Improper Reporting in Pittsburgh VA Legionnaires’ Reports

//Improper Reporting in Pittsburgh VA Legionnaires’ Reports
Article further vindicates copper silver ionization system in Pittsburgh Legionnaires’ outbreak.

 

(via Pittsburgh Post-Gazette): Early in the Centers for Disease Control and Prevention’s investigation into the Legionnaires’ outbreak at the Veterans Affairs Pittsburgh Healthcare System in 2012, epidemiologists wondered if the deadly Legionella bacteria had started to spread because it had developed resistance.

It was a logical idea to test. After all, the copper-silver ionization system in the hospital that had controlled Legionella for decades suddenly wasn’t working. Twenty-two veterans were sickened by Legionnaires’ in 2011 and 2012, leading to six deaths.

 How the CDC analyzed copper and silver concentration samples at the Pittsburgh VA

How the CDC analyzed copper and silver concentration samples at the Pittsburgh VA

But on Feb. 1, 2013, a CDC microbiologist, Claressa Lucas, emailed her colleagues and told them that her experiments at the CDC’s laboratory in Atlanta found no resistance to the minute amounts of copper and silver that were released in the system to control Legionella.

She said that the optimal manufacturer’s target of 400 parts per billion (ppb) of copper and 40 ppb of silver in water had resulted in a “significant decrease in viability” in Legionella “at” 24 hours.

Saying that copper-silver was effective at controlling Legionella would be a stunning admission for the CDC to make.

As revealed in emails the Pittsburgh Post-Gazette obtained 3½ years after requesting them through a Freedom of Information Act request, the CDC’s top officials have long believed that copper-silver systems are not effective at controlling Legionnaires’ disease. That is despite much counter evidence, including that many hospitals — including other VA hospitals — swear by the systems that are considered the gold standard in filtration.

Those beliefs extend not only to deriding copper-silver systems, but anyone who believes they do work, particularly two prominent Legionnaires’ researchers, Victor Yu and Janet Stout. They used to work at the Pittsburgh VA in Oakland and have long battled the CDC about many aspects of Legionnaires’ disease, including whether copper-silver is effective in preventing it. Other CDC emails also show CDC officials attacking them and deriding their research.

Holding a bias against the two researchers or copper-silver systems is not itself a problem, science ethics experts say.

“There is a certain amount of professional competition and skepticism about what it is that is being promised by the copper-silver science,” Stephanie Bird, editor of the Science and Engineering Ethics journal, said after viewing the CDCs emails cited in the Post-Gazette’s stories. “But competition is competition and it leads people to try to present their data in the best light that also presents skepticism about contrary claims.”

But if that bias plays a role in how research is conducted or reported, it could raise questions.

“Whenever there is a change in data from what you report to your colleagues [privately] to what you put in your [public] reports, the question of intent is always there,” said Zubin Master, a research ethics expert and an assistant professor of bioethics at Albany Medical College in Albany, NY.

Dr. Master considers changes in interpretation to be a a change in data.

“At” versus “within”

The CDC’s reporting of its test results raises questions.

The 2013 lab results showed that a 400/40 copper-silver solution “at” 24 hours had found a “significant decrease in viability.” But that’s not what the CDC reported in a 2015 article in the Clinical Infectious Disease journal.

Instead, the journal article said that amount of copper-silver “failed to reduce viability of any strain within 24 hours.”

This change in reporting raises a question: How did the test results go from seeming to demonstrate that copper-silver worked against Legionella bacteria to appearing to not work?

The lead and senior authors of the 2015 journal article were, respectively, Alicia Demirjian, a CDC epidemiologist who was the leader of the Pittsburgh investigation, and Lauri Hicks, the CDC’s Legionnaires’ expert, with Ms. Lucas listed as one of the co-authors. The CDC would not make either Dr. Demirjian or Dr. Hicks available to answer questions for this story.

In an inteview, Ms. Lucas, who oversaw the testing, said the CDC’s resistance testing reflected in both her email from 2013 and the 2015 journal article are correct.

“The key word in that [journal] sentence is ‘within,’ ” Ms. Lucas wrote in an email response to the Post-Gazette.

In an interview, she said that is because the last time before 24 hours that she tested for reduced viability was 3 hours into the test. By then, at 3 hours, the copper-silver solution had not yet shown significant reduction, and would not show significant reduction until the next test at 24 hours.

As a result, she said, it was accurate to say in the journal article that the copper-silver had not reduced it significantly “within” 24 hours, even though it did “at” precisely 24 hours.

From “possible” to “definite”

There was another issue in which the journal article does not match with the CDC’s investigation results as shown in its emails.

During its investigation, the CDC looked back before the outbreak period to see if there had been hospital-acquired cases at the Pittsburgh VA before 2011. In the five years between 2006 and 2010, it found 10 cases going back to 2006 but none likely originated at the Pittsburgh VA, according to a CDC email. The same email indicated none of the 10 patients stayed longer than one day at the VA.

But two years later in the 2015 journal article, that data was reported this way: “Our retrospective review revealed 1 definite and 9 possibly health care association cases of LD associated with VAPHS in 2006-2010.

How one of those 10 ccases went from being “possibly” health care-associated in the email, to “definite” in the journal article is not clear.

Under the CDC’s definition, a “definite” case could be confirmed only if: an analysis of the Legionella bacteria recovered from the patient matched the bacteria found in the hospital’s water; or the patient had been continuously hospitalized at the Pittsburgh VA for the entire 14 days prior to onset of Legionnaires’.

The CDC, however, would not answer questions about the change.

Choosing the right sample sites

Another issue with the data the CDC reported had at least one prominent investigator scratching his head in 2013.

In both the 2013 report to Congress, and the 2015 journal article, the CDC reported during its investigation in Pittsburgh it had tested 11 water samples drawn from various spots at the VA.

In seven of those 11 samples, it found copper-silver levels above the manufacturer’s target levels of 200 parts per billion copper and 20 ppb silver. And the “mean,” or average concentration of those 11 sites was rounded to 300 ppb for copper and 20 ppb for silver — again, both at or above the manufacturer’s minimums.

But the owners of Liquitech and Enrich, the two companies that — respectively — made and maintained the copper-silver systems in use at the Pittsburgh VA, said this was the one area of the investigation they always disagreed with most. That’s because four of the 11 sample sites were located “prior to” or “immediately after” copper-silver units in the building.

The water near the copper-silver units would not be properly circulated and would typically read much higher than, say, a bathroom sink in another part of the hospital.

And that is just what the CDC’s samples showed: Four of the five highest copper readings were from those four sites near the copper-silver systems, and the top two highest silver readings were from there, too.

If those four sites near the copper-silver units were removed from the calculations, the average for copper drops to 240 ppb — just above the manufacturers’ minimum — and the average for silver drops to 15 ppb — or below the minimum.

Of the remaining seven samples, four of them have measurements below the manufacturer’s minimums.

Jerome Herbers, the VA official who lead the VA’s Office of Inspector General investigation of the outbreak, said he was bothered by the CDC’s use of those four central sample sites in 2013.

“This was our primary disagreement with them,” he said. “We said, ‘You sampled from the wrong place.’ But they disagreed.”

Though he never understood why the CDC took the stance they did on sampling, “Now that you show me the background [in the emails] it makes sense to me: They had already written it off.”

In an email she sent on Jan. 3, 2013, Ms. Lucas warned her colleagues that they should think about including still other samples the CDC took in the hospital because they would “skew your results.”

“I went ahead and did the math including them as distal [situated away from the center] in the report, but I’m thinking it’s not entirely fair or should at least be mentioned,” she wrote.

In the end, the CDC did not include these other samples.

Mr. Master, the research ethics expert, said he was bothered by Ms. Lucas’ email as well as the other use of data reflected in the CDC emails.

“You don’t cherry pick the data points so that it fits your hypothesis,” he said. “You can drop or not drop certain data, but only if they met conditions you set before you test.”

Click here to read the full article on the Pittsburgh Post-Gazette. 

2016-12-13T23:18:23+00:00