Is a low-cost latex immunoturbidimetric assay suitable for routine 25(OH)D screening? Analytical comparison with chemiluminescent immunoassay and LC-MS/MS

Low-cost latex immunoturbidimetric 25-hydroxyvitamin D [25(OH)D] assays for clinical chemistry analysers have recently been introduced, but their agreement with reference methods at clinical decision limits is undocumented. We compared one such assay with a chemiluminescent immunoassay (CLIA) and liquid chromatography–tandem mass spectrometry (LC–MS/MS). Serum from 42 osteoporosis-clinic outpatients was analysed in parallel by the latex assay (Beckman Coulter AU5800), the Access 25(OH) Vitamin D Total CLIA (DxI 800) and LC–MS/MS. Imprecision was verified according to CLSI EP15-A3; agreement was assessed by Passing–Bablok regression, Bland–Altman analysis, a ± 25% total-error criterion, and diagnostic accuracy and kappa at the 50 and 75 nmol/L decision limits, with LC–MS/MS as reference. Within-laboratory CVs of the latex assay were 5.17% and 1.84% at 70 and 116 nmol/L. Against LC–MS/MS the Passing–Bablok slope was 0.79 (95% CI 0.71 to 0.90), mean bias −7.4 nmol/L (−9.1%), limits of agreement −33.6 to 18.8 nmol/L; 81.0% of results were within ±25% of LC–MS/MS (CLIA 71.4%). For deficiency (<50 nmol/L) sensitivity was 100.0% and specificity 86.7%; three-category agreement was 66.7% (kappa 0.51), all discordant results falling into a lower category. The assay is precise and detected every LC–MS/MS-defined deficient sample, but a negative proportional bias of 12–15% at the decision limits makes it over-call insufficiency. It may serve as a rule-out screening test if results near the decision limits are confirmed by LC–MS/MS, but it is not interchangeable with LC–MS/MS and its calibration requires standardisation before routine use.