September 23, 2003

Mother's Milk: Appendix 1: Methods & sample analysis

Sample collection

Between November 2002 and June 2003, EWG recruited 20 healthy, breastfeeding mothers from 14 states to participate in a breast milk study. Participants collected a breast milk sample into chemically clean, 4 oz study jars, between 7 and 100 days postpartum. Ten participants hand expressed the breast milk sample and ten participants used sterilized, commercially available breast pumps. Study jars were frozen, packed with ice and sent overnight delivery to the study center.

Sample analysis

Study samples were analyzed by AXYS Analytical Services, of Sidney, British Columbia, using high-resolution gas chromatography/high resolution mass spectrometry. AXYS analyzed an average of 2.5 gram of lipid, for 44 PBDE congeners. Reportable levels of 35 PBDE congeners were detected in the breast milk samples.

Sample preparation and analysis: After addition of surrogate standard solution, the milk samples were liquid/liquid extracted with 2:1 acetone/hexane. The extract was reduced in volume and cleaned up using gel permeation, acid/base silica, Florisil and alumina chromatographic columns. The final extracts were reduced to a volume of 20 uL and spiked with 2 uL of the labeled recovery (internal) standard for a final volume of 22 uL; 1 uL was analyzed.

Analysis was conducted by high resolution gas chromatography/high resolution mass spectrometry (HRGC/HRMS) on an AUTOSPEC ULTIMA high resolution MS equipped with an HP 6890 gas chromatograph, a CTC auto-sampler, and an Alpha data system running Micromass software. A 30 m chromatography column was coupled directly to the MS source.

Detection of trace congeners

Samples were analyzed in 2 batches with 3 laboratory blanks per batch. The reporting limit was definted as 2 times the 99% confidence level for background contamination. The 99% confidence level was calculated as three times the standard deviation of the three background samples, assuming that non-detects were 100% of the detection limit. We included only the reported levels that were >2 times the 99% confidence level when calculating a woman's body burden of PBDEs. All but 3 reported values were more than double the 99% confidence level. These 3 values were omitted from our analysis (PBDE- 71, 206 and 208 in three separate participants.)