Table 1.
Trade-off between number of clusters and cluster size for the case study
| No of clusters in each arm | 80% power | 90% power | ||
|---|---|---|---|---|
| Cluster size | Total sample size | Cluster size | Total sample size | |
| 6 | 191 | 2292 | NA | NA |
| 7 | 89 | 1246 | 1383 | 19 362 |
| 8 | 58 | 928 | 191 | 3056 |
| 9 | 43 | 774 | 103 | 1854 |
| 10 | 35 | 700 | 70 | 1400 |
| 11 | 29 | 638 | 54 | 1188 |
| 12 | 25 | 600 | 43 | 1032 |
| 13 | 22 | 572 | 36 | 936 |
| 14 | 19 | 532 | 31 | 868 |
| 15 | 17 | 510 | 28 | 840 |
Sample size needed to detect a difference between two proportions of 0.60 and 0.45 at two sided significance level of 5%, assuming normal approximations (formula in Appendix 1).
Assumes 228 in each arm needed for 90% power and 171 for 80% power.
ICC assumed to be 0.03.
Rounding has occurred at some levels.
NA=not achievable