Ghost publications in the pediatric surgery match.

INTRODUCTION: Pediatric surgery fellowship is considered one of the most competitive subspecialties in medicine. With fierce competition increasing the stakes, publications and first authorship are paramount to the success rate of matching. We analyzed Electronic Residency Application Service applications for verification of authorship to determine rate of misrepresentation. METHODS: After institutional review board approval, the bibliographies of fellowship applications from 2007-2009 were reviewed to allow time for publication. Only peer-reviewed journal articles were evaluated. A Medline search was conducted for the article, by author or by title. If the article could not be found, other authors and journal were used as search parameters. If the article was still not found, the website for the journal was searched for abstract or manuscript. Finally, an experienced medical sciences librarian was consulted for remaining unidentified articles. Differences between misrepresented and accurate applications were analyzed, including: age, gender, medical and undergraduate school parameters, advanced degrees, other fellowships, number of publications, first author publications, American Board of Surgery In-Training Examination scores, and match success. RESULTS: There were 147 applications reviewed. Evidence of misrepresentation was found in 17.6% of the applicants (24/136), with 34 instances in 785 manuscripts (4.3%). Manuscripts classified as published were verified 96.7% of the time, were not found in 1.4%, and had incorrect authors or journal in less than 1% each. "In press" manuscripts were verified 88.3% of the time, 6.4% could not be found, and 4.3% had an incorrect journal listing. Number of publications (P = 0.026) and first author publications (P = 0.037) correlated with misrepresentation. None of the remaining variables was significant. CONCLUSIONS: The pediatric surgical pool has a very low incidence of suspicious citations; however, authorship claims should be verified.