The main findings of this nationwide longitudinal study include the following: (1) For Israeli adolescent males of Ethiopian origin with normal or high BMI, the risk for incident early-onset diabetes is 3-4-fold higher than for native Israeli males. This difference is unlikely to be accounted by differential weight gain in adulthood and was not observed among immigrants to Israel from the former USSR countries, who arrived to Israel at the same period as immigration from Ethiopia. (2) Diabetes risk is accentuated in second-generation Israeli-born men of Ethiopian origin. Point estimates were stable following adjustment to sociodemographic background and extensive sensitivity analyses.
A solid body of evidence indicates an association of black ancestry with increased diabetes risk compared to white Caucasians. Examples are the disproportionate increased risk for early-onset diabetes among persons of African origin in the UK,(15,16) Canada,(17) and the US.(18) HRs for developing type 2 diabetes for middle-aged African Americans (2,19–23) and Ethiopian immigrants to Israel,(24,25) compared to native populations, were reported in the range of 1.3 to 2.9. Among young adults and adolescents, point estimates ranged between 2.1 (23) and 6.2.(26,27) Methodological differences relating to the nature of population selection, the level of adjustment to confounders, and the wide age range at study entry may account for the differences observed. Additionally, in several US studies, African Americans were aggregated into a homogenous group despite diversity in ethnic and cultural background.(7) In this respect, our study is population-based, with a narrow age range at study entry, and a homogenous ethnic origin. In addition, a series of sociodemographic variables were included that were systematically obtained without a selection bias, given the mandatory nature of the medical screening.
Repeated evidence supports an association between genetic polymorphism and diabetes risk among persons of African ancestry. The thrifty gene hypothesis (28) suggests that certain populations have alleles that are adaptive to the feast and famine cycles of paleolithic human existence. These may promote rapid weight gain that may be advantageous during times of limited food availability but may be metabolically deleterious at times of continuing affluence. In agreement, several examples of novel and African ancestry-specific disease loci have been discovered in association with insulin resistance and inflammation. For example, polymorphisms in tumor necrosis factor-alpha (TNF-α),(29) or the anti-inflammatory adipokine, adiponectin,(30) were more prevalent among persons of African ancestry and were linked to an increased risk for type 2 diabetes. Notably, young African Americans, who largely represent West African ancestry, have been shown to have an increased risk to develop diabetes regardless of BMI,(23) similar to Israelis of Ethiopian origin. This suggests that the increased genetic predisposition for type 2 diabetes risk is largely driven by Pan-African ancestry. This contrasts with the genetic predisposition for end-stage renal disease that affects people of West African, but not Ethiopian origin.(31)
We report higher risk for type 2 diabetes among second-generation Israeli-born Ethiopian men than among those who immigrated; this association correlated with the time interval from immigration. Our findings corroborate previous evidence that diabetes risk correlates with the time since immigration among immigrants from various ethnic backgrounds to Western countries,(32,33) and is higher among second- than first-generation immigrants by 2- to 6-fold.(32,34) Similarly, African immigrants to the US were shown to be more likely to develop pre-diabetes or diabetes than African Americans, despite their having a higher level of education and lower BMI.(4) Notably, no association was found between diabetes incidence and immigration status regarding immigration from the former USSR, which occurred in parallel to the Ethiopian immigration. Additionally, previous studies underscored lifestyle changes among Ethiopian immigrants to Israel, who changed their diet to a more Western one,(35) and became less physically active and with more obesity (13) in a manner that correlated with the time since immigration.(36)
Our findings have implications to other high-income countries that are recipients of immigrants from Ethiopia. In the US, evidence supports the need to disaggregate the African American category into its ethnic subcommunities based on differences in genetic admixture, culture, and metabolic risk.(4,5,37) Of note, as of 2018, the US population included over 730,000 immigrants from East Africa. Nearly half of them were from Ethiopia, and they are considered to be the fastest-growing sub-African community in the US. This trend is more evident in some European countries such as Germany, where African immigrants constitute nearly 1% of the population.(38) The observation that the prevalence of high BMI among Israeli-born Ethiopian male adolescents has tripled over the last decade and reached nearly 20%(13) echoes with US data,(39) and predicts an increased burden of type 2 diabetes in young adults from this population. The lower prevalence of type 2 diabetes in Ethiopia, by over 3-fold, compared to Israel, the US, and most other Western countries, further emphasizes the contribution of immigration to diabetes risk at young ages in high-income countries.(1) Importantly, undiagnosed diabetes among young adults may be over 10-fold more prevalent among blacks than whites, as exemplified in the US population.(27) Collectively, these data emphasize the importance of clinical awareness, education for healthy lifestyle, and reduction of other preventable risk factors, especially in regard to adolescent obesity, for Ethiopian and possibly other East African immigrants.
This study has several limitations. First, only a single BMI measurement, at study entry, was available; and early longitudinal measurements throughout childhood and later during adulthood were unavailable. Nevertheless, BMI measurements among those who developed type 2 diabetes support that differential excess weight gain among Israelis of Ethiopian origin could not explain the results; this concurs with previous reports.(24) Additionally, the time interval from medical assessment at late adolescence to the mean age of diabetes onset was approximately 10 years, which is a reasonable time frame for initiating intervention strategies. Second, we lacked data regarding pre-diabetes and undiagnosed diabetes, which may be especially prominent in a young population. Third, information on lifestyle measures including dietary intake, physical activity, and smoking were not available to us. Fourth, we lacked baseline information regarding plasma glucose, cholesterol, and other measures of adiposity such as waist circumference. Of note, BMI is considered the preferable method for screening adolescents according to the U.S. Preventive Task Force,(40) and our dataset was based on a systematic medical screening process that enabled mitigating confounding by coexisting morbidities. Finally, we are underpowered to discuss the association between BMI and type 2 diabetes among young women with high BMI due to the small number of incident cases. However, findings in women with normal BMI resembled those in men. The strengths of this study include its systematic data collection at a narrow age range, the absence of population selection bias, and linkage between two nationwide databases and outcome cases of type 2 diabetes that were predominantly diagnosed before age 30 years. The ability to compare the findings with those of immigrants from other regions to Israel is advantageous.