1
$\begingroup$

For example, if I have 3 different trials each for 3 different drugs and I used meta analysis (sub group analysis) to generate effect sizes. How can I interpret the effect sizes? All drugs are being compared to placebo e.g.

Drug A - 3 RCTs - pooled odds ratio 1.5 Drug B - 3 RCTs - pooled odds ratio 1.7 Drug C - 3 RCTs - pooled odds ratio 3

I am aware of network meta analysis but I was suggested to avoid this as it’s not necessary for my project. I am able to make simple indirect comparisons as long as I state the limitations.

Firstly, would I be able to say, drug C had the greater effect compared to placebo (at a particular outcome) so can be considered the better of the three drugs, though head to head trials are necessary to confirm this.

What are the general limitations of this method?

I have tried to ensure the studies I have used are as similar in the design as possible. All comparing to a common comparator, placebo.

Also should I make comments on the heterogeneity between the 3 studies used for each drug.

Also should I do this using a random or fixed effects model?

$\endgroup$
1
$\begingroup$

To test whether the odds ratios are different, you could run a meta-analysis on all 9 RCTs and test drug type as a moderator, with drug 3 set as the reference group. If it's significant, you could say it has the greater effect relative to placebo. Personally, I would comment on heterogeneity for this single analysis. You can't conclude that these odds ratios are different from each other without at least examining their confidence intervals.

Whether you want to use a fixed or random effects model depends on whether you believe that the studies all estimate one true effect size and the only source of differences is random error, in which case you'd want fixed effects, or whether you believe it's possible that the true effect size differs from study to study. Assuming the studies aren't sampling the same population, you likely want to use random effects meta analysis.

Source: Borenstein, Hedges, & Rothstein, 2007

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.