Benefits of antiviral usage for influenza treatment: The ability of antiviral drugs to reduce symptoms and shorten their duration are well documented, particularly when antivirals are given early in the course of illness [1] , [2] , [3] .There is evidence that early usage of antivirals for treatment of infected individuals can reduce transmission to others [4] , [5] , [6] ; [4] in particular applies to the current pandemic H1N1pdm strain, though the findings in that study are not statistically significant. A 2003 study suggests a 60% reduction in influenza hospitalization rates for patients who received early antiviral treatment [7] .

Data in [8] show that among hospitalized patients, severe outcomes (ICU admission and/or death) were less likely in patients who received antiviral treatment within two days of symptom onset; the same conclusion is true for lethal outcomes for hospitalized patients in http://jama.ama-assn.org/cgi/content/short/302/17/1896?home . We have attempted to extract further information on the effects of antiviral usage on severe outcomes for the current H1N1pdm strain by an ecological comparison between two neighboring Southern Hemisphere countries that have experienced a complete H1N1pdm season: Chile and Argentina.

Chile vs. Argentina: Chile had an active antiviral usage policy during the course of the epidemic, with almost 650,000 courses distributed (~ 4 courses for every 100 persons in a country of 17 million) by August 16 [9] . The current number of confirmed fatalities in Chile is 136, i.e. 0.8 deaths per 100,000 inhabitants [10] . In Argentina, treatment was reserved for hospitalized patients until after the peak of the epidemic, at which point policy changed to include treatment of pregnant women and members of high risk groups; only near the end of the epidemic was treatment of all ILI cases recommended ( [11] , p. 25). Argentina had 580 confirmed fatalities in a population of 40.4 million [12] , or 1.44 deaths per 100,000 inhabitants.

One might argue that the difference could be due to better ascertainment of fatalities in Argentina. However, overall ascertainment of cases appears to be higher in Chile than in Argentina: Chile has 12,257 confirmed cases and 136 confirmed fatalities [10] , while Argentina has 9,119 confirmed cases and 580 confirmed fatalities [12] . It is thus likely that a higher percentage of H1N1-realted deaths were ascertained in Chile than in Argentina.

A second line of ecological evidence for the benefits of early antiviral treatment appears within the Argentinean data in [11] , p. 18, in which the percentage of fatalities among the confirmed H1N1 infections in pregnant women has dropped significantly after guidelines for treatment of high risk individuals were adopted. More precisely, the data shows 174 pregnant women with confirmed H1N1 infection with symptom onset prior to the adoption of those guidelines; 31 of them (17.8%) died. Among the 69 pregnant women with confirmed H1N1 infection with symptom onset after those guidelines were adopted, there were 4 fatalities (5.8%). The difference is statistically significant (Fisher’s exact test p-value =.015; 95% CI for the odds ratio of a pregnant woman’s death following infection after the guidelines adoption vs. prior to the guidelines adoption is (0.07, 0.85)).

High vs. low risk individuals: In [13] we have argued for pre-dispensing antivirals to high risk individuals, aiming to initiate treatment early, to reduce their risk of progression to severe disease. On Sep. 8, 2009 CDC has revised its guidelines for antiviral usage (subsequently updating those on Sep. 22 and Oct. 16), further encouraging early treatment of individuals with a high risk of complications, and indicating that treatment should not wait for laboratory confirmation of influenza [14] . At the same time, those recommendations state [14] :

“Most healthy persons who develop an illness consistent with uncomplicated influenza, or persons who appear to be recovering from influenza, do not need antiviral medications for treatment or prophylaxis.”

This has, in effect, discouraged antiviral treatment for many individuals. This is a lost opportunity to reduce severe morbidity among “low-risk” individuals, who make up a sizeable percentage of the severe cases (hospitalizations and fatalities), though exact percentages vary between 15%-50% in different studies [15] , [8] , [10] , [16] , [17] , [18] . We are not aware of data assessing this lost opportunity so far, save for some anecdotal evidence [19] .

Current situation in the US and an argument for policy change: CDC data up to week 41 (ending October 17) suggests a rise in all indicators related to the epidemic’s progression in the US, including hospitalizations and confirmed fatalities [20] . Pneumonia and influenza mortality rates in the 122 city reporting system have been above the epidemic threshold for 3 consecutive weeks [20] . While not all of those deaths are necessarily due to H1N1, the percentage of influenza positive tests among the ILI patients keeps rising and stands at 37.5% during week 41 [20] . Pediatric mortality from H1N1 already exceeds annual counts from previous years.

In the current H1N1 influenza pandemic, antiviral drugs are actively prescribed to symptomatic individuals in some countries [21] , [9] . Under the circumstances of elevated mortality and hospitalization rates in the US, with a high proportion of ILI patients actually having H1N1, we think it is worthwhile to consider a revision to the current antiviral usage recommendations that will encourage physicians to prescribe antivirals at their discretion to individuals with moderate to severe symptoms who show up for treatment.

The cons of antiviral usage: Having so far discussed the pros of antiviral usage, we shall now list some of its potential cons.

Antiviral resistance: Oseltamivir resistance levels in H1N1 strains in the US are currently low, standing at 0.5% during week 41 [20] ; those levels have not increased in recent weeks. Resistance levels are even lower (0.17%) in the UK [22] , where more extensive antiviral usage takes place, with a higher proportion of the infected individuals being treated with antiviral drugs. Given the fact that sizeable antiviral resistance has not emerged so far in the UK with higher treatment levels than in the US, we think it is unlikely that if the US treatment levels increase to match the UK ones, significant levels of resistance will emerge in the upcoming weeks before large quantities of vaccine become available. Moreover, prior mathematical modeling studies [23] , [24] have shown that if the epidemic is already large (as it is now) at the time that large-scale antiviral treatment begins, then resistant strains will be unlikely to ascend to high frequency in the population before the epidemic is over, even in the absence of vaccination.

Stockpile depletion: It seems likely that a small percentage of infected persons have sufficiently serious symptoms to seek care. An estimate from New Zealand [25] is that only one out of 18 did, although survey data in the same population suggested that the proportion seeking care was closer to 25% (MG Baker, personal communication). US data suggest a similar range of uncertainty, with self-reports of 42-58% of persons with ILI seeking care [26] , but comparisons of ILI to physician visits suggest much lower numbers [27] , [28] . In Chile, 650,000 antiviral courses were dispensed in a population of 17 million, representing coverage of 3.8% of the population. Such demand can be readily met with the US stockpile, unless demand levels increase sharply.

Excessive demands on health care provision: This phenomenon may be exacerbated by availability of more antivirals and cannot be fully predicted. We note, however, that more pro-active antiviral distribution policies did not overwhelm healthcare systems neither in Chile nor in the UK.

Adverse effects of antiviral treatment: Some adverse effects to antiviral treatment exist [29] and should be discussed between doctors and patients. Those effects should be weighted against alleviation of symptoms and decreasing the risk of deterioration into a severe condition.

Cost: The cost of more extensive use of antivirals would fall to a number of parties, including federal and state governments (who would incur costs in replenishing stockpiles drawn down for treatment), and individuals and insurance carriers, who would pay for privately obtained courses. These costs should be weighed against the potential health benefits, described above, and economic benefits of reduced symptom duration, obtained from more aggressive treatment.

Funding Information: Research is supported by the Center of Excellence award from MIDAS.

Competing interests: ML discloses consulting fees from the Avian/Pandemic Flu Registry (Outcome Sciences), funded in part by Roche. EG declares no competing interests.

Acknowledgement : We thank Leon Danon and Martin Lajous for help in preparation of this manuscript.