Yonah C Ziemba, Suhyeon Yoon, Harvey W Kaufman, William A Meyer, Laura Gillim, Nkemakonam Okoye, Cheryl B Schleicher, Shahidul Islam, Cristina P Sison, Ligia A Pinto, Lynne Penberthy, James M Crawford
Complex and heterogeneous test name and result conventions can represent a significant barrier to understanding the clinical relevance of real-world laboratory data. The generalizable methodology described herein permitted successful harmonization of SARS-CoV-2 diagnostic and serologic test data from multiple laboratory sources with administrative claims data, vital status data, and structured data on cancer diagnostics and treatment. This methodology supports exploration of the clinical utility of semiquantitative serologic data, especially in patients with immunocompromising conditions.
BACKGROUND: The COVID-19 Real World Data infrastructure (CRWDi) was established to study the impact of coronavirus disease 2019 (COVID-19) on patients with immunocompromising conditions. A key challenge was harmonizing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) laboratory test results.
METHODS: There were 27 different test names for SARS-CoV-2 nucleic acid amplification tests (NAATs), and there were 34 and 26 for SARS-CoV-2 qualitative and semiquantitative serologic tests, respectively. We validated a strategy to eliminate the multiplicity of test names and results, and harmonized SARS-CoV-2 semiquantitative serology test results by target antigen and antibody using published conversion factors to report Binding Arbitrary Units (BAU) on a uniform scale.
RESULTS: For 5 200 000 patients, the numbers of unique SARS-CoV-2 test events were 4 865 431 NAAT, 3 092 198 qualitative, and 834 487 semiquantitative serology tests, of which 378 522 were antispike (anti-S) semiquantitative serology. In achieving harmonization of semiquantitative serologic results, semi-log cumulative ordinal plots demonstrated that test results from patients with transplantation exhibited a significantly higher proportion of negative-and-submaximal SARS-CoV-2 anti-S BAU values when compared with patients with cancer or systemic autoimmune or rheumatic diseases (SARDs) or with a general nonimmunocompromised population.
CONCLUSIONS: Complex and heterogeneous test name and result conventions can represent a significant barrier to understanding the clinical relevance of real-world laboratory data. The generalizable methodology described herein permitted successful harmonization of SARS-CoV-2 diagnostic and serologic test data from multiple laboratory sources with administrative claims data, vital status data, and structured data on cancer diagnostics and treatment. This methodology supports exploration of the clinical utility of semiquantitative serologic data, especially in patients with immunocompromising conditions.