Reconcilable differences: The use of reference material to reduce methodological artifacts in the reporting of organochlorine pesticides and polychlorinated biphenyls

Abstract

Numerous long-term monitoring programs have assessed spatial and temporal trends of organochlorine (OC) pesticides and polychlorinated biphenyls (PCBs). Changes in analytical approaches (e.g., gas chromatography coupled with electron capture detection [GC-ECD] versus mass spectrometric detection [GC-MSD]) can reveal artifacts in the reported concentrations. In-house reference material (RM) was used to determine the analytical artifacts in the measurement of OCs and PCBs in Great Lake herring gull eggs previously analyzed from 1994 to 1996 (GC-ECD) and 1997 to 1999 (GC-MSD). Approximately 19.0% of the variability of PCB congeners in gull eggs was associated with analytical artifacts, and differences among colonies were obscured. Although the discrepancy in sum PCBs (SigmaPCBs) was fairly small (2.1%), some congeners varied considerably between methods (> 60%). After statistically removing the artifacts, only 1.4% of the variability in PCBs of herring gull eggs was associated with artifacts, and differences among gull colonies became apparent. After excluding OCs near the detection limit in the RM, statistically removing the artifacts reduced some of the differences between methods for OCs. Analytical artifacts may potentially render inferences difficult, confounded, and erroneous. When combining contaminant data obtained using different methods, the methods should be assumed to give different results unless demonstrated otherwise. Assessments of the compatibility of analytical methodologies should be made using an appropriate RM.

Impact of changes in analytical techniques for the measurement of polychlorinated biphenyls and organochlorine pesticides on temporal trends in herring gull eggs

“…Although we support the use of conversion factors [8,27], they should be used cautiously, because incorrect conversions could make the situation worse. Some previous attempts at creating conversion factors revealed that the conversion factors may vary with matrix; relationships between Aroclor-based and congener-specific methods of quantification differed among water, sediment and aquatic animals, and even species [31].…”

“…Similarly, dioxins and non-ortho PCB concentrations in food for human consumption measured at different laboratories were not numerically comparable [7]. Our earlier study demonstrated differences in quantification of OC compounds between detectors [8] and also differences in technique (see Materials and Methods), and extraction and cleanup also may contribute to discrepancies in the measured concentrations.…”

“…After adjustment, the variation attributed to methodological artifacts ($19.0%) decreased to approximately 3.3%. Consequently, differences in the PCB congener profile became attributable to colony or lake effects [8].…”

“…Not only were there differences in the measurement of organochlorine pesticides and PCBs in herring gull eggs between the two analytical methods (GC-ECD and GC-MSD), but these differences were large enough to obscure real biological phenomena, at least in some contexts [8]. For example, multivariate techniques showed differences between GC-ECD and GC-MSD in PCB congener profiles among 15 colonies of herring gulls that obscured variation attributed to colony location.…”

“…and thus standardized the results for all other samples analyzed using the two methods [8]. After adjustment, the variation attributed to methodological artifacts ($19.0%) decreased to approximately 3.3%.…”