Basic HTML Version
26
Indication of Comparability
Related Questions
Rules for excluding
certain types of observations
should
be similar for all data sets
Qualification and/or validation of data: What criteria were used to qualify or validate
the data? If criteria were not consistent across data sets, the same qualifications may
have different meanings. What QA and QC information is available from the
laboratories?
Sensitivity
Sensitivity
is the capability of a test method or instrument to discriminate between measurement
responses representing different levels (e.g., concentrations) of a variable of interest. Both the precision
of the instrument and the slope of the calibration curve limit sensitivity.. If two methods have equal
precision, the one having a steeper calibration curve will be the more sensitive. Sensitivity can also be
evaluated from the standard deviation of replicate analyses at any concentration level. Sensitivity is
addressed primarily through the selection of appropriate analytical methods, equipment, and
instrumentation.
The sensitivity indicators of primary interest to MERC are indicators that relate to limits of detection.
The detection limit (DL) is a concept concerning the capability of an analytical method to distinguish
samples that do not contain a specific analyte or biological variable from samples that contain low
concentrations. DLs vary by variable and by matrix.
The sensitivity of microbial analyses utilizing an IDEXX® analytical method is reported in the product
literature for each type of analysis (i.e.,
E. coli
,
Enterococcus spp.
, total heterotrophic bacteria).
MERC uses two sensitivity indicators to define detectability for chemical analyses: method detection
limit (MDL) and practical quantification limit (PQL) or reporting limit (RL).
The MDL is the minimum value which the instrument can discern above background but no certainty to
the accuracy of the measured value. MDLs facilitate the determination of whether a single observation
represents a true signal as opposed to noise. Two approaches have generally been recommended to
determine MDLs, “Single Concentration Designs" and "Calibration Designs". The single concentration
design assumes variability at a certain concentration to be equal to the variability at the true MDL. The
calibration method utilizes prediction intervals to model MDL variance with concentration (Hubaux and
Vos, 1970).
MERC utilizes the single concentration design estimator to determine MDL, which is recommended by
the US EPA:
1)
Measurements are taken on at least seven samples of the prepared solution. Results are tabulated and
the standard deviation of the data set is taken:
2)
Using the degrees of freedom from the data set and the appropriate confidence level (usually 1%), the
critical t-value is looked up using reference tables: