Error analysis of one sample clearence methods
Dear Sir,The interesting error analysis of one-sample clearance methods by Kotzerke et al. (1992) should not obscure the fact that the errors are small when compared with other methods for measuring renal function that are in widespread clinical use, such as creatinine clearance or blood-free gamma-camera methods.The first single-injection, single-sample method of which we are aware was described by Alestig, H o o d and Vikgren (1966) using unlabelled inulin. A one-sample method using radiolabelled orthoiodohippurate (OIH) was described by Heiskanen, Weber and Graesbeck (1968). The method of Tauxe, discussed at length by Kotzerke et al. (1992), followed in 1971. Since then m a n y similar papers have appeared (our files contain 27 such papers) using not only these agents but also chromium-51 ethylene diamine tetra-acetic acid (EDTA), ytterbium-169 diethylene triamine penta-acetic acid (DTPA), technetium-99m DTPA, iodine-131 diatrizoate and 99mTc mercaptoacetyl triglycine (MAG3).By now the capabilities of this approach seem well established on experimental grounds. Theoretical arguments such as those of Kotzerke et al. must be interpreted in the context of this body of experimental evidence. Patients with oedema or ascites are usually excluded from such studies. We find no mention of such exclusion in the paper by Kotzerke et al., which may explain their outliers. One or two outliers in a series of 150 patients would not in any case be too surprising, and the correlation coefficient of 0.966 [including the outliers, and covering a range of effective renal plasma flow (ERPF) from 50 to 800 ml/min] indicates an adequate performance for clinical use over a wide range of renal function. Kotzerke's data therefore do not justify his conclusion that the one-sample method only holds for normal patients.The statement by Kotzerke et al. that °'The individuals which were used in the study by Tauxe (Tauxe et al. 1971 represent healthy adults of normal weight and kidney function" is simply incorrect. The 1982 paper included a range of E R P F from 0 to 900 ml/min, and patients were not selected for size. It is an empirical fact, though perhaps counter to intuition, that patient size has little effect in applying this method to adults if the sample time is chosen appropriately. (The optim u m sample time of 44 min for O I H serves to minimize the effect of patient size for intermediate values of ERPF.) We have examined this in several ways, but not until we included children in the study population (Rus-sell et al. 1991) did we find that scaling for size had a decisive influence on the results. We have described two-sample methods (Russell et al. 1985(Russell et al. , 1989a for use in research and under other circumstances when special accuracy is needed but continue to use one-sample methods for routine patient care. They are not without error, but in our opinion the errors are acceptable.The insights provided by Kotzerke et al. should be regarded as an aid in interpreting the on...