Reference | Number of studies | Validation study | Population | N events/total n | Outcome | Predicted years | Discrimination (AUC) | Calibration (p value Hosmer–Lemeshow) |
Models for diabetes population | ||||||||
Kengne 2011 (ADVANCE)24 | 124 | Kengne 201124 | NIDDM with microalbuminuria/proteinuria from 16 countries | 183/1836 | CVD | 4 | 0.69 | 0.03 |
Davis 2010 (Fremantle)25 | 125 | Davis 201025 | NIDDM from Australia | 24/180 | CVD or death | 5 | 0.84 | 0.85 |
Elley 2010 (DCS)26 | 126 | Elley 201026 | NIDDM from New Zealand | NR/12 598 | CVD | 5 | 0.69 | Good |
Donnan 2006 (DARTS)31 | 131 | Donnan 200631 | NIDDM from UK | NR/NR | CHD | 5 | 0.69 | 0.54 |
Kothari 2002 (UKPDS risk engine)22 | 267 68 | Kengne 200967 | NIDDM from 20 countries | 288/7502 | CVA | 4 years | 0.61 | <0.001 |
Davis 200968 | NIDDM from Australia | 23/791 | Stroke | 5 | 0.86 | >0.05 | ||
Stevens 2001 (UKPDS risk engine)23 | 97 9 67–73 | Kengne 200967 * | NIDDM from 20 countries | 407/7502 | CHD | 4 | 0.66 | <0.001 |
Models for general population | ||||||||
McGorrian 2011 (IHMRS)36 | 136 | McGorrian 201136 | IFG, IGT or NIDDM from 21 countries | 95/18 838 | MI | 3.25 | 0.69 | 0.93 |
D'Agostino 2008 (Framingham)41 | 367 71 73 | Kengne 200967 | NIDDM from 20 countries | 1003/7502 | CVD | 4 | 0.80 | <0.001 |
Simmons 200873 | NIDDM from UK | 69/272 | CVD | 10 | 0.73 | 0.02 | ||
Guzder 200571 | Newly diagnosed NIDDM from UK | 98/428 | CVD | 10 | 0.67 | <0.001 | ||
Ferrario 2005 (CUORE)50 | 19 | Pellegrini 20109 | NIDDM | 228/1532 | CVD | 10 years | 0.64 | Risk underestimated |
Decode study Group 200452 | 174 | Coleman 200774 | Newly diagnosed diabetes | NR/5102 | Fatal CVD/fatal CHD | 10 | 0.67 | Risk overestimated |
Assmann 2002 (PROCAM)56 | 17 | Stephens 20047 | NIDDM from UK | 332/798 | CVD and CHD | 10 | CHD: 0.65CVD: 0.67 | CHD: p=0.05CVD: p=0.006 |
Menotti 2002 (Riscard 2002)58 | 19 | Pellegrini 20109 | NIDDM | 228/1532 | CVD | 10 years | 0.59 | Good |
Wilson 1998 (Framingham)62 | 49 68 72 75 | Pellegrini 20109 | NIDDM | 228/1532 | CVD | 10 | 0.68 | Risk overestimated |
Davis 200968 | NIDDM from Australia | 38/791 | CHD | 5 | 0.59 | <0.001 | ||
Van der Heijden 200972 | NIDDM from the Netherlands | 28/125 | CHD | 10 | 0.63 | Risk overestimated | ||
Game 200176 | NIDDM from UK | NR/956 | CHD | 10 years | NR | High sensitivity, low specificity | ||
Anderson 1991 (Framingham)65 66 | 367 74 77 | Kengne 200967 | NIDDM from 20 countries | 1003/7502 | CVD | 4 | 0.56 | <0.001 |
Coleman 200774 | Newly diagnosed diabetes | NR/5102 | Fatal CVD/fatal CHD | 10 | 0.76 | Risk underestimated | ||
McEwan 200477 | NIDDM and IDDM from UK | 172/938 | CHD | 4 | Men: 0.65Women: 0.68 | NR |
↵* Only the study with the greatest study population has been displayed, an overview of all validation studies of the UKPDS risk engine is available in the online appendix table 3.
AUC, area under the curve; CHD, coronary heart disease; CVA, cerebrovascular accident; CVD, cardiovascular disease; IDDM, insulin-dependent diabetes mellitus; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; NIDDM, non-insulin-dependent diabetes mellitus; NR, not reported; UKPDS, United Kingdom Prospective Diabetes Study.