As the death toll has reached 1,000, there has been a lot of press coverage in the US this past week concerning the shortages in the availability of the swine flu vaccine. Planned vaccination clinics for the millions of high-risk individuals in the US, including pregnant mothers, children under the age of 10, and healthcare workers across the country, have been postponed en masse with no word on when they will be rescheduled.
The mess has turned into a major political problem for the Obama administration, which admits that it had been "too optimistic" in regard to promising large quantities of the vaccine would be available by now. In fact, the US Centers for Disease Control (CDC), in what then looked like a very conservative estimate, said in July that it expected that there would be at least 50 million doses of swine flu vaccine available by mid-October; in actuality, there were only 23 million doses available by the end of October.
As the political heat has increased, the administration has sought to blame the vaccine manufacturers for not being honest about the true state of their vaccine production, while the manufacturers are saying that isn't true. The US Congress is blaming both the administration's management of the program and the manufacturers, while the American people are saying a pox on all your houses.
As I have watched this situation play out (see my earlier Advisor observations: "As Swine Flu Pandemic Lurks, Confusion About Strategy Reigns," 30 July 2009, and "A Flu Pandemic: To Risk or Not Risk — That Is the Question," 7 May 2009), I keep getting this sense of déjà vu as if I am involved in the risk assessment of a big-bet IT project.
To understand why, let's rewind the clock to February of this year, for that is when the decision was made about flu strains that would be making up the 2009-2010 seasonal flu vaccine.
If you recall, the US 2008-2009 flu season was relatively mild, with a good match between the flu vaccine and the major strains of flu then in circulation. Flu vaccines are considered a good match when they are 70% to 90% effective against the seasonal flu; because of a poor mismatch, the 2007-2008 flu season was severe, as the vaccine produced had an effectiveness of only 20% against the A strains of flu virus and 2% against the B strains in circulation.
There was also plenty of flu vaccine available last flu season, with 140 million doses produced (and 113 million distributed -- the highest ever) for the US, as opposed to shortages in previous years.
In February, the three strains of flu virus to be vaccinated against for the 2009-2010 season were selected (an A/Brisbane/59/2007 {H1N1}-like virus, an A/Brisbane/10/2007 {H3N2}-like virus, and a B/Brisbane/60/2008-like virus), and an estimated 115 million does were to be produced. The decision was also made to attempt to get the vaccine distributed toward the start of school in September, before the flu season really started to hit, which is usually in early December, with its peak occurring in February. In past years, vaccines have arrived late or in small quantities, thus limiting their general effectiveness.
However, in April, a wrench was thrown into plans for flu production. The first reports started coming out of Mexico of a new swine flu — called novel H1N1 — which seemed to spread more rapidly than seasonal flu, and looked alarmingly dangerous from first reports.
Government health organizations around the world, especially those in the Southern Hemisphere, where it was approaching winter and seasonal flu season, immediately turned on their contingency plans to mass-produce a swine flu vaccine as quickly as possible. In fact, the US 2005 National Strategy for Pandemic Influenza called for every American to be vaccinated within six months of a pandemic onset. Many of these plans had been developed as a result of the SARS epidemic in 2003 and the bird flu scare of 2005-2006.
Even though H1N1 was soon to be seen by health officials as not being as lethal as the 1918 pandemic, most people under the age of 55 had no immunity to it at all; the last pandemic flu was in 1968. Worse, there was a good probability that H1N1 might mutate into something more lethal as it started to spread into the general population.
So, the US government, like others, decided in late April to place a big bet: bulk-produce and test for effectiveness and safety a new flu vaccine within six months, and distribute the vaccine in massive quantities across the county, while simultaneously producing and distributing a very large quantity of seasonal flu vaccine.
To say the least, that plan was ambitious and unprecedented, and required a lot of concurrent activities to occur successfully. The plan was also going to require more than a little luck.
Normally, it takes nine months to develop and test a flu vaccine. You have to choose the right flu strains to target, which is always guesswork. Then you have to culture the strains using hundreds of millions of eggs — one egg per strain of flu virus. Then, when the vaccine is shown to be safe and effective, you still have to fill up the syringes with vaccine and ship them out. Problems can — and often do — occur at every step of what is acknowledged to be a high-risk process even in the best of times. In addition, there are a limited number of vaccine suppliers because of the business risk involved.
Problems that may arise include the possibility that the yield from the cultured strains may be low, or the culture may take longer than is typical, as happened in 2001 and 2007. Or there may be packaging problems, as in 2004. The likelihood of something going wrong is always present, and the resultant risks are an accepted part of the process as well as the business. Delays of a few weeks may mean the difference or being proactive or reactive in the healthcare community's response to a flu outbreak.
In late April, only weeks after initial reports of the novel H1N1 flu, the US government set into motion a major plan to produce enough vaccine to initially vaccinate the 155 million Americans that met the high-risk designation in the span of one or two months. That alone would require more than 300 million total vaccine doses, since it was believed at the time that two doses would be needed per person to provide immunity. Another 300 million doses would then be needed to cover the remainder of the US population. Producing, distributing, and then physically giving that many vaccinations in either number or time frame had never been done before.
In early May, the requirements for a massive H1N1 vaccination plan were being developed, based on the best science at the time, which was still pretty uncertain. But time was not a luxury to be had. H1N1 cultures had to be developed, contracts had to be entered into with vaccine manufacturers, and distribution logistics figured out.
In June, the World Health Organization (WHO) declared a pandemic situation as the H1N1 influenza virus quickly spread around the world. The US government evaluated different vaccine production scenarios in conjunction with the vaccine manufacturers that it had contracted with in May to deliver an initial 195 million shots, on their capability and capacity to develop between 120 and 160 million doses of the vaccine by October, with 80 million more a month produced until 600 million doses were produced. It looked doable, although more study was required.
In mid-July, things starting looking a bit problematic in achieving the planned vaccine dose totals. Manufacturers were saying that the yields of the vaccine cultures were much lower than for a normal flu — as much as 70% lower — meaning lesser amounts of vaccine would be available by autumn unless production could be ramped up even more. Also, the WHO was saying that the H1N1 spread at a rate nearly four times faster than similar viruses in the past, putting added pressure on producing the vaccine quickly.
However, the US government reported that, despite the production problems, there should still be 120 million doses available by the middle of October, with 80 million more each month after that, through March 2010 if necessary. The CDC said in late July that at the very least, 50 million doses of flu would be available by October, and that number might be sufficient, since the uptake they expected initially was for 40% of the 155 million in the high-risk populations to opt to get inoculated. Since a person had to wait four weeks or so between their first and second shot, there would be time for any production shortfalls to catch up.
By mid-August, the US government started backing away from those estimates as production problems were cropping up. Not only were low yields a problem, but also the seasonal flu production was taking longer than expected and there were syringe-filling/packaging problems. The government still was publicly confident, however, that 45 million-52 million shots would be available by mid-October, 85 million by the end of October, and 195 million by the end of the year.
The good news was that the H1N1 virus wasn't mutating in any major way.
In early September, things started looking up. Clinical data showed that one shot was all that was needed for adults, while for children under 10, two shots was still the planning scenario. The total number of shots required would drop dramatically, as would the consequences of any further vaccine production problems.
On 15 September, the US government was still confidently saying that 50 million vaccine doses would be available by mid-October. In fact, just a few days earlier, it had announced that not only would there be ample supplies of vaccine, but that it would reach doctors a few weeks earlier than planned.
To the public, it looked like while there had been some hiccups, H1N1 vaccinations would be starting soon and fears of a nasty flu season averted. The government approved the vaccine as being safe and effective in mid-September.
However, in early October, the promised vaccine was not showing up to the doctors' offices that had ordered them. In addition, vaccination clinics were running out of vaccine quickly, as 10% or less of their orders were being fulfilled. Other clinics were being cancelled around the country. The public, the press, and Congress started asking, "Where's the vaccine?"
In mid-October, the US government finally admitted that there would only be 28 million doses available by the end of the month, and that less that 11 million of the 40 million promised had been shipped by 15 October.
In the ensuing outcry, the government defended its actions, saying that it had basically done the impossible: create a novel flu vaccine and get it out to the public in less than six months. That was true, but it was lost out in the uproar that the government had also repeatedly promised lower and lower vaccine amounts that weren't now being delivered.
Only after getting beaten up badly in the press did the government admit that it had been "overoptimistic," but then it immediately placed all the blame on the vaccine manufacturers. The manufacturers responded by saying they were in daily contact with the US government, and let it know exactly what was happening all along. They also pointed out that the government had been slow in developing quality control tests on the batches of vaccine they produced.
The current expectation is that about 28 million doses will have been produced by the end of October, with another 10 million a week or more from then on -- again, assuming no other glitches crop up. That way, some 150 million doses would be available by the end of the year.
The US government bet big, but did not communicate very well to the public the risks involved. Nor did it manage expectations very well. Previous histories of much smaller runs of seasonal flu production indicated that it was unlikely that the initial targets of producing 120 million doses would be achieved, for instance. But I believe there was extreme pressure to try to act "upbeat and confident" so that the American people wouldn't be worried about not being able to get a flu shot. Keeping the risk of panic down no doubt played a role in the optimistic judgments being made.
In fact, the US was very lucky in that few of the risks involved in speeding up the vaccine development by concurrently producing and testing the vaccine appeared, and that the H1N1 virus didn't mutate and become more lethal. The vaccine produced looks safe and effective. In addition, it now looks like children will require only one dose instead of two, thereby relieving the shortfall even more.
The government was also lucky that nothing went wrong with the seasonal flu vaccine production. If something had gone wrong, there were few if any back-up plans other than to wash your hands a lot. The government hasn't highlighted how lucky it — and the American people — have been, which might have reduced the current angst somewhat.
Big-bet projects require absolute honesty about the risks and problems involved, and the constant reminder that while you are trying your best, you may fail. It is, after all, a big bet — and big bets can go wrong very quickly and in spectacular fashion. If you are willing to spend the time, you can see that the government did warn that there might be vaccine production problems, but those were lost in its continued optimistic forecasts meant to reassure the public that it was working hard on its behalf. The US government didn't completely fail the risk communication test, but they didn't get an A grade either — more like a C or at best a C+.
This pandemic has shown many weaknesses in the production and distribution of vaccines as well as the communication of the risks involved. I hope that the lessons learned now in a low-lethality flu situation, will be applied the next time — and there will be a next time when another big bet has to be made. We may not be so lucky then.
