Long-term reproducibility

The errors mentioned above represent the reproducibility obtained on the same microtiter plate when the sample molecule is assayed in several different wells. When the reproducibility of Pe measurement is assessed on the basis of assays performed at different times over a long period of time, more systematic sources of errors show up, and the reproducibility can be about 2-3 times worse. Figure 7.56 shows reproducibility of standard compounds taken over a period of about 12 months. Carbamazepine show a long-term reproducibility error of 15%. The other compounds show somewhat higher errors. 

In a attempt to compensate for poor long-term reproducibility in a longterm identification study, Chun et al.128 applied ANNs to PyMS spectra collected from strains of Streptomyces six times over a 20-month period. Direct comparison of the six data sets, by the conventional approach of HCA, was unsuccessful for strain identification, but a neural network trained on spectra from each of the first three data sets was able to identify isolates in those three datasets and in the three subsequent datasets. 

In the late 1980 s, methods for precise Li isotope determination (i.e., <2%o long-term reproducibility) by a variety of TIMS methods became estabhshed. These techniques developed different strategies for dealing with the significant thermal mass fractionation of Li from the surface of a hlament from measurement of heavier ionic species (e.g., LijBOj Chan 1987, Sahoo and Masuda 1995 LijE Green et al. 1988 NaLiBOj Chan et al. 1992) to evaporation of a Li molecule and subsequent ionization and measurement of Li as metal ions (Michiels and DeBievre 1983 Xiao and Beary 1989 Moriguti and Nakamura 1993 Clausen 1995 You and Chan 1996 Sahoo and Masuda 1998). 

We have seen that one of the key aspects of Fourier transform NIR analyzers is their control of frequency accuracy and long-term reproducibility of instrument line shape through the use of an internal optical reference, normally provided as a HeNe gas laser. Such lasers are reasonably compact, and have acceptable lifetimes of around 3 to 4 years before requiring replacement. However, we also saw how the reduction in overall interferometer dimensions can be one of the main drivers towards achieving the improved mechanical and thermal stability which allows FT-NIR devices to be deployed routinely and reliably in process applications. 

None. To confirm long-term reproducibility, we control the slope of the standard curves. 

The long-term reproducibility of this type of analyzer is under evaluation this part of the experiment will make it possible to complete the IMGC procedure by determining a suitable period for re-calibration or by suggesting an expansion of the uncertainty linked to the time elapsed since instrument calibration. 

Rehkamper, M., and Halliday, A. N. (1998) Accuracy and long- term reproducibility of lead isotopic measurements by MC-ICPMS using an external method for correction of mass discrimination. Int. J. Mass Spec. Ion Proc., 181, 123-133. 

Indeed, the nature of the sample seems to be a critical parameter in the reproducibility equation. Short-term and long-term reproducibility of retention times and peak areas for CEC separation of cannabinoids have been measured by Lurie et al. [66]. The RSDs (seven replicates) for retention times are around 0.5% for short-term and vary from 1.49% to 2.03% for the long-term study, while peak areas showed up to 39% variation. Significantly improved reproducibility values were obtained when one compound, cannabinol, was used as a reference. The RSDs for relative peak areas varied from 2.3% to 3.95% for short-term and 1.65% to 8.1% for the long-term study. 

Long-term reproducibility is the prime quality of an analytical method used for the study of stability. Stability must be performed on the element or substance to be certified. Extrapolations on the stability of tracers are of little interest. If the uncertainty of the method for an analyte is poor, even for large sample intakes, the analyst must refer to the experience on similar materials with higher contents or to the general chemical properties of the substance or element. In any other situations certification may be impossible. Such difficulties are mainly encountered in organic or organo-metallic analysis. Experience has shown that some substances may be stable in a matrix but unstable in another even similar one [46-47]. All analytical methods suffer from long term reproducibility. In some cases analysts have developed tools and tricks to minimise this effect. 

The method performance study focuses on the analytical method. In order to achieve a constructive work the organiser should assure that the method to be used by the participants has been investigated in detail prior to the start of the interlaboratory study. This preliminary investigation should be undertaken by a senior analyst. It should lead to a detailed draft analytical protocol, to a clear definition of the type and number of samples to be distributed and analysed, to indications on the repeatability of the method and its long-term reproducibility. In parallel, a study on the production of proper samples should be undertaken. Additional materials can also be prepared, e.g. calibration tools (pure calibrants or solutions), or blanks with a similar matrix than the test sample, spiked materials, etc. 

The mussel and tuna analyses allowed us to test the long term reproducibility of the laboratories. It was noted that interferences were systematically higher with mussel tissue but the higher CV obtained (17.4% in comparison with 13.7% for tuna fish) could also be due to the fact that the MeHg content in mussel was much lower than the one of tuna fish ((0.14 0.01) mg kg" as MeHgCl in mussel and (4.33 0.11) mg kg as MeHgCl in tuna fish). 

Analyses of extracts led to difficulties, mainly due to a lack of good long-term reproducibility for many laboratories. Capillary GC was found to offer good possibilities but its use was hampered by the absence of commercially available columns. Furthermore, sources of error were likely due to losses of MeHg. A Youden plot of raw and spiked extract demonstrated systematic errors (Figure 1) which was illustrated by the high CKs found between laboratories (Table 4.1 16.6 and 17.4% for raw and spiked raw extracts, respectively). Better results were obtained for the cleaned extract (12.5%) but the spread was still considered to be too high. However, a consequent improvement was obtained for the aqueous solution analysis (CF of 8.4%). 

Diepinigaitis PV (2003) Short- and long-term reproducibility of capsaicin cough challenge testing. Pulm Pharmacol Ther 16 61-65... 

Davenport PW, Chan PY, Zhang W, Chou YL (2007b) Detection threshold for inspiratory resistive loads and respiratory-related evoked potentials, J Appl Physiol 102(l) 276-285 Dicpinigaitis PV (2003) Short- and long-term reproducibility of capsaicin cough challenge testing, Pulm Pharmacol Ther 16(l) 61-65... 

Long-term reproducibility is affected by eveutual deterioration of resistive filaments or sample wires. All components exposed to sample during pyrolysis (GC injection port liners and quartz sample tubes if used) often require acid cleaning, solvent washing, and oven drying. Active pyrolyzer elements (coils and ribbons in filament pyrolyzers) can be heated without sample to ranove contamination (lOOO C for 2 sec is usually adequate). Curie-poiut wires are inexpensive enough to be discarded after use. 

---