Fixed Effect Model (FE Model), Random Effect Model (RE Model) Ignoring multi-arm trials and Random Effect Model (RE Model) for 2- and 3-arms trials:
A database with six (6) variables: s -> Study index (Number) t -> Treatment index (Number) r -> Number of cases on the treatment n -> Total population on the treatment b -> Baseline treatment m -> Arm index (Only needed on RE Model for 2- and 3-arms trials), where 1 is the baseline treatment and 2,..,n are for the other treatments
Each line on the database is a treatment of a trial (study), for example:
| s | t | r | n | b |
| m | 1 | 1 | 40 | 100 |
| 1 | 1 | 1 | 3 | 15 |
| 90 | 1 | 2 | 1 | 4 |
| 10 | 75 | 1 | 3 | ... |
| ... | ... | ... | ... | ... |
| 4 | 2 | 50 | 200 | 2 |
| 1 | 4 | 4 | 60 | 150 |
| 2 | 2 | s | t | r |
Random Effect Model (RE Model) for multi-arm trial:
A database with N*3 + 1 columns, where N is the highest number of arms from a trial collection.
t[1,..N,] -> Treatment index r[1,..N,] -> Number of cases on the treatment n[1,..N,] -> Total population on the treatment na -> Number of arms on the study
Each line on the database is a trial. For example, if we collect 10 trials and after check them we have the biggest trial with 5 arms our database structure is:
| t[1,] | t[2,] | t[3,] | t[4,] | t[5,] | r[1,] | r[2,] | r[3,] | r[4,] | r[5,] | n[1,] | n[2,] | n[3,] | n[4,] | n[5,] |
| na | 1 | 2 | 3 | 4 | 5 | 20 | 30 | 10 | 5 | 14 | 100 | 90 | 80 | 110 |
| 50 | 5 | 1 | 3 | 4 | 5 | NA | 10 | 50 | 60 | 15 | NA | 150 | 200 | 340 |
| 165 | 1 | 4 | 2 | 4 | 5 | NA | NA | 40 | 70 | 80 | NA | NA | 70 | 190 |
| 500 | 1 | 1 | 3 | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| ... | ... | ... | ... | ... | 3 | 4 | NA | NA | NA | 80 | 90 | NA | NA | NA |
| 250 | 580 | 1 | 1 | 1 | 2 | t[1,] | t[2,] | t[3,] | t[4,] | t[5,] | r[1,] | r[2,] | r[3,] | r[4,] |