Table 3 Methods for handling missing data in sensitivity analysis in 27 trials
From: Handling missing data in RCTs; a review of the top medical journals
Sensitivity method | Assumption | Primary analysis | Assumption | N | Total N (%) |
|---|---|---|---|---|---|
Complete case | MCAR | Simple imputation | MCAR | 3 | 6 (22) |
MI | MAR | 2 | |||
Mixed Model | MAR | 1 | |||
Simple imputation1 | MCAR | GEE | MCAR | 1 | 4 (15) |
MI | MAR | 1 | |||
Mixed model | MAR | 2 | |||
GEE (un-weighted) | MCAR | Complete case | MCAR | 1 | 1 (4) |
MI2 | MAR | Complete case | MCAR | 6 | 10 (37) |
Simple imputation | MCAR | 1 | |||
GEE | MCAR | 1 | |||
Mixed model | MAR | 2 | |||
Mixed model | MAR | Complete case | MCAR | 2 | 2 (7) |
Adjustment using auxiliary data | MAR | Mixed model | MAR | 2 | 2 (7) |
Unclear | Complete case | MCAR | 1 | 2 (7) | |
Simple imputation | MCAR | 1 |