Artifactual increase in journal self-citation

After submission of a manuscript to a peer-reviewed anesthesia journal, several authors were asked to cite additional references from the journal to which they submitted. We hypothesized that there were differences among the anesthesiology journals in both the total number of self-citations and the proportion of self-citations to the total number of references in each manuscript for the years 2005 and 2010.

We conducted a review of a sample of manuscripts from 2005 and 2010 to examine the number and rate of self-citations. As a secondary analysis, we reviewed impact factor (IF), rate of self-referencing, and contribution of self-citations to IF in the population of manuscripts published in 8 anesthesia journals between 2000 and 2009 using the ISI Journal Citation Reports.

The number (P < 0.0001) and rate (P < 0.0001) of self-citations among the different journals were significantly different in 2005, with similar results for 2010 in the number (P < 0.0001) and rate (P = 0.0002) of self-citations. The mean range of number of self-citations ranged from 0.45 (95% confidence interval [CI], 0.06 to 0.84) to 3.95 (95% CI, 2.2 to 5.7) in 2005 and from 0.25 (95% CI, -0.05 to 0.55) to 4.5 (95% CI, 2.2 to 6.9) in 2010. On a per-journal basis, no difference in the number of self-citations was noted between 2005 and 2010. Analysis of the ISI Journal Citation Reports from 2000 to 2009 suggested a general decline in the contribution of self-cites to the IF over time for the aggregate journals (Spearman correlation coefficient (Rs) -0.25 (95% CI, -0.45 to -0.03), P = 0.02), with the exception of the journal in question (Rs = 0.59 (95% CI, -0.1 to 0.88), P = 0.05). Positive correlations were found between self-cited rate and IF (Rs 0.52, 95% CI, 0.34 to 0.66, P < 0.0001), percentage of self-cites to years used in IF calculation and IF (Rs 0.41, 95% CI, 0.21 to 0.58, P < 0.0001), and δ-IF and IF (Rs 0.89, 95% CI, 0.84 to 0.93, P < 0.0001).

Although the number and rate of self-citations differed among anesthesia journals, the contribution of self-citation to IF has declined over time for most anesthesia journals. These results suggest periodic reassessment may be important to ensure that the publication process remains transparent and impartial to bias.