Comparison of representative ranges based on US patient population and literature reference intervals for urinary trace elements

Reference intervals for trace elements are very hard to obtain because of t he difficulty of defining a nonexposed reference population. However, repre sentative ranges for trace elements obtained from a general patient populat ion can provide useful information in interpreting laboratory results. We h ave used urine specimens submitted for trace metal analysis from patients r esiding in the United States to calculate representative ranges for 25 urin ary trace elements, and to compare them to reference values taken from the literature. All urine analytes were measured by inductively-coupled plasma- mass spectrometry except chromium, which was measured by graphite furnace a tomic absorption spectroscopy. For representative range calculation two app roaches were used. In the non-parametric calculation first, the top 10% of results were discarded assuming that those specimens came from individuals with unusually high trace element exposures. Next the central 95% of the re maining data was taken as the reference interval. In the parametric calcula tion the specimens from exposed or not healthy individuals were assumed to appear as outliers and were discarded. The mean and S.D. were calculated, a nd used to determine representative ranges. The two approaches yielded very similar results, and worked remarkably well for 14 analytes. There were mi nor discrepancies for 7 analytes, and major for 4 analytes. All analyses of urinary trace elements included a urine creatinine value, which was used t o express urinary trace element concentrations in terms of creatinine ratio . This corrects for differences in urine concentration that affects the res ults Tor random specimens. (C) 2000 Elsevier Science B.V. All rights reserv ed.