To test the hypothesis that articles with negative results are more likely than articles with positive results to be published in journals with lower impact factor.
Design and Setting
We selected all of the randomized, placebo-controlled trials conducted during the neonatal period between October 1, 1998, and October 1, 2003. Trials were classified as having positive results or negative results (significant or no significant difference, respectively). Only studies dealing with primary outcomes (efficacy) were included.
Main Outcome Measures
The impact factor of each journal was determined, and the sample size for each study was noted.
There were 233 articles that fulfilled the inclusion criteria. There was a significant difference between the 2 groups in terms of impact factor (P = .03) but not sample size (P = .30). Impact factor correlated with both sample size and the type of study results (positive results vs negative results; P<.05).
Articles with negative results are more likely than articles with positive results to be published in journals with lower impact factor.