To investigate the consistency of glomerular filtration rate (GFR) calculation from plasma sampling in the UK.
Ten patients' data sets from plasma sampling measurements of GFR were distributed throughout the UK. The data included count rates from four samples taken between 2 and 4 h after injection, a diluted sample of injected dose for standardisation, the patient's height, weight, age and sex. Participants were asked to use the routine method to calculate GFR and express the results in absolute terms (i.e. in millilitres/minute) and normalized for body surface area (ml/min/1.73 m2). Supplementary data were also requested relating to workload, method used and normal range. Intercentre variability was assessed by calculating the root median square (RMedS) deviation of each GFR from the median for that data set. Centres using a particular analysis method were grouped together and the RMedS deviation of each result from the median for that group and that data set was calculated. The influence of using normalized data and number of samples was also studied.
Seventy-nine returns were received. For the normalized data, the overall RMedS variability was 5.8 ml/min/1.73 m2. This decreased significantly to 0.6 ml/min/1.73 m2 when results were grouped by analysis method. Results were similar for non-normalized data. A small but significant decrease in error with the number of samples was observed.
Considerable variability in GFR values obtained at different centres in the UK for a given set of data was observed. Nearly all this variability was due to different methods of analysis. If methodology were standardized then intercentre variability in GFR analysis could be reduced dramatically. Radionuclide techniques are confirmed as being the method of choice if an accurate value of GFR is required.