Last flu update Again this week the flu level has decreased.  The relevant numbers are provided in the table at the CDC flu site (temporary link and permanent link *). Of the 6,224 samples tested, only 956 (15.4%) were positive for influenza A. Both the total and the percentage are down from 1,880 (20.5%) last week. Of these 708 (74.5%) were H1N1 which is slightly lower than the same percentage last week 1,478 (78.9%). So the flu epidemic in the U.S. seems to be subsiding which is also in agreement with google flutrends data. As mentioned previously, I suspect we will see a second peak of non-H1N1 flu in January-February. One comment on the CDC site that I had not seen before was
Seventeen influenza-associated pediatric deaths were reported. Twelve of these deaths were associated with 2009 influenza A (H1N1) virus infection and five were associated with an influenza A virus for which the subtype was undetermined.
The question is: what is the probability that the undetermined subtypes were in fact H1N1? Let's assume the absence of subtype information is not a result of the subtype, e.g. randomly those 5 were chosen to not be subtyped. Then an estimate of this probability is 74.5% (the empirical percentage of H1N1 of all influenza A viruses). Should we also use the information that the 12 that were subtyped were ALL H1N1? Yes, we should. The reason is that this provides information about the virulency of H1N1 vs non-H1N1, e.g. how deadly the different viruses are. To then calculate the above probability, we need to use Bayes Rule and estimates of 1) the probability of a death if an infant contracts H1N1 and 2) the overall death rate for infants contracting any influenza A strain. At this point, I don't have this information. * The temporary link will work until CDC updates the page with the new week information. At this point, the temporary link will point to week 48 (and a week later to week 49 and so on). The permanent link currently does not exist, but as soon as the CDC updates the weekly information, it should come alive. How does the CDC not create a permanent page immediately?