Definition

The indicator Prevalence of known diabetes is defined as the proportion of people in the population who report they have ever been diagnosed with diabetes by a doctor, or who have a documented current antidiabetic medication.

The indicator Prevalence of unknown diabetes is defined as the proportion of people in the population who do not have known diabetes and who currently have an HbA1c value (long-term blood glucose level) of 6.5% or higher.

Operationalisation

The following information was taken into account for the assessment of known diabetes:

Self-report by the participants on a medical diagnosis of diabetes throughout their life collected during a computer-assisted medical interview in answer to either:

• GNHIES98: ‘Have you ever been diagnosed by a doctor with one of the following illnesses or disorders?: Diabetes with insulin treatment/Diabetes without insulin treatment.’

  • Yes
  • No
    

• DEGS1: ‘Have you ever been diagnosed by a doctor with diabetes?’

  • Yes
  • No
  • I do not know

  or

• Documented intake of antidiabetic medication (ATC code A10) using automated identification of medication taken in the last seven days

The following information was taken into account for the assessment of unknown diabetes:

• No medical diagnosis of diabetes and
• No use of antidiabetic medication and
• Measured long-term blood glucose level (HbA1c) of 6.5% or higher

Reference population

Resident population in Germany, aged 18 to 79 years 

Data source

Nationwide RKI interview and examination surveys 1997-1999 (GNHIES98) and 2008-2011 (DEGS1) based on a population registry sample and self-completed questionnaire, medical interview, automated medicine registration and examination.

Number of cases

• GNHIES98: n = 7,124
• DEGS1: n = 7,115 (of which n = 2,923 had also participated in GNHIES98)

Calculation

• Description: For the indicator, the figures for total, women and men are provided and are stratified by age group, residential area and education as far as the number of cases available for the figure is ≥ 5 and the statistical uncertainty in the estimate of the indicator is not considered too large (a coefficient variation ≤ 33.5%).
• Stratification: The geographical classification of the residence of the participating person was carried out by region (north-east, north-west, middle-east, middle-west and south). Educational status was determined using the CASMIN index, which takes information on both school and vocational training into account and allows a categorisation into a low, medium and high education group.
• Weighting: In order to correct for deviations from the underlying reference population due to different participation rates or sampling probabilities, weighting factors were used when calculating the indicator. These adjust the surveys to the population structure of the reference population with regard to sex, age, federal state, German citizenship (yes / no), community type and education as of 31 December 1997 (GNHIES98) and 31 December 2010 (DEGS1). In DEGS1, the different participation probability of re-participants from GNHIES98 was also taken into account in the weighting.
• Absolute values: Number of persons with known and unknown diabetes in the population aged 18 years or over (reference population), determined by extrapolating the number in the sample to the number in the reference population.
• Age standardization: Age standardization and trend weighting was carried out by calculating the weighting factor in GNHIES98 using the age, sex and federal state structure of the reference population as of 31 December 2010.

Data quality

RKI interview and examination surveys provide representative results for the 18- to 79-year-old resident population of Germany. The population aged 80 years and over will only be included in future survey waves. As is the case in all population-based studies, underrepresentation of the seriously ill and those living in institutions must be assumed. Although the HbA1c threshold used is a guideline-based diagnosis criterion for diabetes, as a single blood glucose parameter it underestimates the prevalence of unknown diabetes in population-based studies (Cowie et al., 2010).