Natural history of influenza infection in human volunteers

If you were infected with seasonal influenza virus, how long before you start having symptoms (the incubation period)? How about how long before you start shedding virus so that you can infect others (the latent period)? When are you most infectious? Expert opinion. based on experience and the scientific literature, says that the incubation period is about 2 days but you might start shedding virus after only a day, i.e., before you get any symptoms. You are most infectious on day 2 of your illness. But will everyone who is infected really "come down with" the flu? We know that a good proportion of people are actually asymptomatic with their flu infections. What proportion? Maybe as much as half. But the basis for these expert opinions is often vague. How about if you could actually experiment on human volunteers, especially those with low levels of flu antibodies? Numerous studies of this type have been done and now a French team has surveyed the English language literature that appeared between 1965 and 2005 and given us a nice summary. One of the surprises is that there are no surprises. There is also some new information.

A broad literature search by the French researchers turned up 71 usable studies in which people were challenged with wild type influenza virus (various subtypes) and some outcome measured. Bird flu (subytype H5N1) has never been given to volunteers. Current case fatality rates (over 60%) make this unethical. But all other subtypes that infect humans on a seasonal basis have been studied in experimental settings with volunteers. The 71 papers described only 56 different studies, comprising 79 subgroups and 1280 subjects in all. Because of the nature of volunteer studies almost all participants were young adults, between the ages of 18 and 40 to 50 years old (a single study had subjects up to the age of 65). The virus was given either as nose drops (most studies), throat spray (3 studies) or aerosol (1 study). The doses ranged over three orders of magnitude, so some examination of dose was possible. Most subjects had low antibody titers against the flu hemagglutinin protein. Where subgroups were reported to have had some immunity, the data were not used in summarizing the viral shedding results. The subjects were almost always kept isolated for a week and studied during that interval for symptoms, signs and viral shedding. Follow-ups ranged from a short 3 ays to 14 days.

The results of this systematic summary of the literature on infected volunteers confirmed current expert opinion. Most volunteers got infected (almost 90%), at least as revealed by a rise in antibody. Viral shedding was also very high: over 90% with A/H1N1 and AH3N2. If these volunteers were exposed they got infected and they shed virus. But not all of them got sick. Only two of three infected got sick and this didn't differ by subtype. Getting infected and shedding virus does not mean you are clinically ill, something we've known for a long time (but often forget). In fact the rate of asymptomatic infection is quite high, over 30%. On the other hand, if you were asymptomatic you shed much less virus, in quantity, than those who were sick.

If you do get sick, what is the most common symptom? Again, no surprises. Upper respiratory symptoms like a stuffy and runny nose, sore throat, sneezing, hoarseness, ear pressure/ear ache were at the top of the list (60%). Only about 20% had lower respiratory symptoms like cough, chest discomfort or trouble breathing. Subtype didn't seem to matter. Fever was more common than lower respiratory illness, with a third reporting a body temperature above 37.8 degrees. C. (100 degrees F.). A curious and unexplained finding was that the proportion with fever was higher in the studies that used lower doses.

As for duration of illness, this peaked on days 2 or 3 and were mainly back to baseline by day 8. Average length of illness was 4- 5 days. Fever and muscle and body aches subsided before the respiratory symptoms. These are the acute symptoms, however. There is no report on the oft reported enervation after flu infection.

What did these studies who about viral shedding? On average, detectable virus shedding appeared about a day after inoculation (83% of subjects), the rest (only one in six) on days two or rarely three. That means that after someone sneezes in your face during flu season you can do the same to others a day, probably at most two days, later and get the same result. There is a sharp increase in shed virus the first day following inoculation, reaching a peak on day two. You keep shedding virus for about 5 days. Most people were done throwing off virus after a week, but occasional people kept shedding for a few days more. These studies indicate a mean generation time (infection to infection) of only 2.5 days, shorter than usually estimated.

This is a valuable review of an important literature. There still remain some questions as to how generalizable these volunteer studies are, however. First, there were no studies of two vulnerable subpopulations, children and the elderly, for obvious ethical reasons. Second, inoculation was mainly by intranasal drops. Does this make a difference, say in the proportion of lower respiratory symptoms? We don't know. Were the viral strains used in these studies more or less (or differently) virulent than the circulating viruses? We don't know, although other information suggests they weren't very different.

How do the authors sum this all up?

Optimistically, vial shedding, the surrogate marker for infectiousness, was of moderate duration, and its dynamics largely overlapped those of systemic symptoms, thus (in theory) permitting efficient isolation of infectious individuals. Pessimistically, viral shedding peaked rapidly, infections were rarely "typical," and symptoms or signs widely used for influenza case definitions (e.g., fever or cough) would be unreliable for identifying infectious individuals. (Carrat F. et al. Am J Epidemiology 176:775-785, 2008).

This is a very interesting paper on one of the major uncertainties in the influenza world: what is the "natural history" of the disease.

More like this

A reader (hat tip River) sent me a link to a New York Times piece quoting a physician who recently saw swine flu cases in Mexico City. He called attention to what seemed like an anomalous clinical presentation of many cases. Besides a higher proportion of gastrointestinal symptoms (nausea, vomiting…
How fast flu spreads is related to how many susceptible people an infectious person can infect (a measure called R0) and also something called the serial interval. The serial interval is the average length of time between the start of one infection and the start of the infection of that case's…
In 1988 a 32 year old woman, 36 weeks pregnant, checked into a community hospital in Wisconsin. She'd had flu-like symptoms with a moderately high (spiking to 102 degrees F.) fever for the previous week. Three days before admission she started a cough that brought up sputum and a day before started…
Influenza/A viruses naturally infect aquatic wildfowl like ducks and there are a lot of influenza/A subtypes besides the H5N1 that has been in the news. Some people are heavily exposed to wild ducks, namely serious duck hunters and game handlers. Why don't they get infected with some of the other…

A curious and unexplained finding was that the proportion with fever was higher in the studies that used lower doses.

Influenza is well known to have mechanisms that suppress interferon production (avian flu apparently has further mechanisms that confer resistance to interferon as well). Perhaps at the high doses this anti-interferon mechanism was more effective and prevented cytokine release, which in turn dampened the fever.

Thanks revere, the prominence of 'atypical' signs and symptoms was a nice nugget they pulled out of all the data.

I like to be optimistic but I think the pessimistic side wins out here... The atypical symptoms I think would preclude any reasonable attempts at isolation of infectious individuals. The lack of fever, cough and muscle aches in most cases make the average infectious case look just like the common cold.... Although H5N1 wasn't tested we'd probably have to assume that it would behave similarly if it became easily transmissible...

""How do the authors sum this all up?

--Optimistically--,
vial shedding, the surrogate marker for infectiousness, was of moderate duration, and its dynamics largely overlapped those of systemic symptoms, thus (in theory) permitting efficient isolation of infectious individuals.

--Pessimistically--,
viral shedding peaked rapidly, infections were rarely "typical," and symptoms or signs widely used for influenza case definitions (e.g., fever or cough) would be unreliable for identifying infectious individuals. (Carrat F. et al. Am J Epidemiology 176:775-785, 2008). ""

By medmatters (not verified) on 01 Apr 2008 #permalink

There is an unusual variant seen with H5N1--the lack of mild cases of avian flu. This prompted a look at antibody levels in asymptomatic people who had been exposed to overtly ill people. Turns out that there really haven't been any mild cases--only severe cases, or actual non-infection. Revere, please comment.

" Most volunteers got infected (almost 90%)"

compare that with the 0% (zero) of two studies
done in 1918.

Roseman,M.J.,Keegan,W.J.,Goldberger,J,and Lake,G.C.:
Experiments upon volunteers to determine the cause and mode
of spread of influenza ,
Boston,November and December,1918,Hyg.Lab.Bull.123:5(1921)

McCoy,G.W.,and Richey,De.W.:
Experiments upon volunteers to determine the cause and mode
of spread of influenza ,
San Francisco,November and December,1918,Hyg.Lab.Bull.123:42(1921)

Annie: Unfortunately this isn't a variant. No seroprevalence studies have shown a significant degree of asymptomatic infection, which means that as far as we know, the current Case Fatality Ratio is not a gross overestimate. However it wouldn't take many asymptomatic cases to knock that down, so this still remains an open question. If you search the blog for "seroprevalence" you should pick up many of our previous comments on this.

anon: Not sure what the point is. We know the infection rate was not 0% for 1918 flu. In 1918 they didn't have the means to determine if someone was infected or not (they didn't know it was a virus). Clinical illness is not the same as infection.

assume even 50% asymptomatic cases,
0 out of more than 100 becoming sick should be very unlikely.
And they tried very hard to infect them, several methods.
Today they manage to infect 90%.
Explain.
If we don't know how to infect people by natural h2h,
how can we know how to avoid infection ?

anon: Haven't seen the numbers and the papers but assume that by this time virtually everyone would have antibodies. They had no way to know if they were infected, either, as I explained. They didn't know it was a virus. The 90% number is based on detecting the virus or antibody, neither of which could be done in the 1918 period.