Debate on CFR from swine H1N1v is valid, comments invited.
Recently, a commenter named H1N1 Watcher posted a comment regarding my blog on the Case fatality Rate from swine flu. Even though I posted the following as a response, I felt it worthy of listing as a separate blog entry. The scientific debate is valid, and I felt it worthy of attention.
"However, since he appears to be using published numbers from the Centers for Disease Control and Prevention (CDC) and the World Health Organization (WHO) his method is consistent. We cannot propose a better method,
"Can't we ?
"As for sound evidence based estimates of the CFR,
plase (sic) take a look at the recently published Wilson paper [1] that provides reasonable estimates of upper and lower bounds of pandmic (sic) H1N1 case fatality rate.
"According to that Study, which, as a meta study takes into account several independent estimation methods the upper bound of the CFR (i.e. the most pessimistic estimate) is as low as 0.06 % (that is 6 in 10.000) which is one order of magnitude less than previous estimates of 0.5% and, BTW perfectly in line with seasonal flu CFR.
----------------
[1] N. Wilson
"The emerging influenza pandemic: estimating the case fatality ratio"
Free full text available at
http://www.eurosurveillance.org/ViewArticle.aspx?ArticleId=19255
July 29, 2009 | h1n1_watcher "
Dear H1N1 Watcher,
First, let us all hope it is from your keyboard to God's ears on the CFR!
A few comments.
First, a low CFR is what we are hoping for, but not necessarily what we should be planning for. In my blogs at Computerworld, I explain the concept of planning for one category higher than the event we are told to prepare for (the hurricane analogy). Even CDC is telling us the CFR could be from the low thousands to the hundreds of thousands, and we have to listen to and respect our own public health establishment.
Second, I cannot find a pandemic in the era of even remotely reliable medical history (1830s pandemic to present) where the second wave was milder than the first. Assumptions have to side with the theory that the virus becomes more efficient and more virulent as it progresses.
Third, the Wilson study does not, as far as I can tell, take into account misdiagnosed causes of death. An earlier blog of mine attempts to take that into consideration. I believe that as we get into larger and larger numbers, a pattern emerges. One cannot assume there are a higher number of "milder" cases and then take the total reported number of deaths as gospel. You have to account for missed deaths due to H1. The CFR was climbing as the virus was reaching its tendrils across the USA, stabilizing somewhere around .005, plus or minus. This was with massive swabbing and lagging results and further lagging of reporting those results.
To quote from the study (you referenced):
"Sophisticated statistical methods have been proposed for obtaining adjusted CFR estimates using data from the early phase of an epidemic [20], and these result in adjustment for various time lags and an upward shift of the CFR. However, such adjustments would probably have little effect on the estimates presented in this article which are based on data from country epidemics which have progressed well beyond their early stages (e.g. the Canadian data). There is also the potential for under-recognition of deaths attributable to influenza in those with serious co-morbidities, but this can only be addressed by careful research studies and post-epidemic modelling to determine total excess deaths."
My takeaway from that is there is no reliable math to calculate deaths that were not correctly attributed to H1 in real-time. There actually is a way to do that, which the CDC performs. You take your baseline death rate, look for spikes in excess mortality and attribute that to the virus. But that reporting also lags, sometimes by weeks. So we are constantly steering by our own wake and not looking forward.
For all these reasons, it is absolutely prudent to plan for a more serious event.
I wanted to thank you again for the link. I shall read this again and attempt to obtain the document referenced in #20.
Scott
I think this is a valid course of discussion. I think we all have this feeling that the actual CFR was not as bad as the numbers indicated. Nonetheless, all the classic warning signs of pandemic mortality -- deaths of pregnant women, younger adults and older children, the lack of infection in the elderly, viral attacks upon brain, stomach and intestines -- these are screaming at us to be very wary of this new strain.
Ultimately, we will never know the actual CFR until months to years after the event. That is the realm of historians and statisticians. But in the near term, we will have to deal with the impact of the virus, moderate to severe CFR or no CFR. And that reality is a third of our population will get sick and our ability to conduct our business will be challenged. A disproportionate number of young adults will become very, very ill. Schoolchildren will be severely impacted. Pregnant women will die; the exact number unknown, but each death felt and great sadness will result.
It is absolutely prudent to get the message out and prepare.
Your comments welcomed.
Reader Comments (3)
Despite the encouraging numbers so far I too think that we must take this pandemic seriously.
The thing that worries me most at that time is the results of the three independent animal studies published so far (ferrets, mice, ...) that show significantly more widespread respiratory tract damage caused from the pandmic H1N1 virus compared to seasonal H1N1.
And this is a marked incompatibility to the very low CFR estimated in the said Wilson paper.
However, as those studies are about small number of animals I rather stick to the very large numbers of humans that the CFR estimates of the Wilson paper are based on.
But nonetheless, those animal studies are disturbing and a reminder to take this pandmic seriously.
And I Agree 100% that it is better to prepare for the worse case than to be caught by surprise after underestimating the risk.
Hi Scott -
Good advice to prepare for what has historically been a worse Wave 2.
I have a question for you: Given that we both suspect, per the Englishman, that Wave 1 victims harbor something more than antibodies, do you think the Chinese are asking for BIG trouble in their intensive efforts to suppress Wave 1--or might that turn out to have been rather clever?
Just curious as to what you think at this moment. I watch the sequestrations with interest--perhaps a big experiment in motion.
Best, Peter
I've read this article and then corresponded with the author.
I said to him:
"In the USA, for seasonal influenza the CDC uses post-epidemic modelling to estimate total excess deaths (in the USA) and estimates an average figure of 36,000. In 2005 for example influenza only appeared on 1,812 death certificates."
"If the CDC used surveilance data here to make their estimates they would be getting an underestimate of more than an order of magnitude. I agree that a younger population and better surveilance methods might change this somewhat but can you be sure that your CFR calculations are not still a large underestimate?"
He responded as follows:
"I agree that this is an area of substantial uncertainty and our article really does stress the uncertainty of all the methods used. Nevertheless
of note is that in our own country (NZ) there seems to be a very high index of suspicion by clinicians in that 2 people who died were apparently only diagnosed as being infected with this new virus as a result of blood tests taken at autopsy. This is likely to be associated with the youthful nature of most of the deaths where clinicians appear to pay far more attention to getting a correct diagnosis. But under-ascertainment of deaths in the 65+ age group is very likely to be a problem, so your point
is particularly relevant to that group.
Nevertheless the approach we described that extrapolated from the under 65 year age group (based on work by Thompson et al) was based on the modelled
excess deaths. Hence this approach should not be subject tounder-ascertainment of deaths."
In my opinion this response does not adequately adress the issue raised (and the issue affects 3 of 4 of his estimates). The effects that Wilson notes are probably significant but seem unlikely to be sufficient to make detected deaths a valid estimate for total deaths (after all it is normally out by about a factor of 20). And the fourth method he says is unaffected is not an independent calculation of CFR. Rather it assumes that the CFR by age group is the same as seasonal flu and works from that starting point. But thats a dubious assumption for a novel flu strain as no one has immunity to it and there is no vacination for it (which normally substantially reduces the CFR).