• English
  • Español

A3O2 | Reporting of effect direction and size in abstracts of systematic reviews

Abstract text
Background: Access to the full text of journal articles is often limited by cost and difficulty of internet access, whereas abstracts are freely available. Abstracts need to be clear, and contain sufficient detail to enable use of the results. The abstract of a systematic review should clearly state the direction, size and statistical uncertainty of estimated intervention effects, and should not require expert statistical knowledge to interpret.

Objective: To find out how many abstracts of systematic reviews present the results clearly and unambiguously.

Methods: Two authors extracted details from abstracts on written and numeric presentation of results of 182 systematic reviews (64 Cochrane and 118 non-Cochrane) published in 2009. Selected text was extracted to provide examples of good and poor wording of results.

Results: The direction of the intervention effect was described in words in 105 of 182 (58%) abstracts, with a further 34 (19%) deducible from the numeric results, and 43 (24%) not deducible from the abstract. The size of effect was given in words in only 22 (12%) abstracts. Cochrane and non-Cochrane reviews had similar quality of reporting results. 105 abstracts (58%) met a minimum standard for reporting of results, but only 11 (6%) included both direction and size of effect in words, size of effect numerically, and confidence interval. Reviews that had statistically significant results were reported better than non-significant ones (76% vs 29% meeting the minimum standard). Across all abstracts 45 (25%) did not present any numerical summary of the results.

Conclusions: The majority of abstracts of systematic reviews fail to communicate their message with regards to the size and direction of the intervention effect and its statistical precision. Examples of better reporting demonstrate that results can be expressed clearly and succinctly. Simple improvements to abstracts of systematic reviews would enhance their value.
Beller E1, Glasziou P1, Hopewell S2, Altman D2
1 Bond University, Australia
2 Centre for Statistics in Medicine, UK
Presenting author and contact person
Presenting author: 
Elaine Beller
Contact person: 
Elaine Beller (Contact this person)
Date and Location
Oral session A3O2
Friday 21 October 2011 - 11:35 - 12:05