The contents of 3 orthopaedic journals [JBJS (Am), JBJS (Br) and CORR] during 2001 and 2011 were compared for publication bias.

There were total of 2028 articles. After exclusion 1662 scientific articles were analysed for statistical results, clinical conclusion, sub-speciality topics studied, the geographical region the study been conducted and the statistical method used. The articles classified into 7 categories: THR, TKR, Basic sciences, Trauma, Spinal disorders, Paediatric disorders and Tumour.

91% of articles on THR and 95% of articles on TKR were positive studies in 2001. Articles dealing with trauma had the lowest proportion of positive studies (74%) as compared to all other topics. We noted that JBJS (Br) published more negative studies as compared to JBJS (Am) and CORR. In 2011 less articles on THR and TKR had positive studies (68% and 76% respectively). Spinal surgery articles report less number of non significant studies nowadays (24% in 2001 and 2% in 2011).

There is a significant change in the trend towards reporting more negative studies in relation with THR and TKR (p < 0.05). Articles dealing with Basic sciences, Trauma, Paediatric disorders and Tumour did not have any significant change in reporting negative studies in the last decade. Significant findings in spinal disorders were 3.8 times more likely to be published than non significant stdies. Overall, JBJS (Br) continued to publish more negative studies as compared to JBJS (Am) and CORR.

Journals seem to prefer reporting more significant results with spinal disorders and more non significant results in relation with Hip and Knee arthroplasty in last ten years. This might be because of authors' perceptions of the importance of their findings and journals preferences for significant results.