Arm vs Contrast level data
RevMan can manage two types of results data in the analyses:
- Arm level data: When data for all study intervention arms are available, such as the number of cases or events for each arm or mean effect sizes for each arm.
- Contrast level data: If only effect estimates for comparing the intervention arms are available, e.g. mean difference, (adjusted) odds ratio, hazard ratios. It is possible to use either of these sets of data, or combine them together.
With automatic analyses (under ‘Data source’), you can choose to include:
- Only arm-level data
- Only contrast-level data
- Contrast- and arm-level data (preferring arm-level data where both exist)
- Contrast- and arm-level data (preferring contrast-level data where both exist)
Section 5.3.6 of the Cochrane Handbook for Systematic Reviews of Interventions provides more guidance on the different types of data and in what circumstances either might be preferred in your analyses.
Here is an example of arm level data:
Study ID | Intervention | Arm | Events | Total |
Herne 1980 | Xibornolics | Xibornolics | 3 | 22 |
Tetracycline | Tetracycline | 4 | 24 | |
Placebo | Placebo | 10 | 22 | |
McKerrow 1961 | Tetracycline | Tetracycline | 5 | 15 |
Placebo | Placebo | 8 | 18 | |
Taylor 1977 | Co-trimoxazol | Co-trimoxazol | 12 | 29 |
Placebo | Placebo | 3 | 59 |
Here is an example of contrast level data:
Study ID | Intervention | Arm | Reference arm | Log[Odds ratio] | Standard error | Covariance |
Hoaglund 1950 | Aeromycin | Aeromycin | No | -1.2 | 0.3 | 0.03 |
Tetracycline | Tetracycline | No | -1.4 | 0.5 | - | |
Placebo | Placebo | Yes | - | - | - | |
Kaiser 1996 | Co-amoxiclav | Co-amoxiclav | No | -0.7 | 0.9 | - |
Placebo | Placebo | Yes | - | - | - | |
Lexomboon 1971 | Penicillin V | Penicillin V | No | -0.9 | 0.2 | - |
Placebo | Placebo | Yes | - | - | - |