Laboratory performance in albumin and total protein measurement using a commutable specimen: results of a College of American Pathologists study.

CONTEXT

Discrepant results for serum constituents were observed among peer groups in the College of American Pathologists Comprehensive Chemistry Survey.

OBJECTIVES

To assess the performance of serum albumin and total protein measurement procedures and to evaluate the commutability of the conventional survey specimens.

DESIGN

A fresh frozen, off-the-clot serum sample was included along with 4 conventional survey specimens. The fresh frozen, off-the-clot serum sample was prepared in a manner expected to confer commutability with native clinical samples.

RESULTS

For the fresh frozen, off-the-clot serum sample, the mean values for 17 peer-groups were -0.07 to 0.32 g/dL from the bromocresol green albumin designated comparison method, whereas 4 VITROS (Ortho Clinical Diagnostics, Rochester, New York) peer groups differed by -0.29 to -0.37 g/dL (15 of 21 differences [71%] had P < .001). For bromocresol purple albumin methods, the mean differences from the designated comparison method from 8 peer groups were 0.25 to 0.47 g/dL (all had P < .001). For total protein methods, 23 peer group mean values were -0.07 to 0.15 g/dL from the reference measurement procedure (12 of 24 [50%] had P < .001). The Beckman (Fullerton, California) Synchron LX20 had a bias of -0.30 g/dL (P <.001). The commutability of the conventional specimens was acceptable for 23 of 24 bromocresol green method-material combinations (96%) and 13 of 16 bromocresol purple albumin method-material combinations (81%). All (100%) of the 36 method-material combinations had acceptable commutability for total protein.

CONCLUSIONS

One (2.2%) of the instrument systems (Synchron) using bromocresol green and none (0%) of the instrument systems using bromocresol purple had satisfactory total-error performance for albumin measurement. Differences in results between bromocresol green and bromocresol purple methods precluded using common reference intervals for interpreting results for serum albumin. Eight of 9 instrument systems (86.5%) had satisfactory total-error performance for total protein measurement.