Data

The American Academy of Forensic Sciences and the Society of Forensic Toxicologists recommend that before results are reported, each batch of analytical data should be reviewed by scientific personnel who are experienced with the analytical protocols used in the laboratory.1 At a minimum this review should include chain of custody data, validity of analytical data and calculations, and quality control data. The review should be documented within the analytical record.

5.10.1 Chain-of-Custody Data

Review of the chain-of-custody documentation ensures that the analytical result represents the correct sample. The data necessary to accomplish this task include the dates and identification of individuals performing the sample collection and transportation to the laboratory; receipt; transfer of specimens or aliquots within the laboratory; chemical analysis; and analytical report.

5.10.2 Analytical Data

The first task of an analyst who wants to evaluate a procedure is to collect relevant data of measurements using that procedure. The second task is to reduce the number of measurements, remove irrelevant and erroneous data, and convert the measurements into statements pertaining to the condition of the procedure under control. To convert the data into a form that can be handled, data reduction procedures are applied.

Analytical instruments are usually equipped with data filters or data handlers that smooth and reduce the data collected, relieving the analyst of such arduous tasks. For this reason, the concepts of data production, information theory, data reduction, data handling, analysis of variance, pattern recognition, and system optimization are not discussed in this chapter. However, the analyst should have some understanding of how instruments perform data analysis. This knowledge will help the analyst identify corrupt data, correctly set thresholds for proper peak integration, and determine signal-to-noise ratios. A thorough discussion on data production and data reduction has been presented by Kateman and Pijpers.28

A thorough review by responsible supervisory personnel of the raw analytical data and the calculations derived therefrom should be performed before a report is issued.

5.10.3 Quality Control Data

A review of the results obtained from the analysis of control material is essential to evaluate the performance of the analysis. This is accomplished by deriving control charts for each control material so that control rules or decision criteria can be applied to determine whether the method performed within expected limits of variation.29-32

The control chart is a graphical representation of the arithmetic mean and the control limits calculated for the control material as described in Section 5.6.3. The x-axis is scaled to provide appropriate time period intervals. Horizontal lines are drawn corresponding to the mean (center) and multiples of the standard deviation above and below the mean, as shown in Figure 5.4. These graphs are sometimes referred to as Shewhart or Levey-Jennings charts.

The control rules or criteria described here are based on statistical properties arising from single control measurements, rather than replicate measurements, since they are more common. They provide a low level of false rejection, improved capability for detecting analytical errors, and some indication of the error type to aid in problem solving. Rejection of an analytical run occurs when the control value obtained violates one or more of the following rules, symbolized for brevity and convenience:

• (13s) A value outside 3 standard deviations

• (22s) Two consecutive values outside 2 standard deviations; one value outside 2 standard deviations should be considered a warning, requiring inspection of a second control

• (R4s) Two consecutive values that differ by 4 standard deviations or more

• (41s) Four consecutive values on the same side of the mean that are more than 1 standard deviation from the mean

• (10x) Ten consecutive values on the same side of the mean

Blood Ethanol Control

Lot No. 23

Blood Ethanol Control

Lot No. 23

Date

Figure 5.4 Quality control chart.

Date

Figure 5.4 Quality control chart.

The particular rule violated may give some indication of the type of analytical error occurring. Random error will most often be detected by the 13s and R4s rules. Systematic error will usually be detected by the 22s, 41s, or 10x rules and, when very large, by the 13s rule. However, these rules may be difficult or impossible to properly apply in forensic toxicology work where, with the exception of ethanol and a small number of drugs, the frequency of analysis is relatively low.

0 0

Post a comment