how to write a paper (hall/how to write a paper) || discussion

4
Chapter 5 Discussion George M. Hall Department of Anaesthesia & Intensive Care Medicine, St George’s, University of London, London, UK Many authors find this section of the paper to be the most difficult. However, it should be an exercise in logic and discipline, and a satisfactory discussion can be based on the format shown in Box 5.1. Poor discussions have no structure, try to cite all publications found during the literature search and induce acute boredom in the reader. Keep it short, snappy and relevant. A useful rule is – if in doubt cut it out. You are most unlikely to have a manu- script rejected just because the discussion is too short. Principal findings The reader has just finished a detailed presentation of the results so it is important to remind them of the key findings. A good start to the discussion is two or three sentences that summarise the results. These should be clear and unambiguous, the ‘take home message’, and can often be used in the abstract. Further analysis of the data should not be undertaken in the discus- sion. If you missed something important out of the results then you will have to go back and rewrite this section. Methodology It is most unlikely that the methods used in the study were perfect so a brief appraisal is necessary in the discussion. A common problem is the sample size, and the power calculation described in the methods may have been optimistic. There is no point in trying to hide this from editors and assessors. It may be necessary to downgrade your study from the definitive clinical trial in this area to a pilot or preliminary study that will enable other researchers to undertake a correctly powered investigation. 29 How to Write a Paper, Fifth Edition. Edited by George M. Hall. © 2013 John Wiley & Sons, Ltd. Published 2013 by John Wiley & Sons, Ltd.

Upload: george-m

Post on 30-Nov-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: How to Write a Paper (Hall/How to Write a Paper) || Discussion

Chapter 5 Discussion

George M. Hall Department of Anaesthesia & Intensive Care Medicine, St George ’ s, University of London, London, UK

Many authors fi nd this section of the paper to be the most diffi cult. However, it should be an exercise in logic and discipline, and a satisfactory discussion can be based on the format shown in Box 5.1 . Poor discussions have no structure, try to cite all publications found during the literature search and induce acute boredom in the reader. Keep it short, snappy and relevant. A useful rule is – if in doubt cut it out. You are most unlikely to have a manu-script rejected just because the discussion is too short.

Principal fi ndings

The reader has just fi nished a detailed presentation of the results so it is important to remind them of the key fi ndings. A good start to the discussion is two or three sentences that summarise the results. These should be clear and unambiguous, the ‘ take home message ’ , and can often be used in the abstract. Further analysis of the data should not be undertaken in the discus-sion. If you missed something important out of the results then you will have to go back and rewrite this section.

Methodology

It is most unlikely that the methods used in the study were perfect so a brief appraisal is necessary in the discussion. A common problem is the sample size, and the power calculation described in the methods may have been optimistic. There is no point in trying to hide this from editors and assessors. It may be necessary to downgrade your study from the defi nitive clinical trial in this area to a pilot or preliminary study that will enable other researchers to undertake a correctly powered investigation.

29

How to Write a Paper, Fifth Edition. Edited by George M. Hall.© 2013 John Wiley & Sons, Ltd. Published 2013 by John Wiley & Sons, Ltd.

Page 2: How to Write a Paper (Hall/How to Write a Paper) || Discussion

30 How to write a paper

Unusual study designs often alarm assessors so you should explain pre-cisely why you chose this design and, if possible, provide supporting citations using similar methodology. To use a sporting analogy, ‘ get your retaliation in fi rst ’ . In essence, you are trying to deal with any criticisms from editors and assessors by showing that you had already thought of the diffi culties inherent in the study design.

On the other hand you may be able to emphasise here any strengths of the methods used. For example, you may have developed a more sensitive and specifi c assay for plasma rhubarb concentrations that has enabled you to fi nd changes during routine surgery that other investigators failed to observe. Criticism of the methodology of previous investigators may be appropriate, but make sure that you remain objective and scrupulously fair.

Previous work

A key part of the discussion is the comparison of your results with other published studies. You should cite only major relevant work, both confi rma-tory and contradictory. Do not simply repeat the sentences you used in the introduction when defi ning the research question, and never, never quote what you have not read. There is the temptation to cite every paper written on the subject to show the assessors the thoroughness of your literature search. Resist the temptation; a surfeit of references is a sign of insecurity not scholarship. You will know who are the major research groups so concentrate on them. Do not ignore previous literature that disagrees with your fi ndings. This ‘ selective citation ’ will be spotted very quickly by assessors and you will lose credibility as a consequence. When dealing with previous work, be impartial; there are often good reasons why results cannot be exactly replicated, and you may be able to explain some of the discrepancies.

Box 5.1 Discussion: overall format

• Statement of principal fi nding(s)

• Appraisal of methods

• Comparison with previous work

• Clinical and scientifi c implications (if any)

• Further work

• Conclusion (optional)

• Acknowledgements

Page 3: How to Write a Paper (Hall/How to Write a Paper) || Discussion

Discussion 31

Implications

If your results may change clinical practice, then this should be discussed. Most investigators never make a major breakthrough, so do not exaggerate the importance of your work. It is likely to be just a small contribution to a limited area of knowledge, but it is still important to state how our under-standing has increased as a result of your work. If there has been no progress, it indicates that the study was of little value. Similarly, if the study was non - clinical, any basic scientifi c implications should be discussed.

Further studies

After you have dealt with your fi ndings in relation to previous work and any clinical or scientifi c implications, you are ready to suggest further work in the area. Some editors dislike the speculation that this entails, and you may fi nd that this paragraph(s) is subsequently deleted. For other research workers this is often the most interesting part of the discussion as it gives ideas for future research. However, before you parade your best ideas in public, you are advised to have started the work; otherwise you may fi nd that other research groups publish fi rst. If you do not intend to continue working in the area, then this part of the discussion may be useful for claiming prec-edence of ideas.

Conclusion

It has been traditional to fi nish the discussion with a brief concluding para-graph, which is a succinct r é sum é of the major fi ndings. This is increasingly omitted, or deleted by the editor, as it often repeats information that has already appeared in the structured abstract, results and at the start of the discussion. Many authors still prefer to fi nish the discussion in this way, but you need to ensure that it is not a direct repetition of previous parts of the manuscript.

Acknowledgements

Many journals expect the source of funding for the research and any confl ict of interest to be given at the beginning of the paper, even the title page. If the instructions to the authors are not explicit about the matter, then these should be stated clearly in the acknowledgements. Funding bodies must be listed and any commercial links given. If you are unsure whether the study could have been infl uenced by any current or previous commercial

Page 4: How to Write a Paper (Hall/How to Write a Paper) || Discussion

32 How to write a paper

undertakings, then declare everything and let the editor decide what should be included.

You should also acknowledge any person who enabled the study to proceed but did not achieve authorship (see Chapter 7 by Richard Horton). For example, medical colleague, technician, research nurse and statistician, and permission to cite these people should be obtained before the paper is sub-mitted. Do not use the acknowledgements to fl atter colleagues, such as the head of the department.

Finally, sometimes assessors and editors feel that they have made a greater contribution to the fi nal published manuscript than the authors and yet they are never acknowledged!