Automation of chromatographic peak review and order to result data transfer in a clinical mass spectrometry laboratory

Introduction: Mass spectrometry-based assays have increasingly been implemented in clinical laboratories for their multiplexing capacity and high specificity and sensitivity. However, these methods are often associated with labor-intensive and error-prone data-related workflows, due to the volume of data generated that is often manually reviewed and resulted. We aimed to establish a system within our clinical mass spectrometry laboratory to facilitate data flow from electronic medical record order to result and to automate processes for chromatogram peak review. The processes and validation are described for a 25-hydroxyvitamin D assay. Methods: Automating chromatogram review and order to result data transfer required flat file interfacing, file transfers of standardized data formats, barcode scanning, and software for peak processing and review. Validation of the automated workflow involved (1) correlation of quantified results generated by two chromatogram analysis methods: Waters TargetLynx and Indigo Bioautomation ASCENT, (2) manual verification of quality assurance flags applied in ASCENT, and (3) testing data flow and integrity across all the systems from order to result. Efficiency and quality improvements were assessed through calculation of batch review times and rates for autoverification and manual manipulations. Results: The correlation of TargetLynx and ASCENT quantitation methods for 25-hydroxyvitamin D2 in patient samples yielded slope of 0.99 (95% CI: 0.989 to 0.996), intercept of 0.46 (95% CI: 0.363 to 0.565), with r = 0.999. The correlation for the D3 fraction showed Deming regression slope of 0.98 (95% CI: 0.969 to 0.989), intercept of 0.06 (95% CI: -0.115 to 0.313), and r = 0.995. Results from both quantitation approaches were also compared to the assigned value in CDC reference samples. The mean bias relative to the CDC was 4.6% for ASCENT and 2.5% for TargetLynx. The median time for chromatogram review of a full 96-well plate of vitamin D results is reduced from approximately 2 h to 14 min and 80% of batches were reviewed within 30 min. Instead of 100% peak review, technologists review only the peaks that have been flagged by the system based on applied rules. Analysis of full plate batches showed that 220% of peaks per batch were flagged for manual review. Manipulations made by technologists during chromatogram review were reduced by 75% when using the automated versus manual system. Conclusions: We describe a system to facilitate data flow from electronic order to result and to automate chromatogram peak review in a clinical liquid chromatography mass spectrometry assay for 25-hydroxyvitamin D. This eliminated manual result entry, repetitive transcription, and unnecessary review of high quality data while enabling systematic evaluation of data quality indicators. The new processes were accurate, improved the data review and processing times, and helped to reduce manual manipulations during chromatogram review.