Short steps to reading a paper; Part 14. The discussion
You have now read the methods and results. Now it is time for the discussion. These features are what I look for when I read the discussion section of a paper.
When I was preparing this post, I found it rather hard to write. This is also true of the discussion part of a paper. I have always found this section difficult to write. This is because you have said all that you are going to say in the introduction most of the time. As a result, many discussions suffer from being a repeat of the introduction, with a few results being liberally sprinkled through it. As a result, many are difficult to read or not too useful. Here are my tips.
- Look to see if the authors have stated why their findings are important and clinically relevant.
- Find the section where they describe if the effect size is clinically significant.
- Do not spend too long on some large sections comparing their results with many other studies. If there is a section on this, take care to check that they compare similar study designs. If they are not, the comparison is redundant.
- My general advice is skim-read the discussion. If you do not find these three factors, then move straight to the conclusions.
You may feel that this that this section is too brief. However, by the time you have read the introduction, methods and results sections, you should have made your own interpretation of the paper and its findings. As a result, reading the discussion may not be necessary.
Emeritus Professor of Orthodontics, University of Manchester, UK.
Have your say!
I always taught my residents that a discussion is an update of the review, given that there now is one additional paper. No repetition of things previously said. If the review was thorough and to the point, the discussion is already off to a good start.
How do you decide if the effect size is clinically significant (in presence of statistical significance)? For example, in a hypothetical study, a significant difference of 0.15mm is found – is this regarded as clinically significant? If not – why not?