How to Interpret a Research Paper

If you’re lucky enough to get access to the full text of a research paper, the next hurdle physios often run into is interpreting this research! This blog breaks down the anatomy of a research paper, telling you what you should and shouldn’t do with each section. Ultimately, by reading this blog you should have a better understanding of how to interpret a quantitative research paper. Let’s get into it!

Abstract

What is it good for: For you to work out whether you want to read the whole paper.

What to do with it: If the question looks important and the results interpretable, then go ahead and get hold of the paper if you can.

What not to do with it: Do not form your opinions on the conclusions based on reading this. Do not post something on twitter if you’ve only read the Abstract!

What you need to know: You cannot assess the risk of bias from the abstract, which means that you cannot judge how much confidence you should have in the findings. The information in the Abstract may not be trustworthy because researchers often misrepresent the findings of their study.

At Physio Network, we’re big believers in reading more than just the abstract. That’s why we started our Research Reviews, where industry experts break down clinically relevant papers for you into 5 minute reads.

Learn more about these Research Reviews HERE

Introduction/Background

What is it good for: Describing why the study is important and setting out the research question.

What to do with it: If the research question is clear and relevant to you, then read on.

What not to do with it: Do not be too convinced by what the researchers say in this section, focus on whether you think the research question is important.

What you need to know: The section is essentially a cherry-picked literature review that includes all the stuff that supports the researchers’ view and ignores the rest. Peer-reviewers and journal editors generally do not insist that researchers present a balanced picture of the field in the Introduction.

Methods

What is it good for: Describing what the researchers actually did. After reading this you should have a clear picture of who was in the study (participants), how participants came into the study (recruitment), what data were collected (descriptive variables and outcomes), and how the data were analysed (analysis).

What to do with it: If the participants, data and analysis are clear and match the research question, then read on.

What not to do with it: Do not get too impressed by fancy analysis methods. Simple analyses that are described in a way that is understandable are appropriate for most clinical research questions, and if the analysis is too complex you won’t be able to interpret the results.

What you need to know: Pay attention to recruitment methods, they are important to judge generalizability. Pay attention to how outcomes are measured, the study tells you nothing if measures are not sufficiently reliable and valid.

Results

What is it good for: Providing the answer to the research question, and describing the sample.

What to do with it: If the numbers can be mapped directly to the analysis in the Methods and directly to the question in the Introduction, smile. You can read on if you like, but you’ve actually already got what you really need to know.

What not to do with it: Place most confidence in primary analysis (especially if there is a published protocol or register). Have less confidence in the secondary outcomes and very little at all in posthoc or exploratory analyses. Ignore results of analyses that do not appear in the Methods. It is reasonable to be sceptical if the sample size is small but don’t be seduced by a big one; results are not valid, meaningful or important simply because a sample is large. Judge whether you think the effect or association is clinically meaningful (big enough to matter) for your patients.

What you need to know: The size of effect estimates or associations (and confidence intervals) are what matters, p-values give you very little useful information. Researchers will not always present the primary analyses most prominently; the most common example is an RCT that reports within-group changes for both groups in an RCT, rather than between-group differences.

Discussion

What is it good for: Stating the results in an interpretable way and giving an overview of related research on the topic.

What to do with it: If the statement about the main findings reflects what is in the results and directly addresses the research question in the Introduction, you have your answer. The other main part of the section should be an outline of related literature which will be helpful if: a) you are not aware of what has been published, and b) the researchers present a complete and balanced overview. If you have some sense of a) and you’re not confident about b) then you can just interpret the findings yourself.

What not to do with it: Ignore all the stuff that restates the methods or says stuff about this being the “the first…”, it doesn’t help you. Don’t pay too much attention to the parts that address anything other than the primary research question/outcomes. Be wary if the results of the study are quite different to other studies that address similar questions, regardless of the researchers’ explanations for such differences.

What you need to know: Like the Introduction, this section is a cherry-picked literature review, alongside the results of the study. While some consideration of the implication of the findings is appropriate, researchers often commit large chunks of the Discussion to speculation that goes well beyond discussion of the actual findings of the study. Limitations often appear in a contained section by themselves, and if they are serious enough they need to be incorporated into interpretation of the findings.

Conclusion

What is it good for: A concise answer to the research question.

What to do with it: If the conclusion directly answers the research question, using the results of the primary analysis, your work here is done!

What not to do with it: If the section goes on for more than a couple of sentences, it includes stuff that does not belong in the conclusion. Read on if you’re interested but don’t hold too tightly to what is in there.

What you need to know: Research sometimes coughs up findings that are disappointing to the people running the study. When this happens there is a temptation for the authors to over-reach in their conclusions.

Hopefully this blog has helped shed some light on how to interpret a research paper, so that you’re better able to implement research / research reviews into your practice in the future.