- Email: [email protected]
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults Graziella Bruno, MD1, Teresa Spadea, MSc2, Roberta Picariello, BSc2, Gabriella Gruden, PhD1, Federica Barutta, PhD1, Franco Cerutti, MD3, Paolo Cavallo-Perin, MD1, Giuseppe Costa, MD2,4, Roberto Gnavi, MD2, and Piedmont Study Group for Diabetes Epidemiology* Objective To examine the potential role of 2 early-life socioeconomic indicators, parental education, and crowding index, on risk of type 1 diabetes (T1DM) in patients up to age 29 years to test heterogeneity by age at onset according to the hygiene hypothesis. Study design The study base was 330 950 individuals born from 1967 to 2006 who resided in the city of Turin at any time between 1984 and 2007. Data on their early life socioeconomic position were derived from the Turin Longitudinal Study; 414 incident cases of T1DM up to age 29 years were derived from the Turin T1DM registry. Results Socioeconomic indicators had opposing effects on risk of T1DM in different age at onset subgroups. In a Poisson regression model that included both socioeconomic indicators, there was a 3-fold greater risk of T1DM (relative risk 2.91, 95% CI 0.99-8.56) in children age 0-3 years at diagnosis living in crowded houses. In the 4- to 14-year subgroup, a low parental educational level had a protective effect (relative risk 0.50, 95% CI 0.29-0.84), and the effect of crowding nearly disappeared. In the 15- to 29-year subgroup, neither crowding nor parental educational level was clearly associated with the incidence of T1DM. Conclusions We provide evidence of heterogeneity by age at onset of the association between early-life socioeconomic indicators and the risk of T1DM. This finding is consistent with the hypothesis that infectious agents in the perinatal period may increase the risk, whereas in the following years they may become protective factors (hygiene hypothesis). (J Pediatr 2013;162:600-5).
T
he incidence of type 1 diabetes (T1DM) is increasing worldwide; this rapid change in incidence cannot be explained by evolving genetic susceptibility, and environmental causes are believed to play a predominant role.1 Social environments may influence the risk of exposure to infections, and in the last few decades the authors of several epidemiologic studies have assessed the association between socioeconomic factors and T1DM, reporting conflicting results. The Northern Ireland Diabetes Study Group has shown a reduced risk of diabetes in children living in areas with a greater level of household crowding.2 In a study from Montreal, the risk among children ages 0-14 years was slightly greater in wealthier as opposed to poorer classes.3 On the contrary, an ecological study performed in North Rhine–Westphalia showed that the risk of T1DM was greater in children living in socially deprived areas.4 The heterogeneity of T1DM by age at onset is well established. Patterns of autoimmune insulitis markers vary among subgroups with different age at onset, with prevailing positivities of glutamic acid decarboxylase antibody in young adults and positivities of antibodies to protein tyrosine phosphatase (IA2) and insulin autoantibody in children.5,6 Residual b-cell function at disease onset is lower in children than in adolescents and adults with T1DM.5,7 Finally, the effect of diabetes-susceptible human leukocyte antigen haplotypes and Sardinian genetic heritage on the risk of T1DM is greater in children than in young adults.8,9 Whether heterogeneity in age at onset reflects heterogeneity in the etiopathogenetic mechanisms underlying the development of T1DM is unknown.10 However, a shift to younger age at onset recently has been reported in both Sweden11 and East-Central Europe,12 suggesting that factors in very early life, both in utero and in the perinatal period, may be important in the etiology of T1DM. However, a protective role of early-life infections also has been hypothesized13 and, according to the “hygiene hypothesis,” a decreased early-life exposure to infectious diseases may enhance the risk of later development of immune-mediated disorders, including T1DM.14-17 The city of Turin, Northern Italy, presents a unique opportunity to investigate, From the Department of Medical Sciences, University of in detail, early life socioeconomic factors associated with T1DM by age at onset. Turin, Turin, Italy; Epidemiology Unit, ASL TO3, The Registry of T1DM has recruited all incident cases of the disease from birth up Piedmont Region, Grugliasco (TO), Italy; Department of 18 Pediatrics, and Department of Clinical and Biological to 29 years of age since 1984, and the Turin Longitudinal Study database has Science, University of Turin, Turin, Italy collected individual data on socioeconomic indicators derived from censuses *List of members of the Piedmont Study Group for Diabetes Epidemiology is available at www.jpeds.com for all people who have been resident in the city of Turin since 1971.19 In this (Appendix). 1
2
3
4
RR T1DM
Relative risk Type 1 diabetes
Supported by the Piedmont Region, Ricerca Sanitaria Finalizzata 2007. The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2013 Mosby Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2012.09.010
600
Vol. 162, No. 3 March 2013 study, we analyzed population-based data derived from both data sources to examine the potential role of 2 early-life socioeconomic indicators assessed around birth, parental education and crowding index, on risk of T1DM in different age groups up to 29 years of age.
N=1372 Patients born between 1967 and 2006, with onset of T1DM between 1984 and 2007 resident in the Province of Turin
N=797 Patients excluded as not resident in the city of Turin
Methods The Registry of the Province of Turin recruited incident cases of patients with T1DM ages 0-29 years during the period 1984-2007 (n = 1551) through the following sources of ascertainment: (1) diabetes clinics from both public and private hospitals to which patients with diabetes are referred after diagnosis (primary source); and (2) files of all subjects who obtained exemption from payment of drugs, syringes, and glucose monitoring strips because of a diagnosis of diabetes mellitus (secondary source).18 For all people who were recruited, demographic data were checked through the demographic files of towns of residence. Clinical charts were reviewed in all patients to assess clinical features at diabetes onset (glycemia, ketonuria, hospital admission). A diagnosis of T1DM was determined on the basis of permanent insulin treatment within 6 months of diagnosis, fasting C-peptide levels #0.20 nmol/L (assessed in 86.6% of patients, 99.8% ages 0-14 years and 79.9% ages 15-29 years), and positivities for either islet cells antibody or glutamic acid decarboxylase antibody (assessed in 64.5% of patients, 92.2% ages 0-14 years and 51% ages 15-29 years). The Turin Longitudinal Study database contains demographic information on all people residing in the city of Turin since 1971 (approximately 860 000 inhabitants at the 2001 census) through a linkage between the historical Turin Population Register and the socioeconomic information of census.19,20 To test the impact of early life socioeconomic status on risk of T1DM, individual information was derived from the census (1971, 1981, 1991, 2001) closest to birth, that is, within 5 years before or after the birth of each individual, yielding 414 incident cases of T1DM who were born between 1967 and 2006. The Figure shows the steps followed to select the study population. Consistent with selection of the cases, the population base for the analyses was defined as people born from 1967 to 2006 who resided in the city of Turin at any time between 1984 and 2007 with individual data on their early-life socioeconomic position. This definition resulted in 330 950 individuals, corresponding to more than 4 800 000 person-years. Socioeconomic status was defined by 2 indicators of 2 different dimensions of individual resources, namely cultural and material resources.21,22 The first dimension (cultural resources) was represented by parental education, defined as the greatest educational level attained within the couple classified according to 3 levels: high (university/high school, ie, $13 years of education), medium (middle school, ie, up to 12 years of education), and low (primary school or less, ie, #8 years of education). The second dimension (material assets) was synthesized by the crowd-
N=575 Patients resident in the city of Turin
N=56 Patients excluded as no parent was found in the Turin Population Register N=519 Patients
N=105 Patients excluded as parent(s) were not linked to census N=414 Patients included in the analyses
Figure. Selection of patients with T1DM recruited from the Registry of the Province of Turin and included in the study.
ing index (square meters/number of family components), classified into 3 quantiles of the continuous distribution: <14 m2 per person (high crowding), 14-20 m2 per person (medium), and >20 m2 per person (low). The crowding index also was used as a proxy of early-life exposure to infectious diseases21-23 to test the hygiene hypothesis on risk of T1DM. Of 414 cases, information about parental education was available for 413, and information about crowding was available for 328 subjects. There were no differences between these 2 groups either by sex (P = .126), age at onset (P = .783), or parental education (.359). Therefore, although univariate analyses were performed by the use of 2 different databases, multivariate analysis was limited to 327 patients that included both indicators. Age at diabetes onset was classified into 3 groups (0-3, 414, 15-29 years) on the basis of the level of incidence rates at each year of age, to test the a priori hypothesis of heterogeneity of the disease by age at onset. The first group (0-3 years) included very young cases with a relatively low incidence rate, <10/100 000 person-years; in the second group (4-14 years), the incidence rate reached a peak of 18/100 000 at 11 years; in the last group (15-29 years), rates started to decrease again below 10/100 000 person-years. The impact of socioeconomic factors on risk of T1DM was assessed by means of multivariate Poisson regression models with the use of SAS version 9.1 (SAS Institute, Cary, North Carolina). 601
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 162, No. 3
Results
Discussion
Table I shows the characteristics of incident cases of T1DM from the Turin Registry (n = 414) who fit selection criteria and the record-linkage process and person-years. Results of the univariate analysis on the role of socioeconomic indicators in the risk of T1DM by age at onset are shown in Table II. Compared with the greatest educational level, the lowest parental educational level conferred an increased risk of T1DM in children with 0-3 years of age at onset (relative risk [RR] 2.17, 95% CI 0.71-6.59), although this finding did not reach statistical significance. In the subgroup of children ages 4-14 years at onset, a low level of parental education was associated with a statistically significant 40% lower risk of diabetes (RR 0.60, 95% CI 0.40-0.89). In young adults (15-29 years), the role of parental educational was neutral (RR 1.09, 95% CI 0.73-1.64). With regard to crowding, in children aged 0-3 years at onset, a high index conferred an increased risk (RR 3.24, 95% CI 1.20-8.75). No association was found between crowding and risk of T1DM in the 4- to 14-year age group, but a tendency towards a protection of greater crowding was found in the 15- to 29-year age group (RR 0.76, 95% CI 0.47-1.21). Results were confirmed in multivariate analyses that included both socioeconomic indicators and were stratified by age at onset (Table III). In the 0- to 3-year age group, there was a 3-fold greater risk of T1DM (RR 2.91, 95% CI 0.99-8.56) in children living in crowded houses. A low parental educational level also increased the risk in this group, although this was not significant. In the subgroup with 4-14 years of age at onset, a low level of parental education had a statistically significant protective effect (RR 0.50, 95% CI 0.29-0.84), and the effect of crowding nearly disappeared. In the 15- to 29-year age group, neither crowding nor level of parental education was clearly associated with risk of T1DM.
Our results provide evidence of heterogeneities by age at onset of the association between 2 different individual socioeconomic indicators at birth and the risk of having T1DM up to age 29 years in a large industrialized city. Indeed, living in crowded houses, which may be a proxy of exposure to infectious diseases, conferred an almost 3-fold greater risk of having T1DM up to 3 years of age, had a neutral effect in children with 4-14 years of age at onset, and showed a tendency towards a protective effect in young adults. However, our study shows that a low parental educational level had a protective effect in subjects who developed the disease at 4-14 years. Parental education may indicate a variety of health-related factors, including dietary patterns, residential environmental exposures, exposures to infectious agents, and the knowledge of and compliance with healthy lifestyles. Consistent with our results, in an ecological study from Germany investigators showed that socioeconomic deprivation, assessed by a composite deprivation index including household crowding, increased the risk of T1DM in children ages 8 years and younger and that this effect progressively diminished with increasing age.4 Collectively, these findings support the hypothesis of heterogeneity in etiopathogenesis of T1DM by age at onset. In very young children, the increased risk of T1DM in subjects with a low socioeconomic status might be attributable to differences in lifestyle, nutrition, and exposure to perinatal
Table I. Characteristics of the study populations T1DM
Age at onset, years 0-3 4-14 15-29 Educational level High Medium Low Missing Crowding Low Medium High Missing Total 602
Turin population
n
%
Person-years
%
35 239 140
8.5 57.7 33.8
629 812 1 818 709 2 365 045
13.1 37.8 49.1
160 161 92 1
38.7 39.0 22.3
1 749 830 1 733 855 1 310 908 18 974
36.5 36.2 27.3
105 117 106 86 414
32.0 35.7 32.3
1 220 689 1 248 360 1 320 084 1 024 433 4 813 567
32.2 32.9 34.8
100.00
100.0
Table II. Univariate analysis of the role of socioeconomic indicators on risk of T1DM in the Turin Registry, by age at onset Age at onset Educational level 0-3 years High Medium Low 4-14 years High Medium Low 15-29 years High Medium Low Crowding 0-3 years Low Medium High 4-14 years Low Medium High 15-29 years Low Medium High
n
RR (95% CI)
P for trend
14 17 4
1.00 1.60 (0.79-3.25) 2.17 (0.71-6.59)
.10
107 102 30
1.00 0.99 (0.76-1.30) 0.60 (0.40-0.89)*
.03*
39 42 58
1.00 0.94 (0.61-1.46) 1.09 (0.73-1.64)
.63
6 11 11
1.00 2.23 (0.82-6.03) 3.24 (1.20-8.75)*
.02*
66 66 58
1.00 1.07 (0.76-1.50) 1.01 (0.71-1.44)
.93
33 40 37
1.00 1.03 (0.65-1.63) 0.76 (0.47-1.21)
.22
Interaction between age of onset and parental education: P = .154. Interaction between age of onset and crowding: P = .006. *Statistically significant.
Bruno et al
ORIGINAL ARTICLES
March 2013
Table III. Multivariate analysis of the role of socioeconomic indicators on the risk of T1DM in the Turin Registry by age at onset Age at onset 0-3 years
4-14 years
15-29 years
Socioeconomic indicators Educational level High Medium Low Crowding Low Medium High Educational level High Medium Low Crowding Low Medium High Educational level High Medium Low Crowding Low Medium High
RR (95% CI)
P for trend
1.00 1.13 (0.49-2.61) 1.49 (0.39-5.71)
.59
1.00 2.13 (0.77-5.91) 2.91 (0.99-8.56)*
.05*
1.00 0.99 (0.72-1.38) 0.50 (0.29-0.84)
.03*
1.00 1.13 (0.79-1.61) 1.21 (0.82-1.79)
.27
1.00 1.08 (0.65-1.82) 1.33 (0.78-2.25)
.26
1.00 0.92 (0.56-1.61) 0.65 (0.38-1.12)
.10
Educational level and crowding reciprocally adjusted. *Statistically significant.
infections.1 Duration of breast-feeding in developed countries has increasingly become shorter in mothers of lower social classes and is associated with an enhanced risk of T1DM.24 Case-control studies have shown a greater risk of prenatal infectious in children with diabetes compared with children without diabetes,25,26 and a poor socioeconomic status with overcrowding present may expose one to a greater risk of perinatal infection with diabetogenic micro-organisms triggering autoimmunity. These results are relevant because epidemiologic studies have shown a shift to earlier age of onset of T1DM in East-Central Europe11,12 and a doubling of incidence in children younger than 5 years is predicted between 2005 and 2020.27 Although an increase in incidence in very young children has not been observed in Italy,18,28 environmental factors operating in this age group might be similar. In older children and young adults, a low socioeconomic status, particularly a lower level of parental education, was associated with a reduced risk of T1DM. This result is in keeping with the hygiene hypothesis, which suggests that reduced exposure to infections in people with greater socioeconomic status impairs the maturation of the developing immune system and increases the risk of T1DM.14 Several experimental studies have provided evidence supporting this hypothesis.29,30 An ecological study conducted in Sweden showed a greater incidence in areas with a high proportion of small families and greater socioeconomic levels.31 A lower risk of T1DM was found in people with low socioeconomic indexes and in those living in rural compared with urban areas.2,4
Case-control studies have shown a lower risk of diabetes with greater birth order and degree of social mixing in infancy, proxy measures of early childhood infections.32,33 Finally, a recent prospective population-based study showed that individual indicators of greater socioeconomic levels were associated with increased risk of latent autoimmune diabetes, supporting the hypothesis of increased later onset of autoimmunity in people with lower probability of infection occurrence in early life.34 A number of limitations of our study should be mentioned. The number of incident cases in each age group was quite low, and the effect of chance on our results cannot be ruled out. Information on early life socioeconomic position was retrieved for 72% of the cohort (414/575), and the remaining 28% of patients was excluded because of missing data of parents or failures in record linkage with census data. The availability of data depended upon completeness of the Turin Population Register and on the likelihood that the family was resident in Turin when the child was younger than 5 years of age. Both factors, however, are not likely to be associated with socioeconomic status and risk of diabetes; thus, the effect on our results, if any, is likely negligible. We compared age at diagnosis and sex distribution in the 2 groups (included and excluded from the study); although no sex differences emerged, subjects included were significantly younger (mean age 12.9) then those excluded (mean age 17.6). This difference is likely attributable to the difficulty in retrieving information about parents for older patients, possibly living outside their original family; again, this factor depends on the structure of the Turin Population Register and is unlikely to have biased our results. Finally, even if parental education and crowding are partially correlated (r Pearson = 0.45), they are not so collinear to bias our results. Our study has also several strengths. First, the populationbased study design allowed us to compare incidence rates of T1DM in a large cohort. Second, we were able to assess indicators of both crowding and parental education at the individual level, thus allowing to exclude the potential effect of ecological bias1,4,23,31,35; family crowding can be considered a better proxy of exposure to infectious disease than residence in areas with high population density.4,29,35-37 Furthermore, the use of socioeconomic indicators from census data during a 40-year period avoids recall bias, affecting questionnaire-based ascertainment of data. Finally, the study population is geographically homogeneous, being constituted by residents of a large Italian city, thus excluding potential environmental effects that affect mixed urban-rural populations. T1DM is likely a multifactorial disease, with different latency from exposure to disease onset in different age groups38; moreover, viruses may act as major players, either increasing or reducing the risk of the disease.29,39 In the light of these data, the association between socioeconomic indicators and T1DM may differ by age. Exposure to infectious disease (operationally measured through the crowding index) could have a direct effect in the very first years of life due to exposure to infectious agents in the perinatal (or even in
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults
603
THE JOURNAL OF PEDIATRICS
www.jpeds.com
the prenatal) period, whereas in the following years it may become a protective factor as the result of its immunogenic effect (hygiene hypothesis). Finally, as already reported in other studies, the association with socioeconomic factors diminishes with increasing age.4 A better understanding of the association between socioeconomic position and the incidence of the disease requires further research to disentangle the interplay between exposures, age, and socioeconomic position. n Submitted for publication Apr 10, 2012; last revision received Jul 13, 2012; accepted Sep 5, 2012. Reprint requests: Graziella Bruno, Department of Medical Sciences, University of Turin, corso Dogliotti 14, I-10126 Turin, Italy. E-mail: [email protected]
References 1. Forlenza GP, Rewers M. The epidemic of type 1 diabetes: what is it telling us? Curr Opin Endocrinol Diabetes Obes 2011;18:248-51. 2. Patterson CC, Carson DJ, Hadden DR. Epidemiology of childhood IDDM in Northern Ireland 1989-1994: low incidence in areas with highest population density and most household crowding. Diabetologia 1996;39:1063-9. 3. Siemiatycki J, Colle E, Campbell S, Dewar R, Aubert D, Belmonte MM. Incidence of IDDM in Montreal by ethnic group and by social class and comparisons with ethnic groups living elsewhere. Diabetes 1988;37: 1096-102. 4. Du Prel JB, Icks A, Grabert M, Holl RW, Giani G, Rosenbauer J. Socioeconomic conditions and type 1 diabetes in childhood in North Rhine– Westphalia, Germany. Diabetologia 2007;50:720-8. 5. Bruno G, Cerutti F, Merletti F, Cavallo-Perin P, Gandolfo E, Rivetti M, et al., Piedmont Study Group for Diabetes Epidemiology. Residual betacell function and male/female ratio are higher in incident young adults than in children: the registry of type 1 diabetes of the Province of Turin, Italy, 1984-2000. Diabetes Care 2005;28:312-7. 6. Steck AK, Johnson K, Barriga KJ, Miao D, Yu L, Hutton JC, et al. Age of islet autoantibody appearance and mean levels of insulin, but not GAD or IA-2 autoantibodies, predict age of diagnosis of type 1 diabetes: diabetes autoimmunity study in the young. Diabetes Care 2011;34: 1397-9. 7. Bruno G, Runzo C, Cavallo-Perin P, Merletti F, Rivetti M, Pinach S, et al., Piedmont Study Group for Diabetes Epidemiology. Incidence of type 1 and type 2 diabetes in adults aged 30-49 years: populationbased registry in the Province of Turin, Italy. Diabetes Care 2005;28: 2613-9. 8. Bruno G, Arcari R, Pagano A, Cerutti F, Berrino M, Pagano G. Genetic heterogeneity by age at onset of type 1 diabetes: higher prevalence of patients with 0 susceptible heterodimers in adults than in children in the registry of Turin, Italy. Diabetologia 2000;43:260-1. 9. Bruno G, Pagano G, Faggiano F, De Salvia A, Merletti F. Effect of Sardinian heritage on risk and age at onset of type 1 diabetes: a demographical case-control study of Sardinian migrants. Int J Epidemiol 2000;29:532-5. 10. Dahlquist G. Can we slow the rising incidence of childhood-onset autoimmune diabetes? The overload hypothesis. Diabetologia 2006;49:20-4. 11. Dahlquist GG, Nystr€ om L, Patterson CC, Swedish Childhood Diabetes Study Group, Diabetes Incidence in Sweden Study Group. Incidence of type 1 diabetes in Sweden among individuals aged 0-34 years, 19832007: an analysis of time trends. Diabetes Care 2011;34:1754-9. 12. Jarosz-Chobot P, Polanska J, Szadkowska A, Kretowski A, BandurskaStankiewicz E, Ciechanowska M, et al. Rapid increase in the incidence of type 1 diabetes in Polish children from 1989 to 2004, and predictions for 2010 to 2025. Diabetologia 2011;54:508-15. 13. Von Herrath M. Can we learn from viruses how to prevent type 1 diabetes? The role of viral infections in the pathogenesis of type 1 diabetes and the development of novel combination therapies. Diabetes 2009;58:2-11. 604
Vol. 162, No. 3 14. Bach J- F. The effect of infections on susceptibility to autoimmune and allergic diseases. N Engl J Med 2002;347:911-20. 15. Rook GA. Review series on helminths, immune modulation and the hygiene hypothesis: the broader implications of the hygiene hypothesis. Immunology 2009;126:3-11. 16. Fleming J, Fabry Z. The hygiene hypothesis and multiple sclerosis. Ann Neurol 2007;61:85-9. 17. Cooke A. Review series on helminths, immune modulation and the hygiene hypothesis: how might infection modulate the onset of type 1 diabetes? Immunology 2009;126:12-7. 18. Bruno G, Novelli G, Panero F, Perotto M, Monasterolo F, Bona G, et al., Piedmont Study Group for Diabetes Epidemiology. Incidence of type 1 diabetes is increasing both in children and young adults in Northern Italy: 1984-2004 temporal trend. Diabetologia 2009;52: 2531-5. 19. Marinacci C, Spadea T, Biggeri A, Demaria M, Caiazzo A, Costa G. The role of individual and contextual socioeconomic circumstances on mortality: analysis of time variations in a city of north west Italy. J Epidemiol Community Health 2004;58:199-207. 20. Spadea T, Zengarini N, Kunst A, Zanetti R, Rosso S, Costa G. Cancer risk in relationship to different indicators of adult socioeconomic position in Turin, Italy. Cancer Causes Control 2010;21:1117-30. 21. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Commun Health 2006;60:7-12. 22. Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: concepts, methodologies, and guidelines. Ann Rev Public Health 1997;18:341-78. 23. Staines A, Bodansky HJ, McKinney PA, Alexander FE, McNally RJ, Law GR, et al. Small variation in the incidence of childhood insulindependent diabetes mellitus in Yorkshire, UK: links with overcrowding and population density. Int J Epidemiol 1997;26:1307-13. 24. Gerstein HC. Cow’s milk exposure and type I diabetes mellitus. A critical overview of the clinical literature. Diabetes Care 1994;17:13-9. 25. Dahlquist G, Ivarsson SA, Lindstr€ om B, Forsgren M. Maternal enteroviral infection during pregnancy as a risk factor for childhood insulin dependent diabetes mellitus—a population based case-control study. Diabetes 1995;44:408-13. 26. Hyoty H, Hiltunen M, Knip M, Laakkonen M, V€ah€asalo P, Karjalainen J, et al. A prospective study of the role of coxsackie B and other enterovirus infections in the pathogenesis of IDDM. Childhood Diabetes in Finland (DiMe) Study Group. Diabetes 1995;44:652-7. 27. Patterson CC, Dahlquist GG, Gy€ ur€ us E, Green A, Soltesz G, EURODIAB Study Group. Incidence trends for childhood type 1 diabetes in Europe during 1989-2003 and predicted new cases 2005–20: a multicentre prospective registration study. Lancet 2009;373:2027-33. 28. Bruno G, Maule M, Merletti F, Novelli G, Falorni A, Iannilli A, et al., RIDI Study Group. Age-period-cohort analysis of 1990-2003 incidence time trends of childhood diabetes in Italy: the RIDI study. Diabetes 2010;59:2281-7. 29. Atkinson MA, Bluestone JA, Eisenbarth GS, Hebrok M, Herold KC, Accili D, et al. How does type 1 diabetes develop? The notion of homicide or b-cell suicide revisited. Diabetes 2011;60:1370-9. 30. Coppieters KT, Wiberg A, Tracy SM, von Herrath MG. Immunology in the clinic review series: focus on type 1 diabetes and viruses: the role of viruses in type 1 diabetes: a difficult dilemma. Clin Exp Immunol 2012; 168:5-11. 31. Gopinath S, Ortqvist E, Norgren S, Green A, Sanjeevi CB. Variations in incidence of type 1 diabetes in different municipalities of Stockholm. Ann N Y Acad Sci 2008;1150:200-7. 32. McKinney PA, Okasha M, Parslow RC, Law GR, Gurney KA, Williams R, et al. Early social mixing and childhood type 1 diabetes mellitus: a casecontrol study in Yorkshire, UK. Diabetes Med 2000;17:236-42. 33. Jarosz-Chobot P, Polanska J, Polanski A. Does social-economical transformation influence the incidence of type 1 diabetes mellitus? A Polish example. Pediatr Diabetes 2008;9:202-7. 34. Olsson L, Ahlbom A, Grill V, Midthjell K, Carlsson S. High levels of education are associated with an increased risk of latent autoimmune
Bruno et al
ORIGINAL ARTICLES
March 2013 diabetes in adults: results from the Nord-Trøndelag health study. Diabetes Care 2011;34:102-7. 35. Holmqvist BM, Lofman O, Samuelsson U. A low incidence of type 1 diabetes between 1977 and 2001 in south-eastern Sweden in areas with high population density and which are more deprived. Diabet Med 2008;25:255-60. 36. Miller LJ, Pearce J, Barnett R, Willis JA, Darlow BA, Scott RS. Is population mixing associated with childhood type 1 diabetes in Canterbury, New Zealand? Soc Sci Med 2009;68:625-30.
37. Grigsby-Toussaint DS, Lipton R, Chavez N, Handler A, Johnson TP, Kubo J. Neighborhood socioeconomic change and diabetes risk: findings from the Chicago childhood diabetes registry. Diabetes Care 2010;33: 1065-8. 38. Todd JA, Knip M, Mathieu C. Strategies for the prevention of autoimmune Type 1 diabetes. Diabet Med 2011;28:1141-3. 39. Boettler T, von Herrath M. Protection against or triggering of Type 1 diabetes? Different roles for viral infections. Expert Rev Clin Immunol 2011;7:45-53.
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults
605
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 162, No. 3
Appendix Members of Piedmont Study Group for Diabetes Epidemiology include: S. Cianciosi (Avigliana); A. Lesina (Carmagnola); C. Giorda (Chieri); A. Chiambretti and R. Fornengo (Chivasso); V. Trinelli (Cirie); A. Caccavale (Collegno); R. Autino and P. Modina (Cuorgne); L. Gurioli and L. Costa Laia (Ivrea); C. Marengo and M. Comoglio (Moncalieri); T. Mahagna (Nichelino); M. Trovati and F. Cavalot (San Luigi Hospital, Orbassano); A. Ozzello and P. Gennai (PineroloPomaretto-Torre Pellice); D. D’Avanzo (Rivoli); S. Davı and M. Dore (Susa); S. Martelli and E. Megale (Giovanni Bosco Hospital, Turin); A. Blatto (Maria Vittoria Hospital, Turin); P. Griseri and C. Matteoda (Martini Hospital, Turin); A. Grassi and A. Mormile (Mauriziano Hospital, Turin); G. Grassi, A. Bruno (Molinette Hospital, Turin); G. Petraroli (Ophthalmologic Hospital, Turin); F. Cerutti and I. Rabbone (Regina Margherita Pediatric Hospital, Turin); A. Clerico (Valdese Hospital, Turin); and G. Bendinelli and A. Bogazzi (Venaria).
605.e1
Bruno et al
T
he incidence of type 1 diabetes (T1DM) is increasing worldwide; this rapid change in incidence cannot be explained by evolving genetic susceptibility, and environmental causes are believed to play a predominant role.1 Social environments may influence the risk of exposure to infections, and in the last few decades the authors of several epidemiologic studies have assessed the association between socioeconomic factors and T1DM, reporting conflicting results. The Northern Ireland Diabetes Study Group has shown a reduced risk of diabetes in children living in areas with a greater level of household crowding.2 In a study from Montreal, the risk among children ages 0-14 years was slightly greater in wealthier as opposed to poorer classes.3 On the contrary, an ecological study performed in North Rhine–Westphalia showed that the risk of T1DM was greater in children living in socially deprived areas.4 The heterogeneity of T1DM by age at onset is well established. Patterns of autoimmune insulitis markers vary among subgroups with different age at onset, with prevailing positivities of glutamic acid decarboxylase antibody in young adults and positivities of antibodies to protein tyrosine phosphatase (IA2) and insulin autoantibody in children.5,6 Residual b-cell function at disease onset is lower in children than in adolescents and adults with T1DM.5,7 Finally, the effect of diabetes-susceptible human leukocyte antigen haplotypes and Sardinian genetic heritage on the risk of T1DM is greater in children than in young adults.8,9 Whether heterogeneity in age at onset reflects heterogeneity in the etiopathogenetic mechanisms underlying the development of T1DM is unknown.10 However, a shift to younger age at onset recently has been reported in both Sweden11 and East-Central Europe,12 suggesting that factors in very early life, both in utero and in the perinatal period, may be important in the etiology of T1DM. However, a protective role of early-life infections also has been hypothesized13 and, according to the “hygiene hypothesis,” a decreased early-life exposure to infectious diseases may enhance the risk of later development of immune-mediated disorders, including T1DM.14-17 The city of Turin, Northern Italy, presents a unique opportunity to investigate, From the Department of Medical Sciences, University of in detail, early life socioeconomic factors associated with T1DM by age at onset. Turin, Turin, Italy; Epidemiology Unit, ASL TO3, The Registry of T1DM has recruited all incident cases of the disease from birth up Piedmont Region, Grugliasco (TO), Italy; Department of 18 Pediatrics, and Department of Clinical and Biological to 29 years of age since 1984, and the Turin Longitudinal Study database has Science, University of Turin, Turin, Italy collected individual data on socioeconomic indicators derived from censuses *List of members of the Piedmont Study Group for Diabetes Epidemiology is available at www.jpeds.com for all people who have been resident in the city of Turin since 1971.19 In this (Appendix). 1
2
3
4
RR T1DM
Relative risk Type 1 diabetes
Supported by the Piedmont Region, Ricerca Sanitaria Finalizzata 2007. The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2013 Mosby Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2012.09.010
600
Vol. 162, No. 3 March 2013 study, we analyzed population-based data derived from both data sources to examine the potential role of 2 early-life socioeconomic indicators assessed around birth, parental education and crowding index, on risk of T1DM in different age groups up to 29 years of age.
N=1372 Patients born between 1967 and 2006, with onset of T1DM between 1984 and 2007 resident in the Province of Turin
N=797 Patients excluded as not resident in the city of Turin
Methods The Registry of the Province of Turin recruited incident cases of patients with T1DM ages 0-29 years during the period 1984-2007 (n = 1551) through the following sources of ascertainment: (1) diabetes clinics from both public and private hospitals to which patients with diabetes are referred after diagnosis (primary source); and (2) files of all subjects who obtained exemption from payment of drugs, syringes, and glucose monitoring strips because of a diagnosis of diabetes mellitus (secondary source).18 For all people who were recruited, demographic data were checked through the demographic files of towns of residence. Clinical charts were reviewed in all patients to assess clinical features at diabetes onset (glycemia, ketonuria, hospital admission). A diagnosis of T1DM was determined on the basis of permanent insulin treatment within 6 months of diagnosis, fasting C-peptide levels #0.20 nmol/L (assessed in 86.6% of patients, 99.8% ages 0-14 years and 79.9% ages 15-29 years), and positivities for either islet cells antibody or glutamic acid decarboxylase antibody (assessed in 64.5% of patients, 92.2% ages 0-14 years and 51% ages 15-29 years). The Turin Longitudinal Study database contains demographic information on all people residing in the city of Turin since 1971 (approximately 860 000 inhabitants at the 2001 census) through a linkage between the historical Turin Population Register and the socioeconomic information of census.19,20 To test the impact of early life socioeconomic status on risk of T1DM, individual information was derived from the census (1971, 1981, 1991, 2001) closest to birth, that is, within 5 years before or after the birth of each individual, yielding 414 incident cases of T1DM who were born between 1967 and 2006. The Figure shows the steps followed to select the study population. Consistent with selection of the cases, the population base for the analyses was defined as people born from 1967 to 2006 who resided in the city of Turin at any time between 1984 and 2007 with individual data on their early-life socioeconomic position. This definition resulted in 330 950 individuals, corresponding to more than 4 800 000 person-years. Socioeconomic status was defined by 2 indicators of 2 different dimensions of individual resources, namely cultural and material resources.21,22 The first dimension (cultural resources) was represented by parental education, defined as the greatest educational level attained within the couple classified according to 3 levels: high (university/high school, ie, $13 years of education), medium (middle school, ie, up to 12 years of education), and low (primary school or less, ie, #8 years of education). The second dimension (material assets) was synthesized by the crowd-
N=575 Patients resident in the city of Turin
N=56 Patients excluded as no parent was found in the Turin Population Register N=519 Patients
N=105 Patients excluded as parent(s) were not linked to census N=414 Patients included in the analyses
Figure. Selection of patients with T1DM recruited from the Registry of the Province of Turin and included in the study.
ing index (square meters/number of family components), classified into 3 quantiles of the continuous distribution: <14 m2 per person (high crowding), 14-20 m2 per person (medium), and >20 m2 per person (low). The crowding index also was used as a proxy of early-life exposure to infectious diseases21-23 to test the hygiene hypothesis on risk of T1DM. Of 414 cases, information about parental education was available for 413, and information about crowding was available for 328 subjects. There were no differences between these 2 groups either by sex (P = .126), age at onset (P = .783), or parental education (.359). Therefore, although univariate analyses were performed by the use of 2 different databases, multivariate analysis was limited to 327 patients that included both indicators. Age at diabetes onset was classified into 3 groups (0-3, 414, 15-29 years) on the basis of the level of incidence rates at each year of age, to test the a priori hypothesis of heterogeneity of the disease by age at onset. The first group (0-3 years) included very young cases with a relatively low incidence rate, <10/100 000 person-years; in the second group (4-14 years), the incidence rate reached a peak of 18/100 000 at 11 years; in the last group (15-29 years), rates started to decrease again below 10/100 000 person-years. The impact of socioeconomic factors on risk of T1DM was assessed by means of multivariate Poisson regression models with the use of SAS version 9.1 (SAS Institute, Cary, North Carolina). 601
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 162, No. 3
Results
Discussion
Table I shows the characteristics of incident cases of T1DM from the Turin Registry (n = 414) who fit selection criteria and the record-linkage process and person-years. Results of the univariate analysis on the role of socioeconomic indicators in the risk of T1DM by age at onset are shown in Table II. Compared with the greatest educational level, the lowest parental educational level conferred an increased risk of T1DM in children with 0-3 years of age at onset (relative risk [RR] 2.17, 95% CI 0.71-6.59), although this finding did not reach statistical significance. In the subgroup of children ages 4-14 years at onset, a low level of parental education was associated with a statistically significant 40% lower risk of diabetes (RR 0.60, 95% CI 0.40-0.89). In young adults (15-29 years), the role of parental educational was neutral (RR 1.09, 95% CI 0.73-1.64). With regard to crowding, in children aged 0-3 years at onset, a high index conferred an increased risk (RR 3.24, 95% CI 1.20-8.75). No association was found between crowding and risk of T1DM in the 4- to 14-year age group, but a tendency towards a protection of greater crowding was found in the 15- to 29-year age group (RR 0.76, 95% CI 0.47-1.21). Results were confirmed in multivariate analyses that included both socioeconomic indicators and were stratified by age at onset (Table III). In the 0- to 3-year age group, there was a 3-fold greater risk of T1DM (RR 2.91, 95% CI 0.99-8.56) in children living in crowded houses. A low parental educational level also increased the risk in this group, although this was not significant. In the subgroup with 4-14 years of age at onset, a low level of parental education had a statistically significant protective effect (RR 0.50, 95% CI 0.29-0.84), and the effect of crowding nearly disappeared. In the 15- to 29-year age group, neither crowding nor level of parental education was clearly associated with risk of T1DM.
Our results provide evidence of heterogeneities by age at onset of the association between 2 different individual socioeconomic indicators at birth and the risk of having T1DM up to age 29 years in a large industrialized city. Indeed, living in crowded houses, which may be a proxy of exposure to infectious diseases, conferred an almost 3-fold greater risk of having T1DM up to 3 years of age, had a neutral effect in children with 4-14 years of age at onset, and showed a tendency towards a protective effect in young adults. However, our study shows that a low parental educational level had a protective effect in subjects who developed the disease at 4-14 years. Parental education may indicate a variety of health-related factors, including dietary patterns, residential environmental exposures, exposures to infectious agents, and the knowledge of and compliance with healthy lifestyles. Consistent with our results, in an ecological study from Germany investigators showed that socioeconomic deprivation, assessed by a composite deprivation index including household crowding, increased the risk of T1DM in children ages 8 years and younger and that this effect progressively diminished with increasing age.4 Collectively, these findings support the hypothesis of heterogeneity in etiopathogenesis of T1DM by age at onset. In very young children, the increased risk of T1DM in subjects with a low socioeconomic status might be attributable to differences in lifestyle, nutrition, and exposure to perinatal
Table I. Characteristics of the study populations T1DM
Age at onset, years 0-3 4-14 15-29 Educational level High Medium Low Missing Crowding Low Medium High Missing Total 602
Turin population
n
%
Person-years
%
35 239 140
8.5 57.7 33.8
629 812 1 818 709 2 365 045
13.1 37.8 49.1
160 161 92 1
38.7 39.0 22.3
1 749 830 1 733 855 1 310 908 18 974
36.5 36.2 27.3
105 117 106 86 414
32.0 35.7 32.3
1 220 689 1 248 360 1 320 084 1 024 433 4 813 567
32.2 32.9 34.8
100.00
100.0
Table II. Univariate analysis of the role of socioeconomic indicators on risk of T1DM in the Turin Registry, by age at onset Age at onset Educational level 0-3 years High Medium Low 4-14 years High Medium Low 15-29 years High Medium Low Crowding 0-3 years Low Medium High 4-14 years Low Medium High 15-29 years Low Medium High
n
RR (95% CI)
P for trend
14 17 4
1.00 1.60 (0.79-3.25) 2.17 (0.71-6.59)
.10
107 102 30
1.00 0.99 (0.76-1.30) 0.60 (0.40-0.89)*
.03*
39 42 58
1.00 0.94 (0.61-1.46) 1.09 (0.73-1.64)
.63
6 11 11
1.00 2.23 (0.82-6.03) 3.24 (1.20-8.75)*
.02*
66 66 58
1.00 1.07 (0.76-1.50) 1.01 (0.71-1.44)
.93
33 40 37
1.00 1.03 (0.65-1.63) 0.76 (0.47-1.21)
.22
Interaction between age of onset and parental education: P = .154. Interaction between age of onset and crowding: P = .006. *Statistically significant.
Bruno et al
ORIGINAL ARTICLES
March 2013
Table III. Multivariate analysis of the role of socioeconomic indicators on the risk of T1DM in the Turin Registry by age at onset Age at onset 0-3 years
4-14 years
15-29 years
Socioeconomic indicators Educational level High Medium Low Crowding Low Medium High Educational level High Medium Low Crowding Low Medium High Educational level High Medium Low Crowding Low Medium High
RR (95% CI)
P for trend
1.00 1.13 (0.49-2.61) 1.49 (0.39-5.71)
.59
1.00 2.13 (0.77-5.91) 2.91 (0.99-8.56)*
.05*
1.00 0.99 (0.72-1.38) 0.50 (0.29-0.84)
.03*
1.00 1.13 (0.79-1.61) 1.21 (0.82-1.79)
.27
1.00 1.08 (0.65-1.82) 1.33 (0.78-2.25)
.26
1.00 0.92 (0.56-1.61) 0.65 (0.38-1.12)
.10
Educational level and crowding reciprocally adjusted. *Statistically significant.
infections.1 Duration of breast-feeding in developed countries has increasingly become shorter in mothers of lower social classes and is associated with an enhanced risk of T1DM.24 Case-control studies have shown a greater risk of prenatal infectious in children with diabetes compared with children without diabetes,25,26 and a poor socioeconomic status with overcrowding present may expose one to a greater risk of perinatal infection with diabetogenic micro-organisms triggering autoimmunity. These results are relevant because epidemiologic studies have shown a shift to earlier age of onset of T1DM in East-Central Europe11,12 and a doubling of incidence in children younger than 5 years is predicted between 2005 and 2020.27 Although an increase in incidence in very young children has not been observed in Italy,18,28 environmental factors operating in this age group might be similar. In older children and young adults, a low socioeconomic status, particularly a lower level of parental education, was associated with a reduced risk of T1DM. This result is in keeping with the hygiene hypothesis, which suggests that reduced exposure to infections in people with greater socioeconomic status impairs the maturation of the developing immune system and increases the risk of T1DM.14 Several experimental studies have provided evidence supporting this hypothesis.29,30 An ecological study conducted in Sweden showed a greater incidence in areas with a high proportion of small families and greater socioeconomic levels.31 A lower risk of T1DM was found in people with low socioeconomic indexes and in those living in rural compared with urban areas.2,4
Case-control studies have shown a lower risk of diabetes with greater birth order and degree of social mixing in infancy, proxy measures of early childhood infections.32,33 Finally, a recent prospective population-based study showed that individual indicators of greater socioeconomic levels were associated with increased risk of latent autoimmune diabetes, supporting the hypothesis of increased later onset of autoimmunity in people with lower probability of infection occurrence in early life.34 A number of limitations of our study should be mentioned. The number of incident cases in each age group was quite low, and the effect of chance on our results cannot be ruled out. Information on early life socioeconomic position was retrieved for 72% of the cohort (414/575), and the remaining 28% of patients was excluded because of missing data of parents or failures in record linkage with census data. The availability of data depended upon completeness of the Turin Population Register and on the likelihood that the family was resident in Turin when the child was younger than 5 years of age. Both factors, however, are not likely to be associated with socioeconomic status and risk of diabetes; thus, the effect on our results, if any, is likely negligible. We compared age at diagnosis and sex distribution in the 2 groups (included and excluded from the study); although no sex differences emerged, subjects included were significantly younger (mean age 12.9) then those excluded (mean age 17.6). This difference is likely attributable to the difficulty in retrieving information about parents for older patients, possibly living outside their original family; again, this factor depends on the structure of the Turin Population Register and is unlikely to have biased our results. Finally, even if parental education and crowding are partially correlated (r Pearson = 0.45), they are not so collinear to bias our results. Our study has also several strengths. First, the populationbased study design allowed us to compare incidence rates of T1DM in a large cohort. Second, we were able to assess indicators of both crowding and parental education at the individual level, thus allowing to exclude the potential effect of ecological bias1,4,23,31,35; family crowding can be considered a better proxy of exposure to infectious disease than residence in areas with high population density.4,29,35-37 Furthermore, the use of socioeconomic indicators from census data during a 40-year period avoids recall bias, affecting questionnaire-based ascertainment of data. Finally, the study population is geographically homogeneous, being constituted by residents of a large Italian city, thus excluding potential environmental effects that affect mixed urban-rural populations. T1DM is likely a multifactorial disease, with different latency from exposure to disease onset in different age groups38; moreover, viruses may act as major players, either increasing or reducing the risk of the disease.29,39 In the light of these data, the association between socioeconomic indicators and T1DM may differ by age. Exposure to infectious disease (operationally measured through the crowding index) could have a direct effect in the very first years of life due to exposure to infectious agents in the perinatal (or even in
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults
603
THE JOURNAL OF PEDIATRICS
www.jpeds.com
the prenatal) period, whereas in the following years it may become a protective factor as the result of its immunogenic effect (hygiene hypothesis). Finally, as already reported in other studies, the association with socioeconomic factors diminishes with increasing age.4 A better understanding of the association between socioeconomic position and the incidence of the disease requires further research to disentangle the interplay between exposures, age, and socioeconomic position. n Submitted for publication Apr 10, 2012; last revision received Jul 13, 2012; accepted Sep 5, 2012. Reprint requests: Graziella Bruno, Department of Medical Sciences, University of Turin, corso Dogliotti 14, I-10126 Turin, Italy. E-mail: [email protected]
References 1. Forlenza GP, Rewers M. The epidemic of type 1 diabetes: what is it telling us? Curr Opin Endocrinol Diabetes Obes 2011;18:248-51. 2. Patterson CC, Carson DJ, Hadden DR. Epidemiology of childhood IDDM in Northern Ireland 1989-1994: low incidence in areas with highest population density and most household crowding. Diabetologia 1996;39:1063-9. 3. Siemiatycki J, Colle E, Campbell S, Dewar R, Aubert D, Belmonte MM. Incidence of IDDM in Montreal by ethnic group and by social class and comparisons with ethnic groups living elsewhere. Diabetes 1988;37: 1096-102. 4. Du Prel JB, Icks A, Grabert M, Holl RW, Giani G, Rosenbauer J. Socioeconomic conditions and type 1 diabetes in childhood in North Rhine– Westphalia, Germany. Diabetologia 2007;50:720-8. 5. Bruno G, Cerutti F, Merletti F, Cavallo-Perin P, Gandolfo E, Rivetti M, et al., Piedmont Study Group for Diabetes Epidemiology. Residual betacell function and male/female ratio are higher in incident young adults than in children: the registry of type 1 diabetes of the Province of Turin, Italy, 1984-2000. Diabetes Care 2005;28:312-7. 6. Steck AK, Johnson K, Barriga KJ, Miao D, Yu L, Hutton JC, et al. Age of islet autoantibody appearance and mean levels of insulin, but not GAD or IA-2 autoantibodies, predict age of diagnosis of type 1 diabetes: diabetes autoimmunity study in the young. Diabetes Care 2011;34: 1397-9. 7. Bruno G, Runzo C, Cavallo-Perin P, Merletti F, Rivetti M, Pinach S, et al., Piedmont Study Group for Diabetes Epidemiology. Incidence of type 1 and type 2 diabetes in adults aged 30-49 years: populationbased registry in the Province of Turin, Italy. Diabetes Care 2005;28: 2613-9. 8. Bruno G, Arcari R, Pagano A, Cerutti F, Berrino M, Pagano G. Genetic heterogeneity by age at onset of type 1 diabetes: higher prevalence of patients with 0 susceptible heterodimers in adults than in children in the registry of Turin, Italy. Diabetologia 2000;43:260-1. 9. Bruno G, Pagano G, Faggiano F, De Salvia A, Merletti F. Effect of Sardinian heritage on risk and age at onset of type 1 diabetes: a demographical case-control study of Sardinian migrants. Int J Epidemiol 2000;29:532-5. 10. Dahlquist G. Can we slow the rising incidence of childhood-onset autoimmune diabetes? The overload hypothesis. Diabetologia 2006;49:20-4. 11. Dahlquist GG, Nystr€ om L, Patterson CC, Swedish Childhood Diabetes Study Group, Diabetes Incidence in Sweden Study Group. Incidence of type 1 diabetes in Sweden among individuals aged 0-34 years, 19832007: an analysis of time trends. Diabetes Care 2011;34:1754-9. 12. Jarosz-Chobot P, Polanska J, Szadkowska A, Kretowski A, BandurskaStankiewicz E, Ciechanowska M, et al. Rapid increase in the incidence of type 1 diabetes in Polish children from 1989 to 2004, and predictions for 2010 to 2025. Diabetologia 2011;54:508-15. 13. Von Herrath M. Can we learn from viruses how to prevent type 1 diabetes? The role of viral infections in the pathogenesis of type 1 diabetes and the development of novel combination therapies. Diabetes 2009;58:2-11. 604
Vol. 162, No. 3 14. Bach J- F. The effect of infections on susceptibility to autoimmune and allergic diseases. N Engl J Med 2002;347:911-20. 15. Rook GA. Review series on helminths, immune modulation and the hygiene hypothesis: the broader implications of the hygiene hypothesis. Immunology 2009;126:3-11. 16. Fleming J, Fabry Z. The hygiene hypothesis and multiple sclerosis. Ann Neurol 2007;61:85-9. 17. Cooke A. Review series on helminths, immune modulation and the hygiene hypothesis: how might infection modulate the onset of type 1 diabetes? Immunology 2009;126:12-7. 18. Bruno G, Novelli G, Panero F, Perotto M, Monasterolo F, Bona G, et al., Piedmont Study Group for Diabetes Epidemiology. Incidence of type 1 diabetes is increasing both in children and young adults in Northern Italy: 1984-2004 temporal trend. Diabetologia 2009;52: 2531-5. 19. Marinacci C, Spadea T, Biggeri A, Demaria M, Caiazzo A, Costa G. The role of individual and contextual socioeconomic circumstances on mortality: analysis of time variations in a city of north west Italy. J Epidemiol Community Health 2004;58:199-207. 20. Spadea T, Zengarini N, Kunst A, Zanetti R, Rosso S, Costa G. Cancer risk in relationship to different indicators of adult socioeconomic position in Turin, Italy. Cancer Causes Control 2010;21:1117-30. 21. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Commun Health 2006;60:7-12. 22. Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: concepts, methodologies, and guidelines. Ann Rev Public Health 1997;18:341-78. 23. Staines A, Bodansky HJ, McKinney PA, Alexander FE, McNally RJ, Law GR, et al. Small variation in the incidence of childhood insulindependent diabetes mellitus in Yorkshire, UK: links with overcrowding and population density. Int J Epidemiol 1997;26:1307-13. 24. Gerstein HC. Cow’s milk exposure and type I diabetes mellitus. A critical overview of the clinical literature. Diabetes Care 1994;17:13-9. 25. Dahlquist G, Ivarsson SA, Lindstr€ om B, Forsgren M. Maternal enteroviral infection during pregnancy as a risk factor for childhood insulin dependent diabetes mellitus—a population based case-control study. Diabetes 1995;44:408-13. 26. Hyoty H, Hiltunen M, Knip M, Laakkonen M, V€ah€asalo P, Karjalainen J, et al. A prospective study of the role of coxsackie B and other enterovirus infections in the pathogenesis of IDDM. Childhood Diabetes in Finland (DiMe) Study Group. Diabetes 1995;44:652-7. 27. Patterson CC, Dahlquist GG, Gy€ ur€ us E, Green A, Soltesz G, EURODIAB Study Group. Incidence trends for childhood type 1 diabetes in Europe during 1989-2003 and predicted new cases 2005–20: a multicentre prospective registration study. Lancet 2009;373:2027-33. 28. Bruno G, Maule M, Merletti F, Novelli G, Falorni A, Iannilli A, et al., RIDI Study Group. Age-period-cohort analysis of 1990-2003 incidence time trends of childhood diabetes in Italy: the RIDI study. Diabetes 2010;59:2281-7. 29. Atkinson MA, Bluestone JA, Eisenbarth GS, Hebrok M, Herold KC, Accili D, et al. How does type 1 diabetes develop? The notion of homicide or b-cell suicide revisited. Diabetes 2011;60:1370-9. 30. Coppieters KT, Wiberg A, Tracy SM, von Herrath MG. Immunology in the clinic review series: focus on type 1 diabetes and viruses: the role of viruses in type 1 diabetes: a difficult dilemma. Clin Exp Immunol 2012; 168:5-11. 31. Gopinath S, Ortqvist E, Norgren S, Green A, Sanjeevi CB. Variations in incidence of type 1 diabetes in different municipalities of Stockholm. Ann N Y Acad Sci 2008;1150:200-7. 32. McKinney PA, Okasha M, Parslow RC, Law GR, Gurney KA, Williams R, et al. Early social mixing and childhood type 1 diabetes mellitus: a casecontrol study in Yorkshire, UK. Diabetes Med 2000;17:236-42. 33. Jarosz-Chobot P, Polanska J, Polanski A. Does social-economical transformation influence the incidence of type 1 diabetes mellitus? A Polish example. Pediatr Diabetes 2008;9:202-7. 34. Olsson L, Ahlbom A, Grill V, Midthjell K, Carlsson S. High levels of education are associated with an increased risk of latent autoimmune
Bruno et al
ORIGINAL ARTICLES
March 2013 diabetes in adults: results from the Nord-Trøndelag health study. Diabetes Care 2011;34:102-7. 35. Holmqvist BM, Lofman O, Samuelsson U. A low incidence of type 1 diabetes between 1977 and 2001 in south-eastern Sweden in areas with high population density and which are more deprived. Diabet Med 2008;25:255-60. 36. Miller LJ, Pearce J, Barnett R, Willis JA, Darlow BA, Scott RS. Is population mixing associated with childhood type 1 diabetes in Canterbury, New Zealand? Soc Sci Med 2009;68:625-30.
37. Grigsby-Toussaint DS, Lipton R, Chavez N, Handler A, Johnson TP, Kubo J. Neighborhood socioeconomic change and diabetes risk: findings from the Chicago childhood diabetes registry. Diabetes Care 2010;33: 1065-8. 38. Todd JA, Knip M, Mathieu C. Strategies for the prevention of autoimmune Type 1 diabetes. Diabet Med 2011;28:1141-3. 39. Boettler T, von Herrath M. Protection against or triggering of Type 1 diabetes? Different roles for viral infections. Expert Rev Clin Immunol 2011;7:45-53.
Early Life Socioeconomic Indicators and Risk of Type 1 Diabetes in Children and Young Adults
605
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 162, No. 3
Appendix Members of Piedmont Study Group for Diabetes Epidemiology include: S. Cianciosi (Avigliana); A. Lesina (Carmagnola); C. Giorda (Chieri); A. Chiambretti and R. Fornengo (Chivasso); V. Trinelli (Cirie); A. Caccavale (Collegno); R. Autino and P. Modina (Cuorgne); L. Gurioli and L. Costa Laia (Ivrea); C. Marengo and M. Comoglio (Moncalieri); T. Mahagna (Nichelino); M. Trovati and F. Cavalot (San Luigi Hospital, Orbassano); A. Ozzello and P. Gennai (PineroloPomaretto-Torre Pellice); D. D’Avanzo (Rivoli); S. Davı and M. Dore (Susa); S. Martelli and E. Megale (Giovanni Bosco Hospital, Turin); A. Blatto (Maria Vittoria Hospital, Turin); P. Griseri and C. Matteoda (Martini Hospital, Turin); A. Grassi and A. Mormile (Mauriziano Hospital, Turin); G. Grassi, A. Bruno (Molinette Hospital, Turin); G. Petraroli (Ophthalmologic Hospital, Turin); F. Cerutti and I. Rabbone (Regina Margherita Pediatric Hospital, Turin); A. Clerico (Valdese Hospital, Turin); and G. Bendinelli and A. Bogazzi (Venaria).
605.e1
Bruno et al