| Â | Assumed % correct diagnoses | Â | Â | Â | Â | Required number |
|---|
| Â | C | I | Alpha | N0 (total) | ICC | VIF | Mean per practice | N patients | N practices |
|---|
Primary |
 Diabetes | 0.75 | 0.85 | 0.05 | 500 | 0.103 | 21.12 | 192 | 10,560 | 56 |
Key secondary |
 CKD | 0.50 | 0.70 | 0.05 | 186 | 0.32 | 9.1 | 25 | 1693 | 68 |
- Required total number of practices to have at least 80% power for the comparison of two proportions based on a logistic regression model correcting for the clustering of patients within GP practice. ICC and resulting VIF were obtained from historical data. The alpha level has been set at 5% for the primary endpoint and also at 0.05 for the key secondary endpoint (following a hierarchical closed testing procedure)
- C control group, I intervention, Alpha alpha-level, N0 total number of required diagnoses (patients) based on a χ2-test (hence, ignoring clustering within GP practices), ICC intra-class correlation, VIF variance inflation factor (taking into account differences in cluster size), Cluster size number of diagnoses per practice, Mean per practice mean denominator per practice, N patients total number of required diagnoses taking into account the clustering, N practices required total number of practices. The number of practices is rounded upwards such that an equal number of practices can be allocated to the control and the intervention group
