Table 4.
Reclassification of Risk According to the Genotype Score for Subjects in Whom Diabetes Developed and for Those in Whom Diabetes Did Not Develop*
| Model without Genotype Score | Model with Genotype Score | |||
|---|---|---|---|---|
| 0 to <2% Risk | 2 to <8% Risk | ≥8% Risk | Total at Risk | |
| number (percent) | ||||
| Model 1: adjusted for sex | ||||
| Subjects in whom diabetes developed | ||||
| 0 to <2% risk | 0 | 0 | 0 | 0 |
| 2 to <8% risk | 1 (0.4) | 242 (94.9) | 12 (4.7) | 255 |
| ≥8% risk | 0 | 0 | 0 | 0 |
| Total no. | 1 | 242 | 12 | 255 |
| Subjects in whom diabetes did not develop | ||||
| 0 to <2% risk | 0 | 0 | 0 | 0 |
| 2 to <8% risk | 65 (1.1) | 5731 (97.6) | 79 (1.3) | 5875 |
| ≥8% risk | 0 | 0 | 0 | 0 |
| Total no. | 65 | 5731 | 79 | 5875 |
| Net reclassification improvement — % | 4.1 | |||
| P value | 0.004 | |||
| Model 2: adjusted for sex and self-reported family history | ||||
| Subjects in whom diabetes developed | ||||
| 0 to <2% risk | 0 | 0 | 0 | 0 |
| 2 to <8% risk | 2 (0.9) | 202 (94.8) | 9 (4.2) | 213 |
| ≥8% risk | 0 | 17 (40.5) | 25 (59.5) | 42 |
| Total no. | 2 | 219 | 34 | 255 |
| Subjects in whom diabetes did not develop | ||||
| 0 to <2% risk | 0 | 0 | 0 | 0 |
| 2 to <8% risk | 252 (4.6) | 5102 (93.8) | 87 (1.6) | 5441 |
| ≥8% risk | 0 | 218 (50.2) | 216 (49.8) | 434 |
| Total no. | 252 | 5320 | 303 | 5875 |
| Net reclassification improvement — % | 2.60 | |||
| P value | 0.22 | |||
| Model 3: adjusted for simple clinical risk factors | ||||
| Subjects in whom diabetes developed | ||||
| 0 to <2% risk | 20 (83.3) | 4 (16.7) | 0 | 24 |
| 2 to <8% risk | 4 (7.4) | 45 (83.3) | 5 (9.2) | 54 |
| ≥8% risk | 0 | 2 (1.1) | 175 (98.9) | 177 |
| Total no. | 24 | 51 | 180 | 255 |
| Subjects in whom diabetes did not develop | ||||
| 0 to <2% risk | 3924 (97.5) | 101 (2.5) | 0 | 4025 |
| 2 to <8% risk | 156 (12.2) | 1063 (83.4) | 56 (4.4) | 1275 |
| ≥8% risk | 0 | 57 (9.9) | 518 (90.1) | 575 |
| Total no. | 4080 | 1221 | 574 | 5875 |
| Net reclassification improvement — % | 2.13 | |||
| P value | 0.17 | |||
Net reclassification improvement for 255 cases of diabetes among 6130 person-examinations was based on pooled logistic-regression models with generalized estimating equations including or not including the genotype score. Net reclassification improvement is better if more people in whom diabetes develops are reclassified as being at higher risk when the genotype score is added to the model and more people who remain free of diabetes are classified as being at lower risk when the genetic score is added. The net reclassification improvement is worse when there is erroneous reclassification — for example, if many people in whom diabetes develops are classified as being at lower risk when the genetic score is added to the model. The number and proportion of people who were not reclassified by the genotype score are indicated by the bold font.