unique criterion for all studies. Meta-analysis is partiIntervention-Difference in protein intake cularly suitable for studies of rare events in, for between control and treated groups of at least 0*2 g example, ponderous or longlasting trials that do not protein/kg/day. include many patients.9 The quality and impact of a Main outcome measure-Number of renal deaths meta-analysis depends on the selection process for the (the necessity to start dialysis or death of patient criterion, which must be well defined at the beginning during study).of the meta-analysis, and easily collected by nonResults-156 renal deaths were recorded, 61 in the specialist observers. low protein diet group and 95 in the control group, Definition ofa common event-We defined renal death leading to an odds ratio of low protein to control of as either the necessity to begin dialysis or the death of 0-54 with a 95% confidence interval of 0 37 to 0 79. patients during the study. Patients receiving grafts Conclusions-This result, obtained on a large before starting dialysis were counted as having had population of patients suffering from chronic renal renal death. Patients who were lost to follow up or who insufficiency, strongly supports the effectiveness of stopped taking the diet were not counted as having had low protein diets in delaying the onset of end stage renal death. These criteria were applied independently renal disease.of the randomised treatment by the authors on the basis of published data or complementary data kindly provided by the investigators.
IntroductionCollection of data and study selection-To avoid During the past 10 years a lot of experimental data omission of results from less well known studies we have suggested that dietary protein restriction may scanned research reports with a computer search of retard or even halt the development of non-specific Medline files and abstract books from the International glomerular lesions and hence the