Effect Measure

CDC’s Advisory Committee on Immunization Practices (ACIP) recently rolled out their 2009 Recommendations. It’s for seasonal flu, for which a vaccine exists, not for swine flu, for which there is (as yet) no vaccine. There is a lot to say on the subject of vaccines (see what we’ve said over the years under the vaccine category), but this seems like a good time to review some basic terminology, including what is meant by vaccine efficacy or effectiveness and how it is measured or estimated. There’s a lot to say, so we’ll split this into two posts, and in this one we’ll go over some terms and concepts we’ll need later and that are also used in thinking about infectious disease. We’ve done this before, of course. But it’s like the Henny Youngman joke. Man goes to the doctor complaining of pain in his shoulder. “Did you ever have it before?” the doctor asks? “Yes,” says the patient. Doctor: “Well you have it again.” Once again, we talk about transmissibility, pathogenicity and virulence.

“Transmissibility” would seem to be pretty straightforward for contagious diseases. Contagious means one person gives it to another. It doesn’t say how a person gives it to someone else, and it turns out for flu, we aren’t quite sure what the main modes of contagion are. Is it large droplets that fall to the floor within a couple of feet of the cougher/sneezer/wheezer? Or are infectious viral particles ensconced on tiny particles of mucus or water or whatever that are so small (less, say, then 5 millionths of a meter or 5 microns in aerodynamic size) that they remain suspended in the air for hours or maybe days at a time? What about inanimate objects like door knobs or airplane armrests? Virus will remain replicable on hard, non-porous surfaces for days or weeks at a time. But of these three main possibilities, which of them really is more than a theoretical or rare means of transmission? What’s the “mix” of transmission modes? Nor is it just a physical problem. There’s some serious biology involved here, too. Suppose you get a good load of bird flu virus up your nose? Does that mean you will become infected? Apparently not. Because, so far, the influenza virus that infects birds doesn’t transmit very well to humans. It happens, and when it does it has devastating effects. But mostly tranmission doesn’t occur. We don’t know the exact reasons. Compare that to another flu virus that jumped from animals, the current swine flu. Not only did it transmit well from animals (at some point, we don’t know when or where), but once it did, it transmitted well from person to person. Here, when speaking of “transmission” I’m talking about a virus that infected the host cells of one animal or person that then moves to another animal or person and infects their host cells as well. So even when a bird flu virus makes the jump from a bird to a human, it still transmit very poorly between humans. It’s not very transmissible in people. Transmissibility is both physical movement from one host to another and it is the biological ability to infect the second host once it gets there. It’s a process that can be affected at various points along the way and then can depend on the biological environment it finds when it gets there. Maybe the person it lands on is immune? But we’re getting ahead of ourselves. We need first to discuss two more terms, pathogenicity and virulence.

Notice that in talking about transmissibility I did it in terms of whether the second host got infected or not. I didn’t say anything about whether the second host got sick or not. A virus is an obligate intracellular parasite. In less fancy terms, it only reproduces inside another living cell, a host cell, whose genetic and protein making machinery it hijacks to make copies of itself. Without the host cell it can’t do anything. It doesn’t eat, it doesn’t move on its own, it doesn’t grow or excrete or anything else. It is inert. Once it gets inside a host cell and starts making copies of itself it can have a variety of effects, some of which the host organism hardly notices at all and others of which might be so severe the host organism dies or gets sick. It’s estimated that close to half of all infections with seasonal influenza are asymptomatic, i.e., the infected hosts aren’t sick. The ability of a virus to make people sick — say to give them symptoms or signs of dysfunction or pathology — is what is called pathogenicity. There are lots of infecting viruses that are totally non-pathogenic in that they infect human host cells but never make us sick; many more that do so sometimes or a lot of the time (influenza is a good example), and some that almost always make us sick, sometimes fatally so (rabies). So that’s pathogenicity. It’s what biostatisticians call a conditional measure. Its measurement is conditional on the host being infected first. One measure of pathogenicity, then, would be the proportion of all those infected who felt sick.

What about virulence? The discussion about pathogenicity didn’t specify how sick the person got. Virulence is a measure of the severity of the illness produced by a pathogen. “Pathogenic” means the ability to cause disease. “Virulence” means the ability to cause severe disease. That’s the standard usage. Unfortunately in the bird flu context, it has become standard to talk about “highy pathogenic” avian flu viruses. There are two things to say about this. The first is that the correct term should be “highly virulent,” not “highly pathogenic.” But here, as elsewhere, usage is the final arbiter, so highly pathogenic bird flu viruses are the standard terminology. The second thing to say is that it refers to birds, not humans. A highly pathogenic bird flu virus causes severe disease in birds. It may also do so in humans, but that’s not part of the definition. Virulence is a bit harder to talk about than transmissibility or pathogenicity, because in either of those cases the question is “yes” or “no,” did the virus transmit or did it cause illness. With virulence we’re talking about graded responses of various kinds. Virulence is also a conditional measure, but exactly what is being measured can vary. The most common measure in the flu world is case fatality ratio, the proportion of those infected who die (note the conditional that it is measure only in those that are first infected). This produces other problems, of course, which we’ve discussed here. For example, if you die of a heart attack 35 days after you got the flu, was flu an underlying cause? And of course there is serious outcomes short of death, like winding up on a vent in the ICU for 3 weeks.

What does all this have to do with vaccines? We’ll see in the next post (assuming I actually sit down to write it, which I haven’t yet), that vaccine efficacy is a comparison of outcomes in two groups. Each of transmissibility, pathogenicity and virulence can be used as outcomes. We started this post with a Henny Youngman joke, so we’ll finish with one. Friend: “Hey, Henny. How’s your wife?” Henny: “Compared to what?”


  1. #1 raven
    July 27, 2009

    CDC’s Advisory Committee on Immunization Practices (ACIP) recently rolled out their 2009 Recommendations. It’s for seasonal flu, for which a vaccine exists, not for swine flu, for which there is (as yet) no vaccine.

    May be a case of sticking one’s finger in the wrong dike.

    There is some reason to suspect that swine flu may crowd out seasonal flu .

    1. There is a principle in biology that only 1 species can occupy an ecological niche at one time in the same area.

    2. More to the point, In Argentina, 90% of the flu cases seen are swine flu. Not sure about the other Southern Hemisphere countries.

    IWe will have to wait and see but it could well happen.

  2. #2 Paul
    July 27, 2009

    Now, take my wife… Please!

  3. #3 Ken
    July 28, 2009

    “it turns out for the flu, we aren’t quite sure what the main modes of contagion are”

    Are scientists able to study this for viruses and bacteria? I am generally familiar with exposure assessment methods that are used for chemical exposures (i.e. standard industrial hygiene). I know that there are people who study exposure to endotoxin. I assume people use mathematical models to estimate exposure pathways… But are there actual exposure studies done on the flu?

    You’d think that it would be possible, given how common the flu is, to devise some sort of exposure assessment method, or even have some type of intentional dosing study (given ethical issues are addressed)…

  4. #4 revere
    July 28, 2009

    Ken: It’s possible, but difficult. It’s not just a matter of exposure but of transmission of infection from one host to another. We’ve posted on this quite a few times (search the blog for “transmission”), most recently a study by NIOSH on flu genetic material in the air of emergency rooms. The problem is that detecting genetic material is not the same as viral particles able to replicate. Think about how you would do this for a bit, and you’ll begin to see the problems.

  5. #5 Ken
    July 28, 2009

    Thanks, Revere.

New comments have been disabled.