Table 3.
Sensitivity analysis comparing results of ITT analysis of complete cases with ITT analyses, with missing IPSS data imputed
| No imputed | No used in analysis | Overall mean (SD) | Difference in means* (95% CI) | P value | ||
|---|---|---|---|---|---|---|
| Intervention arm | Usual care arm | |||||
| Complete case ITT | 0 | 0 | 887 | 12.79 (6.64) | −1.81 (−2.66 to −0.95) | <0.001 |
| Best case scenario | 77 | 78 | 1042 | 10.89 (7.63) | −1.89 (−2.89 to −0.88) | <0.001 |
| Worst case scenario | 77 | 78 | 1042 | 14.82 (7.82) | −1.14 (−2.08 to −0.20) | 0.02 |
| MICE† | 82 | 80 | 1077 | — | −1.61 (−2.57 to −0.66) | 0.001 |
CI=confidence interval; IPSS=International Prostate Symptom Score; SD=standard deviation ITT=intention to treat; MICE=multiple imputation by chained equations; SD=standard deviation.
*
Analysis adjusted for baseline IPSS and minimisation variables.
†
Data are imputed using IPSS at baseline and six months, treatment arm, practice size, centre, and deprivation (index of multiple deprivation). To allow for clustering in Stata, imputations were performed separately for each practice.