I have a Meta analysis report & I have to present the findings of the report to non-statisticians. There are 2 treatment groups & 5 endpoints.
How do I explain pooled odds ratio, forest plot, CI etc. in a simple graphical format? Is it okay to use bar diagram or line diagram etc. to show odds ratio? Should I convert pooled odds ratio to Pooled relative risk as odds ratio is somewhat tricky to understand?
I would show them the forest plot. Then take them through the components of the plot, the estimates (and say why the blobs are of different sizes), the confidence intervals (and why they are truncated, if any are), and the summary estimate assuming there is one. I would leave any discussion of exactly what the odds ratio is until someone asks. If you explain it in terms of relative risk you are not really going to seriously mislead them but if you feel uncomfortable doing that then do you analysis in terms of risk ratios instead. Have some example two by two tables up your sleeve with OR and RR ready calculated. I think that should be enough for half an hour.
The approach typically recommended by many experts in evidence based medicine is to convert pooled odds ratios (OR), relative risks (RR) or hazard ratios (HR) to number needed to treat (NNT) and number needed to harm (NNH). Otherwise you can rerun your meta-analysis using risk differences (RD), but they may be less mathematically suitable.
You can give an estimate of the uncertainty of the pooled effect estimates by reporting the two extremes of the confidence/credible interval, referring to them as the extremes of reasonable certainty for the average effect estimate.
It is always useful to try applying your findings to real-world scenarios, such as a patient at low risk of event or another at high risk.
As a reference, the reporting recommendations of the Cochrane Library are always insightful:
Archived Link.
