One of the eight investigations Mr. Johnson mentioned was a 2005 audit by the DC Office of the Inspector General (OIG) that was carried out in response to a 2004 request by then DC Council Member Carol Schwartz for an "independent corroboration" of WASA's lead-in-water data. Written on February 27, 2004 -- only weeks after DC residents learned that for two and a half years the city's drinking water dispensed hazardous levels of lead -- Ms. Schwartz's letter to the DC OIG explained that such corroboration was necessary because the public had "lost faith" in the utility's credibility. In contrast to the other seven investigations -- which exposed gross breaches of responsibility and management deficiencies -- the DC OIG's 2005 audit seemed to redeem WASA. In the words of Mr. Johnson, it "affirmed [the] Authority's lead water sample findings."
Indeed, the DC OIG report stated that only 6% (as opposed to 21% in 2004) of 272 homes had lead-in-water concentrations above the Environmental Protection Agency (EPA) action level of 15 parts per billion. These findings were hailed by WASA as proof that its much questioned water sampling methodology was sound, and that its claim about declining lead-in-water levels due to the new chemical "orthophosphate" were correct. "Report Supports Integrity of DC WASA Testing Program," declared WASA's press release of June 20, 2005, days after the announcement of the DC OIG findings.
The water utility's press release quoted Mr. Johnson saying:
"We're extremely pleased with the Inspector General's findings, as it validates the hard work that this agency and its partners have done to address this issue. [...] Moreover, this report affirms our recent announcements of declining lead levels, as well as the integrity of our test sampling process."Turns out this wasn't the last we were going to hear about the DC OIG audit. WASA brought it up again on February 20, 2009, this time at a DC Department of the Environment (DDOE) meeting about the District's latest effort to evaluate independently our drinking water for lead. At this meeting, a WASA board member asserted that WASA's lead-in-water monitoring practices -- which continue to be questioned by independent experts as well as public health and environmental organizations -- were validated by the DC OIG in 2005. This validation, the board member suggested, freed the DDOE of any need to focus its current investigation on the integrity of WASA's monitoring methods.
In our work on lead in the District's drinking water we have learned that you cannot trust WASA's statements without triple-checking the facts. Prompted by the assertions of WASA's board member, we decided to take a closer look at the DC OIG 2005 audit.
The DC OIG 2005 audit, which was supposed to detect the presence of lead in DC's drinking water by re-sampling "high risk" homes in WASA's own testing pool, used a water-sampling methodology that departs significantly from the standard EPA protocol and, by design, is not able to capture normally available lead at the tap. The methodology is rendered useless by the following inexplicable instructions:
1. No first-draw samples. A first-draw sample represents the water that is collected immediately after a minimum stagnation period of 6 hours. Compliance with federal regulations necessitates that 90% or above of first-draw samples at "high-risk" homes measure below 15 parts per billion of lead.
2. "Second-draw" samples after 4 minutes of flushing. Second-draw samples are not required for compliance with federal regulations, but they can provide useful information about lead that is released from lead service lines outside many homes. For single-family residences EPA suggests collecting second-draw samples by "allowing the water to run until there is a significant change in temperature which would be indicative of water that has been standing in the lead service line." Water corrosion expert at Virginia Tech and 2007 MacArthur Fellow Marc Edwards says that, in practice, EPA's guidance translates to a flush of approximately 30 seconds to 1 minute. A 4-minute flush, Dr. Edwards told us, "is completely worthless in assessing any aspect of corrosion control, water safety, or veracity of prior EPA monitoring data. Whoever wrote that instruction guaranteed that the results obtained would be favorable to DC WASA, because the collected samples had no chance of detecting a lead hazard even if it were present."
What DC residents have painfully learned in the last few years is that there are sampling protocols that find lead problems when they exist, and there are sampling protocols that miss lead problems when they exist. The DC OIG 2005 audit, which comprised the first independent evaluation of DC's drinking water for lead following the two-and-a-half years when WASA kept the contamination under wraps, used a sampling protocol that misses lead problems. It is noteworthy that the DC OIG findings directly contradicted concurrent research in Washington DC by a University of California, Berkeley team, which in the Fall of 2005 reported that:
"Despite reports that lead levels have fallen significantly over the past year, 63% of all homes tested during the second year of our study  still contained lead levels that exceeded the EPA limit. In addition, drinking water collected from a well-used fountain at an elementary school site that a local government reported as lead free contained lead concentrations greater than 5 times the EPA action limit two years in a row."Indeed, the fact that 16 of the 272 homes sampled for the DC OIG investigation were found to have elevated levels of lead after 4 minutes of flushing (and during cold months of the year, when lead-in-water levels are not at their peak) suggests that in 2005 DC had a much more serious lead-in-water problem than the DC OIG data revealed. Contrary to WASA's claims, therefore, the DC OIG audit affirmed neither declining lead levels, nor the integrity of WASA's sampling methods.
One, of course, has to wonder how the DC OIG came up with the flawed sampling protocol. The DC OIG told us that it was given the impression that this protocol was what WASA was using at the time and that it complied with EPA requirements. If this is what the DC OIG was told, then it was led astray.
The protocol WASA was using in 2004 included a first draw sample, as EPA requires for compliance with federal regulations and the DC OIG report's comparison table demonstrates (pp. 10-16). Moreover, to our knowledge, it instructed homeowners to take a second draw sample after feeling a significant change in the temperature of the water, which is the language recommended by EPA. Excessive flushing prior to second draw sampling does not meet EPA requirements, and WASA knew this long before DC OIG began its investigation.
Although we haven't yet confirmed who exactly gave DC OIG the flawed sampling protocol, we know that improper flushing prior to testing -- either on the eve of sampling or immediately before collecting a sample -- is one of WASA's signature practices:
- Prior to July 2003 and as part of its water monitoring program for compliance with federal regulations, WASA did in fact instruct homeowners to run the water for five minutes before collecting a second draw sample (Covington & Burling report, p. 108). In July 2003, however, WASA realized that excessive flushing prior to second-draw sampling was not included in EPA's guidelines. As a result, it changed its protocol to reflect EPA's language (i.e., allowing the water to run until there is a significant change in temperature). But it did not tell EPA that 400+ lead-in-water measurements below 15 parts per billion it had submitted for compliance with federal regulations had been obtained with the illegal sampling method. EPA Region III learned about WASA's erroneous methodology from a different investigation one year later. In January 2005, the federal agency issued an administrative order requiring WASA to, among other things, notify the residents in those 400+ homes about the unreliability of their 2004 water test results and the need to use lead-removing water filters.
- In February 2004, two weeks after the Washington Post broke the story about the two-and-a-half year contamination, WASA tested DC public schools (DCPS) by running all taps for 10 minutes immediately before collecting first draw samples -- another version of the flushing practice that lowers lead at the tap and goes against the standard EPA protocol. The water utility's press release declared that "the vast majority of the DCPS and facilities have extremely low levels of lead in the water." Indeed, 98.94% of the taps tested measured below 15 parts per billion for lead. Following complaints by safe water activists and an official admission by WASA that 10 minutes of flushing gets rid of much of the lead, then Mayor Anthony Williams pressed WASA to repeat the school testing without the 10-minute flush. Lead-in-water levels this time came back markedly higher. Later, however, it was discovered that this time again WASA had recommended another pre-sampling flush. Rather than letting the water run immediately before sampling, WASA advised DCPS to flush all buildings for 10 minutes the night before. "If I did not want to find a lead problem where a serious problem existed, this is the protocol I would write," said Dr. Edwards. Another engineer commented, "They flushed the devil out of those schools." Three years later, in 2007, WASA instructed DCPS to flush all schools for 45 minutes and all taps for 5-15 minutes the night before sampling. Still, results showed that 3/4 of DC schools had at least one tap with lead levels that exceeded 15 parts per billion.
- Similarly, from 2005-2008, WASA instructed homeowners in its water monitoring pool to flush their taps for 10 minutes the night before sampling for compliance with federal regulations, and refused to admit this to inquiring members of the public until asked by the Washington Post in mid-2008. Despite the objections of public health and environmental organizations and EPA's September 2008 determination that flushing on the eve of sampling goes against the intent of the law, WASA continues to recommend running the tap on the eve of sampling, albeit now for 2 minutes instead of 10. This, despite data from Virginia Tech that a 2-minute flush on the eve of sampling eliminates almost as much lead as a flush that lasts for 10 minutes.