Authors should not have included data from one study

Editor–I wish to draw attention to an important omission in the report of a meta-analysis of observational studies of serious gastrointestinal complications of non-steroidal anti-inflammatory drugs.1 The paper was based on both published and unpublished data, and this process was assisted by a workshop held in September 1993 in Newcastle, Australia. This was attended by many of the authors of the original studies. Among those attending was Dr David Kaufman, from the Slone Epidemiology Unit at Boston University School of Medicine.

Kaufman and his colleagues had previously published an international study on major gastrointestinal complications of non-steroidal anti-inflammatory drugs.2 At the workshop Kaufman presented unpublished data from a follow up of their original study. Kaufman and his colleagues were not prepared to be coauthors of the report arising from the workshop because they objected to the use of meta-analysis during systematic reviews of non-randomised studies. However, they agreed to provide access to their data on the understanding that they would write a commentary to accompany the final report, which would outline their concerns. I undertook to facilitate this, but during a protracted and complicated analysis and …