What's new in the other general journalsBMJ 2005; 331 doi: https://doi.org/10.1136/bmj.331.7513.369 (Published 11 August 2005) Cite this as: BMJ 2005;331:369
- Christopher Martyn, associate editor (email@example.com)
How information on the performance of diagnostic tests is presented doesn't seem to matter
There's something of an industry among evidence based medicine experts in devising ways to present to practising clinicians numerical information on the performance of diagnostic tests. How useful they are in everyday practice is another matter. In a randomised experiment at a continuing medical education conference, 183 physicians were asked to estimate probabilities of common illnesses from vignettes of six patients. Each vignette described the symptoms and risk profile of the patient and contained the result of a commonly used diagnostic test. The accuracy of the test was described either in terms of sensitivity and specificity or as a likelihood ratio or as a simple graphic representation of the likelihood ratio. Participating doctors were randomly allocated to receive one of these descriptions.
Differences in estimates of the probability of disease between the groups assigned to different presentations were very small after the test. The authors admit that their findings are preliminary but think that presenting test performance as a likelihood ratio does not have much influence on the way doctors interpret the results. Another possibility, of course, is that clinicians are so familiar with the business of applying the results of diagnostic tests to patient management that they simply do not need these heuristics.
Mortality in people with end stage renal disease depends on method of dialysis
People with end stage renal disease may receive treatment with either peritoneal dialysis or haemodialysis. Which is better? In the absence of data from randomised controlled trials, we cannot know for sure, but follow-up of a large group of such patients found that mortality was higher in those receiving peritoneal dialysis, although the difference became statistically significant only during the second year of treatment.
More than 1000 patients at 81 centres in 19 states of the United States took part. A quarter of the patients receiving peritoneal dialysis and 5% of those receiving haemodialysis switched methods …