Effect Measure

Yesterday one of the questions we asked was whether swine H1N1 would replace seasonal viruses this season. In previous pandemics one subtype completely replaced its seasonal predecessor: in 1957 H2N2 replaced the H1N1 that had been coming back annually at least since 1918; only 11 years later, in 1968, a pandemic with H3N2 replaced the H2N2. H2N2 is no longer circulating but in 1977 an H1N1 returned and has been co-circulating with H3N2 since then. This was a new situation. We could ask why this hadn’t happened before with H2N2 and H1N1 or H2N2 and H3N2 or all three together; or we could ask why it happened after 1977 with H1N1 and H3N2. We still aren’t sure of the answer, but the question has special pertinence at this point because it bears on whether we will be seeing three viruses this fall in the north (swine H1N1, seasonal H1N1, seasonal H3N2) or just the seasonal viruses or just the swine virus, or the swine virus and one of the others. We have a vaccine to cover some of the seasonal virus (there is a lingering question of a good match for the H3N2 component) but not yet one for swine flu. We can’t wait to find out and I, for one, will be getting the seasonal vaccine as soon as it’s my turn (which will be shortly at my medical center) and then get the swine flu vaccine when it’s my turn in the queue as it becomes available. To me that seems the most prudent way to cover my bets, although I won’t know until later whether it was the right choice. Meanwhile there is science to be done to help us understand all this, and within hours of yesterday’s post appearing there was a paper published in PLoS Currents/Influenza [PLoS Currents: Influenza. 2009 Aug 25 [revised 2009 Aug 27]:RRN1011] that provided a tantalizing data point on the subject of the dynamics of infection with combinations of swine flu and the seasonal flu A subtypes.

Researchers at the University of Maryland and colleagues tested combinations of swine flu and seasonal flu in ferrets. As most readers here know, the ferret is considered a reasonable animal model for transmission and infection for human flu, better in many ways than more traditional lab animals like mice or rats. This was a very small study but the results were clear cut enough that they provide a reasonable (although not ironclad) way to interpret them. The design was straightforward. A ferret was inoculated with swine flu virus (the original California isolate) or one of the seasonal H1N1 or H3N2 (Brisbane) viruses, or a combination of swine flu and each of the seasonal viruses (no ferret got all three or just the two seasonal ones). So that’s three five inoculated ferrets. Then two more “naive” ferrets (i.e., not inoculated) were housed with each of the infected ones, one in the same cage area, the other separated from the infected and direct contact by a wire mesh wall that allowed respiratory droplets but no direct contact. The idea was to see if the viruses alone or in combination could be transmitted. The number of animals is a bit unclear as the write-up is not very good, so I am not sure exactly what groups there were or how many. But the basic idea is obvious and the initial results are very interesting.

First, the swine virus is very contagious, at least as or more so than seasonal influenza. It also seems to have a shorter incubation period (look at Figure 1, top) and the authors state the animals developed higher viral titers, although this isn’t obvious from the data (there are no statistical tests given and the figures are only slightly suggestive). All the inoculated animals got infected by all three viruses, both one at a time and swine with each of the seasonal viruses, and shed virus for the usual 5 – 7 days. There was some evidence that swine flu infected more of the respiratory tree and the authors and evidence from the photomicrographs (Figure 2) show a more severe pathology for swine flu. There is also a suggestion that the combination of swine flu and seasonal H3N2 co-infection might be a bad combination, but this requires a great deal of further study. All of the animals inoculated with two viruses were infected with both. Swine H1 did not prevent infection an animal with season H1 or seasonal H3. What is happening within the tissues isn’t known, however. Can two viruses infect the same host cell? This happens with seasonal flu and is the way reassortment happens. But on an initial look there was no evidence of reassortment in this small experiment. We’ll come back to that in a moment. But as of now, each subtype stayed the same and there was no mixing and matching of the genetic segments.

So if co-infection is possible, how is it that swine seems to be crowding out the seasonal viruses? In this set-up, the evidence suggests it is happening at the level of transmission. If the animal is infected with swine and seasonal virus, it can infect other animals but it in this case only with the swine flu virus. There was transmission from inoculated animals to contact and respiratory droplet animals caged with them (as described), but the transmitted cases were only swine virus. In other words, although an inoculated animal was infected with two viruses, only the swine virus is passed on. How this works we don’t know yet. The authors only say that the swine virus is more genetically fit, giving it a competitive advantage, but this is only a description of what they are seeing, not an explanation of how it works. We need to understand this much better to know how general it is. What is quite clear is that the swine virus is fully adapted to humans and, as the authors remark, is less likely to reassort because the combination it has is better than the seasonal viruses. That may be why we have yet to see a change in virulence. A more virulent virus would only be selected for if it was more transmissible and this virus seems to be plenty transmissible without any help. A change in virulence could even decrease transmissibility by decreasing contact rates.

Finally, about the virulence (clinical severity) of the swine virus, which in this ferret model seems to be worse than the seasonal viruses. Not by a lot but by a noticeable amount. As the authors point out, the ferrets did have evidence of a small amount of pre-existing immunity to both H1 and H3. It wasn’t a lot and wasn’t enough to prevent the animals from being fully infected. There remains a chance that it somehow modulated either the transmission or severity of the infection. But as the authors also observe, this is also a good model of the situation the human population finds itself in. There is quite a lot of background immune experience with H1 and H3 viruses, and this may be the reason that swine H1N1 has a competitive advantage. With a contagious and rapidly replicating virus like flu, it may be that this is just enough of an edge to tip the disease dynamics to allow swine flu to trump seasonal flu. That’s speculation. There’s still a lot to figure out here.

This is not one of the better written papers I’ve seen, but it is extremely pertinent to the current situation and the work is well done by very experienced investigators. We’re learning a lot but most of our knowledge will probably not come in time to affect how we handle whatever swine flu has in store for us over the next year. As Hippocrates said, the Art is long, but life is short.


  1. #1 GCP Training
    September 2, 2009

    I am also getting my vacine asap. I think many people underestimate the situation…and the vacines…

  2. #2 Brenda Gundry
    September 2, 2009

    Reveres: Welcome back from vacation! Are you going to CIDRAP in Minneapolis later this month?

    Thank you for all your clear writing and indepth analysis.

  3. #3 JBH
    September 2, 2009

    Revere: You state that “the ferrets did have evidence of a small amount of pre-existing immunity to both H1 and H3…(and that)…There remains a chance that (this immunity) somehow modulated either the transmission or severity of the infection.”

    Do you mean the ferrets had some immunity to SEASONAL H1 (and not swine H1)and this potentially lessened the severity of infection? So swine H1 may not be necessarily a more severe virus than swine H1 (on its merits), but that the ferret immune response to swine H1 was simply weaker than seasonal H1?

  4. #4 anon
    September 2, 2009

    why should ferrets have _any_ immunity ?
    unless they are being used for multiple studies

  5. #5 JBH
    September 2, 2009

    Anon: Good question! I guess they were good little ferrets and got their flu shots last year. They probably have universal ferret health care and get vaccinations for free.

  6. #6 revere
    September 2, 2009

    JBH, anon: Yes, that’s what the authors state in the paper. As to how or why they got any immunity, we must assume that there is seasonal virus in their environment, perhaps from lab handlers or the suppliers. What I reported in the post can be found in the paper. The link is provided so you can take a look and interpret it yourself.

  7. #7 Dylan
    September 2, 2009

    Revere: In your second paragraph you state, “A ferret was inoculated with swine flu virus (the original California isolate) or one of the seasonal H1N1 or H3N2 (Brisbane) viruses, or a combination of swine flu and each of the seasonal viruses (no ferret got all three or just the two seasonal ones). So that’s three inoculated ferrets.”

    Actually, that would be five inoculated ferrets, as I read it; novel H1N1, seasonal H1N1, seasonal H3N2, novel H1N1 + seasonal H1N1, and novel H1N1 + seasonal H3N2. No combination of all three, and no combination of just two seasonal strains. The experiment would have been incomplete, without all five of the above I believe.

  8. #8 melbren
    September 2, 2009

    I feel so badly for these poor researchers who had to study pandemic flu transmission on a few measly ferrets–didn’t they get the email? The rest of the world gets to use tens of thousands of American college students as their test subjects!

    Of course, human college students are larger than ferrets, and require larger cages. (We call them “dormitories.”) But, unlike ferrets, we can leave the doors to their cages open and they won’t run away-no matter how sick they get! (Thanks to sex drive, keg parties, teenage angst, lack of transportation, and the fact of their just-recent escape from 17-20 previous years of captivity.)

    And–just in case one of the infected little buggers does try to escape-we have grad student research assistants (we call them “student health clinic staff” and “school administrators”) at the laboratories (or, “clinics”) to kick them back into play by telling them they are only sick if they have detectable fevers. No fever = No problem!

    And–for those pesky “symptoms” that so-often accompany not being sick, they are given free Advil, Tylenol, Sudafed, Benadryl, Delsym, Mucinex, and Luden’s throat lonzenges for pressing a lever at the “student pharmacy.”

    (And free face masks for those who plan to do any kissing!)

    And–just to ensure transmission to the not-quite-yet infected population of test subjects– the “student health clinic staff” advise the a-febrile, virus shedding, infected students to reassure their roommates, classmates, friends, janitorial staff, professors, and concerned parents that they are NOT contagious because they haven’t had a fever in 24 hours, and because they have been coughing/sneezing/sleeping/breathing/talking into their elbows. Like Elmo.

    And–best of all–and also unlike ferrets, the parents of these mammals actually pay for them to be part of the experiment!

    Silly ferret-researchers! Americans want statistically significant data–and we’ll spare no treasure to get it! (And we want it by mid-October; after then, we’ll be too busy packing up and shipping vaccine to the southern hemisphere.)

  9. #9 revere
    September 2, 2009

    Dylan: Oh, shit. You are quite right. There are 8 possible combinations (2^3) of which they used 5. That was a slip on my part. Thanks for pointing it out. Will correct.

  10. #10 h1n1_watcher
    September 2, 2009

    Interesting study. However, sometimes it is also instructive to “just look out of the window” as the Reveres nicely put it in a former blog article.
    Look at the onging real-life experiment “flu season 2009 in the southern hemisphere” involving not a handful of ferrets but millions of humans.
    Each of the main points of the ferret study we already know from the real-life experiment:

    1. Swine H1N1 rapidly crowds out both seasonal H1N1 and H3N2 as was uniformly observed everywhere from Argentina to New Zealand.

    2. Swine flu is more pathogenic than seasonal flu:
    While death caused by direct viral pneumonia is very rare with seasonal influenza for the young & middle aged (estimated at one in a million(!) cases in that age group) the picture is very different with swine H1N1 that already proved capable of killing hundreds of young & healthy people by severe viral pneumonia alone.

    See [1] for a scientific study showing that with respect to the capability of causing direct viral pneumonia the swine H1N1 is about 100 times more lethal than seasonal flu.

    This is also consistent with all recent animal studies that uniformly show that swine H1N1 infection always involves the lungs whereas seasonal H1N1 almost always restricts itself to the nose and throat.

    I am no “swine flu alarmist” and fortunately “100 times more lethal than 1 in a million” is still a low CFR (1:10.000) but nevertheless it shows that the common perception of “swine H1N1 = just another strain like seasonal flu” is not justified. This new virus is indeed “a different animal” and should be taken seriously.


  11. #11 revere
    September 2, 2009

    h1_n1: I don’t think it is quite so easy. For one thing, there are many more people in the younger age group infected so there will naturally be more severe illness in this age group which is noticed more easily. More importantly, however, is not whether the other viruses are being crowded out (if it turns out they are, which is still unclear) but how. If it is at the level of infection, that’s one thing but if it is at the level of transmission it is another. In this case it seems to be at the level of transmission. The role of residual immunity in all this is also important for vaccine strategy. So while I am in favor of looking out the window for some things, I still think that looking at the barometer gives additional and often important information.

  12. #12 Ralph
    September 2, 2009

    This notion that one type of flu “crowds out” others has been intruiging me for some time – thank you Revere for chosing to cover it (was thinking of asking!). I think there are still more interesting questions than answers though.
    I’m a physicist, so rather out of my field here, but the maths feels wrong to me for just coinfection to result in crowd out. Clearly seasonal flu is finely balanced with respect to survival – its well honed transmissability countered by an only slowly replenishing group of infectables – but surely not quite so finely balanced that the number of host similtaneously infected with both novel and seasonal flu would completely crowd it out. We are talking multiplying two very small percentages (other things being equal you would need the square root of the US population – still a reasonable number – to have seasonal, and the same number to have novel at a given moment, just to produce *one* person who at that moment is similtaneously infected with both – some crowd!).
    THere are lots of possible answers to this:
    1. Crowd out isn’t really happening in most countries (yet – it may as numbers get higher)
    2. Massively more people have either or both types of flu than anyone thinks (presumably they have no significant symptoms)
    3. There is some cross immunity (you do not need to be co-infected – if you have had swine flu you are less likely to catch seasonal)

    But the one that really intruiges me (even if it is probably the least likely without the evidence) is

    4. There is no cross immunity to infection, but some resource or other in the host used to infect the next victim is used up by the first infection. This would amplify any crowd out effect as the incumbant flu would have to do without all of the potential hosts cumulatively infected with swine flu, rather than the snapshot

    With sufficient statistical data it should be possible to determine if the crowd out effect relates to current or cumulative infections (this would not distinguish between 3 and 4 of course).

    As for a mechanism permit me a flight of fancy – what if certain types of cells in the throat or lungs are predominantly responsible for shedding and these are completely used up then taken out (some graphics show cells bursting open to release virus – not sure this is really accurate) by a flu infection – and (presuming these are non essential to survival, in the short term at least) the body takes its time replacing them (or like my brain cells they never come back!). This could mean that after one flu infection you are no less likely to *catch* a different one, but much less likely to pass it on.

    What do you think?

  13. #13 revere
    September 2, 2009

    Ralph: I don’t know the answer. One could think of crowd out as a simple bifurcation at some reproductive parameter (e.g., if the number of new infections is greater or less than one per host). It was my view that co-infection was the bottle neck, i.e., that co-infection was not independent, and that may be true, but in the ferret study there was no difficulty infecting the animals with two viruses at once. The bottleneck came at transmission: only one transmitted. If the transmissibility of swine is greater (how much greater it would have to be I’m not sure; it would depend on the dynamics, which are almos certainly non-linear) then one could easily see a displacement. I am more puzzled by the fact that we don’t see it with seasonal H1 and H3.

  14. #14 silqworm
    September 2, 2009

    Why is everyone here completely divorced from reality? The death rate from Baxter H1N1 (aka swine flu) is much lower from older strains, and you can see at the CDC website that it has displace the other strains, so there is no need for the seasonal vaccine, and by the time the Baxter H1N1 vaccine gets here, the population will have already gone through the epidemic, which was obvious before they spent tens of billions of dollars on this scam. Maybe a lot of you have been getting your seasonal vaccine and are now so mercury poisoned, you are incapable of thinking sanely.

  15. #15 revere
    September 2, 2009

    silqworm: I advise Reynolds Wrap. I’m sure you will look quite nice with it on.

  16. #16 bc
    September 2, 2009

    A couple other interesting articles in the PLOS Currents: Influenza series:

    First estimation of direct H1N1pdm virulence
    By Antoine FLAHAULT
    which suggests risk of ARDS from pandemic H1N1 is about 100x greater than from seasonal flu


    Swine origin influenza A (H1N1) virus and ICU capacity in the US
    By Marya D. Zilberberg, Christian Sandrock and Andrew Shorr

  17. #17 anon
    September 2, 2009

    Ralph, no.
    revere,not simultaneous co-infection but wait some
    days, weeks. There should be a period (~1week – 10 weeks ?)
    where people are protected against other flu-infections.
    Why else would waves decline sharply after the peak ?

  18. #18 h1n1_watcher
    September 3, 2009


    Once both viruses begin to reproduce in the body cells, some (slight) residual immunity wrt. to the seasonal strain gives a slight edge to the pandemic strain.

    Think of it as a slightly reduced “inter-cell reproductive rate” ‘R_s’ of the seasonal strain versus ‘R_p’ for the pandmic strain.

    For a very simplisitic exponential growth model, lets assume R_s=3 and R_p=4.

    After the first 5 virus generations, the pandmic strain will already have outnumbered the seasonal one by a factor of 4 (4^5/3^5) and after 10 generations there will be about 20 times more pandmic than seasonal virus in the body. If transmission to another host occurs at that time, then it is already 20 times more likely that the transmitted virus is of the pandemic type.

  19. #19 h1n1_watcher
    September 3, 2009


    I’m sorry: after having read your post again, I see that my answer missed your main point.
    So you are right, there seems to be no straight forward explanation why there is not more “independent” co-circulation of the different viruses among different hosts as long as the total attack rate of both strains remains too low to make co-infection sufficiently likely to help the fitter stain “play out” its reproductive advantage within the same host.

  20. #20 HSR0601
    September 3, 2009

    Genes included in the new swine flu have been circulating undetected in pigs for at least a decade, according to researchers who have sequenced the genomes of more than 50 samples of the virus.

    The findings suggest that in the future, pig populations will need to be monitored more closely for emerging influenza viruses, reported a team led by Rebecca Garten of the federal Centers for Disease Control and Prevention in a report released by the journal Science.

    Additionally, a simple action like brushing teeth following each and every meal could make a big difference in our immune system,I believe.

    Thank You !

  21. #21 Ralph
    September 4, 2009

    Thanks h1n1_watcher – you understood (in your second post) what I was trying (in my roundabout way) to say. I dare say in a year or two’s time we will have a much beter idea (provided we are careful to collect the data!)

  22. #22 Phillip Huggan
    September 5, 2009

    I didn’t think nothing of this research at first, but I just had what could be good idea, or nothing, or a bad idea:

    This Swine Flu overpowers seasonal flu. I wonder if it would overpower more deadly strains. That is, should this flu be kept in reserve in case some 5% deadly strain, or 75% deadly strain, or 99% deadly strain appears one decade. If the armageddon strain appears, this flu strain could be held in reserve in all cities and released in an attempt to neutral the deadly strain. Of course maybe the deadlier strain doesn’t behave like Jan-Mar 2009 seasonal strain and maybe this makes things work. But if so, the goal here would be to get people as sick as possible as quick as possible with Swine Flu. I don’t know if there would be enough time to test this on the fly…

  23. #23 Phillip Huggan
    September 5, 2009

    …but if people are immune to the 2009 strain, maybe this specific tactic doesn’t work for some/most people. I’m not sure if fighting fire with cooler fire is wise or effective here…it would involve trying to make people sick on purpose.

  24. #24 Phillip Huggan
    September 6, 2009

    Obvious typo is first post. I guess this is a suggestion for a really immediate and crappy vaccine.

  25. #25 Phillip Huggan
    September 6, 2009

    Re-rethinking this our knowledge of immunology might be too weak for this to ever be useful. But maybe not. If you have ten different candidates and there is a very deadly flu outbreak you can impose draconian social distancing and immediately do human testing, immediately unleash a milder pandemic.

  26. #26 anon
    September 7, 2009

    create and spread milder flu-viruses to crowd out the bad ones

    ..or to prevent new ones from taking over.

    E.g. create H3N2s with truncated PB1, let then reassort
    with other strains’ PB1s , test,evolve in labs,animals.
    They should be less virulent.
    H3N2 is apparantly worse than H1N1.

    Colds are less virulent, yet more prevalent in humans.
    FluB is specialized on humans and is also milder.

    Mild viruses seem to have some evolutionary advantage.
    But flu-A comes from birds, when evolving in humans it doesn’t give the
    evolutionary experience back to birds so it can be used for further evolution.

    Viruses coming from birds were often virulent at the beginning but
    became less severe over the years.

    Identify the features which make the viruses less virulent and
    better spreading and speed up the process.

    Use these viruses to crowd out new possible dangerous pandemic
    viruses (H5N1?)

    is it ethical ? You were creating viruses which infect humans,
    cause pain and death – but for the benifit of preventing/reducing
    other viruses to do even more damage !

    I haven’t seen papers on this, is it being considered ?

    I remember, with H5N1 we’d had the idea of a milder
    artificial “counter-pandemic” with a milder virus like H5N2
    or the same H5N1 with reduced cleavage site.

    We would support its spread, getting infected deliberately
    and deliberately trying to infect others…

    assume some company comes up with a life
    reassorted virus,
    evolutionary fit in humans, well spreading among humans,
    mild as a cold but with
    current seasonal HA and inducing immunity as a real
    infection with seasonal flu –
    cold they sell it and make $$$ ?

    Maybe some are already working on this…

    benefits over vaccine are:
    1.) much cheaper
    2.) spreads automatically, no vaccination required
    3.) induces better immunity, from real infection – not just vaccine
    T-cell+B-cell + innate(short term)

    we usually have a H1N1-season or a H3N2-season but rarely both
    so, if one strain is around, there usually is herd-immunity against
    the other flu-A

    it _is_ influenza. Just mild influenza.
    Mild seems to be superior, flu gets milder over the decades, flu-B had more
    time to evolve in humans and is milder than flu-A.
    Colds are even more successful respiratory viruses
    and are milder.
    Flu in ducks is generally mild.
    Virulence seems to an error due to evolution in the wrong host
    then species jump and not enough time to adapt.

  27. #27 Phillip Huggan
    September 7, 2009

    Pshhh. I thought I might have an original idea here.

    Here what you do if possible and doesn’t impair fitness or whatever: you tack on a trait that makes the “Skynet” virus mucus-membrane symptomatic before sheds en masse. Probably not original either….damn health/defense budgets…

  28. #28 Phillip Huggan
    September 7, 2009

    ….or a fever or anything self-reportable.