1. Estimate an appropriate sample size, especially if a result of no difference or no relationship is reported. Any time a sample is used, a power analysis should be performed to estimate a sample size adequate to detect the effect of interest. This is crucial for “negative” results to ensure that they are not the result of a Type II statistical error. The chapter and tables on sample size estimation in Hulley SB, Cummings SR. Designing clinical research: An epidemiologic approach. Baltimore: Williams & Wilkins, 1988. are an excellent resource on this topic.

  2. Indicate the magnitude of a relationship or difference where possible. P-values only reflect the probability that a measured result was found by chance. Reporting the measures themselves (e.g. means and standard deviations, proportions, coefficients) helps the reader to assess the clinical or practical significance of the result.

  3. Be sure results support conclusions. Statistics or statistical tests supporting conclusions should be reported in Results.

  4. Avoid irrelevant statistics. In the abstract, statistics that are not related to results or conclusions need not be reported.

  5. When more than one statistical test is used, identify which test produced which result.

  6. When reporting proportions or percentages, always report the denominator on which they are based.

  7. Relating the results back to clinical practice. The results could be expressed as the number needed to treat or effect a result or the number needed to screen or detect one more case or prevent one more death, etc. The difference between two groups with a p-value of 0.000005 still lacks significance, if it is the difference of 0.2 patients out of a thousand clinic visits. It is the ill – defined “so what” factor of disease oriented medicine vs. the more clinically relevant patient outcome measures that are finally getting some attention. A beautifully designed study that is well-executed and produces numerical results but has no applicability is of less relevance than one with POEM that is immediately applicable.


For those who want to read more on how to write a good abstract, the following are excellent resources:

  • * Haynes RB, Mulrow CD, Huth EJ, Altman DG, Gardner MJ. More Info Abstracts Revisited. Ann Intern Med. 1990; 113:69-76

  • * Why Not Say It Clearly: A Guide To Scientific Writing. L.S. King, Little Brown 1978. How to Write Better Medical Prose; written by a physician with illustrative examples.

  • * How To Write And Publish Papers In The Medical Sciences. E.J. Hugh, Philadelphia ISI Press 1990. Written by the former editor of Annals of Int the best reference on preparing a medical paper publication.

  • * How To Write And Publish A Scientific Paper. R.A. Day, Philadelphia ISI Press, 2nd Ed., 1988. Another “How-to” manual in writing medical articles.