In-depth evaluation of automated non-contact reflectance-based hematocrit prediction of dried blood spots.

Abstract

Dried blood spot(s) (DBS) microsampling has increasingly attracted interest as a patient-centric alternative to conventional blood withdrawal. Despite the many advantages associated with DBS sampling, its widespread use in clinical practice is still hampered, which is mainly caused by the hematocrit (Hct) effect. One approach to cope with this issue is the Hct prediction of DBS using ultraviolet-visible (UV-Vis) spectroscopy. Recently, a UV-Vis-based Hct prediction module has been incorporated into the automated CAMAG® DBS-MS 500 HCT system. However, although a proof-of-principle yielded promising results, there is no formal in-depth evaluation of the performance of this module. Hence, it remained to be established to what extent automated Hct prediction of DBS via this module can universally be applied and generates acceptable results. Using authentic patient samples, we set up and validated a calibration model and evaluated whether this could serve as a 'generic' calibration model for different, independent Hct prediction modules. A quadratic calibration curve with 1/x2 weighting was established. The bias, intra-day and total precision were below 0.025 L L-1, 2.2% and 2.7%, respectively. Additionally, the influence of storage and the robustness of the method was evaluated. Moreover, a lab-lab comparison of the performance of the Hct module of two independently operated instruments demonstrated that the validated model can be used as a generic calibration model. Finally, application of the method to venous DBS (n = 48) prepared from patient samples in the context of therapeutic drug monitoring of tacrolimus revealed a good concordance between the actual (i.e. Sysmex-based) and UV-Vis-based predicted Hct.