Source
Faculty of Medicine, Institute of Social and Preventive Medicine, University of Geneva, Switzerland. Jean-Francois.Etter@imsp.unige.ch
Abstract
OBJECTIVE:
To study variations in the number of times trials of nicotine replacement therapy (NRT) were cited, and which characteristics of trials predicted the number of citations and the impact factors of journals in which articles were published.
STUDY DESIGN AND SETTING:
We used all 105 randomized controlled trials in the Cochrane review of NRT for smoking cessation. We obtained impact factors from the Journal Citation Reports and the number of citations from ISI Web of Knowledge and Google Scholar.
RESULTS:
Trials were cited from 0 to 632 times (median 23 times). Trials were cited more often when results were statistically significant than when they were not (median=41 vs. 17 times, P<0.001), and when impact factors were higher (10.2 more citations per impact factor point, P<0.001). Patch trials were cited more often than gum trials (median=29 vs. 17 times, P=0.001), and trials funded by the pharmaceutical industry were cited more often than other trials (median=28 vs. 16.5 times, P=0.001). Trials with statistically significant results were published in journals with higher impact factors than trials with nonsignificant results (median impact factor=2.80 vs. 1.81, P=0.011).
CONCLUSION:
Citations were biased toward trials with positive results and toward trials published in high-impact-factor journals.