High analytical variability exists with multiple methods of serum 25-hydroxyvitamin D analysis in dogs

Abstract

Vitamin D deficiency, defined as low serum 25-hydroxyvitamin D (25D), is a highly prevalent finding in dogs with several disease states. Significant analytical variability between 25D methods of analysis in humans is well-documented; it is unknown if this variability exists in dogs. Our primary goal was to evaluate agreement of 2 chemiluminescent immunoassays (CLIA1, CLIA2); 2 liquid chromatography with tandem mass spectrometry (LC-MS/MS1, LC-MS/MS2) assays, from 3 laboratories; and a point-of-care lateral flow assay (LFA) across a wide range of expected 25D results in dogs. The null hypothesis was that all methods of analysis would have clinically acceptable agreement (<25 nmol/L difference). The tests were assessed for agreement via Bland–Altman analysis, Passing–Bablok regression, and Lin correlation coefficients. In Bland–Altman analysis, all inter-test comparisons exceeded the a priori clinically acceptable ≤25 nmol/L mean difference, except for LC-MS/MS1 vs. CLIA1 (mean difference −7.2 nmol/L). The range of mean bias across all test comparisons was −186% (LC-MS/MS2 vs. CLIA2) to 254% (CLIA2 vs. LFA). Tests of comparable methodology did not produce more similar results. The mean bias between the 2 CLIA assays was 54%, and the mean bias between the 2 LC-MS/MS tests was 125%. The significant analytical variability of canine 25D tests suggest that they cannot be used interchangeably, and caution should be used when comparing 25D results presented in canine studies from different laboratories and methods of analysis.