False identification results

Figure 7.4. The significance level of an identification result can be determined once the distribution of scores for false identification results is known. Score distributions for true results can vary between experiments and are typically unknown, in contrast with the distribution of scores for false identification results, which can be derived by various methods (see text for details). A score that is in a region with little overlap with the distribution for false results yields a good significance level (the gray area is small).

Independent of the type of scoring system used, there is always a risk of obtaining a false identification result. Each measured spectra can match the frequencies of several different molecules in addition to the match with the molecule actually present in the sample. False results are caused by the accidental matching of the frequencies of a chemical with no properties similar to the unknown. An incorrect result is also obtained when the score due to random matching cannot be discerned from the score match due to matching of the real spectra in the sample. It is therefore useful to include other information when interpreting the quality of the match, such as the chemistry involved in the unknown sample. The likelihood of accidental matches (and hence the likelihood of false identification) increases if the spectral reproducibility is reduced. 

Many diseases, including anthrax, are most effectively treated before actual manifestation of the symptoms is observed. Presently a presumptive identification of Bacillus anthracis can be made in about 3 hours however, if a full laboratory response network (LRN) confirmation procedure is utilized, the theoretical time increases substantially to approximately 48 hours. During the recent anthrax cases 72 to 96 hours were common to complete the entire LRN protocol. In the meantime antibiotics were administered as a precaution based on the presumptive results to individuals thought to be exposed to B. anthracis spores or with anthrax symptoms. The mass administering of antibiotics from a cost standpoint, as well as from medical prudence to prevent the rise of antibiotic-resistant strains, is not the optimal answer to the anthrax infection problem. Therefore it is important that early tests be rapid and reliable with a minimum number of false positive and false negative results. 

For tests designed to detect the presence or absence of an analyte, the threshold concentration that can be detected can be determined from replicate measurements over a range of concentrations. These data can be used to establish at what concentration a cut-off point can be drawn between reliable detection and non-detection. At each concentration level, it may be necessary to measure approximately ten replicates. The cut-off point depends on the number of false negative results that can be tolerated. It can be seen from Table 4.7 that for the given example the positive identification of the analyte is not reliable below 100 xg g-1. 

Another important source for false-positive results in the use of the Griess reaction for the identification of explosives is the possible presence of nitrate ions, together with some accidental reducing substances. In this situation, the nitrate ions (NOs ) could be reduced to nitrite ions (NO2 ), giving a positive result in the Griess reaction. 

Recognizing the fact that shifts in retention time of individual compounds may cause false negative results, laboratories use retention time windows for target analyte identification. Retention time windows are experimentally determined retention time ranges for each target analyte. To minimize the risk of false positive results, EPA methods require that chromatography analysis be performed on two columns with dissimilar polarities. This technique, called second column confirmation, is described in Chapter 4.4.3. It reduces the risk of false positive results, but does not eliminate them completely. 

In individual compound analysis with the FID, analyte identification relies solely on retention time. As a result, single compound determinations (for example, individual phenols or PAHs) are usually gravely affected by interferences that often render the results unreliable or unusable. Because the FID methods for individual compound analysis are so susceptible to false positive results, they should be used with caution and only for interference-free matrices. 

If BTEX components and MTBE are not sufficiently resolved in a complex gasoline mixture, their identification and quantitation with a PID becomes questionable. False positive results and high bias are common in gasoline component analysis with EPA Method 8021. 

Second column confirmation must be used in pesticide, PCBs, and chlorinated herbicide analyses by EPA Methods 8081, 8082, and 8151, respectively. In these methods, two columns with dissimilar polarities and two ECDs provide compound identification and quantitation. This technique produces a lower rate of false positive results, but does not eliminate them completely. This is particularly true for low concentrations of pesticides and herbicides, where non-target compounds, such as constituents of the sample matrix or laboratory artifacts listed in Table 4.3, produce chromatographic peaks on both columns. These interference peaks cannot be distinguished from the target analytes based on retention time only and cause false positive results. 

Direct analysis of water samples and extracts has the advantage that no time-consuming sample preparation is required, as it would be required for the indirect GC analysis. This precludes identification of artifacts incurred during evaporation, derivatiza-tion, or by hot injection ports, which could result in false identifications. 

The current penalty system is strict, clear, and straightforward however, it does not allow for minor errors. In fact, minor errors are treated equally to a false negative result, as demonstrated with the following example Laboratory 1 failed to identify one of the spiking chemicals in a test. Laboratory 2 found all spiking chemicals however, it made a minor reporting error in the identification of one of them the evidence provided was adequate to indicate the presence of the particular chemical, however, did not fulfill all the reporting requirements. 

Therapy becomes problematic when antibodies and markers of HCV infection are evident. It is imperative to complete the unambiguous identification of HCV antibodies by determination of HCV RNA. (The positivity of HCV antibodies may well be a methodologically related false-positive result.). If HCV RNA is detected, i. e. in florid HCV infection, interferon therapy (or a combination with ribavirin) should be considered. This antiviral therapy may be indicated despite positivity of LKM 1 following critical evaluation and close-meshed controls. There is a possibility of AIH activation under antiviral therapy. In this case, INF therapy has to be stopped as quickly as possible. It is not currently known whether this risk can be avoided by simultaneous administration of low-dose prednisolone. The same problem arises in the (very rare) combination of AIH with coexistent florid HBV infection. 

ST53 Butch, A.W., Goodnow, T.T., Brown, W.S., McCellan, A., Kessler, G. and Scott, M.G. (1989). Stratus automated creatine kinase-MB assay evaluated Identification and elimination of falsely increased results associated with a high-molecular-mass form of alkaline phosphatase. Clin. Chem. 35, 2048-2053. 

The sensitization test selected should identify substances with significant allergenic potential and minimize false-negative results. Several animal models have been developed for identification of sensitizers, including... 

The Rule Section contains the definitions of rules that are applied to a substructure. The rule is enclosed in quotation marks and uses the logical operators AND, OR, and NOT to return either TRUE or FALSE as result. Rules are organized in expressions, and their priority of execution is determined by the enclosing brackets. A rule identifier ensures the unique identification and reuse of rules in other expressions. An example is as follows ... 

Pollution unknown or information limited. If no data are accessible on the presence of pollution, bioassays may be used for toxicity screening in order to identify polluted samples. Moreover, if only insufficient data from chemical analysis are available, bioassays may identify samples that require further analysis (presence of undetected or unknown toxicants). Although bioassays respond to a wide range of toxicants, success in identification of contaminated samples may be reduced if false negative results are achieved (see below). 

---