Table 3.
Sensitivity analysis
| Intervention group | Control group | Estimate*(95% CI) | p value | |
|---|---|---|---|---|
| Cluster-level models† | ||||
| Omit cluster with <3 responses | 14; 60·2% (19·7) | 14; 42·9% (13·8) | 15·2% (2·3 to 28·1) | 0·026 |
| Omit cluster with >30% individuals missing data (excluding centre) | 5; 58·5% (14·4) | 5; 44·1% (6·6) | 14·7% (1·1 to 28·2) | 0·040 |
| Including percentage missing and number of responders | 21; 58·4% (21·6) | 21; 40·5% (23·8) | 15·3% (−0·1 to 30·6) | 0·051 |
| Including percentage missing | 21; 58·4% (21·6) | 21; 40·5% (23·8) | 15·0% (0·3 to 29·7) | 0·046 |
| Including number of responders | 21; 58·4% (21·6) | 21; 40·5% (23·8) | 15·0% (0·6 to 29·4) | 0·042 |
| Multiple imputation | .. | .. | 15·0% (−2·0 to 32·0) | 0·076 |
| Hierarchical models‡ | ||||
| Fixed effects for cluster (n=397) | 112/198 (56·6%) | 85/208 (40·9%) | 1·93%§ (1·21 to 3·09) | 0·0058 |
Data are N; mean (SD), or n/N (%), unless otherwise specified. N is the number of pharmacies.
*
The estimated treatment effect (percentage difference between groups) in proportion with the outcome.
†
Outcome is the percentage of effective contraception uptake and analyses were adjusted for phase, treatment group, centre, mean age, percentage of participants who were in a current sexual relationship, and percentage of participants who had used effective contraception previously.
‡
Outcome is the uptake of effective contraception (binary) and was adjusted for phase, treatment group, and centre.
§
Odds ratio.