Facts, not Fantasy

Amazon Contextual Product Ads

Sunday, October 11, 2009

Why the epidemiology of swine flu matters

From ScienceBlogs:

If you are hesitating to be vaccinated for swine flu this year, perhaps this post will help you make up your mind. If it does, I hope it pushes you to get vaccinated, but whatever persuasion we attempt here will only be from a recital of what we know of the epidemiology of this pandemic. Because it is the different epidemiology that is the main feature, not the clinical characteristics or the virulence of the virus. So far this looks pretty much like a standard influenza A virus -- except for the epidemiology. Since I'm an epidemiologist, you might expect me to think this is important, and I do. Epidemiology is the public health science that studies the patterns of illness in populations. One kind of pattern we study is who is getting sick. And it is a change in this pattern that is one of the big differences between a pandemic strain and a seasonal strain.

Pandemic strains have a greater tendency to infect and make sicker much younger victims. In seasonal influenza it is the over 65 age group that contributes most of the serious illness and deaths, but with pandemic strains (not just the current one), lack of immunity in the population makes those under age 65 a bigger target and they sicken and die proportionately more than in a non-pandemic season. And that's exactly what we are seeing this year. The story that made the headlines on Friday was that 19 more pediatric deaths were added to the pediatric death toll in week 39, although not all of these children died in that week. The dates of death for the 19 stretched between July 19 and October 3 because of the way the tally is done. Starting in the 2003 - 2004 flu season deaths from influenza in people below the age of 18 became a nationally notifiable disease, reported to CDC through state epidemiologists. Each state has its own way of ascertaining the number. Some states are more complete and more timely than others, so notifications that come in a particular week can, and do, represent deaths that occurred over varying time periods, as in the instance of the 19 added this week. Still, it is clear that the young population is being hit particularly hard in comparison to the last three years for which we have data. 76 have died from swine flu since it made its first appearance in April, and 29 of those deaths have come since August 30, i.e., 29 in 5 weeks and those 5 weeks are extremely early in the flu season. In fact they occurred before the official administrative flu season even started (week 40).

The pattern of pediatric deaths and its difference from previous years is dramatically shown in this graph from CDC's weekly flu report:


The green bars are pediatric deaths from seasonal flu. You can see that in green bar terms, this flu season, which was not as bad as the previous year, was pretty typical. But this year, unlike previous years, there was a second flu season that started just as the usual one was finishing (the middle hump). We remarked on this back in July. Then there was a lull (we don't know why) and now we are into flu season and we see a third set of bars.

If you aren't in the pediatric population, here are some more patterns that may help convince you that getting vaccinated is a good idea. This comes from the Emerging Infections Program (EIP), another part of the multicomponent CDC flu surveillance system. These are lab confirmed influenza related hospitalizations in 60 counties in 12 metro areas of 10 states (San Francisco CA, Denver CO, New Haven CT, Atlanta GA, Baltimore MD, Minneapolis/St. Paul MN, Albuquerque NM, Las Cruces, NM, Albany NY, Rochester NY, Portland OR, and Nashville TN). It's a passive surveillance system where the data are obtained by record review of lab and hospital admissions databases and infection control logs. Each panel is an age group, with the top being babies and toddlers and the bottom one people over age 65 (you can see the original here if this is too small).


These are cumulative incidence rates. Think of them, at each week (the horizontal axis), as the chance that you will have gotten the flu by that time this year, i.e., if you look at week 37 the height of the curve reflects the probability that by week 37 you would have gotten flu at sometime before that (not necessarily that week). Obviously the curve can only go up, since it is cumulating (adding onto the pile of previously lab diagnosed hospitalized cases that have occurred up to that week). If you wonder how high it "should" go, based on the average of the last three flu seasons (October to April of each year) that level is the dashed horizontal line in each panel (NB: the scales are different for the top and bottom panels, because those are the age groups, the very young and the old, that usually have the highest risks so the curves wouldn't fit if the vertical scales of the other age groups was used). What this means for interpreting the curves is that if this year behaved like the average of the three previous years, the curve would slowly move upward until it reached the dashed line by the end of the flu season in April or May. Obviously this year is very different.

You once again see the altered age pattern, this time very dramatic when comparing the youngest age group with the oldest. For those of us in the over 65 age group, the seasonal has barely begun. We are just inching our way up from typical summer levels. There's a long way to go before we get to where we would usually be by the end of flu season. My age group looks pretty normal for this time of year. But if you look at the other panels you see that several have already exceeded, in the first week of the official flu season, the level of season risk we would have expected by the end of the season. Babies and toddlers are two thirds of the way there already (remember the scale is different for them so the same height is a higher risk than compared to the 2 to 4 year olds). The 2 to 4 year olds are already there and everyone between 5 and 49 years old has already exceeded their year end risk at a time when the season is usually just starting. The 50-44 year old group is already at seasonal average and then there's the over 65 age group, the exception that proves the rule: this is a pandemic strain.

Where and when it will peak we don't know but there's a long, long way until whatever causes flu to be seasonal is ended (usually sometime in April or May). Swine flu could burn itself out before then (but not before taking more children and adults with it); or it could keep going for the whole flu season; or it could co-circulate with the seasonal flu H1N1 and H3N2 strains. The assumption that swine flu will completely crowd out seasonal flu strains is premature and we could have a normal seasonal flu pattern in the over 65 age group appearing later. Or not. We still don't understand flu well enough to predict with any confidence what is going to happen.

The seasonal and swine flu vaccines are not intrinsically different except for the flu strains they contain. We change flu strains every couple of years routinely and these vaccines are made in the same way as we have been making them for many years. The only difference is a strain change, which is routine. This means that we have had extensive experience with the swine flu vaccine already, extending over years and hundreds of millions of delivered doses. It is not untested. Far from it.

The most rational thing to do at this moment, given what we know and don't know, is to get vaccinated with both seasonal flu and swine flu vaccines. That's what I'm going to do. I already got my seasonal flu shot and I'll wait in line for my turn for the swine flu vaccine and get it as soon as I can. You not only protect yourself but you help prevent spread of flu to others.

No comments:

Post a Comment

Please keep posts here respectful. Those that cross boundaries will be deleted, and then placed in a special place for future ridicule.