I am working on a topic and one of the included studies in my review is a meta-analysis. Would it be possible or "legal" to use the study on my pooled analysis? If it is, should I use the individual data mentioned in the study or the pooled analysis of the study itself? Thanks in advance.
Hitesh Khurana
Thanks for your answer, If i may ask, would you recommend me to use the pooled result of the study or using the mentioned studies included in the meta-analysis.
Hello Lowilius,
To ensure the validity of most, if not all, of the statistical output from your meta-analysis of primary studies, it would probably not be best to include the pooled effect estimate. For instance, the 'pooled study' would have more weight, given the likely larger sample size, but essentially not a true reflection of each individual study.
If all the studies that contribute to the pooled effect estimate meet your inclusion criteria, it would probably be best to include each of the primary studies in your meta-analysis.
I hope you find this useful.
Kindly check https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3049418/
Also check https://himmelfarb.gwu.edu/tutorials/studydesign101/metaanalyses.cfm
Thanks to Chinaza Godswill Awuchi sir... That links are help to get new info
Chinaza Godswill Awuchi Chinaza Godswill Awuchi thank you for your suggestion. I will keep this in mind
You can only make an "Umbrella review of meta-analyses" by checking their quality based on AMSTAR-2 checklist. It is not logical to meta-analyse meta-analyses. You will increase the rate of problems. Redundant included publications for example.
In simple words: a meta-analysis is a secondary analysis of primary analyses. And so, a meta-analysis-on-meta-analysis would be a secondary analysis over a secondary analysis of primary analyses. The problem is: the more you advance the level of posthoc analyses, the more you lose the credibility and validity of your estimates. That is why whenever performing statistics we need to adhere to the most direct analysis as possible. In your case: you can add the included studies of the previous meta to your meta (if they meet your PICO, of course), and it will be an updated meta-analysis. Otherwise, you should not include the final pooled estimate of the previous meta in your pooled analysis, as this has severe bias and statistical flaws.
If the currently published systematic reviews are prevalent; yes, you can do a systematic review of systematic reviews. However, when two or more systematic reviews include overlapping studies, you will have to dismantle the studies and analyze them per conventional systematic review.
It is still better to pool individual studies rather than systematic reviews as your model will be more direct and unbiased. Furthermore, only including systematic reviews in your eligibility criteria may pose imminent risks of 'missed' studies.
I hope this helps.
Best,
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis