Our PLoS H1N1 predictions: right so far (UPDATED)

In August we published a knol in PLoS Currents: Influenza, where, based on the data to date, we modeled hospitalization, respiratory failure and death rates from H1N1. Because "preditions are hard, especially about the future", we could not be sure that we got the numbers right, especially since they were quite staggering.

Well, yesterday, the CDC and ANZIC published the US experience with the first 272 hospitalized patients and the winter flu season Australian-New Zealand ICU experience with H1N1. Here is a table that compares our predictions to what the CDC they found:

	Zilberberg knol (base case)	CDC NEJM study	ANZIC study
ICU admission		25%
ARF (relative to ICU)			65%
Acute Respiratory Failure rate [ARF] (relative to hospitalization)	12%	25% x 63% = 16%
Mortality rate (relative to ARF)	58%	Not reported	17%/65% =26%*
Mortality rate (relative to ICU)			17%
Mortality rate (relative to hospitalization)	12% x 58% = 7%	7%

*Based on assumption that all deaths occurred in ICU patients
Clearly, we got it right at least partly right -- totally consistent with the CDC numbers, and slightly higher on mortality than ANZIC. Also of interest in the CDC study is the data on antibiotics use:

Of 260 patients for whom data were available regarding antibiotic therapy, 206 (79%) received antibiotics. Of 198 patients for whom the date of initiation of antibiotics was available, such therapy was started before admission in 30 patients (15%), on admission in 117 patients (59%), within 48 hours after admission in 44 patients (22%), and more than 48 hours after admission in 7 patients (4%). Patients received a median of two antibiotics (range, one to seven); 70% of the patients received more than one antibiotic. Commonly used antibiotics included ceftriaxone (in 94 patients), azithromycin (in 84 patients), vancomycin (in 56 patients), and levofloxacin (in 47 patients). Seventy-threepercent of patients who had radiographic findings that were consistent with pneumonia were treated with antiviral drugs, and 97% were treated with antibiotics.

It is not surprising that there was such a high rate if antibiotics use, as these patients represent a group with very high severity of illness, where you throw everything at them up front and ask questions later. The rate is also similar to what was reported in the Spanish study, though those patients might have been even sicker since they were all in the ICU.

What concerns me more, however, is the 15% rate of use of outpatient antibiotics in the setting of this viral illness. I am not pointing any fingers here, as it is a decision that has to be made individually in the office by the doctor and patient. But, as I have done in the past, I do want to caution about what indiscriminate use of antibiotics could do to the incidence of C. diff and other resistant infections. We need to be very judicious in our use of these therapies as the onslaught of the H1N1 pandemic continues.