Approach | Main model | Comments |
Qualitative systematic review | Qualitative description of evidence | May be most appropriate if large degree of heterogeneity among studies. |
Meta-analysis overall estimate | Fixed effects model | Assumes a single truth across populations and homogeneity among studies. |
Random effects model | Incorporates between-study heterogeneity. Wider confidence interval than fixed effects model when heterogeneity is present. |
Meta-analysis of multiple interventions simultaneously | Network meta-analysis (mixed or multiple treatment comparison) | Estimates relative effects of multiple interventions against each other, even if no individual study analyzed a comparison between 2 or more specific interventions. |
Meta-analysis subgroup analyses | Fixed or random effects models | Estimates treatment effects for each group. May explain heterogeneity. Subject to ecological fallacy. May not be possible due to limited reported data. Arbitrary selection of subgroups may result in spurious findings. |
Meta-regression | Regression across studies | Tests interaction between subgroup and treatment effects. Can test continuous or categorical variables singly or in a multivariable analysis. May explain heterogeneity. Subject to ecological fallacy. May not be possible due to limited reported data. Arbitrary selection of subgroups may reult in spurious findings. |
Individual patient data meta-analysis | Multivariate regression across individuals | Allows most complete analysis of data and evaluation of heterogeneity. Costly and resource-intensive. |