Errata in medical publications.

BACKGROUND

Information is limited about the communication of corrections or errors in the medical literature; therefore, we sought to determine the frequency and significance of published errata in high impact factor journals.

METHODS

Retrospective evaluation of errata reports for articles published in 20 English-language general medicine and cardiovascular journals (mean impact factor, 12.23; median, 5.52) over 18 months. Each independently adjudicated erratum was categorized by location in the article and qualitative categories of severity. Descriptive statistics and Spearman's rank correlation coefficients were computed to describe the association between author and errata number. Source of error, association between impact factor and errata occurrence, and errata rate by journal were assessed.

RESULTS

A total of 557 articles were associated with errata reports (overall errata report occurrence 4.2 per 100 published original and review articles; mean of 2.4 errors per errata report). At least 1 major error that materially altered data interpretation was present in 24.2% of articles with errata. There was a strong association between impact factor and errata occurrence rate (rho = 0.869, P < .001). Across all errata, 51.0% were not corrected or the report did not specify whether a correction was made.

CONCLUSIONS

The reporting of errata across journals lacks uniformity. Despite published criteria for authorship that mandate final approval of the manuscript by all authors, errors are frequent, including those that may materially change the interpretation of data. Increased vigilance by authors to prevent errata and consensus by journal editors on the format of reporting are warranted.