Hi Deepan,

I would not say that is is invalid to do a systematic review (and eventually a meta-analysis) that has already been done before.

For one, you can in your review cover aspects that were not included in the originl review or even correct things that the authors of the previous review did "wrong" (=could have done better). You can many times improve on methodology, or searched databases/languages/etc..

But apart from that, it is well known and shown that systematic reviews can become relatively quick out-of-date. Judging on the existing evidence, I would say that a review could/should be updated every 5-6 years-although that depends on the field we are talking about. Seeing that the review you mention covered up to 2004 and you have a decade or so of additional studies, I would say that you can go on with your idea.

Regarding your final question, it is surely better to take all existing articles and assess them together. Taking only the articles published after 2004 makes no sense and could lead you to false conclusions.

In any case, I would suggest taking into account things like review registration, CONSORT statement and the Cochrane's Handbook.

Hope it was of help,

Spyros

Thank you spyridon for your valuable suggestions.

This is a recent good article about this.

Overlapping meta-analyses on the same topic

http://www.bmj.com/content/347/bmj.f4501

than you all....... dear ignacio Mora, can you guide me how i get agrre from cochrane collaboration

Hi

To get involved with Cochrane, look for the Cochrane review group that would be interested in the topic. Connect with that review group and discuss registering your title with them. Relatively simple way to find out if you can do the review or if someone already is doing it for Cochrane.

For sure. Meta analyses always ned an update and also your question may be slightly different and include slightly different primary studies based on different inclusion and exclusion criteria. The main thing is that new evidence should be available by now which should be included.