At the beginning of September CDC initiated a new system for monitoring influenza activity. We reported last week that the old system ended on August 30 and that we were now into the new flu season. We even titled the post, “End of Flu Season.” To paraphrase Mark Twain, the reports of the death of the 2008 – 2009 season may have been exaggerated. Well, not really, but the advent of the new system caused some confusion, at least for us, and we think we should clear it up.
There are several interconnected issues here. One is that the way trends in influenza will be monitored has been changed and the changes reflected in the FluView report issue Friday (September 11). The change affects an important part of a multicomponent surveillance system, but it is not the only part, the one that counts influenza and pneumonia deaths and now hospitalizations. The previous counts will be reset to zero and new counting began this week. But there are another 8 or 9 components to the existing influenza surveillance system that has been in place for decades and administratively extended starting from week 40 (October) of the fall in the northern hemisphere to week 20 (April) of the following spring. This is only part of a year but was considered “flu season.” This year the normal surveillance continued throughout the spring and summer because of the advent of the 2009 swine flu pandemic in April. I hope that we will be seeing year round surveillance from now on. Still, because of decades of data, CDC is retaining the October start date for the 2009 – 2010 flu season, even though they acknowledge it looks like the actual flu season may already have started. The reason for this is to allow consistent comparisons.
So a new flu year has indeed started “on the ground” and state reported flu hospitalizations and deaths have been reset to zero. So the idea that the new flu season has begun is correct. But in an administrative sense CDC is retaining week 40 as the start of the 2009 – 2010 flu season (and Sept. 30 as the end of the 2008 – 2009 season), so in that sense we were premature in reporting “the end of flu season.”
These may seem like arcane points of administratrivia, but they are very important to those trying to figure out what is going on. The new state reporting system for influenza and pneumonia hospitalizations and deaths has two sources: laboratory confirmed hospitalizations and deaths from influenza reported to state epidemiologists by hospitals; and hospitalizations and deaths from pneumonia and influenza reported to state epidemiologists. The latter are clinically defined and include all types and subtypes of influenza (i.e., both flu A, seasonal and pandemic and flu B but also deaths that are not clinically distinguishable from an influenza pneumonia caused death even though there is no laboratory evidence to corroborate it and will have been caused by something else. Prior to the pandemic state epidemiologist did not provide weekly reports of all hospitalizations and deaths from pneumonia and influenza, but began to do so specifically for confirmed and probable cases of A/swine flu 2009 when the pandemic started in the spring. That system has now been formalized and all influenza and influenza like pneumonia and flu hospitalization deaths added to the surveillance system. Thus the new system that just started is a continuation of one that started with the pandemic but enlarged to include all pneumonia and influenza hospitalizations and deaths each week as reported by each state’s epidemiologist in charge of CDC notifications (typically two per state).
Here is CDC’s rationale for broadening the system beyond the pandemic strain, indeed beyond influenza. As an epidemiologist I see it as a compromise, but one that makes sense. It is not the only one that could have been made, but I consider it highly defensible:
CDC believes that regular seasonal influenza viruses will co-circulate with 2009 H1N1 influenza and capturing all laboratory-confirmed influenza will provide a fuller picture of the burden of all flu during the pandemic.
There are too many cases of flu to test and confirm so laboratory-confirmed data is a vast underestimate of the true number of cases and this bias would be exacerbated over the course of the pandemic as more and more people become ill.
Influenza and pneumonia syndrome is a diagnostic code used by all hospitals. Capturing this number will reflect a fuller picture of influenza and influenza-related serious illness and deaths in the United States during the pandemic. Influenza and pneumonia syndrome hospitalizations and deaths may be an overestimate of actual number of flu-related hospitalizations and deaths, but CDC believes influenza and pneumonia syndromic reports are likely to be a more sensitive measure of flu-associated hospitalizations and deaths than laboratory confirmed reports during this pandemic.
The rest of the influenza surveillance system will continue as before and will have its numbers reset October 4, not September 1. If this seems a bit confusing, it is. It will get sorted out over time, I expect, but right now it seems a bit discordant. There are good reasons for each decision, but the net result seems a bit like what the computer people call a kludge. Within the regular system there is a laboratory virologic surveillance component involving 150 laboratories (that includes all the state labs). This system will continue to keep track of what types and subtypes of influenza are being seen in clinical laboratories. This means that lumping all the pneumonia and influenza hospitalizations and deaths in the state reports won’t result in a loss of that information because it will continue to be available through the existing system. In addition, the percentage of total deaths from pneumonia and influenza deaths in the network of 122 cities will continue as before, using the same definitions, allowing a comparison to the new state reporting system with what we’ve been seeing in years past by other means. Similarly, the network of physicians who report the proportion of their visits for influenza-like illness (ILI) each week (so-called ILI-Net system) will remain. It is this system which seems to show that the new flu season has started in earnest, with the proportion of ILI visits up considerably from what we would expect at baseline. CDC usually waits for this to happen three weeks in a row before calling a flu outbreak, but that seems certain to happen within the next week or two. So we’re really there, and it’s only the first week in September. In other words, it looks like we are having a very early flu season. That could mean it will also end early or that it will smoulder along all year or that it will get really bad.
Because this is flu and you never know what it will do.
You can read the gory details about the “new” system here. Here’s the ILI-Net for the week. Disentangling the signal from the noise isn’t too hard:
Source: FluView, September 11, 2009