Randomised controlled trials in nephrology: are more and better needed?

Craig J, Wheeler D

Abstract: This study aims to evaluate the quantity of randomised controlled trials (RCTs) in nephrology compared with other medical specialties, the quality of these trials, and to ascertain if an even coverage of RCTs exists across the major nephrology disease categories. Medline (1966-98) was searched to determine the total number of citations and number of RCTs for 13 major specialty-specific subheadings, including kidney diseases. The proportion of citations which were RCTs was then calculated (RCT index). To evaluate the equality of RCT coverage across nephrology, a standard classification of kidney diseases was cross-referenced with the Medline search. Identified RCTs were evaluated for quality using standard criteria. Of the 212,978 citations for kidney disease, 2047 were RCTs (0.96%). This was the lowest RCT index of all 13 major specialties evaluated. The RCT index has increased from 0.14 to 1.8 in the past 30 years (but at a slower rate than other specialties) and is now falling. 33% of nephrology disease categories have no RCTs. Of the major disease categories, the RCT index was above average for urinary tract infection and kidney transplantation and below average for glomerulonephritis. Quality of trials has improved since 1966, but remains suboptimal. In 1998, 40% of trials examined inadequately reported allocation concealment, 20% did not state blinding of outcome assessors, 20% did not conduct or report intention-to-treat analysis and 40% had greater than 20% loss to follow-up. In conclusion, nephrology research has a poor track record of the quantity and quality of RCTs published. Many diseases, particularly glomerulonephritis, have few or no high quality RCTs on which to base treatment decisions. Better mechanisms are required to ensure that more RCTs are done, and done well, in nephrology.