For a meta-analysis, we are interested in comparing the mean difference across time for intervention versus control conditions. This lends itself well to computing an effect size for the intervention. What happens, however, when we have a RCT that compares the effectiveness/efficacy of two or more interventions, and does not include a true control group? Is it possible to calculate a real effect size of an intervention, without having a comparison to a no-treatment control? RCTs of this sort are common in comparative effectiveness studies, including superiority, non-inferiority, and equivalence trials.
It is possible to compute effect size from comparative effectiveness studies. Of course, it has to be clear that these studies (standard-of-care comparisons) cannot offer the type of effect size which placebo comparisons offer. There exist statistical methods to derive placebo-compared effect sizes from comparative effectiveness studies. Attaching one example below, and couple of related issues.
Maybe i think yor are now finding the method of pre-post meta-analysis.
it is possible to make effect size of one arm; treatmet or placebo. However in this methodology could not be consider about placebo effect.
please check the attach file.
Sir, what do you mean by "when we have a RCT that compares the effectiveness/efficacy of two or more interventions, and does not include a true control group?"What do you mean by "true control group"? Result of a comparison is always embedded in certain context; let´s say new intervention vs standard of care. Is it "true effect" when you compare new treatment vs placebo? Yes (efficacy of new treatment vs placebo) Is it reflecting real world clinical practice? Non whatsoever, because in real world only there is (almost) always standard of care (no placebo).
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis