Analysis of logistic regression modelling
| Pathological feature contributing to estimation of probability of recurrence | β | SE β | χ2 |
|---|---|---|---|
| Regression coefficients from binary logistic modelling; β is the regression coefficient, SE β is the standard error of β, χ2 values are from a log likelihood ratio test for comparing two regression equations. Comparing the full model (with all four variables) with a reduced model with margin omitted gives a χ2 value of 4.284, which is significant at p=0.05 (χ2(0.05)=3.841). | |||
| Nuclear grade (NHSBSP) | 0.2379 | 0.4580 | 0.0272 |
| Necrosis | 0.4456 | 0.5537 | 0.718 |
| Margin distance (mm) | −0.1427 | 0.0733 | 4.284 |
| Histological size | 0.0228 | 0.0225 | 0.954 |