City of Chicago Emerging Contaminant Study
Analysis of Endocrine Disrupting Chemicals, Pharmaceuticals,
Personal Care Products, and Hexavalent Chromium

INTRODUCTION.  The City of Chicago Department of Water Management (CDWM) is proud to provide high quality drinking water that exceeds all standards set by state and federal water quality regulators.  Source water taken from Lake Michigan is filtered and treated at Chicago’s two water purification plants: the Jardine Water Purification Plant (JWPP) and the South Water Purification Plant (SWPP).  Having completed the purification process, the finished (fully treated) drinking water is then distributed via pipelines to all of CDWM’s customers.  The reader is encouraged to visit the City of Chicago website and read the annual water quality report on CDWM’s homepage.

The CDWM completed a two-year water quality study to monitor some compounds that have not historically been considered to be contaminants of concern, but have been recently documented at trace concentrations in our nation’s waterbodies.  Water samples were tested for emerging contaminants, including Endocrine Disrupting Chemicals (EDCs), Pharmaceuticals & Personal Care Products (PPCPs), and hexavalent chromium.  EDCs are compounds with potential to interfere with natural hormone systems.  PPCPs are a group of compounds consisting of prescription or over-the-counter therapeutic drugs, veterinary drugs, and consumer products such as sun-screen, lotions, insect repellent, and fragrances.  Hexavalent chromium, which has both natural and industrial sources, is the contaminant made famous in the Erin Brockovich movie.  The reader is encouraged to visit the United States Environmental Protection Agency (USEPA) website to learn more about EDCs, PPCPs, and hexavalent chromium.

Advances in technology over the past several years now allow for the detection of compounds at extremely low concentrations.  Typically, regulated compounds are measured and have standards in the range of parts per million (ppm) or milligrams/liter (mg/L).  However, modern laboratory tests can now detect certain compounds down to levels of parts per trillion (ppt) or nanograms per liter (ng/L) concentrations.  Since it is difficult to conceptualize a trillion of anything, the following example from The MegaPenny Project website may help: it would take approximately 2.6 trillion pennies to fill the Willis Tower (formerly the Sears Tower).  One ppt (1 ng/L) would equal approximately 2 ½ pennies within a Willis Tower made entirely of pennies!

Most of the compounds classified as EDCs and PPCPs are not currently regulated – in other words, drinking water concentration limits have not been set for these compounds and water quality regulators do not require that drinking water providers test for these compounds.  At this time, human health effects have not been demonstrated at the trace levels at which these unregulated compounds are being detected.  Nevertheless, more research is being conducted on the presence and impacts of EDCs and PPCPs in our nation’s waters and on human health (studies are being conducted by groups such as the USEPA and the Water Research Foundation).

SAMPLING PROGRAM.  In response to the growing interest and awareness in EDCs and PPCPs, and recognizing that emerging contaminant research studies may take years to complete, the City of Chicago developed a sampling program that encompassed both temporal and laboratory variability.  Lake Michigan source water and finished drinking water were sampled six times (Table 1). 

Table 1 – EDC-PPCP Sampling Events

Sampling sites include the offshore crib intakes, shore intakes, and finished water outlets at the JWPP and the SWPP, plus one field blank (Table 2). 

Table 2 – EDC-PPCP Sampling Locations

Since most of these compounds are not regulated, EDC and PPCP laboratory tests do not have standardized analyte lists, methods, or reporting limits.  Therefore, CDWM decided to send samples to three independent laboratories with extensive experience doing EDC and PPCP analyses.  This allowed for the evaluation of intra-laboratory variability, for inter-laboratory variability, and for the seasonal patterns and levels of occurrence of a large number of EDCs and PPCPs.  The three laboratories each used different analytical methods, had partially overlapping analyte lists, and had reporting limits in the ppt range (Table 3). 

In addition, hexavalent chromium samples were collected quarterly in 2011 and analyzed by USEPA Method 218.6, following USEPA guidance for voluntary monitoring issued in early 2011.  Samples were collected from the sampling locations listed in Table 2 plus a variety of distribution system locations throughout the City of Chicago, for a total of up to 29 sampling locations per round. 

Table 3 – EDC-PPCP Analytical Methods

EDC-PPCP RESULTS.  A total of 147 unique EDC and PPCP compounds were included in the study and 61 of these compounds were analyzed by more than one method.  Only thirty-four compounds were detected above reporting limits in at least one sample over the course of the study, not including the field blanks (Table 4).  The detections reported were well below any known public health limits. 

Table 4 – Frequency of EDC-PPCP Detections

Low Frequency of Detection (1-10% of Samples)
Analytes that were detected at frequencies of less than 11 percent (<11%) of samples (Table 5) may represent analytical artifacts, although these results were included here because they were not ruled out during our review of the laboratory QA/QC data. 

Table 5 – Compounds with Low Frequency of Detection

The frequency of detection is partially a function of the sensitivity of the method used by a given lab.  About half of the detections of the compounds listed in Table 5 were present at concentrations of < 5 ng/L, and were often below the sensitivity of a second lab if the analyte was on a second lab’s target list. Medium Frequency of Detection (11-30% of Samples)
Eight compounds were detected between 11% and 30% of the time (Table 6).  These were detected either:

• consistently across sites in one or two rounds; or
• in a subset of sites, but not in all rounds (making a point source a potential cause); or
• very close to the reporting limit of the lab and, therefore, may have been present more frequently than shown because the analytical precision at the reporting limit masked additional detects; or
• only during the last few rounds of sampling (for the new analytes added to methods part way through the study), making it difficult to determine the true frequency of occurrence.

Table 6 – Compounds with Medium Frequency of Detection

High Frequency of Detection (31-100% of Samples)
Analytes that were detected more than 30% of the time were usually detected in each round of the study (Table 7).  These compounds were either detected by multiple labs (when lab sensitivity was sufficient in more than one lab) or had occurrence patterns that were consistent with expectations based on geochemical and water treatment considerations.  These considerations include:

• Similar concentrations in all samples from a given plant (e.g. PFOS).
• Higher concentrations in source waters than finished waters from a given plant for compounds known to be oxidized by chlorine (e.g. sulfamethoxazole).
• Analytes that are breakdown products of another analyte (e.g. triazine degradates, such as DEA, DIA, and DACT).

Table 7 – Compounds with High Frequency of Detection

Patterns
There are some clear patterns that emerged during the course of the study.

• Cotinine (metabolite of nicotine) was detected by two labs at similar levels (it was not on the third lab’s list).  Values were generally within 1 ng/L of each other on the same sites.   The compound was seen in most sampling rounds at similar levels, suggesting that it is persistent as a low level contaminant.  It could possibly originate from a non-point source (e.g. airborne transport) because of the persistence and very low concentrations.
• Nicotine (stimulant in tobacco) was detected frequently but was only on the target list for one laboratory.  It is the parent compound of cotinine and it was therefore not unexpected to detect it, considering the detection frequency of cotinine.
• Sulfamethoxazole (antibiotic) was detected by all three labs at similar levels in most of the raw water samples in most sampling rounds.  It was not present above reporting limits in finished water, because it is highly susceptible to oxidation by chlorine.   Its frequency of detection over a broad geographic region makes it likely that it is representing a non-point source, similar to the phenomenon for cotinine.
• Atrazine (triazine herbicide) was detected in nearly all samples by the two labs that include it in their analyte lists.  There was some variation in measured concentrations between the labs, but it is clear that atrazine is persistent.   Measured concentrations were similar to those that have been reported in the scientific literature for the Great Lakes, so it is likely that this represents a non-point source pollutant from agricultural activities in the region over an extended period of time or again airborne transport.
• Triazine degradates (DACT, DIA, DEA) and Simazine (triazine herbicide) were all detected frequently throughout the survey.  In light of the high frequency of detection of atrazine it is not at all surprising to see its degradation products present and it is also not unexpected to therefore see another commonly used triazine (simazine).  
• DEET (insect repellent) appears to be somewhat seasonal in occurrence and is present only at very low levels.  It was detected sporadically in various rounds, consistently at levels near the lab reporting limits.  The fact that it was not detected in every round suggests that it may represent more of a point source pollutant, but it is also possible that it has an airborne source.
• PFOS (fluorosurfactant) has been detected routinely in all field samples at very low levels in each sampling round.  This compound is on the EPA Unregulated Contaminant Monitoring Rule (UCMR3) list, but with a much higher proposed reporting level.  There is extensive literature suggesting that the perfluorinated compounds have extensive airborne transport.
• Compounds detected sporadically may or may not have been real hits; based on the QC data provided by the labs, they appeared to be real (although some were flagged as questionable based upon review of lab QC data).  In most cases where these “trace” level hits are detected they are within a factor of 2x the lab reporting limit and they might therefore be false positives or false negatives, depending upon the lab’s precision and accuracy at those levels.   

It is difficult to compare data directly between labs because of inconsistent overlap among lab analyte lists and/or reporting limits.  In addition, for approximately two-thirds of the detections over the course of the study, the reported hits were at levels of <10 ng/L and also less than 5x the reporting limit of the laboratories, making it is difficult to evaluate agreement between labs as far as absolute concentration differences (e.g., a hit of cotinine from one lab at 1 ng/L and 1.7 ng/L from another lab are considered to be comparable in this assessment).

It is also of great interest to note the compounds that were not detected over the duration of this study, but have been reported to occur in wastewater effluents at high frequency and sometimes at high concentrations, in addition to being resistant to treatment.  These include sucralose, carbamazepine, primidone, meprobamate, and iohexal.  The lack of detection of these analytes supports the suggestion that most of the detected analytes are not emanating from wastewater plants, but rather represent more widespread non-point sources.

Data Summary Tables
The two attached tables summarize the results of the EDC-PPCP study.  Table A shows the results of the Lake Michigan source water samples (JWPP and SWPP crib and shore intakes) and the drinking water samples (finished water from JWPP and SWPP).  Disagreements between laboratories for certain compounds and issues with laboratory quality control have called some of the results into question; Table B shows all of the data that have been flagged as “not trusted” or “questionable.”  Refer to the table footnotes for additional information.

HEXAVALENT CHROMIUM RESULTS.  Low levels of hexavalent chromium (0.16 – 0.25 µg/L) were detected in all samples analyzed, including raw Lake Michigan water, finished water at the plants, and finished water at various distribution points in the City.  This compound appears to be persistent throughout the system at very low levels.  USEPA has not yet established a standard for this compound.

ADDITIONAL INFORMATION.  CDWM is happy to provide the consumers of Chicago’s drinking water with additional information about water quality.  Please address any questions or concerns to   312.742.7499.

 Lake Michigan and Drinking Water Results