Let me throw a monkey wrench into the mix. Most people think of systematic reviews as being costly because it takes a long time to 'properly' complete each task. I think the problem is not that, but that we are very inefficient at completing each task. For example, historically, the average time for me to run a 'thorough' search in PubMed was around 30 min to 1 hour or longer for ~60 lines of search terms. I would copy the search term, paste it into PubMed, run the search, wait for the browser to refresh, look at the warnings from PubMed (if any) and so on. Now I run the full search in under 30 seconds (and still line by line) using an automated program I wrote. Even better yet, it saves the number of citations for each search term, how PubMed interprets the search term, any errors that PubMed has raised, and it even downloads the results for me without ever having to open a web browser. It's all about being more efficient in practice. Once we can optimize each aspect of the review process, we won't need to find ways to cut corners to be cost-effective. Some researchers are developing computer-assisted screening and data extraction tools, etc. Soon, we may be at the point where the only thing the reviewer needs to do is identify the elements of the PICO and everything will be automated from preparing a search strategy to extracting the data and analyzing it.

Ahmed