Systematic review checklist.
If you tried to stay current by reading two articles each day, within 1 year you would fall 55 centuries behind!
Systematic reviews can save you some time by synthesising the literature to draw strong clinical conclusions and make recommendations for practice. It is common for PhD students or clinical academics to undergo this process. Though methodologies may vary- the process is largely the same. So here’s a handy checklist to make sure your systematic reviews are, errm… systematic?
Note- I am by no means an expert, this is just a handy tool that I developed through my own experience.
- Dirty searching. Scope out your topic with a dirty google search. Check for similar reviews and for papers of interest to give you an idea of the volume of literature available. Check PROSPERO for registered reviews that are ongoing and unpublished as not to duplicate the literature.
- Write. A. Protocol. Do not skip this step. You will not figure it out as you go. You will create more work for yourself. A good protocol should cover a short rationale (this will help justify your protocol), search strategy, inclusion and exclusion criteria, data analysis and presentation. You may refine the inclusion or search strategy at a later date but the fundamentals should be there prior to starting.
- Register your review on PROSPERO. As I mentioned this prevents duplication.
- Do your searches. This needs to be per database, and should be built up using conjunctions as you go. Try and start broad and refine as you need to. The more you add, the more you may miss. It’s better to have to screen more but be sure that you haven’t missed key papers. Try not to add too many limiters to your search- ideally do not limit to the English language as this presents a lot of bias
- Screen titles and abstracts. Ideally you would have a second person. If from the abstract you are unsure, you should include in the full text screen until you are confident. Do not mark as “maybe”- stick to include or exclude. If you have nailed your exclusion criteria at the start there will be very little discrepancies at this stage. But if there are any- check with a third reviewer.
- Obtain and screen your full texts. Again with two reviewers and a third for discrepancies.
- Make a data extraction template and pilot it before you extract the data. Think about what information will be useful for your audience. Better to have too much than too little.
- Extract your data. Don’t rush this step. Do it right and you will only do it once!
- Assess risk of bias. This is kind of done alongside the extraction. Make sure you used a recommended tool for RoB. This is different for different study designs (Cochrane, ROBINS, COSMIN). You may need more than one if you are including different study designs.
- Tabulate your data. This will prepare you for synthesis. Put all the information you have into a table so that you can see which studies are similar or not and what the findings were.
- Synthesise your data. Well, that’s much easier to say than to do. But this will vary based on your preferred method and included studies. This may be a narrative synthesis, quantitative synthesis or meta-analysis. I won’t go into too much detail here but maybe watch this space… You may need to transform your data so that it is in the same format (i.e. Mean[CI] or whaterver)- again, too much to cover here.
- Interpret your results and formulate your conclusion. This is helpful to discuss with the team that worked on it with you to ensure you are being accurate in your interpretation.
- Write the paper. This is a topic in its own right. So in brief, just write it all down and make it look pretty.
- Update PROSPERO with your publication details.
Easy peasy eh! If you were looking for expert advice you are in the wrong place. For more practical guidance on how to perform a systematic review check out the cochrane handbook
https://training.cochrane.org/cochrane-handbook-systematic-reviews-interventions
