2
$\begingroup$

I am conducting a project where I comparing a 5 different drugs (anti-psychotics) for a particular treatment group. There are currently no head-to-head trials for me to be able to make direct comparisons.

I have 3-4 trials of each drug comparing them to placebo. On RevMan I have done a subgroup analysis for different outcomes. I do not want to conduct a network meta-analysis. Can I make comparisons on which is the better drug by comparing their pooled odds ratios?

Also when I conduct the subgroup analysis, do I use random-effects or fixed-effect models?

P.s. I am aware of the limitations of just comparing them this way, which I will highlight in my project.

$\endgroup$
  • 1
    $\begingroup$ Are all drugs compared to the same comparator (eg placebo)? $\endgroup$ – Joe_74 Mar 16 '19 at 8:45
  • $\begingroup$ Yes. For example if the pooled odds ratio of one drug is 2. and another is 1.5, both being compared to placebo. Would I be able to say out of the two drugs, the first drug has a better effect? Or is there another way to word this better. $\endgroup$ – user241325 Mar 16 '19 at 16:13
1
$\begingroup$

Comparing the odds ratios vs., say, a placebo is effectively a network meta analysis. It is simply one with extra many unverifiable and likely partially unreasonable assumptions. E.g. differences in trial design, population, adhetence, background care, duration, outcome definition and analysis makes such indirect comparisons very unreliable, but at least you are looking at an odds ratio rather than say a risk difference, or - gasp, horror - a within group change from uncontrolled trials.

If patient populations don't match entirely, but trial conduct is identical, then subgroup results in completely matching subgroups may be useful for seeing the effect of non matching populations.

$\endgroup$
  • $\begingroup$ Thank you for answering!or example if the pooled odds ratio of one drug is 2. and another is 1.5, both being compared to placebo. Would I be able to say out of the two drugs, the first drug has a better effect? Or is there another way to word this better. – $\endgroup$ – user241325 Mar 16 '19 at 16:14
  • $\begingroup$ "X had larger observed effect estimates than Y, but without a more appropriate analysis of results or a head-to-head trial we cannot say whether X is better than Y." would be more traditional. $\endgroup$ – Björn Mar 16 '19 at 21:29

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.