DDW News


Secrets from a DDW® Abstract Reviewer

The abstract submission period for Digestive Disease Week® 2020 is now open, and we teamed up with Peter D.R. Higgins, MD, PhD, MSc to share a few tips to keep in mind before you submit your abstract (by the Dec. 1 deadline, of course). Dr. Higgins is the director of the inflammatory bowel disease (IBD) program at the University of Michigan, and has played an active role in DDW abstract review for the past 14 years. He shares his insights on Twitter: @ibddoctor. If you have additional questions, ask them in the comments below.

1. Use a structured abstract.

  1. Background: include a concise statement of the problem, hypothesis and aims.
  2. Methods: demonstrate rigor and approaches to reduce bias.
  3. Results: this is the meat of your abstract – it should be close to 50% of character count.
  4. Conclusions: be conservative; don’t overstate your conclusions.


2. Keep the background and rationale concise.

The reviewers are experts in the field, and they already know the epidemiology and impact of the problem. Save your character count for the results.


3. Don’t break the blind.

DDW review is blinded. The reviewers quite deliberately do not know who the authors are. Do not deliberately break the blind in your abstract. Avoid stating details about author names, grants, locations (city, county) and institutions. It will count against you.


4. Be specific in your hypothesis and aims.

If you have more than one aim, number them.


5. Methods:

Show us that you know what the weaknesses and potential flaws are, and that you did everything you could to minimize these, including:

  1. Questing for an unbiased random sample.
  2. Avoiding multiplicity of testing.
  3. Avoiding overfitting – a logistic model with 20 deaths should not have more than 2 predictors.
  4. Avoid fancy machine learning on samples of less than 100 (preferably avoid with samples of less than 1000).
  5. Develop your model on a training set, then test its accuracy on a held-out testing set. Report the accuracy on the testing sample, not the training sample.


6. Results:

Avoid vague statements about increases or difference. Provide numbers, measures of centrality (mean, median), measures of spread (95%CI or range or IQR [not SE or SEM]). Provide a graph or two to illustrate your main point. Don’t worry about p values. Confidence intervals will make your point.


7. Conclusions:

Don’t over-sell it. You rarely have a perfectly random sample, you usually have samples from a few centers that are often biased toward academic, tertiary centers, and your findings may be less generalizable than you would like. Recognizing the limitations of your data builds your credibility. You should be proud of your work, but recognize that it is just one abstract. Not yet changing the world.


8. Editing:

Keep cutting to make it concise. Remove any adjectives or adverbs that are not essential. Make lists clear with an Oxford comma. Structure lists with numbers, commas, or semicolons if needed. Use html tags sparingly. Make sure that you have clear topic sentences for each section. The more complex your ideas are, the simpler and shorter your sentences should be.


9. Take note of the deadline.

The Abstract opening bell is on Oct. 17, and the deadline is Dec. 1, which is on a Sunday, and three days post-Turkey in the U.S. Time to get started!


Leave a Reply

Your email address will not be published. Required fields are marked *