode ‘Case-Control’”) or ype 1 diabetes results from an ab- oped countries (2,3) and suggested that (“explode ‘Cohort’”) or (“explode ‘Risk normal T-cell–mediated autoim- there is a positive association between al- Factor’”)). On each database the search T mune response, thought to be lergic and autoimmune diseases in indi- was limited to studies on humans pub- primarily of the T helper type 1 (Th1) vidual patients. This issue has previously lished before March 2003. Abstracts were cells. By contrast, T helper type 2 (Th2) been addressed by others who have screened independently by two investiga- cells predominate in allergic diseases. A sought to clarify the nature of any associ- tors (C.R.C. and C.C.P.) to establish if the review of the role of infections in suscep- ation at the patient level. Some studies studies were likely to provide relevant tibility to both autoimmune and allergic have reported an inverse relationship, but data based on the following inclusion cri- diseases (1) noted the increasing rates of the findings have been conflicting, possi- teria: 1) they identified comparable both disorders in recent decades in devel- bly because of poor design or inadequate groups with and without type 1 diabetes, 2) they ascertained the prevalence of one ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● or more manifestation of atopic disease in From the 1Department of Epidemiology & Public Health, Queen’s University of Belfast, Belfast, Northern 2 these two groups whether at diabetes di- Ireland, U.K; and the Department of Child Health, Queen’s University of Belfast, Belfast, Northern Ireland, agnosis or subsequently, and 3) they re- U.K. Address correspondence and reprint requests to Chris Cardwell, Department of Epidemiology & Public corded data that were primarily related to Health, The Queen’s University of Belfast, Grosvenor Road, Belfast BT12 6BJ, U.K. E-mail: c.cardwell@ the childhood period. Studies that made qub.ac.uk. use of comparative data on atopic diseases Received for publication 17 April 2003 and accepted in revised form 16 June 2003. gleaned from other publications were not Abbreviations: Th1, T helper type 1; Th2, T helper type 2. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion excluded from consideration at this stage. factors for many substances. Citations identified from the more general © 2003 by the American Diabetes Association. MEDLINE search were initially screened

2568 DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003 2569

Table 1—Characteristics of studies that have investigated the association between atopic diseases and type 1 diabetes

Study Case Subjects Control Subjects Exposure

Cardwell and Associates First author, year Age at (reference) Design Country Source onset Age at study No.* Source (matching variables) No.* Method Asthma Eczema Rhinitis Any atopy Prior†

Hermansson, 1971 (9) C-C¶ Sweden Va¨sterbotten county register Ͻ16 yr ? 128 (?) Va¨sterbotten county population 310 (?) Medical records and ———Diagnosis No register examination

Villa, 1988 (28) C-C¶ Italy ? ? 7–18 yr 101 (?) School population 90 (?) Questionnaire ———Symptoms No Siemiatycki, 1989 (27) C-C¶ Canada Montreal registry Ͻ18 yr Ͻ15 yr 161 (100) Friends (age, sex, district) 321 (100) Telephone interview 4OOOOOO Unspecified OOOOOO3 Yes

Blom, 1991 (24) C-C Sweden Swedish registry Ͻ15 yr Ͻ15 yr 264 (67) Swedish population register 388 (49) Questionnaire On —— —Yes (age, sex, county) medication

Stro¨mberg, 1995 (10) C-C¶ Sweden Norrkoping hospital ? 7–18 yr 61 (100) School friends (age, sex) 72 (92) Questionnaire 4OOOOOO Unspecified OOOOOO3 No Douek, 1999, 2000 (11,12) C-C United Kingdom Oxford registry Ͻ15 yr 12–14 yr 157 (76) ISAAC atopy study (age) 2,324 (86) Questionnaire 4OOO Symptoms OOOO3 — No Vargas, 1999 (18) C-C¶ Puerto Rico Puerto Rico registry Ͻ15 yr 13–22 yr 102 (?) Puerto Rican high school 80 (?) Questionnaire Diagnosis or —— —No symptoms Stene, 2000 (31) C-C Norway Vest-Agder registry Ͻ15 yr Ͻ15 yr 85 (94) Vest-Agdar population register 1,071 (73) Questionnaire 4OOO Unspecified OOOO3 — Yes C-C Austria Vienna registry Ͻ15 yr Ͻ15 yr‡ 99 (85) School lists (age) 363 (76) Questionnaire  Ͻ Ͻ   C-C Bulgaria W. Bulgarian registry 15 yr 15 yr‡ 128 (73) School, policlinic lists (age) 440 (78) Interview C-C Latvia Latvian registry Ͻ15 yr Ͻ15 yr‡ 141 (99) Latvian population register 324 (79) Interview   (age) EURODIAB, 2000 (13) Ͻ Ͻ C-C Lithuania Lithuanian registry 15 yr 15 yr‡ 116 (94) Policlinic lists (age) 263 (71) Questionnaire  4OOOO Diagnosis or symptoms OOOO3 Yes  C-C Luxembourg Luxembourg registry Ͻ15 yr Ͻ15 yr‡ 56 (95) Preschool, school lists (age) 159 (85) Interview   C-C Romania Bucharest registry Ͻ15 yr Ͻ15 yr‡ 82 (74) Health service register (age) 276 (81) Interview C-C United Kingdom Leeds registry Ͻ15 yr Ͻ15 yr‡ 208 (89) General practice list (age) 409 (76) Interview   C-C United Kingdom N. Ireland registry Ͻ15 yr Ͻ15 yr‡ 182 (76) General practice list (age) 444 (61) Questionnaire  Huang, 2001 (22) C-C¶ United States Diabetes website Ͻ18 yr ? 403 (?) U.S. health surveys 10,904 (?) Questionnaire 4OOO Unspecified OOOO3 — No Oleson, 2001 (21) C-C Denmark Danish registry Ͻ16 yr 3–15 yr 817 (89) Danish birth register (age) 7,683 (79) Questionnaire 4OOOOO Diagnosis OOOOO3 — No Kero, 2001 (14) Cohort Hospital discharge & social Ͻ8yr Ͻ8 yr 181 Finnish birth register (age) 59,684 Record linkage ICD code or ——— No support register benefit register 2003 Sebastiani, 2001 (26) C-C Italy Rome county registry Ͻ15 yr 11 yr§ 150 (93) Rome county school population 750 (94) Questionnaire 4OOO Unspecified OOOO3 — No (age, sex)

Simpson, 2002 (30) Cross-sectional United Kingdom General practice database Ͻ15 yr Ͻ15 yr 89 Registered general practice 49,051 Consultation in 4O Code for diagnosis (ICD) or symptoms O3 No EPTEMBER patients practice database

Meerwaldt, 2002 (15) C-C Netherlands Dutch Pediatricians Ͻ13 yr 7–12 yr 188 (?) ISAAC atopy study (age) 781 (?) Questionnaire 4OOO Symptoms OOOO3 — No 9, S Matilla, 2002 (16) C-C Finland Finish registry Ͻ15 yr 10–27 yr 361 (42) Finnish population register 483 (59) Questionnaire Diagnosis ——— No (age, sex)

Wright, 2003 (23) C-C¶ United Kingdom 5 North Trent hospitals ? 11 yr§ 226 (?) ISAAC atopy study 5472 (?) Questionnaire On medication ——— No NUMBER Rosenbauer, 2003 (33) C-C Germany German registry Ͻ5yr Ͻ6 yr 760 (71) German population register 1886 (43) Questionnaire 4OOOOOOUnspecifiedOOOOOO3 Yes

(age, sex, area) 26, *Number included in the analysis (percentage response rate for asthma exposure); †exposure referring to period prior to diabetes; ‡age at which atopic exposure was determined; §mean age; exposure referring to the year prior to questionnaire; ¶study design failed to satisfy quality criteria. C-C, case-control; ISAAC, the International Study of Asthma and in Childhood. VOLUME , ARE C IABETES D Meta-analysis of type 1 diabetes and atopy

Figure 1—Meta-analysis of studies of asthma and type 1 diabetes using the random-effects model, ordered by date of publication. ED, EURODIAB; n, number of subjects with asthma; N, total number of subjects in diabetes and control groups. *Test for overall effect: z ϭϪ1.06, P ϭ 0.29; test for heterogeneity ␹2 ϭ 43.6, df ϭ 21, P ϭ 0.003. †Test for overall effect: z ϭϪ2.07, P ϭ 0.04; test for heterogeneity ␹2 ϭ 23.0, df ϭ 16, P ϭ 0.11. to remove obviously irrelevant studies. was considered of adequate design, based Haenszel method was used to adjust the The reference lists of all pertinent articles on three criteria: 1) representative case as- OR for age in 5-year age-groups. Tests for were also examined. Finally, proceedings certainment, e.g., through a population- heterogeneity between studies were con- of recent relevant conferences were based registry, 2) population-based ducted and random-effect models used to reviewed. control selection, and 3) identical ascer- calculate pooled ORs (6). Random-effect Most studies identified were of the tainment of atopy exposure in the diabetic models were deemed more appropriate case-control design, with case definition and control groups. Information about than fixed-effect models because they based on childhood-onset type 1 diabetes atopy prevalence was then extracted. Au- provide a more conservative OR estimate, and atopic diseases considered as poten- thors were contacted to clarify ambigu- representing the mean association in the tial risk factors. No studies were found ities or incompleteness in the published populations, and it was anticipated that when these roles were reversed, reflecting results and in one instance to request in- there would be heterogeneity because of the much lower prevalence of diabetes. formation restricted to the age-group the observational nature of these studies. The atopic manifestations considered Ͻ15 years of age. Study-specific weights in the random- were asthma, eczema, allergic rhinitis, or effects model were calculated and scaled any atopy (defined as any of these three Statistical methods to percentages. Publication/selection bias diseases). Odds ratios (ORs) and exact 95% CIs was investigated by checking for asymme- Eligible studies were assessed inde- were calculated for each atopic exposure try in funnel plots of the logarithm of the pendently by two reviewers using a struc- and for combined atopy where possible. If study ORs against their SEs (7,8). A sub- tured form to abstract information about no exposure occurred in the diabetic or set analysis was conducted on studies that the study (country and year of publica- control group, a correction of one-half met the three quality criteria. Another tion), study subjects (source, area, and was added to both the exposed and unex- analysis was further restricted to studies age at diagnosis), and atopy exposures posed groups to provide an estimate of in which atopy exposure was ascertained (method of ascertainment and definitions the OR (5), and the Woolf approximation before diabetes diagnosis. Exploratory in- used). Studies were also assessed for qual- was used to calculate 95% CIs. In one vestigations of heterogeneity were also ity, and a subgroup was identified that cross-sectional study the Mantel- conducted by subdividing studies by geo-

2570 DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003 Cardwell and Associates

Figure 2—Meta-analysis of studies of eczema and type 1 diabetes using the random-effects model, ordered by date of publication. ED, EURODIAB; n, number of subjects with eczema; N, total number of subjects in diabetes and control groups. *Test for overall effect: z ϭϪ1.52, P ϭ 0.13; test for heterogeneity ␹2 ϭ 53.2, df ϭ 17, P Ͻ 0.001. †Test for overall effect: z ϭϪ1.33, P ϭ 0.18; test for heterogeneity ␹2 ϭ 47.5, df ϭ 14, P Ͻ 0.001. graphic region, incidence level, and atopy with the authors, two further abstracts each atopic exposure, although none at- ascertainment features (e.g., doctors’ di- were excluded; one (32) because informa- tained statistical significance (P Ͻ 0.05): agnosis). As the meta-analysis was con- tion specific to the childhood age range asthma (OR 0.90, 95% CI 0.73–1.10), ec- ducted from summary figures rather than was obtained from an associated study zema (0.82, 0.63–1.06), rhinitis (0.88, individual case records, the ORs could (30) and another (17) because more de- 0.68–1.12), and any atopy (0.80, 0.64– not be adjusted for confounders. Most tailed information was available from an 1.01). Funnel plots (Fig. 4) did not reveal studies incorporated some form of match- article in press (33). any obvious departures from symmetry, ing diabetic and control subjects for age; a The 19 remaining publications, 14 which is suggestive of publication/ few reported ORs adjusted for other con- articles, 3 abstracts, and 2 letters, cor- selection biases. Although the rhinitis plot founders. Statistical analyses were per- responded to 25 studies since 2 publica- suggested a deficit of smaller nonsignifi- formed using STATA 7.0 (Stata, College tions referred to different atopic cant studies, the pooled OR was close to 1 Station, TX). conditions recorded in a single study so the omission of the studies should not (11,12) and 1 publication reported re- result in bias. There was evidence of sta- RESULTS — The searches identified sults from 8 studies (13). Of these, 23 tistically significant heterogeneity be- eight eligible articles using MEDLINE (9– contained data on asthma, 18 on eczema, tween the studies: asthma (P ϭ 0.003), 16), a further six through Web of Science 18 on rhinitis, and 13 on a combined eczema (P Ͻ 0.001), rhinitis (P ϭ 0.002), (17–22), and another from PubMed (23). atopy exposure. The study characteristics and any atopy (P ϭ 0.04). This was con- The more general MEDLINE search iden- are summarized in Table 1. Seven studies firmed by failure of the funnel plots to tified four further articles (24–27). Re- were judged to have inadequate study de- demonstrate the expected funnel shape, view of reference lists revealed three sign, three featured nonrepresentative indicating that even the largest studies additional publications (28–30), and two case ascertainment (10,22,28), three showed differences that could not be ex- more abstracts (31,32) were discovered in non–population-based control subjects plained by chance and suggesting that conference proceedings. (10,18,27), and three ascertained atopy bias and confounding could be present. Six of these articles were excluded differently in diabetic and control groups The meta-analyses were therefore re- from further consideration. One was ex- (9,22,23). peated and restricted to studies with an cluded (25) because more detailed infor- Results from all studies for asthma, adequate design. The resulting pooled mation was available in a previous report eczema, and rhinitis are summarized in ORs were statistically significant for (13), and three meeting abstracts Figs. 1, 2, and 3, respectively. Analyses asthma (0.82, 0.68–0.99) but not for ec- (19,20,29) were excluded in favor of the including all studies suggested weak in- zema (0.82, 0.62–1.10), rhinitis (0.97, subsequent article (15). After contact verse associations between diabetes and 0.82–1.16), or any atopy (0.84, 0.65–

DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003 2571 Meta-analysis of type 1 diabetes and atopy

Figure 3—Meta-analysis of studies of rhinitis and type 1 diabetes using the random-effects model, ordered by date of publication. ED, EURODIAB; n, number of subjects with rhinitis; N, total number of subjects in diabetes and control groups. *Test for overall effect: z ϭϪ1.02, P ϭ 0.31; test for heterogeneity ␹2 ϭ 39.2, df ϭ 17, P ϭ 0.002. †Test for overall effect: z ϭϪ0.30, P ϭ 0.77; test for heterogeneity ␹2 ϭ 14.4, df ϭ 14, P ϭ 0.42.

1.08). There was less evidence of hetero- diabetes. Although the associations with but it is unclear in which direction any geneity for asthma (P ϭ 0.11) and rhinitis eczema, rhinitis, or any atopy did not at- bias might operate. Subsequent to a diag- (P ϭ 0.42), but the findings for eczema tain significance, the results for eczema nosis of diabetes, children have more fre- (P Ͻ 0.001) and any atopy (P ϭ 0.03) and any atopy were nevertheless consis- quent contact with health services that were still significant. Heterogeneity for tent with a small reduction similar in could result in greater likelihood that eczema was dramatically reduced (P ϭ magnitude to that for asthma. Failure to atopic conditions would be diagnosed. 0.52) only after further removing two out- detect any association with rhinitis may Alternatively the burden of coping with lying centers with significant positive as- reflect the difficulty of ascertaining this diabetes could lead to atopic conditions sociations, which resulted in a significant form of atopy reliably by questionnaire. being relegated in importance in parents’ OR (0.66, 0.58–0.75). Funnel plots con- The heterogeneity between the ORs, minds and therefore underreported. Only formed more closely to the expected even for the adequately designed studies, the latter bias could explain our finding of shape when restricted to the adequate could indicate true effect differences be- an inverse relationship between diabetes studies. tween populations, but possible sources and asthma. Analyses of the adequately designed of bias must also be considered (34). The Another potential source of bias was studies with the additional restriction that limitations of a meta-analysis based solely the generally lower response rates in con- exposure was measured before diabetes on summary statistics make it difficult to trol compared with case subjects. If there diagnosis gave findings that were not rule out biases from confounding factors. was an increased likelihood of response markedly altered: asthma (OR 0.83, 95% In six studies, attempts were made to con- for children with atopy, this could lead to CI 0.66–1.05), eczema (0.87, 0.59– trol for various confounding factors an overestimation of atopy prevalence in 1.29), rhinitis (0.99, 0.72–1.37), and any (13,14,16,21,27,33), but few had any ap- control subjects. Although we cannot rule atopy (0.84, 0.64–1.10). None of the preciable impact. It is difficult to predict out this effect, it is primarily a concern in other exploratory subdivisions revealed a what effect confounding by such variables studies where atopy was the main focus of markedly more homogenous grouping as birth weight, breast feeding, family investigation. In one such study (21) a and therefore did not identify sources of size, and maternal age might have, but the comparison of early and late responders heterogeneity. evidence from these studies would sug- showed no difference in atopy status. Co- gest that it may be small. hort studies of young children at high risk CONCLUSIONS — Our synthesis of In the predominantly retrospective for diabetes should be less affected by adequately designed studies showed a studies of this review, it is possible that these sources of bias and may help to re- small but significant reduction in the fre- recall of atopy could differ between the solve the issue, although an early report quency of asthma among children with parents of diabetic and control children, from one such study (35) was inconclusive.

2572 DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003 Cardwell and Associates

Figure 4—Funnel plots of obser- vational studies of atopy expo- sures and diabetes. Studies of adequate (F) and inadequate (E) design are labeled by reference number.

The argues that sion that the hygiene hypothesis provides References early environmental stimulation by infec- the explanation for these inverse associa- 1. Bach JF: The effect of infections on sus- tions is necessary to achieve a mature and tions could be premature because there ceptibility to autoimmune and allergic balanced repertoire of immune responses may be other shared environmental or ge- diseases. N Engl J Med 347:911–920, 2002 (36). Epidemiological studies provide ev- netic risk factors that predispose to one 2. Butland BK, Strachan DP, Lewis S, Bynner idence that frequent exposure to infec- disease and protect against another. J, Butler N, Britton J: Investigation into tions early in life is protective for both The focus of this review has been on the increase in hay fever and eczema at age diabetes (25,37,38) and asthma (39,40). the association between type 1 diabetes 16 observed between the 1958 and 1970 The protective mechanisms induced by and atopy in childhood, and there is evi- British birth cohorts. BMJ 315:717–721, infection are unknown but thought to be dence to suggest that the findings may 1997 related to the production of regulatory T- have differed if the analysis had been 3. EURODIAB ACE Study Group: Variation cells. Complex interactions between var- and trends in incidence of childhood dia- based on older age-groups (30,45). betes in Europe. Lancet 355:873–876, ious components of the In summary, our analysis suggests 2000 control the production of Th1 cells, that children with type 1 diabetes have a 4. Blettner M, Sauerbrei W, Schlehofer B, which are traditionally associated with slight but significant reduction in the risk Scheuchenpflug T, Friedenreich C: Tradi- , and Th2 cells, of asthma, although findings for the other tional reviews, meta-analyses and pooled which are traditionally associated with al- atopic diseases are less conclusive. Some analyses in epidemiology. Int J Epidemiol lergic disease (41,42). Such interactions early reports may inadvertently have ex- 28:1–9, 1999 could explain an inverse relationship be- aggerated the strength of the association 5. Gart JJ, Zweifel JR: On the bias of logit and tween autoimmune and allergic disease between atopic conditions and type 1 its variance, with applications to quantal such that the hygiene hypothesis is con- diabetes. bioassay. Biometrika 54:181–187, 1967 sistent with an inverse association be- 6. Dersimonian R, Laird N: Meta-analysis in clinical-trials.ControlledClinTrials7:177– tween atopic diseases and type 1 diabetes 188, 1986 at the individual level, despite their simul- Acknowledgments— C.R.C. received a 7. Egger M, Smith GD, Schneider M, Minder taneously increasing incidence in the Northern Ireland Department of Education C: Bias in meta-analysis detected by a sim- population. Such inverse associations and Learning grant. ple, graphical test. BMJ 315:629–634, have been reported for other autoimmune Thanks to those who kindly supplemented 1997 disorders (43,44). However, the conclu- their published data. 8. Sterne JAC, Egger M: Funnel plots for de-

DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003 2573 Meta-analysis of type 1 diabetes and atopy

tecting bias in meta-analysis: guidelines trup-Pedersen K: Association between 33. Rosenbauer J, Herzig P, Giani G: Atopic on choice of axis. J Clin Epidemiol and insulin-dependent eczema in early childhood could be pro- 54:1046–1055, 2001 diabetes mellitus: a case-control study. tective against type 1 diabetes. Diabetolo- 9. Hermansson B, Holmgren G, Samuelson Lancet 357:1749–1752, 2001 gia 46:784–788, 2003 G: Juvenile diabetes mellitus and atopy. 22. Huang S, Hitchcock J: Influence of TH1/ 34. Stroup DF, Berlin JA, Morton SC, Olkin I, Hum Hered 21:504–508, 1971 TH2 paradigm: the prevalence of asthma Williamson GD, Rennie D, Moher D, 10. Stromberg LG, Ludvigsson GJ, Bjorksten and allergic diseases in patients with type Becker BJ, Sipe TA, Thacker SB: Meta- B: Atopic and delayed-hypersensi- 1 diabetes (Abstract). J Allergy Clin Immu- analysis of observational studies in epide- tivity in children with diabetes. J Allergy nol 107 (Suppl. S):991, 2001 miology: a proposal for reporting: Meta- Clin Immunol 96:188–192, 1995 23. Wright NP, Wales JKH: The incidence of analysis Of Observational Studies in 11. Douek IF, Leech NJ, Gillmor HA, Bingley hypoglycaemia in children with type 1 di- Epidemiology (MOOSE) group. JAMA PJ, Gale EA: Children with type-1 diabe- abetes and treated asthma. Arc Dis Child 283:2008–2012, 2000 tes and their unaffected siblings have 88:155–156, 2003 35. Ziegler AG, Bonifacio E: No inverse rela- fewer symptoms of asthma (Letter). Lan- 24. Blom L, Nystrom L, Dahlquist G: The Swed- tionship between total IgE levels and islet cet 353:1850, 1999 ish Childhood Diabetes study: vaccinations in children of parents with 12. Douek IF, Leech NJ, Bingley PJ, Gale EA: and infections as risk determinants for dia- type 1 diabetes (Letter). Diabetes Care 23: Eczema and type 1 diabetes (Letter). Dia- betes in childhood. Diabetologia 34:176– 1205–1206, 2000 bet Med 19:174–175, 2000 181, 1991 36. Gale EA: A missing link in the hygiene 13. The EURODIAB Substudy 2 Study 25. McKinney PA, Okasha M, Parslow RC, hypothesis? Diabetologia 45:588–594, Group: Decreased prevalence of atopic Law GR, Gurney KA, Williams R, Bodan- 2002 diseases in children with diabetes. J Pedi- sky HJ: Early social mixing and childhood 37. Gibbon C, Smith T, Egger P, Betts P, Phil- atr 137:470–474, 2000 type 1 diabetes mellitus: a case-control lips D: Early infection and subsequent in- 14. Kero J, Gissler M, Hemminki E, Isolauri E: study in Yorkshire, UK. Diabet Med 17: sulin dependent diabetes. Arch Dis Child Could TH1 and TH2 diseases coexist? 236–242, 2000 77:384–385, 1997 Evaluation of asthma incidence in chil- 26. Sebastiani L, Visalli N, Adorisio E, D’Elia 38. The EURODIAB Substudy 2 Study dren with , type 1 diabetes: R, Conte A, Pozzilli P: [Type 1 diabetes in Group: Infections and vaccinations as risk or : a register study. J Lazio: note 2—case control study of risk factors for childhood type I (insulin-de- Allergy Clin Immunol 108:781–783, 2001 factors] (Article in Italian). Ann Ig 13:523– pendent) diabetes mellitus: a multicentre 15. Meerwaldt R, Odink RJ, Landaeta R, Aarts 530, 2001 case-control investigation. Diabetologia F, Brunekreef B, Gerritsen J, Van Aalderen 27. Siemiatycki J, Colle E, Campbell S, Dewar 43:47–53, 2000 WM, Hoekstra MO: A lower prevalence of RA, Belmonte MM: Case-control study of 39. Ball TM, Castro-Rodriguez JA, Griffith atopy symptoms in children with type 1 IDDM. Diabetes Care 12:209–216, 1989 KA, Holberg CJ, Martinez FD, Wright AL: diabetes mellitus. Clin Exp Allergy 28. Villa MP, Cacciari E, Bernardi F, Cicog- Siblings, day-care attendance, and the 32:254–255, 2002 nani A, Salardi S, Zapulla F: Bronchial re- risk of asthma and wheezing during child- 16. Mattila PS, Tarkkanen J, Saxen H, Pitka¨ni- activity in diabetic-patients: relationship hood. N Engl J Med 343:538–543, 2000 emi J, Karvonen M, Tuomilehto J: Predis- to duration of diabetes and degree of gly- 40. Illi S, von Mutius E, Lau S, Bergmann R, position to atopic symptoms to inhaled cemic control. Am J Dis Child 142:726– Niggemann B, Sommerfeld C, Wahn U: antigens may protect from childhood type 729, 1988 Early childhood infectious diseases and 1 diabetes. Diabetes Care 25:865–868, 29. Meerwaldt R, Odink R, Landaeta R, the development of asthma up to school 2002 Brunekreeft B, Gerritsen J, van Aalderen age: a birth cohort study. BMJ 322:390– 17. Giani G, Herzig P, Rosenbauer J: Atopic WMC, Hoekstra MO: Prevalence of atopy 395, 2001 diseases, infections, and vaccinations as in children with diabetes mellitus type 1 41. Mallia P, Johnston SL: Respiratory virus- risk for type 1 diabetes mellitus in child- (Abstract). Horm Res 50:125, 1998 es: do they protect from or induce hood (Abstract). Diabetologia 42 (Suppl. 30. Simpson CR, Anderson WJA, Helms PJ, asthma? Allergy 57:1118–1129, 2002 1):269, 1999 Taylor MW, Watson L, Prescott GJ, God- 42. Roep BO: The role of T-cells in the patho- 18. Vargas L, Roman W, Vega J, Veray G, den DJ, Barker RN: Coincidence of im- genesis of type 1 diabetes: from cause to Montealegre F, Saenz C, Frazer T: Preva- mune-mediated diseases driven by Th1 cure. Diabetologia 46:305–321, 2003 lence of asthma in IDDM (Abstract). Dia- and Th2 subsets suggests a common aeti- 43. Allanore Y, Hilliquin P, Coste J, Renoux betes 48 (Suppl. 1):A162, 1999 ology: a population-based study using M, Menkes CJ: Decreased prevalence of 19. Meerwaldt R, Brunekreef B, Gerritsen J, computerized General Practice data. Clin atopy in rheumatoid arthritis. Lancet 351: Von Aalderen W, Hoekstra M: A lower Exp Allergy 32:37–42, 2002 497, 1998 prevalence of atopic diseases in children 31. Stene LC, Ulriksen J, Joner G: Atopic ec- 44. Tremlett HL, Evans J, Wiles CM, Lus- with type 1 diabetes mellitus (Abstract). zema and the risk of childhood type 1 combe DK: Asthma and multiple sclero- Diabetologia 43 (Suppl. 1):417, 2000 diabetes in Vest-Agder County, Norway sis: an inverse association in a case- 20. Meerwaldt R, Odink R, Landaeta R, (Abstract). Eur J Endocrinol 142 (Suppl. control general practice population. QJM Brunekreeft B, Gerritsen J, van Aalderen 1):16, 2000 95:753–756, 2002 WMC, Hoekstra MO: Prevalence of atopic 32. Anderson WJA, Simpson CR, Taylor MW, 45. Sheikh A, Smeeth L, Hubbard R: There is disease in children with insulin-depen- Prescott GJ: Prevalence of TH2 mediated no evidence of an inverse relationship be- dent diabetes mellitus (IDDM) (Abstract). allergic disease in patients with TH1 me- tween TH2-mediated atopy and TH1-me- Am J Respir Crit Care Med 159 (Suppl. S): diated insulin dependant diabetes relative diated autoimmune disorders: lack of A146, 1999 to the general population (Abstract). Eur support for the hygiene hypothesis. J Al- 21. Olesen AB, Juul S, Birkebaek N, Thes- Respir J 16 (Suppl. 1): 160, 2002 lergy Clin Immunol 111:131–135, 2003

2574 DIABETES CARE, VOLUME 26, NUMBER 9, SEPTEMBER 2003