Errors in measurement of glomerular filtration rate using the slope-intercept technique and their identification.

Background: GFR measured from plasma sampling may be expressed as slope-intercept GFR (SI-GFR) and scaled to body surface area (mGFR/BSA) or as GFR per unit extracellular fluid volume (mGFR/ECV), which is based only on half-time. Measurement errors comprise 3 categories. Pre-injection error arises from error in administered marker and is suspected when mGFR/BSA and mGFR/ECV disagree. Injection errors include 'tissued' injections. Post-injection errors include inaccurate sample timing, inaccurate pipetting, sample haemolysis and sampling through long IV lines through which marker was administered. The aim of the study was to evaluate the impact of errors on mGFR.

Methods: We compared mGFR/BSA with mGFR/ECV in 898 patients undergoing routine investigation. To investigate post-injection error, we took two further patient datasets with r values (correlation coefficient of the 3-sample fit) of 1.0 and introduced errors, in isolation, into each of the 3 recorded sample values, as follows: pipetting (volume) errors of -20%, -10%, -5%, 5%, 10% and 20%, and timing errors of -15 min, -10 min, -5 min, 5 min, 10 min and 15 min.

Results: The correlation between mGFR/BSA and mGFR/ECV was close and independent of r. Post-injection error depended on the time of the sample in which it occurred. r correlated poorly with error magnitude for both volume and timing errors. When a 'rogue' sample is suspected its error needed to be substantial for it to be identified by single sample estimates applied to the other samples.

Conclusion: SI-GFR is resistant to post-injection timing and volume errors but not to pre-injection error.