Our study used population-based data retrieved from a disease register in a setting with remarkable rates of immigration. We observed a significantly higher prevalence for HMPC citizens than for Italians, and lower prevalence for citizens from HDC countries. The prevalence was particularly high for North Africans and Southern Asians. Our findings are consistent with similar studies conducted in Europe [2,4,7-11], in Canada [3,5], in the USA , and in Australia . The high prevalence of diabetes in specific ethnic migrant groups are likely attributable to a complex interplay of genetic and environmental factors, including acculturation, stress, social isolation, and employment and economic challenges [3,29]. In addition, people from Southern Asia seem to have a genetic susceptibility to Type 2 diabetes [30,31]. Furthermore, some authors have suggested that hypovitaminosis D, particularly relevant in migrant groups with darker skin pigmentation, increases the risk of diabetes in Africans [32,33].
While prevalence of diabetes is higher among men than women in Italians, HMPC women in our study had prevalence rates that were roughly equivalent to or higher than those of men from the same countries. Particularly high prevalence was evident among women from Northern Africa compared to men, and, with a smaller gap, among women from sub-Saharan Africa. This pattern, founded previously in Canada, Sweden, and France [2,3,34], suggests that gender plays a role as effect modifier in the relationship between ethnicity and diabetes in non-western countries. A higher prevalence of diabetes in women can be related to a higher rate of obesity and physical inactivity, especially high among women of non-western origin [35,36].
Women from Northern Africa and Southern Asia usually migrate to Italy with their husbands, who seek job opportunities. For example, in 2007, the percentage of residence permits for family reunion issued to Tunisian women was 83.2% and to Indian women 70.1%, while the general percentage was 48.4% . The host environment may reduce the need for physical activity to housework  and introduce a more caloric diet, thus contributing to the higher risk of overweight, obesity, and diabetes.
Italian diabetes guidelines recommend initial screening for diabetes at age 45 years , although an opportunistic screening at younger age is recommended for patients belonging to an ethnic group at high risk. Our study confirms the rationale of these recommendations and identifies the ethnic groups particularly at risk, as well as North African women and Southern Asians of both sexes.
Disease management and glycaemic control
Our study did not find any differences between Italians and immigrants in the access of diabetes clinics as a whole, although North Africans proved to use the clinics less than did Southern Asians, who used the clinics more than did Italians. This could be due in part to a general difficulty for immigrants in accessing to primary care, and particularly to the GP or family paediatrician . An Irish study conducted among asylum seekers  found that they were more likely to be referred to outpatient services for various medical conditions than were their Irish counterparts.
As regards recent measurement of HbA1c, HMPC were less likely to perform the annual test, the difference being greater for women. Detailed analysis by sub-regions revealed that among HMPC, only Southern Asians had a level of compliance to guidelines similar to that of Italians.
Concerning glycaemic control, HMPC fared worse than did Italians, with the difference once again being more marked for women. For this intermediate outcome, only immigrants from Europe no HDC experienced values similar to those of Italians. Surprisingly, Southern Asians had the worst outcomes in terms of glycaemic control, despite their being the ones who most frequently cared for by specialized diabetes clinics and who seemed to be the most compliant to guidelines for HbA1c testing. Some authors suggest that one important factor contributing to increased Type 2 diabetes in Asian Indians is excessive insulin resistance compared to Caucasians. This difference in the degree of insulin resistance may be explained by either an environmental or a genetic factor or by combination of both [27,28]. Our findings confirm the studies carried out in the UK [13,16,41,42], in Sweden , in the USA [15,17], and in Italy . Gray and Heisler did not find any significant difference in frequency of measurement, while Buja underlined less coverage in terms of HbA1c for HMPC compared with Italians. The other studies, focusing on glycaemic control, found worse values for non-white ethnic groups.
Thabit  argued in his study that poor levels of health literacy can negatively impact the patients’ ability to interpret blood glucose levels, understand educational materials, and read labels on medication, irrespective of the patient’s educational level when operating in the foreign language. The same author suggested that health knowledge and perceptions about managing diabetes may be conceptualized differently in ethnic minorities compared to the majority population.
A possible explanation may be that, in our setting, immigration is a relatively recent phenomenon and consequently the patients from HMPC on average have been into the care of diabetes clinics for less time compared to Italians. As a result, the higher level of HbA1c could reflect a higher proportion of newly diagnosed and newly treated patients. Nevertheless, it must be noted that all the differences in control of disease have been observed even when we adjusted for treatment.
Strengths and limitations
In our knowledge, this is the first study carried out in Southern Europe based on diabetes register that is able to determine both age-standardized prevalence and quality of care indicators for immigrants and for different geographical area of origin, using gender approach.
Nevertheless, as the information about diabetes onset is not completely consistent, we decided not include this variable in the analysis. As proxy of disease severity we used diabetes treatment regimen, distinguishing between subjects using diet only and those treated with antidiabetic drugs. In addition, we had no information about the socioeconomic status of diabetic patients or about their BMI values or other clinical characteristics.