Background: Reference inaccuracy in scientific articles brings the scientific validity of the research into question and may create difficulty when accessing the cited background data. The objective of this study was to examine the reference accuracy in the general surgery literature and its correlation with the journal impact factor.
Methods: Five general surgery journals were chosen with varying impact factors. From the year 2007, one issue was randomly chosen from each journal, and from each issue 180 citations were randomly chosen for review. Three investigators evaluated the chosen references for primary, citational, and quotational errors. The impact factor of each journal was compared to the percentage of errors detected.
Results: The total number of errors per journal ranged from 31.3 to 39.3%, with a total of 35.4% of all citations reviewed containing some type of error. The most common error type detected was incorrect citation of the primary source supporting a statement, the incidence of which ranged from 13.8 to 25.2%, depending on the journal, and accounting for 53.6% of the total errors found. Citational errors, which included incorrect author names, pagination, dates, and issue and volume numbers, ranged from 1.8 to 18.1% and accounted for 20.4% of the total errors detected. Qualitative errors, which occurred when the author misquoted another author's written assertions or conclusions, ranged from 7.4 to 16.0% and accounted for 34.7% of the total errors detected. Quantitative errors (misquoted numerical data) ranged from 3.1 to 8.6% and accounted for 17.9% of the total errors detected. No association between impact factor and error rate was demonstrated.
Conclusions: Reference inaccuracy is common in the general surgery literature. The impact factor has no clear association with the error rate, demonstrating that journal quality does not necessarily correlate with reference quality. Further investigation into potential methods for improving reference accuracy in the general surgery literature is warranted.