Article type
Year
Abstract
Introduction/Objective: To determine what proportion of abstracts of randomised controlled trials (RCTs) presented at 3 cystic fibrosis (CF) international conferences were published, time to publication and factors influencing publication.
Methods: At the end of 1995, the Cochrane CF Group's register of (RCTs) contained 199 abstract reports to 180 RCTs identified by handsearching abstract books of 3 international CF conferences. We searched the CF register for subsequent full publications of these reports. Survival analysis was used to investigate time to publication to enable inclusion of abstracts where time since presentation was insufficient for publication to have occurred. Using the log-rank test, we examined if there was a difference in time to publication between reports where the investigators concluded that the test treatment was as effective or better than the control treatment compared to those where it was not. We examined whether there was a difference in time to publication between large and small sample sized RCTs, stratifying around the median. A Cox's Regression model was used to investigate the effects of both variables, sample size and results, both separately and simultaneously.
Results: The absolute proportion of reports followed by a full publication was 32%. Survival estimates suggested that the proportion published before 12 months was 8.1 %, at 2 years was 29% and at 5 years was 40%. We found no significant difference in time to publication when the abstracts were stratified according to results or sample size. There was no evidence of an association between time to publication, sample size, results or both.
Discussion: Survival analysis provides the best estimate of time to publication in this subject area. The small proportion of abstract reports of RCTs followed by full publication is disappointing but consistent with findings in other areas.
Methods: At the end of 1995, the Cochrane CF Group's register of (RCTs) contained 199 abstract reports to 180 RCTs identified by handsearching abstract books of 3 international CF conferences. We searched the CF register for subsequent full publications of these reports. Survival analysis was used to investigate time to publication to enable inclusion of abstracts where time since presentation was insufficient for publication to have occurred. Using the log-rank test, we examined if there was a difference in time to publication between reports where the investigators concluded that the test treatment was as effective or better than the control treatment compared to those where it was not. We examined whether there was a difference in time to publication between large and small sample sized RCTs, stratifying around the median. A Cox's Regression model was used to investigate the effects of both variables, sample size and results, both separately and simultaneously.
Results: The absolute proportion of reports followed by a full publication was 32%. Survival estimates suggested that the proportion published before 12 months was 8.1 %, at 2 years was 29% and at 5 years was 40%. We found no significant difference in time to publication when the abstracts were stratified according to results or sample size. There was no evidence of an association between time to publication, sample size, results or both.
Discussion: Survival analysis provides the best estimate of time to publication in this subject area. The small proportion of abstract reports of RCTs followed by full publication is disappointing but consistent with findings in other areas.