by revere, cross-posted from Effect Measure

Data from the Emerging Infections Program (EIP), one of the component parts of the CDC national influenza surveillance system, is showing that for some segments of its population the US did indeed experience a second flu season. The segment of particular concern are children between the ages of 5 to 17 years old, and to some extent adults between the ages of 18 and 49. The EIP counts only laboratory-confirmed hospitalizations in 60 counties located in 12 metropolitan areas in 10 states (if you are wondering, the 12 metro areas are 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). Here’s the method:

Cases are identified by reviewing hospital laboratory and admission databases and infection control logs for children and adults with a documented positive influenza test (viral culture, direct/indirect fluorescent antibody assay (DFA/IFA), reverse transcription-polymerase chain reaction (RT-PCR), or a commercial rapid antigen test) conducted as a part of routine patient care. (CDC, Hospital Surveillance)

CDC posts the EIP data on its surveillance web page and the latest version shows different rates for different age groups. The rates are the number of lab confirmed hospitalizations for influenza per 10,000 population and the dotted lines in each of the age strata represents the average rate for flu seasons (October 1 to April 30) in 2005-06, 2006-07 and 2007-08. Normally the rate during the late spring and summer months, shown in the chart below, are close to zero and are nowhere near the flu season rates (the dotted line). But in every age group we see a rise in flu hospitalizations and in the 5 -17 year old group it has been at flu season rates and getting close to that in the 18 – 49 year old group:

EIP26.jpg

Source and larger version here.

That’s one view, anyway. Here’s a different one. There’s another part of the surveillance system called the New Vaccine Surveillance Network (NVSN), also lab confirmed influenza hospitalization rates, but for children less than 5 tears old that live in three specific counties: Hamilton County OH, Davidson County TN, and Monroe County NY. If a child is admitted to one of the NSVN hospitals with fever or respiratory symptoms they are tested via PCR and viral culture and rates estimated based on the catchment areas of the hospitals. This is a different population that the top two panels in the chart above:

NVSN26.jpg

Comparing this year to the last three flu seasons suggests it is like the others except for last year, which was exceptionally bad. But note also that the rates estimated in the NVSN are 3 to 6 times higher than the comparable age groups in the EIP. Why the difference and which to believe? Other than the fact that there are different geographic populations involved, the EIP is a passive system that depends on after-the-fact record review of cases where there was enough suspicion of flu to get lab studies done as part of otherwise routine care. The NSVN is an active prospective system whereby any child with fever or respiratory symptoms is worked up for a diagnosis of flu. So there are major differences in these two systems which makes them difficult to compare with each other quantitatively. I’m not sure if the EIP ran through the spring and summer months of previous years, so the internal comparison there is with rates from the same system during the flu season. Given the heightened concern about swine flu there may be more diagnostic work-ups for flu this summer, too, so a comparison even if possible might not be that informative. The NVSN is a much more restricted system as far as geographic representation is concerned (3 counties). Given the very patchy nature of flu this is also a problem, but at least an internal comparison is possible and doesn’t suggest anything out of the ordinary — in these 3 counties. On the other hand, the EIP and general impressions are that the under 5 year old group is not the one hit hardest but the next age stratum, 5 – 17, which is not included in the NVSN. [NB: Marc points out in the comments on Effect Measure (#10) that while both systems are cumulative rates they cumulate from different time points, which is yet another reason — and arguably the most important reason — why they differ. So while all I said remains true, I think I missed the most critical difference. As politicians sometime say, “Mistakes were made.” And this one was made by me. Thanks to Marc and apologies to readers.]

The influenza surveillance system is a patchwork of different parts and wasn’t designed to follow a pandemic. Indeed this is a circumstance which is historically unique in medical history, watching a pandemic unfold in real time. Surveillance, alas, is the poor step-child of public health and has never gotten its due share of resources or the intellectual respect it deserves. So meanwhile we’ll have to make the best of it and try to read the entrails.

And what I see is a second flu season for the 5 to 17 year olds. But I could be wrong.

    Current ye@r *