Methodology for complications of diabetes

1. Literature review
A Medline search was undertaken for each complication: retinopathy, neuropathy, nephropathy, coronary heart disease, stroke and amputations. Six keywords were used: Diabetic Retinopathy/ep, Diabetes Mellitus/Co [Complications], Diabetic Nephropathies/ep [Epidemiology], Amputations/sn [statistical & Numerical Data], Diabetic Neuropathies/ep [Epidemiology], Diabetic Foot Complications/co [Complications]. For CHD and stroke a combination of key words were used: cardiovascular disease, coronary disease, diabetic angiopathies, myocardial infarction, myocardial ischemia, cerebrovascular disorders, cerebrovascular accident, prevalence and diabetes mellitus.

A final search, for countries without data, was performed using the keyword ‘Diabetes Mellitus’, along with each country name. The references of each paper selected for inclusion were reviewed and relevant articles from the reference lists were also reviewed. In addition, each IDF member country was contacted and asked to provide recent papers or unpublished reports on each complication.

2. Study selection
Studies with ≥100 participants were included. Where studies presented results for subgroups (e.g. type 1 or type 2 diabetes) separately, only results for subgroups with ≥100 participants were included. Where more than one study was available for a country, preference was given to larger and population-based studies, those published after 1989, and those with the fewest restrictions (e.g. on age). Where complete articles were not available or were not translated into English, details from abstracts were used.

Where possible the authors of the abstracts were contacted to verify and complete missing details. Studies were excluded in instances where details from the abstracts were insufficient and the authors could not be located. Unpublished studies which were the only available data for a country have been included.

Prevalences are reported for cardiovascular disease, nephropathy, neuropathy and retinopathy, while incidence and prevalence are reported for lower extremity amputations (with preference given to incidence studies). Where amputation studies presented data as a time trend, the most recent data have been included.

Where possible, the age ranges of the populations are reported. Where the age range of the population was not available, the mean or median age is reported. Where a publication only recorded the age range as groups in a Table, the lower and upper age groups are presented as the age range.

Diagnostic criteria for each complication are recorded, as variation in definitions can affect the prevalences reported. Diagnostic criteria for neuropathy were divided into five categories: clinical, quantitative sensory testing, electrophysiology, autonomic function tests and questionnaire to general practitioners. Where validated clinical scores were used, the name of the scale is recorded. Retinopathy was defined as the presence of at least microaneurysms or haemorrhages or exudates. The types of retinal examination are listed. Criteria for nephropathy are listed (type of measurement and diagnostic values used).

3. Study types
Five types of studies were included: population based, register, clinic (primary care), clinic (secondary care), and clinic (primary and secondary care). For amputation studies reporting incidence, the sources of the amputation data are presented. The sources used for these studies were registers, hospital records, operating theatre records, limb fitting centres, hospital discharge records and hospital discharge codes.

Amputations were recorded as first or all amputations. The lowest (most distal) level of amputation included in each study, and the total numbers of amputees (rather than amputations) are reported. In most of the incidence studies, the authors calculated amputation incidence using the recorded number of amputations and an estimated figure for the total diabetic population. In two studies, a cohort design was used in which a defined population was followed over time allowing a more accurate calculation of amputation incidence.

For some countries, results from more than one study are presented. This is usually because they cover different aspects of the diabetic population such as type 1, type 2, undiagnosed diabetes and previously diagnosed diabetes. Furthermore, since studies vary considerably in design, the presentation of two or more studies can help to build a broad picture of the prevalence of a complication. Studies of small minority groups from populations, e.g. Pima Indians, have not been included, as they do not reflect the general diabetic population of the country.

There are some important differences between the section on diabetic complications and the section on diabetes prevalence. The total numbers of individuals within a country who may have complications are not estimated, nor is a national prevalence. Furthermore, data have not been projected from one country onto other countries. There are two reasons for these differences: first, such calculations require knowledge of the age and sex structure of both the original study population, and of the target (national diabetic) population. In most cases, neither of these is known. Secondly, many studies are clinic based, and so their generalizability is limited.