The prevaccine Hib estimates - Page 4

[runes]

Originally Posted by serenitii

I think the 1 in 1000 risk per year is a mean annual risk. Higher risk in lower years (probably < 3), less risk in older years (4-5).

Check this out (very comprehensive!)
http://www.who.int/vaccines-document...F02/www696.pdf

"Hib meningitis incidence. There were 132 population-based studies with incidence data and 97 hospital-based studies. Eleven of the 132 population-based studies concerned special-risk groups in industrialized countries [Aboriginals (Australia), Alaskan Eskimos (USA), Apache Indians (USA), Keewatin Natives (Canada), and Navajo Indians (USA)]. For the special-risk groups, the mean annual incidence of Hib meningitis in children < 5 years of age was 418.1/100 000, with a range of 34.5 to 530. For the remaining 121 population-based studies, the mean annual incidence of Hib meningitis in children < 5 years of age was 22.8/100 000, with a median of 18.0, and a range 0.9 to 94.6."

and...

"For 74 studies in industrialized countries, the mean annual < 5 year Hib meningitis incidence was 23.9 cases/100 000, with a median of 22.0, and a range of 1.4 to 68.6."

So, to me, this means the highest mean annual risk < 5 (in an industrialized country) would be 68.6 or around 1 in 1500 which would be around 1 in 300 cumulative risk for children < 5

Is this still not accounting for the relative skewing of numbers towards the younger age group, though? IMO that's pretty significant.

[mamakay]

I think the 1 in 1000 risk per year is a mean annual risk. Higher risk in lower years (probably < 3), less risk in older years (4-5).

Right, one in 1,000 in kids under 5. Highest risk in year one (most cases happen in year one) and the risk declines thereafter.

So, to me, this means the highest mean annual risk < 5 (in an industrialized country) would be 68.6 or around 1 in 1500 which would be around 1 in 300 cumulative risk for children < 5

But you still can't multiply that 68.6 per 100,000 (if you want to go with that, 22.0 was the average) by 5 since about 60% (almost 40 per 100,000?) are under age 1.

Look at the middle chart on the right on page 119 here:
http://www.cdc.gov/vaccines/pubs/pin...ds/hib-508.pdf


What's 60% of 68.6? (risk in year one)
30% of 68.6? (risk in year two.)
5% of 68.6 (risk in year three)
3% of 68.8 (risk in year 4)
2% of 68.8 (risk in year five)

You should add them like that for your cumulative per 100,000 incidence. That number is going to be a lot lower than the one in 300 you get by multiplying the total under 5 incidence by 5.

[mamakay]

Ok...is this about correct?

The risk in year one of life was 41.16 per 100,000
The risk in year two was 20.58 per 100,000
year 3 was 3.4 per 100,000
year 4 was 2 per 100,000
year 5 was 1.4 per 100,000

[luminesce]

Originally Posted by kidspiration

Is this still not accounting for the relative skewing of numbers towards the younger age group, though? IMO that's pretty significant.

The way I understand it is that a mean annual risk would account for the relative risk across the age groups.

[luminesce]

Originally Posted by mamakay

Right, one in 1,000 in kids under 5. Highest risk in year one (most cases happen in year one) and the risk declines thereafter.



But you still can't multiply that 68.6 per 100,000 (if you want to go with that, 22.0 was the average) by 5 since about 60% (almost 40 per 100,000?) are under age 1.

Look at the middle chart on the right on page 119 here:
http://www.cdc.gov/vaccines/pubs/pin...ds/hib-508.pdf


What's 60% of 68.6? (risk in year one)
30% of 68.6? (risk in year two.)
5% of 68.6 (risk in year three)
3% of 68.8 (risk in year 4)
2% of 68.8 (risk in year five)

You should add them like that for your cumulative per 100,000 incidence. That number is going to be a lot lower than the one in 300 you get by multiplying the total under 5 incidence by 5.

68.6, in this case is the mean annual risk or the middle point between extremes for all 5 years (like finding the mean risk of that middle chart.) See how it jumps up to 180 per 100,000 at what, around six months and then drops off to near nothing at 5 years? I *think* they are calculating a mean annual risk and then using that to calculate a cumulative risk.

I might be misunderstanding it though.

[luminesce]

Originally Posted by mamakay

Ok...is this about correct?

The risk in year one of life was 41.16 per 100,000
The risk in year two was 20.58 per 100,000
year 3 was 3.4 per 100,000
year 4 was 2 per 100,000
year 5 was 1.4 per 100,000

Doing it like this, I think the risk cumulatively would come out to be ~ 1 in 1500. I bet this is probably the "proper" way versus the "quick & dirty" way.

[mamakay]

Originally Posted by serenitii

The way I understand it is that a mean annual risk would account for the relative risk across the age groups.

But it doesn't. The mean annual risk is reflective of the most at-risk age group for the most part (60% happening in kids under 1) and completely excludes the relative risk across age groups from the equation alltogether.

[mamakay]

Originally Posted by serenitii

68.6, in this case is the mean annual risk or the middle point between extremes for all 5 years (like finding the mean risk of that middle chart.) See how it jumps up to 180 per 100,000 at what, around six months and then drops off to near nothing at 5 years? I *think* they are calculating a mean annual risk and then using that to calculate a cumulative risk.

I might be misunderstanding it though.

Here:

"For 74 studies in industrialized countries, the mean annual < 5 year Hib meningitis incidence was 23.9 cases/100 000, with a median of 22.0, and a range of 1.4 to 68.6."

The "mean" is the average that all the studies found.

One study (the one that found the lowest incidence) found 1.4 per 100,000 kids under 5, and another study (the one that found the highest incidence) found 68.6 cases of invasive hib per 100,000 children under 5. The rest of the studies were in the middle somewhere.

They're not accounting for the age-dependent effect at all there.

[cloak]

Originally Posted by kidspiration

I had to read this paragraph several times to try to figure out what they are saying. It's so unclear. Out of the 1247 average cases per year, 272 were in kids 5 or younger. The serotype was known for 76% of this group but then they jump to say that 341 cases were due to type B. I have to assume that the 341 cases are talking about ALL cases, not in children under 5 because that is greater than 100% of all cases in children under 5, and that is a mathematical impossibility.

That paragraph is confusing but here is what they are saying.

From 1996 through 2000, an average of 1,247 invasive
H. influenzae infections per year were reported to CDC in
all age groups (range 1,162–1,398 per year). Of these, an
average of 272 (approximately 22%) per year were among
children younger than 5 years of age. Serotype was known
for 76% of the invasive cases in this age group.
Three-hundred forty-one (average of 68 cases per year) were
due to type b.

From 1996-2000 (5 years total) there was an average of 1247 cases of all Hi per year. So 6235 cases of Hi total. Of those, 272 per year, or 272*5 years = 1360 cases of Hi in the under 5 age group. 341 of those 1360 cases of Hi in the under 5 age group were due to type b.

Still reading through the thread...

[mamakay]

Ok...this method of estimating cumulative rate was proposed by someone named Day, in 1987. It's generally used with cancer.

But this is the important part...read this:

http://books.google.com/books?id=m39...um=1&ct=result

[Deborah]

So the CDC screwed up, because the risk of Hib is not uniform across the first 5 years of life but varies year by year.

[luminesce]

Originally Posted by Deborah

So the CDC screwed up, because the risk of Hib is not uniform across the first 5 years of life but varies year by year.

The question is - how do you get a reliable risk factor for a pre-vaccine era when reporting was poor and the disease wasn't notifiable? Also, I think that in order to get a mean annual incidence rate you have to know the population for that age group. Isn't census data reported in five year brackets? Maybe they calculated the mean annual incidence rate for 0-5 because they didn't know how many 1,2,3,4 and 5 year olds there were (even if they did know (sorta) the # of cases by age?) -- then used that to calculate the cumulative rate. The mean does somewhat take into account the relative risk for all years so it probably isn't a horrible way to do it given the crappy data.

The problem for me is that the data were not that good for HiB in the pre-vaccination era at all, so regardless of the exact method of calculating incidence and risk rates, it probably isn't all that realistic now - especially now in the post-vaccine era where H. Influenzae likely mutates, colonizes, and exists differently in highly vaccinated (and unvaccinated) populations.

ETA: One thing that I haven't seen discussed often regarding risk is the risk in the vaccinated versus the unvaccinated population for each disease. Many people commonly cite "There were only 24 cases of HiB in the US in x year - what are you chances of getting it?" To me, that isn't a completely logical way to analyze it. If there are 1 million vaccinated children and 10,000 unvaccinated children and in those 24 cases, 12 kids were vaccinated (failure, partial vaccination, etc.) and 12 kids were unvaccinated, then isn't your risk 12 in 1 million or 12 in 10,000? for me, the only way to get around this is to try to find prevaccine risk rates -- but then those lose reliability as you get further and further into the post-vaccine era.

[runes]

Originally Posted by serenitii

ETA: One thing that I haven't seen discussed often regarding risk is the risk in the vaccinated versus the unvaccinated population for each disease. Many people commonly cite "There were only 24 cases of HiB in the US in x year - what are you chances of getting it?" To me, that isn't a completely logical way to analyze it. If there are 1 million vaccinated children and 10,000 unvaccinated children and in those 24 cases, 12 kids were vaccinated (failure, partial vaccination, etc.) and 12 kids were unvaccinated, then isn't your risk 12 in 1 million or 12 in 10,000? for me, the only way to get around this is to try to find prevaccine risk rates -- but then those lose reliability as you get further and further into the post-vaccine era.

Also, we usually are exposed to and build immunity to Hib at some point during early childhood (or, we DID in the pre-vax era). Very often, completely asymptomatically. How does this figure in the numbers at all?

[mamakay]

Maybe they calculated the mean annual incidence rate for 0-5 because they didn't know how many 1,2,3,4 and 5 year olds there were

No, because the US birth cohorts hold relatively steady each year, so the "per 100,000 kids" will be equally distributed in each year of the 1-5 age group.

The mean does somewhat take into account the relative risk for all years

But it doesn't. When they say "mean" they're talking about variations between different studies in different years, not the different incidences in different age groups.

ETA: One thing that I haven't seen discussed often regarding risk is the risk in the vaccinated versus the unvaccinated population for each disease. Many people commonly cite "There were only 24 cases of HiB in the US in x year - what are you chances of getting it?" To me, that isn't a completely logical way to analyze it. If there are 1 million vaccinated children and 10,000 unvaccinated children and in those 24 cases, 12 kids were vaccinated (failure, partial vaccination, etc.) and 12 kids were unvaccinated, then isn't your risk 12 in 1 million or 12 in 10,000? for me, the only way to get around this is to try to find prevaccine risk rates -- but then those lose reliability as you get further and further into the post-vaccine era.

I see what you mean and you're right. BUT...the Hib vax does prevent carriage, and Hib circulation is drastically reduced now in the post-vax era.
So the figues mean a new mom can probably take a while to study the issue and it's not like death by Hib is about to strike at any moment (especially if your breastfeeding).
FWIW, there are probably over a million "non-immune" kids (17 and under) in the US according to the pink book. The only problem with the pink book data is that kids who have had one or 2 shots of whatever are counted as "unvaccinated" when they might have partial protection from their one or 2 shots.

[Deborah]

bumping to help out on another hib thread.