Effect Measure

Swine flu: hospital surveillance data

Influenza surveillance in the US has at least five component parts (depending on how you count it is as many as seven). We discussed the virologic surveillance system in another post. CDC has two surveillance sub-systems that look at hospitalized cases with laboratory confirmed influenza, the New Vaccine Surveillance Network (NVSN) and the Emerging Infections Program (EIP). The NVSN is confined to cases in children less than five years old, while the EIP covers all ages. Let’s take a look at the most recent EIP data.

EIP doesn’t cover the entire country. Instead it collects data from 60 counties and 12 metropolitan areas in 10 states (San Francisco CA, Denver CO, New Haven CT, Atlanta GA, Baltimore MD, Minneapolis/St. Paul MN, Albuquerque NM, Las Cruces, NM, Albany NY, Rochester NY, Portland OR, and Nashville TN). It is an active surveillance system that examines routine hospital laboratory and admissions data for cases that have a record of a positive test for influenza (viral culture, direct/indirect fluorescent antibody assay (DFA/IFA), reverse transcription-polymerase chain reaction (RT-PCR), or a commercial rapid antigen test). The counties and metropolitan areas are chosen to allow an estimate of hospitalization rates for the populations of the hospitals’ catchment areas. Using these data, EIP estimates influenza hospitalization rates every two weeks of the flu season (weeks 40 of the previous calendar year to 20 of the new calendar year).

Here are the most recent data through week week ending May 16 (week 19), graphed separately for different age categories:

i-c56da539f9088172d33e4c4b4e94afd1-EIP19.jpg

Source: CDC Fluview

What’s striking is that this year’s flu season looks pretty much like the mild seasons of 2005-2006 and 2007-2008 and quite unlike last year’s moderately severe season — except for the 5 – 17 year group, where it begins to diverge around week 9 (week ending March 9).

We wonder if this was the first sign of this pandemic virus, whose age distribution is markedly left shifted.

Comments

  1. #1 Grahame Grieve
    May 28, 2009

    Is there any data about asymptomatic carriers?

  2. #2 The Doctor
    May 28, 2009

    Thanks for this thread. Yes, the increase in the cases reported by the EIP for those aged 5-17 for this past season is unusual. The first 12 cases of A(novel H1N1) were seen in week 16 and none before even though the case rate for the 5-17 age group was quite high by the time so the two findings may not be related.

    Another interesting finding from this report is that in weeks 17-19 when new cases of A(novel H1N1) were rising so rapid 61% of all the positive isolates tested then were due to seasonal A(H1) or A(H3). In week 19 the number had fallen to a little less than 20%.

    This statement is derived from the data used to produce the WHO/NREVSS Collaborating Lab table in the week 19 CDC report referenced above [A(H1) + A(H3)] / [A(novel H1) + A(could not be typed)] for the weeks mentioned.

    With that much co-circulation of the various strains going on, the opportunities for genetic exchange were certainly present. As far as I have heard though, there has been no signs of reassortment or recombination. Is this your understanding as well?

    Grattan Woodson, MD

  3. #3 revere
    May 28, 2009

    Gratt: Yes, as far as I know there is no evidence of reassortment. Perhaps the first signal, if it occurs, will be oseltamivir resistance, as H1 seasonal has H274Y on NA while novel H1 doesn’t. But as usual with flu, there is no way to predict with any certainty. We can set out some of the possibilities (but probably not all of them) but right now don’t have a good way to know which ones are the most likely, much less how likely.

    Grahame: No data I know of, but I would be very surprised if there weren’t a fairly sizable pool of asymptomatic or inapparent infections with this virus (unlike what we see with H5N1 so far).

  4. #4 sandy
    May 28, 2009

    i think you have to look at all the history.
    the H1N1 1977 virus also showed age related infections, and it went on to worldwide spread. but it did not show death rates elevated to pandemic levels.

  5. #5 Stoker
    May 28, 2009

    Revere: you say, “I would be very surprised if there weren’t a fairly sizable pool of asymptomatic or inapparent infections with this virus (unlike what we see with H5N1 so far)”, however ProMed has a report of inapparent H5N1 infections (22 May 09): “A survey of more than 600 people in Cambodian villages where 2 children died from the virus shows 7 more were apparently infected, but without having known about it.
    Regards, Stoker

  6. #6 anon
    May 28, 2009

    and there is no way to estimate with any certainety the amount of uncertainety
    in influenza which however is ever increasing the more we blog about it…

    When you have no way to know which one is more likely, then they all have the
    same (un)likelyhood for you, or not ?

    Reassortment is more likely with other Mexflu strains than with seasonal H1N1 and then
    it’s 50:50 whether Mexflu will pick N1 from seasonal H1N1 or not.
    Is it really certain that there are opportunities ?
    No one can predict with any gut-feeling
    whether what’s certain today will
    still be certain tomorrow

  7. #7 revere
    May 28, 2009

    Stoker: Yes, I know this paper. It is precisely what I am talking about. This confirms the idea that there are hardly any inapparent infections that we know of. It is very rare, unlike seasonal flu.

  8. #8 Stoker
    May 28, 2009

    Doesn’t the Cambodian data suggest that there are at least double the number of asymptomatic infections in the community as are apparent (assuming that with a mortality rate of 50-60% there were no more than 3 symptomatic cases, including the 2 who died)? Not as many as suggested for H1N1, granted, but not “rare”.

  9. #9 revere
    May 28, 2009

    Stoker: Only if you compare to the apparent cases. If you compare to the extensive exposure, it appears that there is very little infection, period, clinical or otherwise. I was formerly of the opinion there was a lot of mild or inapparent H5N1, but all the data suggests not. Whenever we look we don’t find it. I am surprised and concerned because it means the apparent CFR may be much closer to the actual CFR than any of us want to believe.

  10. #10 melbren
    May 28, 2009

    Once again I feel compelled to apologize for asking such a basic question, but…

    On the CDC website it states that influenza can be detected if testing is done within several days of symptoms.(In fairness to CDC–this was probably posted pre-2009 swine flu outbreak–if that matters.) But does that mean that if a person had a flu virus that triggered a secondary medical issue a week or two later–such as an upper respiratory infection–would the CDC be unlikely or even unable to detect it by then? For how long is (an) influenza virus generally detectable?

    Is it possible that there a lot of influenza cases that only become severe enough for hospitalization beyond when the virus can even be detected?

    For example, the 89 year old who is still hospitalized in California–if doctors tested her today for the first time, would her swine flu virus still show up?

  11. #11 The Doctor
    May 28, 2009

    Yes the acquisition of H274Y on NA by A(novel H1N1)would certainly suggest some sort of genetic exchange. Since Tamiflu is being widely deployed it makes sense that this may well be one of the first new attributes the novel virus incorporates.

    What concerns me more though is its acquiring some of the lethal polymorphisms present in the various H5N1 strains currently circulating.

    GW

  12. #12 clioepi
    May 28, 2009

    it’s hard (but not impossible) to imagine that CDC could totally miss widespread novel A (H1N1) causing increased hospitalizations like that, and for that long.

    any thoughts on Jon Cohen’s piece (http://blogs.sciencemag.org/scienceinsider/swine-flu/) in Science Insider today, about CDC’s sentinel data and the example of emergency department surveillance looking a little different?

    looks like the 5-17 year old kids are the ones most hard hit.
    makes one wonder how under 2 year and 65 and older are being considered risk factors.

  13. #13 revere
    May 28, 2009

    melbren: It is quite possible to miss flu even though secondary complications remain. There are many reasons for false negatives, including bad sampling (swabbing), no more virus is being shed, the test is not sensitive enough, etc. That’s why estimating the toll that flu takes is such a difficult task. On the other hand, a diagnosis can be made clinically with as much accuracy as the usual diagnostic (quick) test during an outbreak. We don’t depend upon tests to make the diagnosis in most cases.

    clioepi: It is not hard to imagine that this amount of flu could be missed. The under 2 and over 65 are considered risk ages for seasonal flu, but it doesn’t appear that the same pattern holds for this one. Some sources (like NYC) are no longer mentioning this but it remains in other accounts. Haven’t read Jon’s piece yet but will do so. ScienceInsider has had a number of interesting pieces on flu so I’ll take a look (been out to dinner with fellow non-flu scientists so just returning to the keyboard now (10 pm)).

  14. #14 kristin
    May 29, 2009

    Here in the UK, Health Protection Agency (HPA) guidance recommends screening only when patient presents with fever and other flu symptoms, And has had recent exposure to people from high-incident areas (namely the US or Mexico) or direct exposure to people with confirmed cases. If there are significant numbers of cases that don’t manifest with fever, then it stands to reason these are cases that will be missed altogether. Does anyone know if there are data (other than the first-hand observation of the physician visiting in Mexico) regarding how often H1N1 presents with/without fever?

  15. #15 MoM
    May 29, 2009

    Revere (et.al.)

    I have asked before (elsewhere) about the possibility of a serologic test for this version of H1N1. We desperately need denominator data, which can only be garnered by testing a piece of the general population to see if they have been infected and then asking them about their illness, or lack thereof.

    Thoughts?

  16. #16 revere
    May 29, 2009

    MoM: We can get seologic data but it doesn’t usually tell us when the person was infected. In the case of swine flu, we’ll have to wait for a bit to allow convalescent antibodies to develop before we can use this to get info on how widespread infection was (and of course it will have to be a good sampling plan). It is a non-trivial exercise. I expect that plans to do this are well underway but it will take time.

Current ye@r *