Performing the search for the systematic review

I am going to assume that you have developed your basic research question and are frothing at the mouth with anticipation that you will be able to search the literature and get to review possibly thousands of abstracts. Like many people I would suggest that you use PICOS to help determine your search.

What does PICOS mean?

From the capitalization you have probably realised that this is an acronym. It stands for:


By working your way through each item and defining it, you should be able to clarify your question and also define your search. I feel that the best way to show how to do this is to give an example.


For this example my research question is going to be:
“Based upon differences in microvascular outcomes seen in people of South Asian origin with T2DM compared with the Western population, especially the relationship of outcomes with BMI, would SGLT2 inhibitors be more effective in this population?”

The first step is to define the population being investigated. This is “people of South Asian origin with T2DM”. However, when we perform the search we have to remember that not everyone with use the exact phrase that we have, for example South Asia includes a number of countries and we may want to include the populations of the separate countries and the names of the countries in the search as well. We will also want to add a search term for type 2 diabetes. These can then be combined as a search strategy:
(T2DM OR Type 2 diabetes) AND (South Asian OR South Asia OR India OR Pakistan OR…)

The next step is to define the intervention and for our question it would be SGLT2 inhibitors. The search strategy should also include the names of the individual drugs as well as the class. This would make the following search:
(SGLT2 OR sodium-glucose cotransporter 2 OR sodium-glucose co-transporter 2 OR canagliflozin OR dapagliflozin OR empagliflozin OR sotagliflozin…)

We then move onto the comparison. We are comparing different populations, rather than interventions, so this would be Western people with T2DM. For the first search I would not include this as a search term, and instead would keep it in mind when I was filtering the results following the search. However, the decision to include only studies comparing the two populations would help filter the results.

Next is outcomes which is undefined in our search and so would need discussion amongst the authoring group. Would we be looking at efficacy endpoints or cardiovascular outcomes, for example. Again this could be used for filtering the results and might not be included in the original search.

Finally is setting. This question could include large observational clinical data sets or clinical trial data and both would be interesting. Reviews would not be relevant to teh article but could be a useful way to ensure that all relevant references are found.

What next?

We have now defined our search and also some criteria to filter the results. So the next step would be to perform the search and the databases you use are dependent upon the subject you are researching. For example, in medicine it would include PubMed, Embase, Cochrane reviews and It might also include results highlighted in the reference lists for any reviews identified.

The numbers of results should be recorded following teh search and before filtering as this is important for the PRISMA flow diagram.

A good handout to help define a search with PICOS is available from, which can be found here.