The aims of this section were to describe the burden of diabetic complications, and to be able to examine the differences existing between countries and between ethnic groups. In order to do this with confidence, it is necessary that there is a degree of uniformity in the methodology of the different studies. This, however, was not the case.

For each of the complications, a wide range of results was found, but understanding the underlying causes of this diversity is difficult. For example, the low prevalence of neuropathy seen in Mauritius   1  could be due to a low inherent risk for neuropathy in that population, the availability of high quality diabetes care, the relative youth of a population in a developing country, the methods used in the study for defining neuropathy or the population-based study design. In the absence of similar studies, it is almost impossible to determine which of these explanations may be correct.

The problem of accurately describing the burden and making comparisons is made particularly difficult by the relative lack of population-based studies. Studies based on secondary care tend to over-represent those with advanced disease, as those requiring more intensive treatment are generally referred on for specialist care. Furthermore, prevalences in clinic populations will depend on local referral patterns, which are likely to vary widely around the world.

Neither the diagnostic tools nor diagnostic thresholds used to define a complication were equivalent across studies. For example, neuropathy was defined as the absence of ankle reflexes in one study   2 , and as the presence of symptoms and clinical signs (using validated scales) in another   3 . The inconsistent use of diagnostic tools to classify a complication has been shown to dramatically affect the prevalence reported. The Diabetes Control and Complications Trial (DCCT) compared the prevalence of neuropathy using 11 different criteria and found that within the one population, the prevalence of neuropathy varied from 0.3%, using sensory examination, reflexes and symptoms, through to 21.8% using nerve conduction tests   4 .

Another example of the variability brought about by changing methodology is highlighted by a study on amputation. The study compared the incidence of lower extremity amputations in one population, when primary or all amputations were included   5 . The incidence of lower extremity amputations varied from 276 per 100,000 person years (primary amputation) to 388 per 100,000 person years (all amputations). Further differences in amputation incidences are likely to be due to the (rather imprecise) method of estimating the total diabetic population from which those with amputations were drawn.

One large study, EURODIAB, examined the prevalence of retinopathy, neuropathy and nephropathy in type 1 diabetes across several European countries   6   7  . This study used standard methodology, making it possible to gain some insight into whether observed differences in prevalence estimates are due to methodological problems or represent actual differences. The study found the prevalence of complications did vary between countries. The prevalence of retinopathy ranged from 21% in Germany to 60% in Portugal. However, much less variation was seen for neuropathy and microalbuminuria, for which the prevalences clustered fairly tightly around 25% and 23%, respectively.

Any conclusions about the burden of disease attributable to diabetic complications must be very guarded, and comparisons between different parts of the world should be extremely cautious. Nevertheless, some tentative comments can be made:

• The prevalence of retinopathy is probably around 30% in type 2 diabetes.
• Of the seven population-based studies giving figures for neuropathy in type 2 diabetes, two of the three highest prevalences were from the USA.
• Populations from Europe had high rates of heart disease and stroke, while migrant Indian populations (Mauritius and Fiji) also had high rates of heart disease.
• No discernable patterns relating to geographic distribution or study design were apparent for nephropathy or amputations.

In summary, the interpretation of these studies of diabetic complications is severely hampered by the lack of population-based studies, and the wide variability in study design. Nevertheless, the data from EURODIAB would indicate that at least for some complications in type 1 diabetes, genuine differences exist between countries. What is absolutely clear from this review is that there are large parts of the world for which there are no useful data, and that there is a great need for population-based studies, using standardized protocols so that meaningful estimates of the prevalence of diabetic complications can be made.