Assessment of publication bias in the reporting of EUS performance in staging rectal cancer

INTRODUCTION: Endoscopic ultrasound (EUS) has emerged as a promising diagnostic modality for locoregional staging of rectal cancer. However, as with any new technology, publication bias, the selective reporting of studies featuring positive results, may result in overestimation of the capability of EUS. The aim of this study was to systematically assess for publication bias in the reporting of the accuracy of EUS in staging rectal cancer. METHODS: A MEDLINE search for all published estimates of EUS accuracy in staging rectal cancer between 1985 and 2003 was performed. All retrieved studies were fully published in the English literature. Published studies were analyzed and the following information was abstracted: accuracy of EUS, year of publication, number of subjects studied, impact factor of journal, and type of journal (gastroenterology, surgery, radiology, other). RESULTS: Two hundred and two abstracts were reviewed; 41 publications met the stated criteria for inclusion. EUS T-staging accuracy was reported in 40 studies while EUS N-staging accuracy was reported in 27 studies. The experience of 4, 118 subjects was reported with an overall mean T-staging accuracy of 85.2% (median, 87.5%) and N-staging accuracy of 75.0% (median, 76.0%). There was a paucity of smaller studies expressing low EUS accuracy rates. Both T-staging and N-staging accuracy rates also declined over time with the lowest rates reported in more recent literature. CONCLUSION: The performance of EUS in staging rectal cancer may be overestimated in the literature due to publication bias. This inflated estimate of the capability of EUS may lead to unrealistic expectations of this technology.