Abstract
The selenium (Se) concentration of paired blood and serum samples from cattle was determined by 2 methods: 1) atomic absorption spectroscopy using hydride generation (HG-AAS), and 2) inductively coupled argon plasma emission spectroscopy using hydride generation (ICP). Samples from 327 cattle were analyzed by HG-AAS, and samples from 344 cattle were analyzed by ICP. The data were examined by linear regression analysis, and the technique of inverse prediction was utilized to determine prediction intervals for estimating blood Se concentration from known serum Se concentration. The correlation coefficients, by simple linear regression of serum Se on blood Se, were 0.79 (r 2 = 0.62) and 0.88 (r 2 = 0.77) for the HG-AAS data and the ICP data, respectively. For the HG-AAS data, the inverse prediction formula for estimating blood Se when serum Se is known, at the 95% prediction interval, was
For the ICP data, the inverse prediction formula for estimating blood Se when serum Se is known, at the 95% prediction interval, was
The prediction intervals were quite wide, and the accuracy of estimating blood Se from a known serum Se was not useful for diagnostic purposes. The use of serum Se concentration to assess nutritional status of cattle with respect to Se does not appear to be appropriate.