The Vitamin D Standardization Program (VDSP) Manual for Retrospective Laboratory Standardization of Serum...

Low concentrations of total 25-hydroxyvitamin D [25(OH)D], the principal biological measure of vitamin D status, have been associated with clinical and public health outcomes. The determination of levels under which there is an increase in the risk of disease, as well as comparisons across populations, have been difficult to establish due the large assay variability in measuring 25(OH)D. Accordingly, the Vitamin D Standardization Program (VDSP) includes the retrospective standardization of existing 25(OH)D values collected by epidemiological and clinical studies, as well as clinical trials, as one of its main objectives. We introduce methodology developed by the VDSP that can be used to standardize the measurement of time-stable analytes, including 25(OH)D, in samples that have been banked and maintained appropriately. Sample size estimation formulae are first applied to calculate the required number of banked blood samples to be reanalyzed using either of two approaches. In the first approach, existing samples are remeasured using the current measurement procedure, and an equation relating "old" to "current" measurements is obtained. A second set of sera, usually 40-50 single-donor serum samples, are measured with the current measurement procedure and an assay traceable to a reference measurement procedure and/or certified reference materials, which yields a second calibration equation. These two equations are combined to produce standardized levels from the original old values. This approach is necessary when study restrictions prevent serum samples from being shipped to an external laboratory and is illustrated with samples from the Canadian Health Measures Survey. When serum samples are permitted to be shared with other laboratories, or the study investigators can carry out the measurements with a traceable assay, a single calibration equation method is used. Existing samples are selected and remeasured using the available traceable assay. We outline the statistical theory supporting the VDSP protocol and provide implementation examples. The methods proposed are generalizable to any instance in which banked specimens have been properly prepared and stored and the analyte of interest is stable under those conditions.