Article type
Year
Abstract
Background: The bias caused by preferential publication of randomized controlled trials (RCTs) with positive results distorts the scientific record. Poorer outcomes or even harmful effects may be under reported.
Objectives: Our aim was to document the extent to which RCTs published as dermatology conference abstracts were subsequently published as full RCTs in journals. We also wanted to see whether publication was related to whether the abstract reported 'positive' findings.
Methods: We searched the Specialised Skin Register database managed with Procite software containing abstracts from comprehensive handsearching of conference proceedings. The search strategy we used was (#39=RCT AND (#20>= 2000 AND #20 <= 2002) AND #04 = 'Abstract'). In this exploratory study 10% of the conference abstracts were randomly selected from this group. They were sorted into those that had positive, negative and neutral outcomes. MEDLINE was searched for the intervention and disease terms in the title field and for randomized controlled trial publication type in the year of the abstract and subsequent years
Results: Of the 30 conference abstracts 22 were rated as having positive, 2 negative, 6 neutral conclusions (2 of these were ongoing trials). Of the 22 abstracts rated as 'positive' 12 were subsequently published. Of the remaining 10 abstracts, 3 appeared to relate to subsequent publications but there were large discrepancies in the patient numbers reported. One of the 6 abstracts rated 'neutral' had a publication but the ongoing trials had not reached publication. The 2 'negative' abstracts had not resulted in publications.
Conclusions: Less than half (43%) of the conference abstracts in this exploratory study were published in MEDLINE in a 3-5year period. Of those rated as 'positive', 60% reached publication compared with neither of the two 'negative' abstracts, suggesting problems of publication relating to the direction of the study conclusion. We are now proceeding to a much larger study to test these preliminary conclusions. The possibility of publication bias would be avoided by prospective registration of all RCTs in the public domain. We are working towards this in the Cochrane Skin Group with our Prospective Register of Ongoing Trials which is freely accessible on www.nottingham.ac.uk/csg
Objectives: Our aim was to document the extent to which RCTs published as dermatology conference abstracts were subsequently published as full RCTs in journals. We also wanted to see whether publication was related to whether the abstract reported 'positive' findings.
Methods: We searched the Specialised Skin Register database managed with Procite software containing abstracts from comprehensive handsearching of conference proceedings. The search strategy we used was (#39=RCT AND (#20>= 2000 AND #20 <= 2002) AND #04 = 'Abstract'). In this exploratory study 10% of the conference abstracts were randomly selected from this group. They were sorted into those that had positive, negative and neutral outcomes. MEDLINE was searched for the intervention and disease terms in the title field and for randomized controlled trial publication type in the year of the abstract and subsequent years
Results: Of the 30 conference abstracts 22 were rated as having positive, 2 negative, 6 neutral conclusions (2 of these were ongoing trials). Of the 22 abstracts rated as 'positive' 12 were subsequently published. Of the remaining 10 abstracts, 3 appeared to relate to subsequent publications but there were large discrepancies in the patient numbers reported. One of the 6 abstracts rated 'neutral' had a publication but the ongoing trials had not reached publication. The 2 'negative' abstracts had not resulted in publications.
Conclusions: Less than half (43%) of the conference abstracts in this exploratory study were published in MEDLINE in a 3-5year period. Of those rated as 'positive', 60% reached publication compared with neither of the two 'negative' abstracts, suggesting problems of publication relating to the direction of the study conclusion. We are now proceeding to a much larger study to test these preliminary conclusions. The possibility of publication bias would be avoided by prospective registration of all RCTs in the public domain. We are working towards this in the Cochrane Skin Group with our Prospective Register of Ongoing Trials which is freely accessible on www.nottingham.ac.uk/csg