Table 2.
Comparison of prediction models
| Variable | Internal model (hazard ratio) | Framingham model (odds ratio)* |
|---|---|---|
| Age, years | — | 0.99 |
| Sex, male | — | 0.65 |
| Fasting plasma glucose, per 10 mg/dL | 1.98 | 1.15 |
| History of high blood glucose | 1.67 | — |
| Hemoglobin A1c, % | 1.93 | |
| Parental history of diabetes | — | 1.55 |
| Height, cm | 0.985 | — |
| Waist:hip ratio, per 0.1 unit | 1.173 | — |
| Waist circumference, per 10 cm | 1.11 | 1.05 |
| Body mass index | — | 1.04 |
| Systolic blood pressure, mm Hg | — | 1.01 |
| Triglycerides, per 10 mg/dL | 1.02 | 1.00 |
| High density lipoprotein cholesterol, mg/dL | — | 0.96 |
| Intercept | — | −18.607 |
| 3 year baseline survival† | 0.999998 | — |
| C statistic | 0.73 | 0.69 |
| Mean:median risk ratio | 0.72 | 1.09 |
| Pearson’s median skewness coefficient | 1.05 | −0.94 |
| Extreme quarter risk ratio | 6.48 | 2.18 |
*Odds ratios for Framingham model continuous variables are per 1 unit increase.
†Can be used to calculate person’s risk predictions on basis of hazard ratios; practitioners can use nomogram in web appendix.