Needless to say, the extra effort needed to respond to the H1N1 crisis strained the under-staffed, under-funded and often ill-equipped laboratories.
In March, 36 state public health laboratories were using or were eligible to use the CDC’s five-target influenza assay, having both the required ABI platform and trained staff. These labs were in a relatively good position to implement the swine flu panel, needing only to validate the test in-house; a process that could take anywhere from a few days to a week.
But there were glitches. Susan Neill, PhD, MBA, director of the Texas Laboratory Services Section, said, “When we started (testing), we had one instrument we could run the test on. The first thing that happened was it broke.”
The Texas laboratory temporarily diverted specimens to a local public health laboratory in San Antonio, while the instrument was repaired. Neill also received FDA approval to use an older ABI platform and secured two loaner instruments from Applied Biosystems.
Adequate emergency preparedness, said Neill, demands “more than one instrument.”
A second group of 30-odd laboratories had at least one scientist who had just finished an NLTN influenza training course. Most of these labs were in the process of either upgrading an older ABI 7500 FAST or purchasing a new one.
In New York City, a hotspot for novel H1N1 activity, the city public health laboratory was using a manual testing system while awaiting a software upgrade for its ABI 7500 FAST and authorization from state authorities to use the swine flu panel. “Once the authorization was sent to us,” said Beatrice, “we had to train the flu staff in the midst of surge and then we had to train the surge staff... We were required to do validation, training, testing and upgrade the instrument simultaneously. It was a little hairy.”
Finally, half a dozen public health labs had no ABI 7500 FAST and no immediate plans to acquire one due to budget constraints.
Arizona’s Waddell had been trying for some time to find $72,000 to purchase the instrument, but was stymied by the simultaneous cuts in his state funding and federal grants. In fact, Waddell had wanted to participate in the CDC validation study of the original five-target assay, but without the ABI 7500 FAST his lab was ineligible.
As the H1N1 outbreak was unfolding, he said, “First thought I had was, ‘We better be able to get our hands on this instrument or we’re gonna be one of the only states that can’t do this testing.’”
Given Arizona’s proximity to the source of the outbreak and the rapid spread of illness in the US, Waddell was able to impress his interim health director with the urgency of the situation. The health agency allocated funds to place a purchase order for two of the machines.
In the meantime—since it can take weeks for a new instrument to be manufactured—the Arizona laboratory was able to borrow an instrument through a generous loaner program Applied Biosystems made available through APHL to laboratories needing the approved testing platform.
“We actually got the test up and running in a little over a week and a half after the first confirmation of the virus in-state,” said Waddell, “which is pretty much unheard of, considering we didn’t even have the instrument in-house to begin with.”
Despite such difficulties, by early June, every state and the District of Columbia had at least one public health laboratory with the swine flu assay up and running.
Once instrumentation issues were resolved, however, the bottlenecks for many laboratories became the front and back ends: specimen accessioning and processing and results reporting.
This was no surprise.
Last fall CDC contracted with Booz Allen Hamilton to map out all the steps involved in influenza PCR testing and to model how long it would take to go through those steps in about 20 representative state labs. The analysis showed that pre- and post-analytic work—all of the staff-intensive activities that take place before and after testing—were common limiting factors.
At one point, the Texas lab, for example, had 82 people working on outbreak response. Just 23 were actually performing testing, with the remainder assigned support tasks.
In New York City, Beatrice said, “The things that are always challenging for us have to do with appropriate collection, transport and documentation of specimens. So if the samples are the absolutely correct samples that are collected appropriately, stored at the right temperature and transported at the right temperature and all of the paperwork is complete and accurate, then the accessioning and processing go smoothly.”
In this case, with city hospitals overwhelmed at times, almost 40% of the H1N1 specimen submissions had problems. “That meant a great deal of one-on-one effort to reach out to hospitals and solve the problems.”
In Arizona, a general shortage of viral transport media prompted the state laboratory to mass produce media in-house using its own quality control program. “We found ourselves becoming more of a distribution site,” said Waddell.
As one of the laboratories participating in the APHL modeling study, the WSLH had the benefit of a preview of probable chokepoints.
Shult said the modeling analysis, which he received in February, identified two or three critical shortfalls: “We didn’t have enough PCR equipment, didn’t have enough people to do pre- and post-analytic work. We would need to have staff work longer hours and probably go to a seven-day workweek.”
Going into the crisis, he said, “We kinda knew what the gaps would be and we had started to take steps to address these.... Modeling helped to brace us for what was going to happen.”
'The Challenge Was Not Having a Well-Defined Disease'
Given the public health laboratories’ limited testing capacity, a critical question facing every major health agency was how much testing to perform. Said Humes, “Early in the introduction of a new influenza strain, PHLs have to do diagnostic testing because no one else has the capability. But once you know the disease is established in the population, diagnostic needs change and PHLs can switch to surveillance testing.”
The goal of influenza surveillance is not to identify every outbreak case, but rather to gather sufficient information to monitor genetic changes in the pathogen and changes in the epidemiology of the disease to inform public health interventions. While national authorities may offer specimen triage guidelines, in practice every state makes this determination for itself based on past experience and current exigencies.
Wisconsin, for example, experienced a significant H1N1 outbreak in the Milwaukee area and has had the highest case count in the country so far, with 4,273 cases reported to the CDC by June 25, compared to 1,492 cases in California and 65 in Georgia.
Shult said, “If you’re too restrictive (in choosing which specimens to test), you’re not going to judge the full impact of the disease accurately. If you’re too liberal, you run the risk of becoming quickly overwhelmed. Our state’s health department took the approach that we’re going to be more liberal.”
Texas also took a liberal approach, at least initially. Lakey said, “The challenge for us was not having a well-defined disease... We knew there was ongoing replication (in Texas). Then we found out about Mexico’s high fatality rate.”
There was also the sheer size of the state to consider. With 254 counties, Lakey said, “Just because you have cases in one part of the state doesn’t mean you will find them in another part.”
By casting a broad net, however, the Austin branch of the state public health laboratory was quickly swamped with hundreds of specimens, about 90% of which tested negative for H1N1.
“It has been my experience throughout all disasters,” said Lakey, “that there are always opportunities to improve your response. I think learning the most appropriate utilization of the laboratory for public health instead of clinical work was a major lesson.”
The state revised its specimen submission policy several times, making it progressively more restrictive. As of mid-June, the state laboratory in Austin had tested more than 9,000 suspect specimens, with the positivity rate running about 16%.
On the other side of the country, Beatrice was working within a system tested and honed in some of the country’s worst disasters.
She said, “I truly believe our pandemic response did work. There was a great deal of attention to asking, ‘What is the most important public health question that needs to be answered and how do we go about doing that in a way that honors everyone’s resources: the PHL, clinicians, epidemiologists?’ And so if the question was, ‘Is swine flu here?’ or ‘Was there community transmission?’ the most logical process was in place instead of everyone sending in samples.”
Said Beatrice, “The lesson we learned in 2001 around anthrax was if every possible sample is allowed to flow into the lab, then when the highly important samples come in they can get stuck at the back of the queue and the critical answers can be delayed. That didn’t happen here.”
By mid-June, the New York City public health laboratory had received about 1,700 suspect specimens. Nearly 60% were confirmed as novel H1N1 and an additional 5% were seasonal flu.
Limits of All-Out Laboratory Effort
Looking back on the outbreak, laboratory leaders marveled at the timeliness of CDC support and the dedication of their own staffs.
They mentioned other federal help as well. The Centers for Medicare and Medicaid Services, for example, delayed its routine regulatory surveys and suspended influenza proficiency testing in state labs during the crisis. The US Department of Homeland Security not only permitted BioWatch staff working in public health laboratories to shift to H1N1 response, but also paid them overtime.
But even with this support, none disputed the limits of the all-out laboratory effort.
APHL’s Humes said, “You can only have people work so many 16-hour days. At some point, you have to give them time off or mistakes will start to happen. You have to have the capability to rotate people.”
In Texas, the incident command center set up to coordinate response activities had a work schedule of five 12-hour days. Neill said, “They went through five rotations of staff and we still had the same laboratory staff working.”
Eventually, staff burnout led Neill to bring in 25 temporary workers. But she cautions against this approach as a routine strategy. “If testing needs exceed capacity, people and resources have to be planned and put in place ahead of time; not just hire temps three weeks into the situation,” she said.
In Arizona, staffing constraints put Waddell in an impossible situation. Near the height of the crisis, his pregnant wife went into labor, facing what was expected to be a tricky delivery. At the same time, the laboratory was in crisis mode with no managerial staff to spare. Fortunately, one of his assistant bureau chiefs returned from vacation just in time to cover for him. Daughter Lily was born May 6.
One innovation that would have tremendously eased the burden on public health laboratories is standardized electronic messaging.
For several years, APHL and partners have been working on a project to equip all state labs with multi-directional data exchange capabilities with CDC labs and local partners. So far, however, only four state labs have the ability to send electronic influenza test results to the CDC, and 11 are scheduled to be live by the fall 2009 flu season.
In those states, the ABI 7500 FAST communicates directly with a laboratory’s in-house information management system, which, in turn, sends approved influenza data to the CDC, automatically populating data fields in the CDC’s own information systems. The possibility of transcription error is eliminated because there is no transcription at either end and data are transmitted in near real-time.
The laboratories without this capability use more time- and labor-intensive means of communication: fax, phone, e-mail and a web-based portal requiring manual data entry.
Patina Zarcone, MPH, who oversees the project at APHL, said $100 million is needed to develop public health laboratories’ IT infrastructure and implement electronic messaging nationwide.
In the meantime, the Houston public health laboratory, for example, has 13,000 sets of test results that have yet to be entered into the lab’s computer system.
A second, much-needed resource is a national reagent stockpile. Since 2002, APHL has advocated for inclusion of reagents in CDC’s Strategic National Stockpile, a reserve of pharmaceutical and medical supplies maintained for use during a public health emergency.
Although a full-blown shortage of the commercial reagents needed to support PCR testing never quite materialized during this first wave of the H1N1 outbreak, reagents were sufficiently scarce that the CDC sequestered a supply for distribution to public health laboratories in the US and internationally.
Perhaps the most pressing laboratory need, however, is predictable and adequate long-term funding. With current resources, Humes said, “If we had had sustained levels of infection throughout the next few months, it would have compromised every state lab’s ability to deliver routine public health services.”
Although the novel H1N1 outbreak is now officially a pandemic, that designation has little impact on the US response, which has already peaked for the time being. With the onset of summer, the incidence of infection has slowed, but remains unusually high for this time of year, especially in Boston, New York City, Seattle, Chicago and a few other large metropolitan areas.
Shult, who lectures extensively on the epidemiology of influenza, explained that we may be nearing the end of the initial wave of disease, at least in the US, but cannot rule out a second or even a third wave to come.
What will happen when influenza season officially resumes in the fall?
“We can’t predict right now,” said Shult. “But we can come up with a small universe of possible outcomes.”
The virus may weaken and simply die out in the Southern hemisphere, which is now in its flu season, but Shult concedes this is unlikely.
It may undergo antigenic drift and become less virulent or more virulent, less transmissible or more transmissible.
Or it could re-assort its genetic material with other seasonal or novel viruses and pick up more pathogenic elements, maybe even gain antiviral resistance. Egypt, for example, is known to have H5N1 in its poultry population and has had dozens of recent cases of both human H5N1 and H1N1, two viruses that would be dangerous to mix.
The current novel H1N1 has an attack rate of about 18%; meaning about 18 of every 100 people exposed to the virus become ill. A few genetic changes could be sufficient to boost the rate into the 20s or 30s, on a par with past viruses of major concern.
“Probably what will happen,” said Shult, “is somewhere between the best- and worst-case scenarios. However, you sorta have to plan for the worst.”
Regardless of what happens, novel H1N1 will continue to complicate influenza surveillance.
“If it does nothing different, it’s still added to the current group of seasonal viruses,” said Shult. “Swine H1N1 and human H1 and H3 and influenza B have different profiles of severity of illness and antiviral susceptibility. It still has an impact on the lab, because you need to know which virus you’re dealing with.”
For public health laboratories already struggling to make do with scarce resources, this is bad news.
In congressional testimony in May, Daniel Sosin, MD, MPH, FACP, acting director of CDC’s Coordinating Office for Terrorism Preparedness and Emergency Response, said, “With stronger laboratory capacity in states, we could accelerate the detection and study of new viruses such as the 2009-H1N1 virus, helping us better understand and respond to emerging health threats.”
But instead of enhancing laboratory capacity, the US is not even treading water.
Said Waddell, “Until you fail nobody’s going to take notice. But we don’t want to fail. That’s not what we do.”
“Maybe,” he said, “H1N1 will be a wake-up call to the federal government to put some money into the labs before it’s too late.”