Article Text

PDF

Diabetes plus
Prediction scores to identify older adults at high risk for type 2 diabetes
  1. S. G. Wannamethee1,
  2. O. Papacosta1,
  3. P. Whincup2,
  4. C. Carson3,
  5. M. C. Thomas1,
  6. D. A. Lawlor4,
  7. S. Ebrahim3,
  8. N. Sattar5
  1. 1
    Research Department of Primary Care and Population Health, University College London, London, UK
  2. 2
    Department of Community Health Sciences, St George’s, University of London, London, UK
  3. 3
    Non-Communicable Disease Epidemiology Unit, Department of Epidemiology and Population Health, LSHTM, London, UK
  4. 4
    MRC Centre for Causal Analyses in Translational Research, Department of Social Medicine, University of Bristol, Bristol, UK
  5. 5
    Division of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UK

    Statistics from Altmetric.com

    Background and Aim

    The greatest relative increase in type 2 diabetes (T2DM) prevalence in developed countries thought likely to occur over the next 25 years will be in the over 65 age group. The aim of this study was to examine the effectiveness of both simple strategies based on simple clinical parameters (suitable for use in primary care) and more complex scores involving blood markers to identify older individuals at high risk of developing T2DM.

    Methods

    A prospective study of non-diabetic men (n = 3490) and women (n = 3392) aged 60–79 years followed up for a mean period of 7 years, during which there were 298 incident cases of T2DM. Logistic regression was used to develop prediction scores to predict incident cases starting with non laboratory predictors and adding blood markers that predicted the incidence of T2DM. Receiving operating characteristics (ROC) analyses were used to assess improvement in prediction.

    Results

    The area under the curve (AUC) for a non laboratory score which included age, sex, family history of diabetes, smoking status, BMI, waist circumference, hypertension, and recall of doctor diagnosis of CHD was 0.765; sensitivity and specificity in the top quintile of the score was 50.3% and 81.4% respectively. Addition of simple blood markers HDL-C, triglyceride and glucose improved prediction significantly (AUC = 0.817 p<0.0001; sensitivity 63.8%; specificity 82.0%). Addition of gamma-glutamyltransferase increased sensitivity and specificity further to 65.8% and 82.1% respectively. Further addition of CRP made no improvement. Of those who were classified as low risk (defined as those who fell into the bottom 60% based on the non-laboratory score), the majority (88%) remained there even when routine blood markers were used and only 3% would be reclassified as high risk (defined as the cut off for the top quintile of the non-laboratory score) on the basis of blood markers. However, appreciable proportions of those in the top 40% of the non laboratory score were reclassified on the basis of blood markers into higher or lower risk categories (59%).

    Conclusion

    In large population settings and for cost effectiveness, simple non-laboratory measurements could be used in the first instance to identify a subgroup of older adults who could benefit from further testing with routine blood markers to identify those at highest risk for intervention.

    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.