**2. Estimating the epidemiology of type 1 diabetes mellitus**

The epidemiology of type 1 diabetes can be estimated in different ways. In principle, there is the possibility of estimating epidemiologic data by self-report of the patients, longitudinalor cross-sectional studies or different-sized registries.

lation-based EURODIAB registers in 17 countries registered 29,311 new cases of type 1 dia‐ betes in children before their 15th birthday (Patterson, Dahlquist et al. 2009). The World Health Organization program, Multinational Project for Childhood Diabetes (Diabetes Mon‐ diale or DIAMOND), has been developed to investigate and characterize global incidence and mortality of type 1 diabetes and the health care provided for type 1 diabetic patients. Both projects used similar ascertainment methodologies. However, DIAMOND ascertained some data retrospectively. This may have led to some underestimation of incidence rates. The completeness of case ascertainment varied from 35 to 100% in DIAMOND. Most Euro‐ pean nations in DIAMOND had comparable (> 90%) rates of ascertainment to EURODIAB (Vehik and Dabelea 2010). DIAMOND reached the lowest completeness rates in Africa, Cen‐ tral and South America. This reflects a general problem when assessing type 1 diabetes epi‐ demiology: data from developing countries are scarce and may not be fully representative

The Epidemiology of Type 1 Diabetes Mellitus

http://dx.doi.org/10.5772/52893

5

This section provides a comprehensive description of type 1 diabetes incidence, its changes

Mean incidence rates of type 1 diabetes vary considerably depending on the geographic region (Galler, Stange et al. 2010). The worldwide incidence of type 1 diabetes is described to vary by at least 100- to 350-fold among different countries (Karvonen, Viik-Kajander et al. 2000). The high‐ est incidence rates are found in Finland and Sardinia (Italy) and the lowest in South American countries, e.g. Venezuela and Brazil, and Asian countries, e.g. China or Thailand (Karvonen, Viik-Kajander et al. 2000; Borchers, Uibo et al. 2010; Panamonta, Thamjaroen et al. 2011). Apart from regions with low to intermediate incidence rates ranging between 5 and 20 per 100,000 chil‐ dren or adolescents per year, there are areas with incidence rates as high as 27 to 43 per 100,000 children or adolescents per year. Canada and Northern European countries, such as Finland and Sweden, have the highest incidence rates ranging between 30 and 40 per 100,000 children/ adolescents per year. Incidence rates of countries in Central Europe (with the exception of Sardi‐ nia) vary from 8 to 18 per 100,000 children/adolescents per year. The incidence for type 1 diabetes in German children aged 0 to 14 years was estimated at 13 per 100,000 per year for 1987–1998 and at 15.5 per 100,000 per year for 1999–2003. The registry of the former German Democratic Repub‐ lic, which was kept from 1960 until 1989, reported incidence rates between 7 and 14 per 100,000 children/adolescents per year (Galler, Stange et al. 2010). In Mediterranean countries, the inci‐ dence rates of type 1 diabetes also show wide variations, although for some of them, there are still no relevant and reliable data (Muntoni 1999). Summarizing the data on type 1 diabetes incidence, the polar-equatorial gradient does not seem to be as strong as previously assumed. The incidence of type 1 diabetes among different countries is presented in Table 1 and Table 2. When comparing the incidence of type 1 diabetes between countries, it is important to keep the size of the sample and the area of sampling in mind. This is because the incidence of type 1 diabetes may show

over the last years, and its variability in populations and patient subgroups.

due to low rates of completeness.

**3.1. Geographic differences**

**3. The incidence of type 1 diabetes mellitus**

Data gained from self-reporting of diabetic patients have been shown to underestimate the true burden of diabetes (Forouhi, Merrick et al. 2006). Another possibility, but with similar limitations, is to assess data retrospectively (Mooney, Helms et al. 2004). Generally, longitu‐ dinal or cross-sectional studies are often locally or regionally performed. This limits the op‐ portunity to get generalizable results because the epidemiology of type 1 diabetes is known to be heterogeneous regarding geography and ethnicity. Cross-sectional studies do not pro‐ vide information on the time-dependent changes of the epidemiology. Additionally, many studies are limited to special settings, e.g. a general practice setting (Frese, Sandholzer et al. 2008), and although providing useful and necessary information, the reported data may not be representative for the epidemiology of type 1 diabetes.

Especially when estimating the incidence of type 1 diabetes, the latency of onset until diag‐ nosis is important and influences the quality of estimated data. Also the validity of the chos‐ en diagnosis should be critically reviewed. In a recent German investigation, 60 (10.3%) of 580 patients were reclassified at mean 2.4 years after the diagnosis of type 1 diabetes: 23 (38.3%) as type 1 diabetes; 9 (15%) as maturity onset diabetes of the young; 20 (33.3%) as "other specific diabetes forms", and 8 (13.3%) as "remission" of type 2 diabetes (Awa, Schob‐ er et al. 2011). The validity of the chosen diagnosis may differ depending on the data source that affords a correct differential diagnosis, e.g. between type 1 diabetes and malnutrition diabetes in developing countries or type 1 diabetes, type 2 diabetes and maturity onset dia‐ betes of the young in industrial countries, as well as a correct encoding of diagnosis. This is because usual classification systems such as the International Classification of Primary Care or International Classification of Diseases cannot be assumed to be sufficiently complete and valid (Gray, Orr et al. 2003; Wockenfuss, Frese et al. 2009; Frese, Herrmann et al. 2012).

It is conclusive that reliable and valid – and thereby comparable – data on type 1 diabetes epidemiology have to be based on a complete and detailed assessment. Disease registries can be assumed to be probably the best method to estimate and manage standardized data. However, the availability, completeness, quality and accuracy of diabetes registers are again very variable (Forouhi, Merrick et al. 2006). Type 1 diabetes registries were established on different levels: local (Howitt and Cheales 1993), regional (Galler, Stange et al. 2010), nation‐ al or multinational.

Much of our knowledge of the epidemiology of type 1 diabetes in young people has been generated by large collaborative efforts based on standardized registry data: the EURODIAB study in Europe and the DIAMOND project worldwide (Dabelea, Mayer-Davis et al. 2010). In order to provide reliable information about the incidence and geographical variation of type 1 diabetes throughout Europe, EURODIAB was established as a collaborative research project (Fuller 1989; Green, Gale et al. 1992). During a 15-year period, 1989 to 2003, 20 popu‐ lation-based EURODIAB registers in 17 countries registered 29,311 new cases of type 1 dia‐ betes in children before their 15th birthday (Patterson, Dahlquist et al. 2009). The World Health Organization program, Multinational Project for Childhood Diabetes (Diabetes Mon‐ diale or DIAMOND), has been developed to investigate and characterize global incidence and mortality of type 1 diabetes and the health care provided for type 1 diabetic patients. Both projects used similar ascertainment methodologies. However, DIAMOND ascertained some data retrospectively. This may have led to some underestimation of incidence rates. The completeness of case ascertainment varied from 35 to 100% in DIAMOND. Most Euro‐ pean nations in DIAMOND had comparable (> 90%) rates of ascertainment to EURODIAB (Vehik and Dabelea 2010). DIAMOND reached the lowest completeness rates in Africa, Cen‐ tral and South America. This reflects a general problem when assessing type 1 diabetes epi‐ demiology: data from developing countries are scarce and may not be fully representative due to low rates of completeness.
