12.02.2013 Views

Environmental Health Criteria 214

Environmental Health Criteria 214

Environmental Health Criteria 214

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

HUMAN EXPOSURE ASSESSMENT<br />

clearly presented as part of the QA/QC procedures. The term "lower<br />

limit of detection" (LLD) may refer to very different concepts. For<br />

example, the instrumental detection limit is defined as a multiple of<br />

the noise level of the analytical instrument, usually a factor of 3.<br />

However, when sampling and analysing environmental media, the<br />

instrumental detection limit constitutes only a fraction of the true<br />

limit of detection. The LOD takes into consideration the response of<br />

the analytical instrument to the specific analyte in standards of<br />

known concentration or, preferably, a matrix similar to that of the<br />

sample. Since the response of the instrument to repeated analysis of<br />

the same analyte concentration varies, the LOD is estimated as a<br />

function of the standard deviation (SD) of the repeated analysis,<br />

typically as t n-0.01 × SD, where t is the value of the Student<br />

t distribution with n-1 degrees of freedom, alpha = 0.01 is the<br />

type I error and n the number of repeated analyses, usually 7.<br />

However, sampling methods can have a background of one or more of<br />

the target analytes. For example, when using solid sorbent for air<br />

sampling, the sorbent may contain varying levels of background<br />

contamination across samples (e.g., varying benzene concentrations in<br />

Tenax). The method detection limit (MDL) includes the background<br />

contamination in the sampler. It is determined using a similar<br />

statistical approach as the LOD, but using field and/or laboratory<br />

blanks instead of the known analyte concentrations. The MDL can be<br />

defined as the analyte concentration at which we have a given<br />

certainty (e.g., 1-alpha = 0.99) that the sample concentration differ<br />

from background. Finally, the lower quantification limit (LOQ) is<br />

defined as 10 SD, where SD is derived from the LOD determination. LOD,<br />

MDL and LOQ differ across laboratories using the same methods. More<br />

importantly, they vary over time for the same laboratory.<br />

The treatment of values below the MDL in statistical analysis of<br />

the data can have a strong impact on estimates of exposure<br />

distribution parameters. The reader may refer to Gilbert (1987), for<br />

more specific information. There also are upper limits of detection,<br />

defined by the range of linear response of the analytical method. In<br />

that case the detector may be overloaded with the result that it is<br />

not working in its linear detection range.<br />

11.5.2 Internal quality control<br />

The purpose of internal QC is to document that the method is in<br />

statistical control and without systematic errors so that the observed<br />

sampling results consistently fall within established control limits.<br />

Internal QC procedures are used primarily for the control of<br />

analytical precision (Taylor, 1988).<br />

11.5.2.1 Control charts<br />

Sometimes changes in the analytical performance are not abrupt<br />

but take place gradually ("drift"). Such gradual changes are difficult<br />

to perceive from a single central result, but many become evident with<br />

time if the results of control samples are graphically displayed with<br />

respect to time or sequence of measurements.<br />

Control limits are used as criteria for signalling the need for<br />

action, or for judging whether a set of data does or does not indicate<br />

a state of statistical control. Control limits can be based entirely<br />

on the data from the samples, in which case the chart illustrates<br />

whether the method or procedure is repeatable. If the control chart<br />

http://www.inchem.org/documents/ehc/ehc/ehc<strong>214</strong>.htm<br />

Page 193 of 284<br />

6/1/2007

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!