1. starters showed that diclofenac was used most commonly in 1987-1988, followed by naproxen, ibuprofen, indomethacin, and piroxicam. Ketoprofen was used in 6-2% of starters and in 2-5% of the reference group
2. The proportion of non-switchers was similar in starters and in the reference population. However, starters appeared to have a background of higher rates of intensive switching among individual NSAIDS. Among starters, 32110 of patients switched in 1987-1988 to more than two different NSAIDS. The background rate was;1-38-1-84, CI
3. RR= 1-61,1
4. A total of 24-10 of starters had received peptic ulcer therapy in 1987-1988, versus 15 7%/" of the reference population. The rate ratio was;1 54, with
5. It was a conservative manoeuvre to exclude from the analysis the 30 000 one time recipients of an NSAID prescription during 1987-1988. When these patients' data were kept in the analysis, the rate ratio