Abstract Background A high frequency of Roche Cobas delta check instrument errors for alkaline phosphatase (ALP) led to the discovery of non-reproducible and spurious ALP results in a tertiary care hospital. A series of studies were subsequently performed on different Roche instruments at multiple hospital laboratories across Alberta, Canada, to determine the root cause and develop workflow mitigation strategies. Methods Three studies were implemented across different hospital laboratories to quantify frequency and identify cause of ALP fliers. ALP was measured on the Roche Cobas Pro c503 at 12 sites and on Roche Cobas 8000 c702 at 1 site. Replicates that exceeded the total allowable error (TAE) of =150 U/L ± 15 U/L or 150 U/L ± 10% were considered discrepant. Firstly, ALP samples run in duplicate and those exceeding the TAE were aliquoted, respun and reanalyzed in for 1-9 months. In the second study, ALP collections in BD Vacutainer plasma separator tubes (PST) were switched to BD Vacutainer Barricor tubes, which produce cleaner plasma, and run in duplicate for 1 month. Lastly, ALP was diluted (5x) and analyzed in duplicate. If TAE was exceeded then neat samples were aliquoted, respun and rerun. Results The first phase identified 563 ALP repeats outside the TAE across the Pro sites but none on the 8000. Frequency of spurious results varied at sites and ranged from 0.3-12/week and 0.04-1.5% of total ALP testing volumes. Initial duplicate comparisons had an overall mean difference (standard deviation, SD) of 15.1% (38.2%), and difference range of -85 to 270% across sites. Respinning significantly improved comparison results, with the percent mean (SD) decreasing to 0.3% (3.2%) and a narrower difference range of -13.5 to 41%. Barricor tubes reduced overall ALP flier rate by 72% and 5x dilutions reduced them by 89%. Conclusion Clinically significant ALP discrepancies were identified on multiple Roche Cobas Pro instrument across a large provincial laboratory network. These fliers were not observed on the Roche Cobas 8000, and were largely eliminated after respinning, dilution, and use of Barricor tubes, suggesting a preanalytical cause. This also suggests that the Pros may be more prone to interference from plasma contaminants. An automated workflow protocol involving a combination of onboard dilution of all ALP samples with respin only if TAE is exceeded was developed as a cost-effective and streamlined approach to minimize these errors.
Raizman et al. (Wed,) studied this question.