HOW TO WRITE AN ABSTRACT?

Writing an abstract is often the first time the data from a study are analysed. You must have a clear idea of the central finding of the study before starting to write. The first step is to display all the data in draft figures and to discuss them with your co-workers. Review the original study hypothesis and work out the key findings. An abstract cannot contain all the results, so construct a priority list of the material which must be included and the less important material which may be omitted.

Please note the following obligatory body parts of your abstract.

• Background and goal of study
• Material and Methods
• Results and Discussion
• Conclusion

Below you will find explained in detail what each part should comprise.

Title. The title is important. It is the first thing that the reviewer will see from your abstract and gives the first impression. It is the only part of your work that most of the participants at the congress will read. A poorly titled abstract may be overlooked by those working in the same field and you will not reach your intended audience. Spend some time on it! It should be a concise summary of your abstract. Write down the key points of the work as a first step and try to link them together. Avoid redundant words like "Study of" or "Investigation of". Construct several versions of your title and ask your colleagues whether they get the key message. Try to read your title from the perspective of a reviewer. Does it convince you that the abstract is new, interesting, important and relevant to the field? (As you are submitting online, you should not the authors names and affiliation in the abstract submitted. Those informations will be encoded separately).
Background and Goal of study. Abstracts need very short introductions. A short summary of previous work, directly relevant to your study to outline the scientific background and indicate the gap in current knowledge that your study will fill. You must state clearly the hypothesis to be tested by study or the observations that are sought. This is often best as a question that later can be answered in the conclusion section.
Material and Methods. State briefly the methods used. Standard methods need only be named, but uncommon or new methods may need a more detailed description. If you did a clinical study, state the study design (randomised, controlled, blinded, prospective... ) and briefly the patient selection and exclusion criteria before you describe how the study was carried out. In animal experiments, define the species used, and the plan of the experiment (i.e. the time sequence of interventions and measurements). The reader must understand how the data were generated. Indicate how the data were analysed and which statistical tests were used. Reporting negative findings is often a source of problems. The reader needs to know if there is really no important difference or whether the study has simply failed to detect a real and important difference, perhaps because the study was too small (In other words does "absence of proof" necessarily show "proof of absence"). If you report negative findings, you can convince readers and reviewers that this is important by giving the power of the study or the minimum (important) difference that your study was designed to detect with a given alpha and beta error. If you are not sure about this issue, seek statistical help before submitting the abstract. Scientific writing already has too many abbreviations which can confuse and mislead: try to avoid them if at all possible and do not invent new ones. If you need to use abbreviations, use explicit ones (like thio for thiopental ) and always define them in the text.
Results and Discussion. Poor presentation of data often causes rejection. In this part of the abstract the reader must learn how your conclusion is justified by the data. For clinical studies, the patient characteristics should be mentioned first (number studied, age, sex, height and weight). You must give the numbers (summarised as mean or median) and always give an index of variation and say what it is (e.g. standard deviation, 95% confidence intervals)! For data that are not normally distributed, median and quartile values are more appropriate. You should not report effects or "trends" only in descriptive terms, like "blood pressure increased significantly after drug A, but not after drug B". Give the results of the statistical tests. If you report negative findings, you may use the 95% confidence interval as index of variation and report the calculated P-value ("P= not significant" is not appropriate; since there is a big difference between P= 0.06, and P= 0.9). Because of the limited space, you must focus on your most important findings. Some data are best given in a table or a figure. There are no fixed rules, but tables are better for presenting exact numbers, while a pronounced trend or effect is often easier seen in a figure. Always include an index of variation in your figures or tables and to report the results of the statistical test. Define all symbols in the text. If you have some free space, it is helpful to put a summary sentence in the text, stating the main finding from the figure or table, but don't repeat the data. For figures you upload, do not use three dimensional columns, gridlines, plot frames, boxes around labels and remove all other unnecessary artwork. All these additional elements will distract the reader from your data. Crowded ticks on one or both axes also will draw the eye away from the curves and labels. As a rule of thumb, keep the maximum number of major ticks per axis to five. Make any breaks in the axis clearly visible. In the best figure, every line or symbol should convey information, and there will be "no non-data ink". Remember that the size of the figures may be reduced: all items must be big enough to be easily read after reduction. Abstracts need only short discussions. Focus on what can be concluded from the data and explain them in simple words. Avoid any speculations.
Conclusion. Here, come back to the study hypothesis outlined in the first paragraph of the abstract. Give a simple answer to the hypothesis in the light of the new data from your study. You may wish to suggest what needs to be studied next (but avoid just saying "further work is necessary"). Authors often then state some clinical implications, to underline the importance of the study. But be careful! The reviewers will check whether your conclusion is justified by your data.
References.. You may include up to three references, numbered according to order of occurrence in the text, with the corresponding number in the text. Use the following layout and punctuation: Author name up to three (followed by et al if more). Journal title abbreviated as for Index Medicus year of publication; volume number: starting page number-finishing page number. Thus: "Adam AB, Eve Q, Cain OM, et al. Clin. Sci. 1997;99:45-101".
Acknowledgements (optional) You may include any acknowledgement you wish to make. Do not include information that you have entered (or still need to) in the disclosure of conflict of interests; you will be asked to disclose any conflict(s) of interest in a separate step of the submission process.
From draft to final version. After you have written the first draft version of your abstract, you may find that your abstract is helplessly over length. Condensing the information until the abstract fits into the electronic abstract submission boxes is a real challenge and may take some time. First, go through the text slowly and check each sentence to see how the same idea can be expressed more briefly. In the result section, a figure or table may replace a long list of data . Substantial shortening is often possible in the introduction and conclusion section. As a last resort, limit the data given in the abstract. Put away this first "final" version for a few days and then discuss it with your co-authors and also some colleagues who have not been involved in your research. The suggestions of a "naive expert" are often very helpful. Go through the abstract several times and take the perspective of a reviewer. Does your abstract meet all formal requirements? Is it interesting and relevant to the field? Are the data clearly presented? Is your conclusion supported by the data? After you have made the final changes in your abstract, make sure that all of your co-authors agree with the final version.

G.B. Drummond (Br J. Anaesth.)