Effect Measure

Two days ago I went with my daughter to the pediatrician to check out her 20 month old who had a fever and rash. Viral origin, probably. Also an ear infection. Pretty much par for the course at this time of year. But lots of little ones and their older sibs weren’t so lucky this flu season. As we’ve had too many occasions to mention, the severity of the 2009 pandemic has yet to be gauged, but trying to compare it to seasonal flu is misleading as its epidemiology is very different. Nowhere is this seen more clearly than in the melancholy figures for pediatric deaths.

Since the beginning of September CDC has registered 265 flu deaths in children under the age of 18. Here’s how that compares with past seasons:

Source: CDC

This is pretty dramatic, even more so when we look at the distribution within the pediatric age group. 48 deaths were in babies and toddlers (less than 2 years old), 30 in children 2 – 4 years old, 98 in the 5 – 11 year age group and 89 in pre-teens and teens (12 – 17 years old). Thus well over two-thirds of the mortality is in children over 5 years old.

For a parent or a grandparent these are chilling numbers, but they are only numbers. The late epidemiologist Irving Selikoff once referred to a statistic as “people with the tears wiped away.” A friend, referring to someone who lost a child, shook his head and just said, “It’s off the scale.”

I look at my little grandson and I look at that graph and all I can think is: Yes. Off the scale.

Comments

#1 Toronto
March 6, 2010

We had the good fortune to have a flu researcher from Mount Sinai hospital in Toronto give a guest lecture in my undergraduate course last week on this very subject. She said basically (I’m paraphrasing) that the reason graphs such as this are so striking (and so highly quoted in the media) is that the amount of yearly deaths from flu in younger people is so low to begin with that percentage increases in times of even mild pandemics such as this can appear quite high. Likewise, the elderly are still the most affected groups but are less affected than in seasonal flu periods.
#2 revere
March 6, 2010

Toronto: The point is that these graphs point to a very big difference in the epidemiology of this virus — dramatic shifts to the left in age distribution. This is hallmark of pandemic viruses. While pediatric deaths from flu are fairly low in the US, they aren’t low in the developing world so this may be a very significant effect there (we don’t have the data to say at this point and the data may not exist). Moreover it isn’t just a numbers game. Pediatric deaths affect policy and public perception much more than deaths of the elderly. So a doubling or tripling in pediatric deaths are newsworthy for more reasons than just selling newspapers. If you go to the link where the graph came from you will see a panel of age specific graphs that compare the different age groups and you will see the dramatic differences between this year and previous years in the surveillance data. The rates in the oldest age group are roughly comparable to the less severe flu seasons of earlier years and much less than the fairly bad flu season of 2007 – 2008.
#3 momkat
March 6, 2010

Losing my grandmother at the age of 88 (not to flu) was nothing compared to the heartbreak of the loss of my infant daughter. We aren’t supposed to lose our children like this.
#4 Pablo
March 6, 2010

She said basically (I’m paraphrasing) that the reason graphs such as this are so striking (and so highly quoted in the media) is that the amount of yearly deaths from flu in younger people is so low to begin with that percentage increases in times of even mild pandemics such as this can appear quite high.

I’m sure that’s a big consolation for all those parents who have lost children.
#5 revere
March 6, 2010

Toronto: I suspect she wasn’t an epidemiologist but a lab person. In any event, the underlying questions — how do you measusre effect — is well known in epidemiology. If you are making a point about relative risk (ratios), then small denominators can look very big with only small absolute increases. To some extent that’s the situation with pediatric deaths if one were to talk about the percent increase over other years, but the idea of this graph is to illustrate the big difference in the epidemiology.

But as I said, it’s not just numbers. The violent death of 2900 people isn’t much compared to how many people die of firearms or motor vehicle accidents every year, but those deaths in 9/11 loomed large. Similarly 5 anthrax deaths isn’t much either, even considering the expected number os close to zero. But it was still a big deal and occupied the newspapers. On the other hand, the converse of her argument is something that increases cancer by 1%. That’s about 2 people a day that wouldn’t have died of cancer but for that exposure but we don’t read about it in the paper because we only see things that cause increases 100 times that level. So I’m not very impressed by what you report about her dismissive attitude to numbers like this. When it comes to epidemiology as with other things, a little knowledge can be a dangerous thing.
#6 Jody Lanard M.D.
March 6, 2010

I certainly agree that people are right to be more upset about child deaths than about adult deaths – and we are right to be even more upset about easily-preventable child deaths.

But I think people are smart enough to understand both halves of the relative risk data:

1. Children have been significantly more at risk of dying from pandemic flu than from seasonal flu, AND
2. Adults are more at risk of dying from pandemic flu than children are.

The two graphs showing these data should be shown side by side: The graph in your post above, which compares absolute numbers of child deaths over several flu seasons, and graphs like this, which show age-specific death rates from pandemic flu 2009, as of the time when the vaccination campaign was ramping up.

Even though people 50-64 with underlying conditions were prioritized for pandemic vaccine, no public service announcements were specifically aimed at that group during the Fall 2009 pandemic vaccine campaign. The most prominent “get your vaccine” materials emphasized children, young adults, and pregnant women.

And even though people 65 and older were at about the same risk of dying from pandemic flu as adults 25-49, no one 65 and older was prioritized for pandemic vaccine during the early months of the vaccination campaign. They were not told, “This is because we value young people more.” They were told, “This is because older people are much less affected by pandemic flu” – true compared with seasonal flu, but false compared with the relative risk of death from pandemic flu.

I fully support efforts to persuade parents to get their kids vaccinated against influenza. But I have a big problem with giving “older adults” the misimpression that they weren’t really at very much risk this year.
#7 MS, MT(ASCP)
March 6, 2010

I recently showed this type of data in a lecture I was giving about Influenza A biology/pathology and about the pandemic strain. The numbers only tell part of the story. We don’t know the pathology of H1N1 (2009) in children vs. middle ages adults, nor do we know what co-morbidities were or were not present in the pediatric deaths. There was significant mortality in those with asthma and metabolic syndrome co-morbidities.

A study by Vaillant et al (Eurosurveillance, 2009; 14(33)pii=19309) published in July, 2009 indicates that of the deaths attributed to H1N1 (2009), the 0-9 year group (47 out of 448 surveyed cases) had significantly fewer deaths than the 20-49 year group (235/448). The means of collecting the data was not entirely clear, and it represents the worldwide epidemiology of the disease up to about mid July, 2009.

Since diseases do not follow human expectations, the pathology may have changed to reflect changes in transmission or interaction with other seasonal diseases. The data from various sources do not match up too well, and I suspect that it is because we do not have a good handle on exact cases and deaths and because of a possible change in the disease pathology.

What is clear from the CDC graph is that the H1N1 (2009) pandemic has markedly altered the flu season. We should be seeing a peak in flu morbidity and mortality in the next week (week 10 of the year). Instead, we saw the peak way back around week 41 of 2009 and now have just background numbers of flu cases by any strain. That is a very big tail wagging the dog for such a mongrel of a flu A strain.
#8 cmc
March 6, 2010

“a flu researcher from Mount Sinai hospital in Toronto give a guest lecture” ==> “I suspect she wasn’t an epidemiologist but a lab person.”

With all due respect Revere, how often do universities invite lab workers to give guest lectures? On the basis of a bit of paraphrased info from a ugrad, you assume that a woman is not an epidemiologists? . . . though this assumption is so ubiquitous in medicine that I do not know why I am protesting. . .
#9 revere
March 6, 2010

cmc: I said this because I know both the epidemiologists at Mt. Sinai and the flu people and the flu people are mainly in the lab and the epidemiologists in chronic disease. So it is based on more information than was apparent.
#10 Marilyn Mann
March 6, 2010

Revere, do you know your twitter account is suspended? I tried to go to it (was hoping to retweet this post) and got the following message: “Sorry, the profile you were trying to view has been suspended due to strange activity.”
#11 cmc
March 6, 2010

Revere: Feathers now in unruffled position. Thanks.
#12 revere
March 6, 2010

Marilyn: Yes. I deleted it after getting hacked twice.
#13 pft
March 6, 2010

It’s too bad that infectious disease exists, but viruses and bacteria do exist in the real world.

I read somewhere that flu deaths in children were often accompanied by a bacterial infection. Solution, more vaccination, and not earlier treatment with antibiotics (even those vaccinated can come down with flu)

There were 152 pediatric deaths in 2003-2004 (a fairly severe year of seasonal flu) as of May 31 (2 months left in the season). Thats also data from only 40 states. I imagine all 50 states were reporting for 2009-2010 given the massive reporting on H1N1.

http://www.cdc.gov/flu/weekly/weeklyarchives2003-2004/03-04summary.htm

You have to wonder how accurate the numbers are though. In other years, perhaps some of these deaths get attributed to some pre-existing disease. Changes in the guidelines (started in 2009) on how to report deaths may have more impact than changing disease patterns.

http://www.cdc.gov/H1N1flu/hospitalreporting.htm

Doing a search on adverse effects of flu vaccines on children does not reveal much. More transparency in this area would be helpful. The lack of evidence does not mean anything, it could be that funding to research such issues is simply not granted, for the public good of course (and those who produce vaccines).
#14 SusanC
March 9, 2010

I wrote about pediatric deaths early in the pandemic. Pediatric Deaths from H1N1 in the US – April to August 2009. Even back then, the age pattern (81% older than 5) was strikingly different from pediatric deaths in seasonal flu. There was also a very high proportion (61%) with neurodevelopmental disorders, in that early cohort. Don’t know if that pattern persisted (I’ve been offline for a couple of months). Would appreciate if anyone can point me to new data. Thanks!!
#15 SusanC
March 9, 2010

momkat,

Losing my grandmother at the age of 88 (not to flu) was nothing compared to the heartbreak of the loss of my infant daughter. We aren’t supposed to lose our children like this.

As someone who raised a child through one childhood cancer only to lose her at 16 to a second unrelated one, I have to say that there are no words to describe that space, the depths that you plunge. I learned that there are no bottoms to hit. Just as you think things couldn’t get any worse, it can and sometimes it does.

The upside of that is I now live with few illusions, a somewhat painful but mostly liberating kind of clarity about life. I also learned that courage is not something that you either have or you don’t (like brown hair), but it’s something that exists in the deepest layers of our being, covered up in the mundanity of ‘normal’ life, that you didn’t know you had until you are called upon to confront the intolerable. And then, because your children/family needed you, somehow you dig and dig and you bring it to the fore, time after time.

I’m a survivor of child bereavement. Was/am I brave? No, I just had no choice.

We are 2 generations away from when losing a child was a ‘normal’ occurrence. It is an idea so alien to our generation that no parent is ever prepared. It is life-changing, often life-destroying.

Which is why from the public health point of view, pediatric deaths should have a special place in every policy consideration…
#16 anon
March 13, 2010

265 deaths is not much.

When comparing with seasonal flu, maybe
chart the estimated number of life-years destroyed
to account for the ~50 vs. ~80 average age of deaths.