Health watchdogs lacked Legionnaires' data for Pittsburgh VA hospital
Dr. Ronald E. Voorhees, acting director of the Allegheny County Health Department: "Did we have all the information we should have? It's a valid question. ... We should have noticed this."
Share with others:
In theory, the Legionnaires' disease outbreak at Pittsburgh Veterans Affairs facilities that infected 21 men and killed five of them in 2011 and 2012 should have been caught much earlier.
County, state and federal health officials should have had the data about the cases that told them an outbreak was under way as early as the summer of 2011, when the second, hospital-related case occurred.
But a review by the Pittsburgh Post-Gazette reveals that problems with the way those 21 cases were -- or were not -- reported, and the flaws in the National Electronic Disease Surveillance System that should have had all the data, meant that none of the government health watchdogs had all the necessary information in time to help prevent the outbreak from spreading.
"Our goal is to catch things as they happen," said Ron Voorhees, acting director of the Allegheny County Health Department. "Did we have all the information we should have? It's a valid question. Between the Allegheny County Health Department and the Pennsylvania Department of Health, we should have noticed this."
The problems occurred throughout the system, and include:
• The Pittsburgh VA apparently did not report, or properly report, six -- about one-third -- of the 21 Legionnaires' cases into NEDSS, which is run by each state's health department and which then feeds information back to the counties;
• Pennsylvania's NEDSS database only lets Allegheny County health officials see data on cases involving Allegheny County residents, which means Dr. Voorhees' staff only knew about eight of the 21 Legionnaires' cases at the VA, and his staff would not have seen, for example, two of the cases in which patients from Beaver and Westmoreland counties died;
• Though the NEDSS database asks questions about whether a disease was contracted in a hospital and to name the hospital, until mid-2012 the Pennsylvania Department of Health did not analyze those answers to look for outbreaks tied to hospitals;
• Questions about whether a disease was hospital-related are on the long-form version of the Pennsylvania NEDSS disease questionnaire, but the state says the long form is not always filled out by hospital or laboratory employees;
• At the request of the federal Centers for Disease Control and Prevention, the state health department asks field investigators to follow up on Legionnaires' cases by filling out a separate, disease-specific questionnaire that also asks whether a case was contracted in a hospital and to name the hospital and then to fax it to the CDC, but the CDC won't say what it does with that information;
• Even if it knew there was an outbreak at the Pittsburgh VA, the CDC said it does not have the authority to insert itself into a case and can only be invited to help analyze a problem, as it was in this case.
The problems are not particular to the Pittsburgh VA Legionnaires' outbreak. Flaws in the NEDSS reporting database have been the source or concern by health officials for decades -- flaws that newer, existing technology could help correct, but that has not yet been implemented, experts say.
"We are still using 20th-century technology [in the system] that is ineffective and inefficient," said Jeffrey Levi, executive director of the Trust for America's Health, a Washington D.C.-based nonprofit that seeks to make disease prevention a national priority.
'System of the future'
Changes to the system could be addressed if President Barack Obama's National Strategy for Biosurveillance, which would include significant technological improvements in the NEDSS system, is fulfilled. The president proposed the strategy this past July, though Dr. Levi points out it has yet to be funded.
"Part of the difficult decision making that has to occur in the next few years is about the significant amount of investment we already have in health information technology," Dr. Levi said. "Can we harness the support to allocate dollars differently" to improve the system?
Jeffrey Engel, executive director of the Council of State and Territorial Epidemiologists, said changes would have to be made because "NEDSS is not set up for outbreak management."
But, currently, "outbreak recognition now really depends on astute clinicians and good relations among people [like county and hospital officials] not so much information management systems," said Dr. Engel, whose Atlanta-based member organization works closely with the CDC in refining NEDSS annually.
The system Mr. Obama envisions would have all hospitals and laboratories automatically uploading all pertinent information on every reportable disease case to NEDSS, giving government health officials the data to help hospitals locate outbreaks.
"That's the system of the future," Dr. Engel said.
That system wasn't in place during the VA outbreak.
County, state and CDC health officials didn't notice there was a problem until the second week of November 2012, when the VA called them in to tell them it thought it had an outbreak and the VA needed help figuring out what to do.
The reasons it took so long start with problems in the hospital.
It appears that the Pittsburgh VA was not entering -- or was improperly entering -- all its Legionnaires' cases into NEDSS.
An internal Pittsburgh VA document obtained by the Post-Gazette says that by Nov. 9, 2012 -- the second day of the CDC's field investigation in Pittsburgh -- the CDC only found 12 possible Legionnaires' cases at the VA.
That is six fewer cases than had occurred by then -- two more occurred after Nov. 9, and one wasn't confirmed until the patient was at another hospital. The CDC got that figure by viewing the county database, the VA's internal database; and presumably the CDC looked at NEDSS.
Asked about this discrepancy, Pittsburgh VA spokesman David Cowgill wrote in an email: "In terms of the 21 patients reported by the CDC, VA Pittsburgh has entered 20 of the 21 cases using the PA-NEDSS program."
Asked in a follow-up question why, then, there was a discrepancy when the CDC did its first count, Mr. Cowgill declined to comment. The CDC also would not explain why it only found 12 cases in its first count.
"We are not concerned about the hospital's case reporting," CDC spokeswoman Alison Patti said in an email. "It was not identified as a contributing factor to the outbreak."
Holli Senior, spokeswoman for the Pennsylvania Department of Health, the state's NEDSS administrator, would not say whether the Pittsburgh VA did or did not report every one of the 20 Legionnaires' cases that it diagnosed.
If the Pittsburgh VA was a state-licensed facility, it could have its state license revoked or restricted for not reporting its reportable disease cases like Legionnaires' within 24 hours of diagnosis. But, because it is a federal facility, the Pittsburgh VA would not face such discipline even though it voluntarily complies with state law.
The state has no idea how common it is for hospitals or laboratories not to report a reportable disease because it makes no effort to audit the filings, said Kirsten Waller, head of the health department's surveillance section in the division of infectious disease epidemiology.
Even if it had reported all its cases, Ms. Senior said, that wouldn't necessarily have helped health officials detect an outbreak because "all that information is not always filled out" on the forms.
"You can ask anything you want [on the forms], but you're not always going to get the information back that you're asking about," she said.
She said until sometime in mid-2012, the state health department did not attempt to analyze the case questionnaire data to look for outbreaks in specific buildings other than nursing homes, though she would not say why except to say it was not related to the VA outbreak.
The internal Pittsburgh VA document was later updated Nov. 14, noting that in the wake of the CDC investigation, that VA pledged to "enhance reporting mechanisms to infection prevention and to the national surveillance system."
Mr. Cowgill would also not answer questions about why the VA pledged to do that.
'We might have asked more'
Dr. Voorhees of Allegheny County said that to his staff's memory, the county did see the reports of the eight county residents who contracted Legionnaires' at the Pittsburgh VA in 2011 and 2012.
But each time a staff member called to ask about those cases, they were told that Pittsburgh VA infection control experts considered the cases nonhospital related because they could not find Legionella, the bacteria that causes Legionnaires', in the rooms where the patients had been staying, a method other experts said was inadequate.
If the staff had known about the other 13, noncounty residents' cases (including one of an out-of-state resident), he said, "we might have asked more or different questions" of the VA.
In addition, Dr. Voorhees said the information his staff currently sees on NEDSS presents the case data using a confusing coded system, and the answers can include more than one response, which confuses it even more.
"From our perspective, it's really not usable data," he said, which is why his staff calls on every Legionnaires' case.
"We need to sit down with everyone and look at this," Dr. Voorhees said. "At a minimum, there needs to be some way all these surveillance systems aren't hampered from finding out what's going on."
First Published March 10, 2013 12:00 am