BACKGROUND: Methods available to predict cardiovascular disease (CVD) and coronary heart disease (CHD) risk include the Joint British Societies Risk Chart (JBSRC), the CardioRisk Manager (CRM) calculator, the PROCAM calculation and specific to diabetes, the UKPDS risk engine. Our aim was to examine their efficacy in a clinic-based population of diabetic patients. DESIGN: Patients were identified who attended clinic at baseline (1990-1991) and categorised by the presence/absence of CHD/CVD at follow-up (2000-2001). Ten-year risk was calculated using JBSRC, CRM, PROCAM and the UKPDS risk engine. METHODS: A total of 798 patients were identified under follow-up (2000-2001), with sufficient data for risk prediction. Risk prediction methods were assessed by: (1) the Hosmer-Lemeshow test (calibration test); (2) the C-index, derived from the ROC curve [a discriminatory measure ranging from 0.5 (no discrimination) to 1.0 (perfect discrimination)]; and (3) Spearman correlation of the observed and predicted risk. RESULTS: All tests (except PROCAM) demonstrated acceptable discrimination with respect to CHD/CVD, however, all underestimated the risk of future events. With respect to CVD, the JBSRC had a C-index of 0.80, CRM: 0.76, UKPDS: 0.74 and PROCAM: 0.67. With respect to CHD the C-indexes were 0.77, 0.73, 0.65 and 0.76 respectively. Risk prediction by CRM had a stronger relationship with observed events than UKPDS and PROCAM (r=0.97, 0.86, 0.81 respectively). CONCLUSIONS: All scores have reasonable discrimination, but underestimate future events. The CRM showed the strongest correlation between observed and predicted risk with the least amount of scatter from the line of best fit. The CRM, when adjusted by the calibration factor, provides the most accurate method of risk prediction.
BACKGROUND: Methods available to predict cardiovascular disease (CVD) and coronary heart disease (CHD) risk include the Joint British Societies Risk Chart (JBSRC), the CardioRisk Manager (CRM) calculator, the PROCAM calculation and specific to diabetes, the UKPDS risk engine. Our aim was to examine their efficacy in a clinic-based population of diabeticpatients. DESIGN:Patients were identified who attended clinic at baseline (1990-1991) and categorised by the presence/absence of CHD/CVD at follow-up (2000-2001). Ten-year risk was calculated using JBSRC, CRM, PROCAM and the UKPDS risk engine. METHODS: A total of 798 patients were identified under follow-up (2000-2001), with sufficient data for risk prediction. Risk prediction methods were assessed by: (1) the Hosmer-Lemeshow test (calibration test); (2) the C-index, derived from the ROC curve [a discriminatory measure ranging from 0.5 (no discrimination) to 1.0 (perfect discrimination)]; and (3) Spearman correlation of the observed and predicted risk. RESULTS: All tests (except PROCAM) demonstrated acceptable discrimination with respect to CHD/CVD, however, all underestimated the risk of future events. With respect to CVD, the JBSRC had a C-index of 0.80, CRM: 0.76, UKPDS: 0.74 and PROCAM: 0.67. With respect to CHD the C-indexes were 0.77, 0.73, 0.65 and 0.76 respectively. Risk prediction by CRM had a stronger relationship with observed events than UKPDS and PROCAM (r=0.97, 0.86, 0.81 respectively). CONCLUSIONS: All scores have reasonable discrimination, but underestimate future events. The CRM showed the strongest correlation between observed and predicted risk with the least amount of scatter from the line of best fit. The CRM, when adjusted by the calibration factor, provides the most accurate method of risk prediction.
Authors: Benjamin J Gray; Richard M Bracken; Daniel Turner; Kerry Morgan; Michael Thomas; Sally P Williams; Meurig Williams; Sam Rice; Jeffrey W Stephens Journal: Br J Gen Pract Date: 2015-11-05 Impact factor: 5.386
Authors: Dele O Abegunde; Bakuti Shengelia; Anne Luyten; Alexandra Cameron; Francesca Celletti; Sania Nishtar; Vasu Pandurangi; Shanthi Mendis Journal: Bull World Health Organ Date: 2007-06 Impact factor: 9.408
Authors: Benjamin J Gray; Richard M Bracken; Daniel Turner; Kerry Morgan; Stephen D Mellalieu; Michael Thomas; Sally P Williams; Meurig Williams; Sam Rice; Jeffrey W Stephens Journal: Br J Gen Pract Date: 2014-10 Impact factor: 5.386
Authors: S van Dieren; L M Peelen; U Nöthlings; Y T van der Schouw; G E H M Rutten; A M W Spijkerman; D L van der A; D Sluik; H Boeing; K G M Moons; J W J Beulens Journal: Diabetologia Date: 2010-11-14 Impact factor: 10.122
Authors: V Shivakumar; A D Kandhare; A R Rajmane; M Adil; P Ghosh; L B Badgujar; M N Saraf; S L Bodhankar Journal: Indian J Pharm Sci Date: 2014-03 Impact factor: 0.975