This site is intended for health professionals only

At the heart of general practice since 1960

QDScore: a risk prediction tool for diabetes

The prevalence of type 2 diabetes is increasing mirroring the rising tide of obesity.

If left undetected significant complications can result, so early identification of those at risk is paramount. Often GPs will know which patients are at highest risk but a formal risk score would identify those less obvious cases and enable early intervention.

Cardiovascular risk score algorithms such as Framingham and QRISK are familiar to most GPs. They are a useful way of assisting clinical decision making and ensuring that appropriate interventions are matched to appropriate levels of risk.

The QDScore has been developed by the QRISK group as a novel algorithm to estimate the 10-year risk of developing diabetes. It uses readily available information and does not require any laboratory tests. Once again, the QResearch database is used which is derived from the EMIS computer records of more than 11 million patients from 551 practices in England and Wales.

The study followed a prospective open cohort design with patients enrolled between 1993 and 2008. Patients were aged 25-79 years and were all initially free from diabetes. This group was further subdivided into a derivation and a validation cohort. The derivation cohort included 2,540,753 patients from 355 practices of whom 78,081 developed type 2 diabetes. The validation cohort included 1,232,832 patients from 176 practices of whom 37,535 developed type 2 diabetes.

The predictive effect of risk factors for diabetes was analysed in the derivation cohort and used to derive a risk algorithm (QDScore). The following risk factors were studied: patient-assigned ethnicity, age, sex, BMI, smoking status, family history of diabetes, Townsend deprivation score, treated hypertension, presence of cardiovascular disease and current use of corticosteroids. The primary outcome measure was the onset of diabetes recorded in the general practice records although there was no confirmation of the accuracy of this diagnosis.

The results demonstrated marked variation in the risk of type 2 diabetes between different ethnic groups when compared with a white reference group. At highest risk were Bangladeshi men and women with hazard ratios of 4.53 (95% CI 3.67- 5.59) and 4.07 (95% CI 3.24-5.11) respectively. The hazard ratio was 2.54 (95% CI 2.20-2.93) for Pakistani men and 2.15 (95% CI 1.84-2.52) for women. Bangladeshi and Pakistani men had significantly higher hazard ratios than Indian men. Other ethnic groups including Black African men and Chinese women also had an increased risk of diabetes.

There was a marked difference in the risk of type 2 diabetes according to social deprivation with women in the most deprived fifth more than twice as likely to develop diabetes compared with the most affluent fifth. This finding was also seen in men although it was less marked. As would be expected age, obesity, smoking, family history, hypertension, cardiovascular disease and corticosteroid use all increased the risk of developing diabetes.

The team tested the performance of the QDScore by comparing the predicted and observed risk at 10 years in the validation cohort. This showed a 97% correlation.

The QDScore can reliably estimate the 10-year risk of type 2 diabetes using readily obtainable data and without resorting to laboratory or other tests. It also allows for both social deprivation and ethnicity. The algorithm can be incorporated into GP computer systems for easy use in routine practice. It would also lend itself to national screening programmes and can be directly accessed by the public ( However, there was no cut-off point established above which patients would be classified as being at high risk.

Any risk score should be used to complement, and not replace, clinical judgement. This score has been derived and tested on a diverse UK primary care population, which should ensure it is relevant to our daily practice. I am sure that further follow-up studies will emerge to assess the success of the QDScore.


Dr Peter Savill
GPwSI Cardiology, Southampton

Rate this article 

Click to rate

  • 1 star out of 5
  • 2 stars out of 5
  • 3 stars out of 5
  • 4 stars out of 5
  • 5 stars out of 5

0 out of 5 stars

Have your say