The reproducibility of measurements of differential renal function in paediatric 99mTc-MAG3 renography

The aims of this study were to establish the reproducibility of estimates of differential renal function (DRF) obtained using the software supplied by different vendors, assess the effects of age, glomerular filtration rate (GFR) and degree of asymmetry of renal function on reproducibility and ascertain whether the software gives the same estimates of DRF.

A stratified sample of 172 renograms covering a wide range of DRF estimates, age and renal function was drawn from an electronic archive containing raw data from 1416 renograms. The renograms were processed by one operator using seven different methods, five times for each method. For each renogram and each method the DRF for the left kidney and difference between the maximum and minimum of the five estimates of DRF were calculated.

There were differences in reproducibility among the seven methods [Friedman analysis of variance, χ(2)(N=172, d.f.=6)=367.0, P=0.0001]. Reproducibility was good with all methods in most children. The appreciable minority in whom reproducibility was not as good tended to be less than 6 months old or had GFRs below 90 ml/min/1.73 m(2) or both. The median of the DRF estimates of the left kidney obtained using the seven methods differed [Friedman analysis of variance, χ(2)(N=172, d.f.=6)=284.7, P<0.0001]. The largest difference between any two methods was 5%.

Although reproducibility was good in the majority of children, an appreciable minority showed poor reproducibility, which could impact clinical decision making. It is essential that these studies be identified. This can be done by processing each renogram several times, ideally using different methods. Those with poorer reproducibility tended to be less than 6 months old or to have a low GFR.