Garcia Rodriguez et al (Gut 2006;55:1538–44) use data from the General Practitioner Research Database (GPRD) to report that the odds ratios (ORs) for incidences of non-cardia gastric cancer were increased fourfold in those who received acid suppressant treatment for ulcer indications, compared with those who did not, and fivefold for oesophageal adenocarcinoma in those with oesophageal indications. They consider that the increased risks reflect the indications for treatment and not drug toxicity. Although their conclusions may be right, their data raise some important concerns.

The GPRD has particular strengths in recording prescriptions, but case details can be less complete, which may explain why 482 of the oesophageal cancer cases were excluded because of unspecified histology, leaving 287 with adenocarcinoma for comparison with their controls without cancer. The assumption that these …