Article type
Year
Abstract
Background: Conferences provide opportunities for researchers to present their trial results, but the brevity of conference abstracts restricts their usefulness as a dissemination tool1. However, exclusion of conference abstracts from a systematic review may introduce publication bias. A Cochrane methodology review found that only 63% of clinical trials presented at conferences get published in full2. None of the studies included in the review investigated HIV/AIDS interventions or trials done in Africa. Trials conducted in poor regions may face greater obstacles to full publication than trials done in well-resourced regions.
Objectives: 1) To determine how many randomized trials (RCTs) of HIV/AIDS interventions done in Africa presented at conferences get published in full; 2) To determine how many HIV/AIDS RCTs done in North America presented at conferences get published in full; and 3) To compare the African and North American trial publication record.
Methods: We identified abstracts for RCTs of HIV/AIDS interventions done in Africa by searching the electronic conference databases, AIDSearch and NLM Gateway, using a comprehensive African HIV/AIDS search strategy. The search was limited to presentations between 1985 and 2003. We searched for full articles reporting the same trials, using a specialised register of African trials3 and repeated this in MEDLINE. We repeated this process for North American trials limiting the conference search to presentations done in 2000.
Results: For African trials, our search of conference databases retrieved 987 records, from which we identified a total of 65 relevant conference abstracts, reporting on 48 discrete RCTs. Most abstracts (75%) were presented at the International AIDS Conference (IAC). Full reports were identified for 54 of the 65 abstracts (82%) with The Lancet and AIDS publishing full reports most frequently. For those abstracts published in full (N=48), the mean time from the conference to full publication of results was 1.9 years, ranging from less than 1 year to 7 years. This excludes six abstracts for which the trial results had been published in full before presentation at a conference. As of June 2006, the results of trials reported in 11 abstracts do not appear to have been published. For North American trials, we retrieved 865 records from conference databases, from which we manually identified a total of 62 relevant conference abstracts reporting on 62 trials. Full reports were identified for 32 of these trials (52%) with a mean time to publication of 2.2 years (range: 1 to 6 years) with the remaining 30 trials unpublished as of August 2006.
Interpretation and conclusions: Our study is the first to assess the progression of HIV/AIDS trials from conference presentations to full articles, and the first to focus on clinical trials done in Africa. It shows that the proportion published is far higher for African (82%) than North American HIV/AIDS trials (52%). This suggests that location, or factors related to the location of a trial, may be a determinant of the success of its eventual publication. Overall, the findings confirm the need to continue to search for conference abstracts to minimise the effects of publication bias in systematic reviews. Further exploration of the relationship between the location of a trial and its subsequent presentation and publication is warranted in other areas of healthcare.
References:
1. Hopewell S. (2004). Trials reported as abstracts and full publications: how do they compare? [abstract]. Clinical Trials (Suppl 1):216-7.
2. Scherer RW, Langenberg P, von Elm E. (2007). Full publication of research results initially presented in abstracts. Cochrane Database of Systematic Reviews 2007, Issue 2. Art. No.:MR000005.
3. Siegfried N, Clarke M, Volmink J. (2005). Randomised controlled trials in Africa of HIV and AIDS: descriptive study and spatial distribution. BMJ;331:742.
Objectives: 1) To determine how many randomized trials (RCTs) of HIV/AIDS interventions done in Africa presented at conferences get published in full; 2) To determine how many HIV/AIDS RCTs done in North America presented at conferences get published in full; and 3) To compare the African and North American trial publication record.
Methods: We identified abstracts for RCTs of HIV/AIDS interventions done in Africa by searching the electronic conference databases, AIDSearch and NLM Gateway, using a comprehensive African HIV/AIDS search strategy. The search was limited to presentations between 1985 and 2003. We searched for full articles reporting the same trials, using a specialised register of African trials3 and repeated this in MEDLINE. We repeated this process for North American trials limiting the conference search to presentations done in 2000.
Results: For African trials, our search of conference databases retrieved 987 records, from which we identified a total of 65 relevant conference abstracts, reporting on 48 discrete RCTs. Most abstracts (75%) were presented at the International AIDS Conference (IAC). Full reports were identified for 54 of the 65 abstracts (82%) with The Lancet and AIDS publishing full reports most frequently. For those abstracts published in full (N=48), the mean time from the conference to full publication of results was 1.9 years, ranging from less than 1 year to 7 years. This excludes six abstracts for which the trial results had been published in full before presentation at a conference. As of June 2006, the results of trials reported in 11 abstracts do not appear to have been published. For North American trials, we retrieved 865 records from conference databases, from which we manually identified a total of 62 relevant conference abstracts reporting on 62 trials. Full reports were identified for 32 of these trials (52%) with a mean time to publication of 2.2 years (range: 1 to 6 years) with the remaining 30 trials unpublished as of August 2006.
Interpretation and conclusions: Our study is the first to assess the progression of HIV/AIDS trials from conference presentations to full articles, and the first to focus on clinical trials done in Africa. It shows that the proportion published is far higher for African (82%) than North American HIV/AIDS trials (52%). This suggests that location, or factors related to the location of a trial, may be a determinant of the success of its eventual publication. Overall, the findings confirm the need to continue to search for conference abstracts to minimise the effects of publication bias in systematic reviews. Further exploration of the relationship between the location of a trial and its subsequent presentation and publication is warranted in other areas of healthcare.
References:
1. Hopewell S. (2004). Trials reported as abstracts and full publications: how do they compare? [abstract]. Clinical Trials (Suppl 1):216-7.
2. Scherer RW, Langenberg P, von Elm E. (2007). Full publication of research results initially presented in abstracts. Cochrane Database of Systematic Reviews 2007, Issue 2. Art. No.:MR000005.
3. Siegfried N, Clarke M, Volmink J. (2005). Randomised controlled trials in Africa of HIV and AIDS: descriptive study and spatial distribution. BMJ;331:742.