External quality assessment of total urinary protein estimation in the United Kingdom.

Three surveys of total urinary protein quantitation have been carried out in 350 UK laboratories. The seven specimens comprised buffered saline or normal human urine with added human serum albumin or human serum, or urine from individuals with nephrotic syndrome. Principal method groups were: turbidimetry (57%), dye binding (25%) and biuret (15%). For all surveys, overall between-laboratory agreement was poor (CV 22.8% to 57.1%), with ranges of results from 24-fold (0.83-20 g/L) to 366-fold (0.05-18.3 g/L); there was no improvement with time. The most popular method (sulphosalicylic acid turbidimetry) consistently performed the worst, and performance of the direct biuret procedure was also unacceptable; both methods should be discontinued. There were no significant differences in performance between the other major method groups, and none can be specifically recommended. Within the individual calibrant groups, least variation was observed with human serum. A common calibrant for all participants yielded significantly better between-laboratory agreement for all methods except sulphosalicylic acid turbidimetry.