Reference laboratory value

There are three types of data error random error in the reference laboratory values, random error in the optical data, and systematic error in the relationship between the two. The proper approach to data error depends on whether the affected variables are reference values or spectroscopic data. Calibrations are usually performed empirically and are problem specific. In this situation, the question of data error becomes an important issue. However, it is difficult to decide if the spectroscopic error is greater than the reference laboratory method error, or vice versa. The noise of current NIR instrumentation is usually lower than almost anything else in the calibration. The total error of spectroscopic data includes... 

Lewandrowski K, Kratz A. Case records of the Massachusetts General Hospital Normal reference laboratory values. N Engl J Med 1999 339 1063-72. 

The knowledge that in a large number of cases the errors in the optical data due to the samples are also small has made routine the implicit assumption that the error in the optical data is always small compared to the reference laboratory errors, so that the inverse Beer s law formulation is always used thus the optical data are always considered the X data and the reference laboratory values are considered the Y data. 

A dataset to be used for calibration via the multiple regression least-squares method contains data for some number (n) of readings, each reading presumably corresponding to a specimen, and some number (m) of independent variables, corresponding to the optical data. The dataset also contains the values for the dependent variable the analyte values from the reference laboratory. We begin by defining the error in an analysis as the difference between the reference laboratory value of the analyte, which we call Y, and the instmmental value for the constiment, which we call Y (read T-hat ) ... 

The multiple correlation coefficient is, as previously stated, a dimensionless measure of the degree to which the calibration fits the data. The multiple correlation coefficient is the same as the ordinary correlation coefficient between the reference laboratory values and the instrument readings for the samples in the calibration set. Normally a correlation coefficient can have values between —1 and -I-1. In a calibration situation, however, only positive values can exist. A value of this statistic close to zero indicates that the calibration is failing to relate the instrument readings to the reference values. As the correlation coefficient increases, the instrument readings become better and better indicators of the reference values until, when it reaches unity, the instrument values and the reference values are identical in all cases. 

Erom Equation (8.20) it is clear that this statistic becomes large when R is close to one (there is a close relationship between the instrumental values and the reference laboratory values), if n is large (many specimens are used in the calibration) and m is small (few independent variables are needed for the calibration, avoiding the possibility of equations that appear good but are actually overfitting... 

Figure 8.14 illustrates the flow of computation needed to perform a principal component calibration. Starting with a data matrix that consists of n spectra of m wavelengths each, plus the set of reference laboratory values for the constituent of interest in each specimen, the optical data are separated from the constituent values and the m x m sum of cross-products matrix computed as described in Equation (8.26). 

The next step is to compute the scores for each principal component with each data spectrum as described by Equation (8.23), in other words, to compute the principal component transform for every spectrum. Thus there are p scores computed for each specimen, resulting in an x p matrix of scores the set of scores from each specimen is then combined with the reference laboratory value from that specimen. 

The term on the left-hand side is a constant and depends only on the constituent values provided by the reference laboratory and does not depend in any way upon the calibration. The two terms on the right-hand side of the equation show how this constant value is apportioned between the two quantities that are themselves summations, and are referred to as the sum of squares due to regression and the sum of squares due to error. The latter will be the smallest possible value that it can possibly be for the given data. 

Two other remarkable universalities emerge from the value of y. First, at a reference laboratory time scale of 1 h IO tq, we have a universal value of Sc O.Sfe. This implies Sc Tg)/sc Tm) — 0.7, where Sc Tm) is, of course, also the fusion entropy. This relation is independent of the precise identity of the moving subunit and holds very well. A second important universal feature emerges from the universal value of y/Tgi the cooperative size at Tg is nearly universal. 

En numbers are used when the assigned value has been produced by a reference laboratory, which has provided an estimate of the expanded uncertainty. This scoring method also requires a valid estimate of the expanded uncertainty for each participant s result. A score of En < 1 is considered satisfactory. The acceptability criterion is different from that used for z-, z - or zeta-scores as En numbers are calculated using expanded uncertainties. However, the En number is equal to zeta/2 if a coverage factor of 2 is used to calculate the expanded uncertainties (see Chapter 6, Section 6.3.6). En numbers are not normally used by proficiency testing scheme providers but are often used in calibration studies. 

Certification through interlaboratory testing. In this case, the reference/certified value is obtained by pooling results from several laboratories that have demonstrated capability in analyzing the analyte(s) of interest. The various laboratory means are manipulated statistically to determine the best or truest estimate of the value of interest. 

There are two disadvantages with this method. On the one hand much effort is required to ensure the accuracy of the reference measurements. This is a matter of capacity in the reference laboratories and a matter of costs as well. On the other hand there might be doubt among the participants about the assigned value, because nobody is perfect . This is especially true if the mean of the participants results deviates from the assigned value. 

For chemical measurements, a possibly preferable system is illustrated in Fig. 1. Various possible forms of realization of traceability are given. They range from virtual lack of traceability to a fully Si-bonded measurement. The authors tentatively use the term SI-bonded to indicate a direct realization of the SI unit, as opposed to being traceably linked by way of measured values. Any user laboratory must seek a reference laboratory that is capable of providing measurement links of the adequate uncertainty and that provides the direct bond to the SI, if that is needed. The reference laboratory can in turn choose the traceability quality that it wishes to maintain, with the responsibility of fulfilling the corresponding competence requirements. 

To conduct a proficiency test, a reference laboratory will prepare a quantity of material appropriate for distribution to all the laboratories under test. Requirements for the material will include that it is well characterized, and that it is sufficiently stable and homogeneous that no tested laboratory will be put at disadvantage by receiving a sample not representative of the lot. Further, the material should be typical of that of interest, and have constituents with concentrations within the ranges of interest. Additionally, it is important that concentration values be kept confidential during the period of the test, so that no participant will have an unfair advantage. 

In Germany, the Federal Medical Association prescribes the use of IFCC enzyme reference procedures in clinical laboratory practice. Accordingly, the evaluation of participants results in the proficiency testing system is based on IFCC reference procedure values. 

In this context, complex reference materials, where they are technically feasible, are normally developed by NMIs for the validation of reference procedures, and help establish traceability of these measurements in the sense that these measurements are supported metrologi-cally by traceable measurements to the SI units, and are capable of reproducing the value within the acceptable uncertainty. The so-called reference laboratories are expected to be capable of conducting the validation process along with NMI. 

To improve measurement capability of field laboratories, CENAM has also been offering a PT scheme, not only because there are few PT providers in Mexico, but also due to the need to promote traceable measurement by the use of reference value provided by CENAM. Following the successful implementation of a PT program for environmental measurement laboratory assessment made by authorities of three local governments [8], similar efforts have been made to promote among laboratories who could be considered in the future as reference laboratories in food, petrochemical, clinical [9] and industrial sectors. 

Achieving traceability in ISO 17025 accredited laboratories clearly demands more than just provision of suitable calibration materials or values. Hence, traceability demonstrator projects are being undertaken to identify issues, best practice, and key areas to address. It is also necessary to transfer methodology and expertise to UK reference laboratories and to provide guidance on implementation of traceability to UK... 

IMEP provides an independent, CRV of demonstrated quality. This value represents the best estimation of the true value with uncertainty of the analyte under investigation in the matrix. Very often in IMEP the CRVs have a relative expanded uncertainty of less than 2D3 percent. Reference laboratories that contribute to the establishment of the IMEP CRVs, if possible, apply primary analytical methods, like ID-MS, for trace element determination. Routine or held... 

---