The number of people with type 2 diabetes is increasing in every country of the world (International Diabetes Federation Diabetes Atlas, 2013).
To see the rise in diabetes (and obesity) around the globe since 1980, visit this online graphic and map from the Washington Post. In the U.S., the latest data show that the prevalence of type 2 diabetes increased by over 30% in youth between 2001 and 2009 (Dabelea et al. 2014).
In the U.S., 12-14% of adults have type 2 diabetes (in 2011-12). The numbers are higher for blacks, Hispanics, and Asians than for whites. The prevalence of diabetes in the U.S. has increased over the past few decades (it was 9.8% in 1988-1994) (Menke et al. 2015). In addition, nearly 1 in every 5 U.S. teenagers has abnormally high blood glucose levels (Menke et al. 2016).
Really. In Texas, perhaps the youngest person ever to develop type 2 diabetes was 3 1/2 years old. It was caught early and reversed with dietary changes and metformin (Yafi et al. 2015). Also in Texas, a 5 year old was also diagnosed with type 2 diabetes (Hutchins et al. 2016). While these cases are rare enough to merit publication as case studies, the trend is alarming!
There are approximately 500,000 children aged under 15 with type 1 diabetes in the world (Patterson et al. 2014); in 2013 alone, 79,000 more children developed type 1 (IDF Diabetes Atlas 2013). Worldwide, the incidence of type 1 diabetes increased, on average, 3% per year between 1960 to 1996 in children under age 15 (Onkamo et al. 1999). Between 1990 and 1999, incidence increased in most continents, with a rise of 5.3% in North America, 4% in Asia, and 3.2% in Europe. This trend is especially troubling in the youngest children; for every hundred thousand children under age 5, 4% more were diagnosed every year, on average, worldwide (Diamond Project Group 2006).
In the U.S., the latest data show that the prevalence of type 1 diabetes increased by 21% in children between 2001 and 2009 (Dabelea et al. 2014), and the incidence of type 1 diabetes in non-Hispanic whites increased by 2.7% per year between 2002 and 2009 (Lawrence et al. 2014). Those numbers are from the SEARCH for Diabetes in Youth study, which has study centers in 5 U.S. states. The CDC collects nation-wide data on diabetes, but does not differentiate between type 1 and type 2 diabetes. A different study of a large population of U.S. patients with commercial health insurance found that type 1 (and type 2) prevalence increased between 2002-2013 in children (Li et al. 2015). Researchers are figuring out ways to determine exactly how many children have type 1 (or type 2) diabetes in the U.S. using electronic health records (Zhong et al. 2016). I wish them luck, and wish we knew!
Essentially all researchers agree that changes of this magnitude cannot be explained by genetic changes alone. In fact, studies are finding that high risk susceptibility genes for type 1 diabetes are becoming less frequent over time in children, while more children with low to moderate risk genes are developing the disease more now than in years past (Fourlanos et al. 2008; Gillespie et al. 2004; Hermann et al. 2003; Resic-Lindehammer et al. 2008; Steck et al. 2011; Vehik et al. 2008). An interesting study from Poland analyzed susceptibility genes from exhumed skeletons from the Middle Ages, and found that genetic predisposition to type 1 diabetes is lower today than it was 700 years ago (Witas et al. 2010).
It is also clear that the trend varies by year and by location, implying environmental factors are critical. The rates of increase are not uniform within Europe or within countries, suggesting that different risk factors vary over time in different countries (Patterson et al. 2012).
I have been making a list of countries/regions that have documented increases in type 1 diabetes published in scientific journals. These include:
Algeria, Argentina, Australia, Austria, Belarus, Belgium, Bosnia and Herzegovinia, Brazil, Bulgaria, Canada, Chile, China, Colombia, Croatia, Cyprus, Czech Republic, Denmark, Dominican Republic, Egypt, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Japan, Jordan, Kuwait, Latvia, Libya, Lithuania, Luxembourg, Macedonia, Malta, Mexico, Netherlands, New Zealand, Norway, Peru, Poland, Portugal, Romania, Russia, Saudi Arabia, Singapore, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Taiwan, Thailand, Tunisia, Turkey, United Kingdom, United States of America, Uruguay, Uzbekistan, U.S. Virgin Islands. (see the link at the bottom of this page for citations).
Around World War 2, in developed countries, but more recently in later developing countries.
The incidence of type 1 diabetes has been rising in children since about the mid-20th century in many European and North American countries (Gale 2002). What has changed during this time period? A number of things changed that may influence the development of type 1 diabetes, including: breastfeeding rates, diet, height and weight, vitamin D levels, infectious disease, vaccines, earlier puberty, factors relating to gestation and birth, and more. A major change that has garnered less attention in studies of type 1 diabetes is environmental chemicals. Yet perhaps we should pay attention: the historical patterns of contamination are consistent with historical patterns of type 1 diabetes incidence. While correlation is not the same as causation, the pattern is certainly consistent.
The rise in type 1 diabetes incidence is coincidental with the large-scale production and use of many industrial and agricultural chemicals. Like the rising incidence of type 1 diabetes, large-scale chemical production also began around the middle of the 20th century. In 1975 about 60,000 chemicals were manufactured or processed in the U.S.; in 1997 there were over 75,000 (Endocrine Disruptor Screening and Testing Advisory Committee (1998) Final Report U.S. EPA). Chemical production increased during this time as well.
Maybe, depending on where you live, but it is not clear yet.
A study from Sweden provides some hope that the trend is leveling off: it found that incidence was increasing in children born through the year 2000, but after that the trend might be flattening. It is too early to say whether this is in fact the case, or just a blip in the data (Berhan et al. 2011). Data from Norway show that the rising incidence has essentially leveled off since 2004 (Skirvarhaug et al. 2014). Data from Sardinia, Italy, also show that the increase is starting to level off in more recent years (through 2009) (Bruno et al. 2013). Data from Ireland show that the rising trend has leveled off since around 2008 (Roche et al. 2016). And in the Netherlands, the incidence in children under 5 appears to have stabilized around 1996-99 (Spaans et al. 2015). In Poland, the rising incidence over the past 24 years shows a "slight" leveling off around 2010 (Chobot et al. 2017).
Data from Finland, however, show that type 1 incidence appears to be increasing even more rapidly since the mid-1990s than in earlier decades (Harjutsalo et al. 2008). In Denmark, type 1 diabetes incidence continues to rise (through 2014) at about 3% per year (Svensson et al. 2016). Incidence continues to rise in Kuwait as well (Shaltout et al. 2016).
And, data from the U.S. show that as of 2000-2004, incidence of type 1 (and type 2) diabetes in children was still increasing significantly, and has been since the 1980s (in both Philadelphia Lipman et al. 2013, and Colorado Vehik et al. 2007). (See chart above for other U.S. data as well). In U.S. adults, considering type 1 and 2 combined, the incidence doubled during 1990-2008, but leveled off between 2008 and 2012. However incidence continues to rise in blacks and Hispanics (Geiss et al. 2014).
It also appears that type 1 diabetes is progressing faster than before. A large study from the U.S. and Europe found that disease progression seems to have increased over the past 27 years in children newly diagnosed with type 1 diabetes (Max Andersen et al. 2014).
In China, meanwhile, the incidence of type 1 diabetes is skyrocketing(Fu et al. 2013). Between 1997-2011, type 1 diabetes incidence increased 14.2% per year (!) in Shanghai's children. At that rate, the incidence of type 1 will double in just four years (between 2016 and 2020), and prevalence will sextuple by 2025 (Zhao et al. 2014). In Zhejiang, China (just south of Shanghai), incidence increased 12% per year between 2007 and 2013. Even worse, incidence increased by over 33% per year in children under 5 years old! (Wu et al. 2015).
Type 1 diabetes incidence ranges from very low in South America and Asia, to very high in Europe, especially northern Europe (Onkamo et al. 1999). Finland, Sardinia (Italy), and Sweden have the highest incidence of type 1 diabetes in the world (Diamond Project Group 2006, Patterson et al. 2014, Tuomilehto 2013). In fact, the longer you live in Sweden, the higher your risk of type 1 diabetes-- offspring of immigrant women living in Sweden for 11 years or more have a 22% higher risk than offspring of women living in Sweden for 5 years or less (Hussen et al. 2015).
As an example of a country with low incidence, black children in Dar es Salaam in Tanzania have very low incidence of type 1 diabetes, at 1.5 people diagnosed per 100,000 people, which is much lower than rates for black children in the U.S., Virgin Islands, or Cuba. In Tanzania, only one child under age 5 was diagnosed during one 10 year study period (Swai et al. 1993). The Raikas, a tribal group in India, have a far higher genetic risk of type 1 diabetes than other North Indians, yet the incidence of type 1 is almost nil (Bhat et al. 2014)-- implying that genetics do not tell the whole story.
On average, in children under age 15, type 1 diabetes incidence increases as a child gets older. In other words, a person 10-14 years old has a higher risk of developing type 1 diabetes, someone 5-9 years old has a middle risk, and someone 0-4 years old has a lower risk. Someone 10-14 has about twice the risk of developing type 1 diabetes as someone under 5. This trend generally does not vary by gender (Diamond Project Group 2006).
Overall, and especially in Europe, however, the rates of increase, however, have been highest in children under age 5, with a 4% annual increase in this age group (Diamond Project Group 2006).
A European study from 1989-1994 found that the average annual rate of increase was 6.3% in 0-4 year olds, 3.1% in 5-9 year olds, and 2.4% in 10-14 year olds (EURODIAB ACE Study Group 2000). In the U.S., a Colorado study also found that the increase was highest in the youngest age group, 3.5% annually in 0-4 year olds in this study, which covered a 26 year period (Vehik et al. 2007).
In most areas, the highest rates of increase, then, are seen in the youngest children, which in itself is a matter of real concern.
Previously unheard of, children are now developing type 2 diabetes. Even children under 10 years of age are now developing type 2 diabetes (Pettitt et al. 2014).
Within the U.S., there are ethnic differences in diabetes prevalence. In children under age 20, type 1 diabetes is more common (prevalent) among non-Hispanic whites, followed by blacks, Hispanics, and Asian/Pacific Islanders, and lowest in American Indians. Among these same children, type 2 diabetes, on the other hand, is more common among American Indians, followed by blacks, Asian/Pacific Islanders, and Hispanics, and lowest in non-Hispanic whites (Liese et al. 2006, Pettitt et al. 2014). In the U.S., type 1 diabetes is still more common in children than type 2 diabetes, with the exception of American Indian youth age 15-19, where type 2 is more prevalent than type 1 (Pettitt et al. 2014).
In China, type 2 diabetes in children has doubled in prevalence in the past 5 years, and now even surpasses the prevalence of type 1 diabetes in children (Fu and Prasad 2014).
Most studies of type 1 diabetes only consider children under the age of 15. A recent study that looked at all new cases of diabetes diagnosed during a 3 year period in an area of Sweden, and actually tested people of all ages for antibodies, found that type 1 incidence peaked during ages 0-9 and then again at ages 50-80, showing that disease onset is not limited to children. In fact, nearly 60% of new type 1 cases were diagnosed in people over age 40 (Thunander 2008). A review of studies from around the world of type 1 diabetes in people over 15 years of age found that geographical variations mirrored that of type 1 diabetes in children, that more adult males were diagnosed than adult females, and that overall incidence tends to decrease after age 14 (Diaz-Valencia et al. 2015).
What about changes over time? As discussed above, incidence of type 1 diabetes among children is increasing over much of the world. Yet one study from Belgium makes an interesting point. That study was one of the few that included people over 14 years old, and the researchers found that even though there had been an increase in type 1 diabetes incidence for 0-14 year olds, there was no overall increase in incidence over the 12 year period, because fewer people over 14 were being diagnosed. In 0-4 year olds, the annual increase in incidence was 5%. This finding implies that at least in Belgium, the increasing incidence in type 1 in children may be due to an acceleration of the disease process, but not an overall increase in incidence (Weets et al. 2002). A follow-up study found that the rising incidence in Belgian children was largely due to an increase in incidence in boys under 10, but not girls (Weets et al. 2007).
Another study shows a similar finding: that during the period of 1983-1998 in Sweden, the incidence of type 1 diabetes did not show an overall increase in the 0-34 year age group, but instead, the average age at diagnosis decreased. A shift to younger age at diagnosis seems to explain the increasing incidence of type 1 diabetes in Sweden during this time period (Pundziute-Lyckå et al. 2002). More recent data from Sweden, covering the period from 1983-2007, shows that incidence was higher in children under age 15, which peaked and then declined, while there were decreases in the older age groups (25-34 years), suggesting that there was a shift to a younger age of onset, instead of a uniform rate of increase among all age groups (although the overall incidence is still increasing) (Dahlquist et al. 2011). However, a new look at the Swedish data shows that these previous studies may have underestimated the number of people with type 1 diabetes in the 0-34 age range-- the actual number is 2-3 times higher than previously thought (Rawshani et al. 2014).
Long-term data from Norway from the 1930s to the 1970s show not only that the age of diagnosis decreased, but also that overall incidence increased, in people up to age 30. In the 1930s, diabetes was more common in people aged 15-29 than in people age 0-14, and by the 1970s it was more common in the younger age group instead (Gale 2002; Gale 2005).
Until there are more studies that include older age groups, it will be difficult to say how much of the increasing incidence in children is actually due to a decreasing age of diagnosis, and whether or not incidence is also increasing in adults. In the meantime, we can say that type 1 diabetes incidence is increasing in children, especially the youngest children, in countries around the world.
Unlike many other autoimmune diseases, where females are more at risk of disease, boys and girls under age 15 are diagnosed with type 1 diabetes at relatively equal rates. Some populations with a high incidence tend to have more males than females with type 1, while some with low incidence show more females than males, although this varies among studies. In people of European descent diagnosed at ages 15-40, however, there is a clear male predominance: more men than women are diagnosed with type 1 diabetes at these older ages (Soltesz et al. 2007).
Worldwide, incidence trends generally do not differ between genders-- the incidence of type 1 diabetes is rising in children of both genders (Diamond Project Group 2006).
Many of the countries with high incidence are located closer to the polar areas of the globe, both to the north and the south (Soltesz et al. 2007). Even within countries, latitude can make a difference: one Australian study, for example, found that type 1 diabetes was three times more common (prevalent) in more southerly regions of that country than in northerly regions (Staples et al. 2003).
Like all rules, however, there are exceptions. For example, Sardinia, Italy's high incidence of type 1 diabetes does not fit the rule. Variations within countries also do not always correspond to latitude (Soltesz et al. 2007).
Vitamin D, which is produced by the skin when exposed to sunlight, is a possible explanation for this pattern. In a study of 51 regions around the world, Mohr et al. (2008) found that areas with lower levels of ultraviolet B radiation (the main source of vitamin D in humans) had a higher incidence of type 1 diabetes. Vitamin D deficiency appears to be a risk factor for type 1 diabetes, and vitamin D cannot be produced adequately by the skin during the winter in areas closer to the polar regions. In Sweden, a study has found that temperature is more important than sunshine in explaining the higher incidence in the northern parts of that country. Cold weather may increase insulin resistance and exacerbate the disease process (Waernbaum and Dahlquist, 2015). Another possibility is that persistent organic pollutants (POPs) play a role. POPs evaporate and migrate to the polar regions of the earth; some can even interfere with vitamin D synthesis (see the vitamin D page).
Most countries with high incidence are Westernized, developed countries (e.g., see Diamond Project Group 2006), and even within Europe, incidence is correlated to gross national product (GNP) and other indicators of national prosperity (Patterson et al. 2001). Some studies have found higher incidence of type 1 in wealthier areas or in people with higher socioeconomic status within countries as well, such as Chile (Torres-Aviles et al. 2010) and the U.S. (D'Angeli et al. 2010).
Why would wealth make a difference in type 1 diabetes incidence? Differences in nutrition or lifestyle may play a role. These factors could include high growth rates in early life (see the height and weight page), improved hygiene and fewer infections (see the viruses page), or more milk consumption (see the wheat and dairy page) (Patterson et al. 2001).
Another possibility seldom considered is environmental chemicals. A number of toxic chemicals are found in plastics, personal care products, and other conveniences of modern life (see, for example, the pages on bisphenol A and phthalates). In developed countries, exposure to bisphenol A is significant and continuous (Welshons et al. 2006). Historically, chemicals such as PCBs were produced in industrialized countries, and now everyone living in developed countries has PCBs in their bodies (Carpenter 2006). While data from less industrialized countries are scarce, in general, people living in more industrialized countries show higher levels of PCBs and dioxins in their bodies than people in less developed countries. In addition, using a measurement of the total toxicity of multiple persistent organic pollutants, levels are higher in people living in industrialized as compared to less developed countries (Sudaryanto et al. 2005; Tanabe and Kunisue 2007). Recently, however, toxic chemicals, pesticides, and chemical wastes, which previously were found only in developed countries, are now used in low and middle income countries, leading to an increased risk of the diseases associated with these chemicals (Suk et al. 2016).
Type 1 diabetes incidence is now rising even in countries with historically low incidence, suggesting a catch-up phenomenon. High incidence countries (e.g., Norway), where incidence rose many decades ago, do not all show a continuing increase in incidence. It is too early to say whether these high-incidence countries have reached a plateau (Gale 2002). Levels of most persistent organic pollutants have declined recently in developed countries. In developing and some former Soviet countries, however, some persistent organic pollutants (like DDT) are still in use, and contamination due to open dumping is also a concern. Levels of some organochlorine pesticides (such as DDT) are now higher in people living in developing countries than in developed nations (Tanabe and Kunisue 2007). Perhaps contamination resulting from industrialization contributed to the rising incidence in many now-developed, high-incidence countries, and other countries, where contamination began later, are now "catching up."
Central and Eastern European countries are one region where type 1 diabetes incidence may be "catching up" to the higher incidence found in Western Europe. Over the period of 1989-1998, Central and Eastern European countries showed the highest rates of increasing type 1 diabetes incidence within Europe. Why? Some environmental factors associated with societal development may play a role (Green et al. 2001). Yet because type 1 diabetes takes a long time to develop, the factors responsible for these rapid increases may have operated before the political changes in those countries (EURODIAB ACE Study Group 2000). Environmental contamination is a serious problem in Central and Eastern Europe, beginning during the communist years, before the political changes (Fitzgerald et al. 1998).
Within countries, socioeconomic status is also associated with differential chemical exposures. In the U.S., people of higher income levels tend to have higher exposures to mercury, arsenic, perfluorinated compounds, one type of phthalate, and benzophenone-3 (found in sunscreen). People with lower income levels had higher levels of lead and cadmium, bisphenol A and three phthalates (Tyrrell et al. 2013).
Clustering has been found in type 1 diabetes, meaning that some groups of people living near each other during the same time periods show a higher than expected incidence of disease. Some studies have found clusters of children with higher than expected incidence of type 1 diabetes, and a study from the U.K. found clustering in teenagers as well (McNally et al. 2006).
The Massachusetts Department of Public Health has conducted a cluster investigation on type 1 diabetes in families living in Newton, Wellesley, and Weston, MA. The results were released in Feb. 2012, and are available at this Mass. Dept. Public Health webpage. A cluster of type 1 diabetes was found, but only in certain portions of Wellesley and Weston. MDPH is now analyzing various environmental factors, including contaminated sites, chemical spills, and pesticide use (on right-of-ways), that potentially could have contributed to this cluster.
Genetic variations likely explain some of the differing incidence and prevalence rates among people worldwide. Yet even among ethnically similar populations, type 1 diabetes incidence can vary. For example, Finns have a six times higher incidence in type 1 diabetes than Russians living across the border. The genes that confer a high risk of type 1 diabetes, however, are the same in these populations, implying that environmental factors contribute to the differing incidence rates (Kondrashova et al. 2005) Similarly, incidence in northern European countries is higher than in Lithuania and other Baltic states. Yet the genetic risk of type 1 is similar in all of these countries (Skrodenienė et al. 2010).
There is no consistent pattern of type 1 diabetes being more common in either rural or urban areas. Some studies have found higher incidence in rural areas, and some in more densely populated areas. For example, a study from Western Australia found type 1 diabetes incidence to be higher in more urban areas as compared to more rural or remote areas (Haynes et al. 2006). A study from New Zealand found type 1 incidence higher in "satellite urban communities" (Miller et al. 2011). A study from Northern Ireland has found higher incidence of type 1 diabetes in more remote areas (Cardwell et al. 2006), and in Egypt in more rural areas (El-Ziny et al. 2014). There are no clear trends in any of these directions. It may depend on the various exposures encountered in each of these areas. For example, the authors of the Egyptian study hypothesize that the higher incidence in rural areas may be due to higher exposures to pesticides.
A seasonal pattern in type 1 diabetes diagnosis has been seen in some countries, with more people diagnosed during the winter months. The pattern is most apparent in countries with greater differences in summer vs. winter temperatures (Soltesz et al. 2007). A European-wide study found that most children were diagnosed between November and February in all age groups and both sexes, although the youngest children had the least seasonal variation in diagnosis timing (Patterson et al. 2014). A Belgian study also found that the seasonal variations were strongest in the older children and adults, especially males with a certain genetic background, and less so in younger children (under 10) and females (Weets et al. 2004). A study of all Dutch children diagnosed from 2009-2011 found higher rates of type 1 diagnosis during the fall and winter in both boys and girls (Spaans et al. 2016). A "marked" seasonal variation in diagnosis has been confirmed in data from throughout Ireland (Roche et al. 2016). The seasonal variation in diagnosis has even been documented in dogs, with peak diagnosis in winter (Atkins and MacDonald, 1987).
When people are diagnosed with type 1 diabetes, they often have some residual beta cell function, and produce a tiny amount of insulin. Studies from Sweden and Belgium found that there is seasonal variation not only in type 1 diagnosis, but also in residual beta cell function at diagnosis (Samuelsson et al. 2013; Weets et al. 2006).
People with type 1 tend to be born more often during certain months-- for example, in the U.S., children born in the spring had a higher risk of type 1, especially in northern vs southern areas (Kahn et al. 2009). A seasonal pattern has also been identified in the appearance of the first autoantibodies in Finland, highest in the fall and winter, and lowest in spring and summer (Kimpimäki et al. 2001). There is also variation year to year in the timing and height of these antibody peaks (Knip et al. 2005).
These seasonal variations are often attributed to viruses, cold weather, or vitamin D levels. In fact, one study from Denmark found that the association between type 1 diabetes and season of birth disappeared (in males) during the years when margarine was fortified with vitamin D (Jacobsen et al. 2015). A number of environmental chemical exposures can also vary seasonally, such as air pollutant levels (Hathout et al. 2002), nitrate levels in drinking water (Parslow et al. 1997), and agricultural pesticide use. There may be other explanations as well (e.g., children may get less exercise in cold climates in the winter, leading to increased insulin resistance, or their parents eat Icelandic smoked mutton at Christmas... but you'll have to read about that story on the nitrate and nitrite page).
Interestingly, seasonal variations have also been found in other types of diabetes. For example, in gestational diabetes incidence/prevalence (see below), and in season of birth for type 2 diabetes (Si et al. 2017).
In Australia, researchers have found that there is a regular 5 year pattern of type 1 diabetes incidence in children. That is, every 5 years there is a peak or trough in the overall incidence rates. Why this is I have no idea (Haynes et al. 2015; Haynes et al. 2012). A Polish study also found a 5 year fluctuation pattern (Chobot et al. 2017). Huh.
The incidence of gestational diabetes also appears to be increasing, in many countries around the world. For example, throughout the U.S., the prevalence of gestational diabetes increased dramatically between 1989 and 2004 (Getahun et al. 2008). A different study also finds an increase in gestational diabetes prevalence in the U.S., between 1979 and 2010 (Lavery et al. 2016). In Korea, the incidence increased dramatically between 2006 and 2010 (Cho et al. 2015). In Canada, the incidence of gestational diabetes has doubled over the past 14 years (Feig et al. 2014). In Pennsylvania, gestational diabetes incidence rose between 1999 and 2008 (Khalifeh et al. 2014).
Gestational diabetes also shows seasonal variations. In Australia, the prevalence and incidence of gestational diabetes can vary by season, although one study found it peaked in the summer (Moses et al. 2016), and one in the winter (Verburg et al. 2016). In Sweden, peak gestational diabetes incidence is in the summer (Katsarou et al. 2016).
To download or see a list of all the references cited on this page (and many additional references, for example from all of the countries/regions listed above), see the collection Diabetes incidence and prevalence in PubMed. The collection includes over 700 studies from around the world-- almost all of which have found an increasing trend.
And, see the collection Clusters of type 1 diabetes in PubMed for studies on seasonal variations and geographical diabetes clusters.