Stability of thiopurine metabolites: a potential analytical bias.

(URDL) specific for each category used in our laboratory (Table 1). Of the 49 824 creatinine results reported, 15 686 (or 31.5%) exceeded their respective URDLs. The range of analytical error attributable to miscompensation was greatest in the pediatric age categories 2 (0 –1 month) and 3 (1 month to 5 years) (e.g., errors of 29% to 24% in group 2) for which the URDLs are lowest, and albumin concentrations are usually lower than adult values. This magnitude of error invalidates use of the single-value protein compensation method for these age groups. In patient categories 1 and 4, in which URDLs were higher, the range of error observed was still unacceptably high, as indicated by the large proportions of error 3.4% (namely 97% and 38%). In the adult categories 5, 6, 7, and 8 the proportion of error 3.4% were lowest, but the range of error encountered was still too high for ensuring reliable creatinine results for clinical practice. The data presented here indicate unacceptable inaccuracy in creatinine results due to variations in albumin concentrations. Hence, the single-value compensated creatinine Jaffe assay cannot achieve the current desired performance goal of 3.4% bias as recommended by the Laboratory Working Group of the National Kidney Disease Program (1 ), nor the recommended optimum goal of 1.7%. Awareness of this limitation may be helpful in setting clinical guidelines for result interpretation and in making future choices on methodology.