Table 1 Sample size calculations for power of 80% (two-sided, 5% significance level), based on sample of 5000 individuals per community and allowing for 20% losses, for each of the main comparisons (ECF vs non-ECF & HH vs non-HH)
Prevalence (%) in Control arm | Reduction in Prevalence due to each intervention (ECF or HH) individually | Prevalence (%) in ECF or HH arm | Overall number of communities needed (equally split into 4 arms) |
|---|---|---|---|
k = 0.15 | |||
0.6 | 30% | 0.42 | 24 |
0.8 | 30% | 0.56 | 20 |
1.0 | 30% | 0.7 | 18 |
1.0 | 20% | 0.8 | 41 |
1.0 | 10% | 0.9 | 168 |
k = 0.2 | |||
0.6 | 30% | 0.42 | 30 |
0.8 | 30% | 0.56 | 26 |
1.0 | 30% | 0.7 | 24 |
1.0 | 20% | 0.8 | 54 |
1.0 | 10% | 0.9 | 227 |
k = 0.25 | |||
0.6 | 30% | 0.42 | 45 |
0.8 | 30% | 0.56 | 42 |
1.0 | 30% | 0.7 | 39 |
1.0 | 20% | 0.8 | 93 |
1.0 | 10% | 0.9 | 398 |