Background:

The primary objective of most systematic reviews is to report the effectiveness of the treatment intervention, including both efficacy and harms of the test intervention. It is not known how often reports of harms are included in abstracts describing trial results and presented at conferences.

Objectives:

To describe the frequency of reports of harms in abstracts describing results of randomized or controlled clinical trials (CCTs).

Methods:

We searched conference proceedings for CCT reports from the Association for Research in Vision and Ophthalmology
(ARVO) for the years 2001, 2004, and 2005. We extracted information regarding reporting of harms, classified as not reported, reported as primary outcome (i.e. safety of the intervention), or reported as other outcome. We searched PubMed and the Science Citation Index to find full length publications of 2001 ARVO abstracts to examine the association between reporting of harms in an abstract and full length publication.

Results:

We identified 221, 295, and 329 abstracts describing CCTs presented at the 2001, 2004, and 2005 ARVO meetings, respectively. Authors of a majority of abstracts did not report assessment of harms related to treatment intervention (603/845; 71.4%). Complications or harms were the primary outcome in 6.2% (52/845) of abstracts, reported as other outcomes in 22.2% (188/845), and unclear as to type of outcome in 0.2% (2/845). There was a trend for reporting harms less frequently from 2001 to 2005 (ARVO 2001 (36.7%; 81/221); ARVO 2004 (26.8%; 79/295); ARVO 2005 (24.3%; 80/329)). Fifty-five point six percent (45/81) of abstracts reporting harms were published in full, compared with 50.7% (71/140) of those not reporting harms (relative risk = 1.13; 95% CI=0.80 to 1.61).

Conclusions:

Authors report harms in about 30% of ARVO abstracts describing CCTs. There does not appear to be any association between reporting harms and subsequent full length publication.