metabolite concentrations (total atrazine) determined with GC. Statistical and graphical
comparisons of the combined values are presented in figures 2a and 2b. Figure 2a, an
analysis of all the data points, shows a correlation coefficient of 0.8223 indicating that the
GC results track fairly well with the ELISA results. The slope value of 1.1332 suggests
that there is a small bias towards higher values using the ELISA tests. Figure 2b is the
same set of data with one outlier result removed. The outlier sample had an unusually
high GC concentration of deisopropylatrazine, a metabolite the ELISA test has low
sensitivity to. Figure 2a shows a slope of 0.9893 and a correlation coefficient of 0.9118
which is a very favorable ELISA vs. GC comparison. As described above, it is helpful to
stratify the data and further compare the two methods. The stratified data is presented in
table 5. The agreement suggested by the statistical analysis is apparent in this table. The
GC method produced 12 wells above the MCL while the ELISA method produced 11.
The GC method had 47 samples with concentrations of 1.0 ppb or less. In the 1.0 ppb or
less range, the ELISA method had 43 samples in agreement with GC and 10 results that
were higher than this range. This is most likely due to the lower limit of detection
inherent in the ELISA technology. The other strata in the table show good agreement
between the methods.
Additional statistical analyses were performed comparing other possible combinations of
ELISA vs. GC results. The atrazine ELISA result alone was not a good predictor of total
atrazine plus metabolites (figures 3a and 3b). The correlation coefficient of figure 3a was
only 0.4937 and the slope was 4.2420. When the GC outlier result (fig. 3b) is removed,
the slope and correlation coefficient improve only slightly. ELISA diaminoatrazine
(figures. 4a & 4b) is a somewhat better predictor (when compared to the ELISA atrazine
test) of total atrazine by GC results. The correlation coefficient was 0.7848 with a slope
of 1.2880. If the GC outlier is removed, the slope and correlation coefficient improves to
1.1195 and 0.8681 respectively.
Other Considerations
As figures. 2a and 2b indicate, "total atrazine" as measured by conventional GC
techniques (the sum of all the metabolites plus atrazine) was usually close to the sum of
the ELISA results for diaminoatrazine and the ELISA results for the atrazine test. [Note
that it is not chemically correct to add the concentrations of compounds of different
molecular weights, but because (1) the state regulations are viewed in this way, that is,
they consider atrazine and atrazine to be additive when assessing the standard and (2) the
difference in concentration if all concentrations were expressed as atrazine would be
generally minimal]. These data suggest that the occurrence of atrazine plus atrazine
metabolites in a water sample can be accurately screened using the new diaminoatrazine
ELISA evaluated in this study and the established (Strategic Diagnostics) atrazine ELISA
assay. If both ELISA assays were performed on a sample, the cost would be on the order
of $50. This cost is perhaps an order of magnitude less expensive than the gas
chromatographic analysis. Table 6 summarizes the ELISA vs. GC method agreement
above and below the MCL. The ELISA combination of atrazine and diaminoatrazine
correctly predicted whether the sample was above or below the MCL for 73 of the 74
samples. Using the ELISA diaminoatrazine method alone in a stratified fashion would
yield accurate results in 72 out of 74 cases with respect to being above or below the total