Table 2 Description of unexplained discrepancies (N = 89)
Unexplained changes | N (%) |
|---|---|
Analysis population | |
Changed set of patients included by specifying additional exclusions | 12 (13%) |
Analysis model | |
Changed model | 13 (15%) |
Changed method of selecting analysis model | 2 (2%) |
Covariates | |
Changed from unadjusted to adjusted analysis | 7 (8%) |
Changed from adjusted to unadjusted analysis | 4 (4%) |
Changed set of covariates included in analysis | 5 (6%) |
Missing data | |
Changed from complete case to multiple imputation | 1 (1%) |
Changed imputation strategy | 2 (2%) |
Unexplained additions | |
Analysis population | |
Not mentioned in the original analysis plan | 8 (9%) |
Incomplete detail given in the original analysis plan | 7 (8%) |
Allowed analyst to subjectively choose analysis population based on the trial dataset | 1 (1%) |
Analysis model | |
Not mentioned in the original analysis plan | 5 (6%) |
Incomplete detail given in the original analysis plan | 5 (6%) |
Allowed analyst to subjectively choose analysis model based on the trial dataset | 7 (8%) |
Covariates | |
Not mentioned in the original analysis plan | 2 (2%) |
Incomplete detail given in the original analysis plan | 2 (2%) |
Allowed analyst to subjectively choose covariates based on the trial dataset | 3 (3%) |
Missing data | |
Not mentioned in the original analysis plan | 9 (10%) |
Incomplete detail given in the original analysis plan | 3 (3%) |
Allowed analyst to subjectively choose missing data approach based on the trial dataset | 1 (1%) |