Background Selective COX-2 inhibitors (coxibs) have been shown to be advantageous for gastrointestinal safety in several large studies but whether this translates into fewer events at the population level is unclear.

Aim To examine the relationship between prescriptions for nonselective NSAIDs, selective COX-2 inhibitors (coxibs), H2-receptor antagonists (H2RA) and protein pump inhibitors (PPI), and hospital separations for upper gastro-intestinal haemorrhage (UGH) in Australia, over the period 1998–2002.

Methods Interrupted time series study. Monthly hospital separations for UGH for Australia for the period July 1998 to June 2002 were obtained from the Australian Institute for Health and Welfare. Data on prescriptions for coxibs and non-selective NSAIDs, PPIs and H2RAs were obtained from commercial sources for Australia for the same period. Rates of UGH per 1000 prescriptions for all NSAIDs were modelled by Poisson regression. The period before the introduction of coxibs (1998–1999) was compared with the period after their introduction (2001–2002).

Results Rates for UGH per 1000 prescriptions for all NSAIDs significantly declined between the two periods. Adjusting for prescriptions for PPIs and H2RAs improved the fit of the model and after adjustment, the reduction in UGH separation rates per 1000 prescriptions for all NSAIDs was estimated to be 9.5% (95% CI 9.1% to 10.9%).

In this presentation, we will also provide extended results covering the period January 1997 to October 2010, and also including sales of anti-ulcerants and other non-steroidal anti-rheumatics in the model.