Climates on incidence of childhood type 1 diabetes mellitus in 72 countries

We are aimed to systematically assess the worldwide trend in incidence of childhood type 1 diabetes mellitus (CT1DM) from 1965 to 2012 and to discuss whether climate affect incidence of CT1DM. We searched the relevant literatures in detail to judge the effect of different climates on incidence of CT1DM. The climates included Mediterranean, monsoon, oceanic, continental, savanna, and rainforest. According to different climates, we further researched relevant factor such as sunshine durations and latitudes. The overall incidence of CT1DM in 72 countries was 11.43 (95% CI 10.31–12.55) per 100,000 children/yr. The incidence of CT1DM in Oceanic climate [10.56 (8.69–12.42)] is highest compared with other climates; the incidence in 40°–66°34′N/S [14.71 (12.30–17.29)] is higher than other latitude groups; the incidence in sunshine durations with 3–4 hours per day [15.17 (11.14–19.20)] is highest compared with other two groups; the incidence of CT1DM from 2000 to 2012 [19.58 (14.55–24.60)] is higher than other periods; all p < 0.01. Incidence of CT1DM was increasing from 1965 to 2012, but incidence in Oceanic climate is higher than other climates. Furthermore, it is higher in centers with higher latitude and lower sunshine durations. The climates might play a key role in inducing CT1DM.

(2) the number of cases was three or more; (3) the study period was more than a year; (4) T1DM was diagnosed according to World Health Organization definition. Studies met the following criteria were excluded: duplication (the same articles in different database); case reports, and comments; the studies not meeting criteria of inclusion. Eligibility assessment was independently conducted by 2 authors, with all inconsistent questions solved by discussion with other authors.
Description of the data. Incidence data were extracted either from the text or from the tables in the publications. There was no incidence rate of the original articles were presented in the figures. Altogether 87 studies from 72 countries met the inclusion criteria and were finally included in this study (Table 1). In 78 studies the children aged from 0 to 14 years and in 9 studies from 0 to 12, 15, 17, 19 years. The time period of the researches ranged from 1 to 30 years. The degree of case-ascertainment ranged from 85 to 100%. The researches included in this study were from the period 1965 to 2012.
Quality assessment. All abstracts ascertained initial search were screened and the researches in violation of inclusion criteria were excluded by two authors. Full-texts were posteriorly accessed by another two authors, in case of disagreement, a third professor was invited to evaluate such studies and the consensus was achieved via discussion. If original data was missing, the corresponding author of this study was contacted with alone tailored application forms by email.
Climate Style, latitude, and sunshine durations. Mediterranean  Australia. Monsoon is currently defined as a seasonal changing in atmospheric precipitation and circulation associated with the asymmetric heating of land and sea. Oceanic climate is the typical of west coasts in higher middle latitudes of regions, with few extremes of temperature and a relatively narrow annual temperature range, and generally features cool summers and winters. Continental climate is referred to climates with significantly annual variation in temperature, which tended to occur in the middles of continents, mostly occur in the mainland China and the eastern U.S. 22 .
Statistical methods. The incidence of CT1DM for our study was obtained from the individual studies as it was researched in these publications. The incidence rates were calculated per 100, 000 people a year. Age standardization of the incidence rates was calculated using 5-years intervals with age groups 0-4 years, 5-9 years, and 10-14 years as the standard. The latitude groups 0°-23°26′N/S, 23°26′-40°N/S, and 40°-66°34′N/S as the study standard according to the tropic of Cancer/Capricorn, the Arctic/Antarctic circle, and westerlies, which based on geographic meteorology. Statistical analysis was performed using SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). Continuous data that accord with a normal distribution were presented as mean [95% confidence interval (CI)], with least significant difference in parameters between two groups were analyzed by t-test, and the one-way ANOVA was used to assess the multiple groups for continuous variables in normal distribution. A p < 0.05 is considered to be statistically significant difference.

Results
Description of the included studies. After initial screening and removal of duplicates, we reviewed 3602 articles in full, of which 87 eligible studies on the incidence of CT1DM in various countries were included in this study (Table 1) Incidence of CT1DM. The average incidence of CT1DM in 70 countries showed in Fig. 1.
Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Discussion
This study performed firstly systematic estimates of CT1DM incidence among various climates, regions, genders, age-groups, latitude, and sunshine durations. The total countries based on this research consisted of 32% of the all countries in the world. The worldwide incidence of CT1DM was increasing between 1965 and 2012 according to this study. Interestingly, the results of this study suggested there no significant difference was found in CT1DM incidence trended in boys compared with girls in this study. There are consistent results on the difference in incidence of diabetes by gender. Haynes et al. and Stipancic et al. 23,24 displayed a significant increase incidence of T1DM in both boys and girls, and no significant difference was found in boys versus girls. However, others found a higher incidence in girls [25][26][27] . Adverse to our findings, Casu et al. 28 expressed that a higher incidence of T1DM in boys. These divergences might depend on difference in sample size and statistical analysis.
In addition, our study illustrated a significantly increased incidence of T1DM in North America. Most countries in North America are developed countries with a higher per capita GDPs. Muntoni et al. 29 showed that countries with a higher per capita GDPs tended to have higher T1DM incidence. Populations in wealthier countries typically drank more milk or eat more cheese than in poorer countries 30,31 . A high frequency of intake of milk or foods rich in protein may induce the occurrence and development of diabetes in humans [10][11][12]32 . Furthermore, these foods and dinks contain higher proportion of carbohydrate. Studies manifested that dietary carbohydrate could exacerbate postprandial glucose responses, which may be play a key role in blood glucose control 33,34 . Therefore, the higher proportion of carbohydrate may be also a major factor in development of diabetes in these regions.
Furthermore, the incidence of CT1DM of Asia has been increasing in recent years, although lower compared with Europe and America. Especially, the result of this study indicated the CT1DM incidence was higher in inland regions with continental climates compared with monsoon climates in China. The study stated the incidence of Huhehot is about 11 times in Hainan 4 . The higher incidence existed in inland region with high latitude that plays an important role in reducing childhood insulin-dependent diabetes mellitus (IDDM) 35 . Recent years, Figure 2. Incidence of childhood type 1 diabetes mellitus in different regions, age-groups, climates, and sunshine durations. (A, Incidence of childhood type 1 diabetes mellitus in different regions: # indicated North America vs. other regions excluded Oceania, all p < 0.01; B, Incidence of childhood type 1 diabetes mellitus in three age-groups: # represented 0-4 years old vs. 10-14 years old, p < 0.01; C, Incidence of childhood type 1 diabetes mellitus in seven kinds of climates: *represented Monsoon climate vs. other climates excluded Savanna climate and Rainforest climate, all p < 0.01; D, Incidence of childhood type 1 diabetes mellitus in three sections of sunshine durations: # showed 3-4 hours/day vs. other two sections, both p < 0.01; all p derived from one-way ANOVA). the per capita milk consumption and protein intake are increasing, especially in Xinjiang 36 or Nnner Mongolia 37 . However, the overall incidence is low in China may based on individual's diet habit and environmental factors 38 , which may resulted in a lack of public awareness, so could lead to a low quality of life of children in China.
As well, the incidence of CT1DM in regions with higher latitude and lower sunshine durations was higher than low latitude with high sunshine durations. In this study, the average incidence of CT1DM in Finland was 38.11 from 1965 to 1999, in which, latitude was 60°10′ N and the average amount of sunshine durations was only 3.18 hours a day. Eurodiab ACE Study Group 39 had reported a 3-fold incidence increase of childhood IDDM was observed with the increasing latitude in Europe, and a similar result was reported within China 40 . In December, the northern Finland only has 2 hours of sunshine durations every day. Although there exists 23 hours of daylight per day in June, the most of the year exposure to daylight, Vitamin D production in the skin, is low by contrast with southern areas. Vitamin D supplementation is, thus, possibly more significant in this populations than others 41 . In this research, children lacked of adequate Vitamin D, who lived in higher latitude with low sunshine durations. Vitamin D is an immunosuppressive agent 42 , and the study believed the adequate Vitamin D supplementation for children might inhibit autoimmune reaction via damaging the β cells of pancreas and reduce the increasing trend in T1DM 41 . On the contrary, Vitamin D deficiency might induce CT1DM.
Last but not least, the incidence of CT1DM in centers or countries with oceanic climate was higher than other climates. The oceanic climate generally features long, but relatively mild winters and cool and short summers, which have a mean temperature below 22 °C in the warmest month 43 . In coastal areas of the higher middle latitudes (45-60° latitude), the prevailing onshore flow creates the basic structure of most oceanic climates. The previous studies reported the incidence rates of T1DM were associated with geographic variables such as average annual temperature 35 . Muntoni et al. 29 indicated that countries or centers with lower annual temperatures tended to induce high incidence rate of CT1DM.
Nevertheless, this study just researched the incidence of CT1DM in 0-14 years old. Incidence data in older age groups exist from a few individuals. Furthermore, the incidence of childhood is unavailable after 2012 in this study. As well, the incidence of gender missing from Table 1 revealed the populations where development of the new register strategy was desired. Therefore, the continuous community-based registries are needed to access the T1DM incidence in the world, and further research is needed to find out the primary factor to identify prevention measures to stop the increased incidence of CT1DM.   Table 3. The incidence of childhood type 1 diabetes mellitus (per 100,000 children/yr) with different gender in different climate. Data showed as Mean (95% CI); *represented boy vs. girl, all p > 0.05, p derived from the t-test.