 |
 |
From: Richard Hancock <[log in to unmask]> >In my very limited experience of this sort of thing, I think the idea is >to use the figures as a stimulus for reviewing practice. It is therefore >not just a question of "How many do I (or my colleagues) request?", but >"Why do I request them?", "What are the evidence-based guidelines for >requesting this test?", "How does my practice compare with the >guidelines", etc. A common example would be use of lumbar spine x-rays, >which seem to be commonly over-used. But at what point do the figures support a contention that the physician whose data is being reviewed is a member of a different population (eg overusers of lumbar Xrays for instance) from the physicians who are inthe middle of the distirbution, or at the far end? If such figures are used to present a snapshot of what happened last year, then statistics is unimportant, description is sufficient. However if they are presented as indicating - clinical governance - poor performance then some statistical validity seems called for, and if the mechanism for enforcing good behaviour is financial - differing amounts of public money being applied to the doctors at one end of a distribution from tthose at the other - then it is a matter for public interest, and potentially for the Courts. Some assessment of whether the differences are _evidence_ and if so of what and to what extent would be useful here, and appears to be beyond the statistical nous of our local figure producers. One factor to throw into the ring of course is that if figures are generated for practices rather than individuals, as is common in th eUK to "avoid pointing at individuals"[1] then small sample effects will show up - the variance for small practices with 1-4 doctors and 1500 - 10000 patients will be greater than the variance for large practices with 14 doctors and 30000 patients, who can be expected automatically to be closer to the mean regardless of the nature of the doctors in the practices. But as with everything else, vague ideas are easy to come by and what is needed is statistical rigour. [1] speaking as a single-handed GP I say "Ha!" but maybe my assistant is doing all that xraying and prescribing<g> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%