Effect Measure

Yesterday DemFromCT had another in his continuing series at DailyKos on Flu and You (Part VIII). He extended an earlier post (part II) on a critical piece of public health infrastructure, laboratory surveillance. One of the graphics is this chart of influenza positive tests reported to the CDC by the WHO/NREVSS collaborating laboratories:

What you see in this chart is a weekly record of what seasonal influenza types and subtypes are circulating in the community (influenza A/H1N1, A/H3N2, B; swine flu makes a late appearance, far right). Flu seasons differ on dominant subtypes, whether they are well matched to the vaccine, and which ones are circulating where and when. These data are important for health services demand and determining the composition of the next year’s vaccine.

When laboratories can’t match a virus against the seasonal varieties — when it isn’t “typable” — CDC looks to see if it is a “novel” virus. Their own laboratories can often type specimens that are “untypable” by less sophisticated labs, but sometimes this turns up a virus that is truly new. That’s what happened with swine flu (see our posts here and here for more details on how it turned up).

Data like the one in this chart don’t just happen. They are the product of a multicomponent surveillance system, in this case the part that collects information from 80 laboratories that are part of a network of state and some county public health laboratories and large academic medical centers integrated into a World Health Organization (WHO) network of Collaborating Laboratories; and another 70 labs part of CDC’s National Respiratory and Enteric Virus Surveillance System (NREVSS). NREVSS is collects information on other respiratory and enteric viruses, but has integrated its laboratories into the CDC influenza surveillance system.

At any one time many viruses circulate that can cause respiratory syndromes indistinguishable from influenza, but when flu season starts the proportion of specimens that test as flu A or B starts to rise. Each week the 150 or so participating labs send CDC the total number of specimens they tested and the proportion that are positive for influenza A or B. The identity of the other viruses may or may not be known, depending on the virus and the laboratory (NREVSS is a surveillance system for some of them). Because of the concern over avian influenza, most of the US 80 WHO Collaborating labs have now been trained to determine influenza A seasonal subtypes (H1, H3), information they report to CDC along with the patient’s age. Most of the other 70 NREVSS labs don’t have that ability. That’s why the chart shows many specimens as “untyped” (not the same as “untypable”). The data from these 150 laboratories form the basis for the weekly report of the total number of positive influenza tests, by virus type/subtype, and the percent of specimens testing positive for influenza. Some of the specimens from the WHO collaborating labs are also sent to a specialized CDC lab for gene sequencing, antiviral resistance testing and specific strain determination. Information from the laboratory surveillance is also used in some methods to estimate the number of excess deaths due to influenza.

We’ll discuss some other components of the influenza surveillance system in subsequent posts. The salient point DemFromCT has made so well is that the laboratory infrastructure is a critical part of the general public health system. Surveillance is one of those unsexy, silent parts of ;ublic health that doesn’t get much support or attention until you discover you really, really need the data.

If you don’t have the infrastructure by then, it’s too late.

Comments

#1 River the Writer
May 11, 2009

It appears in the graph that there was a huge surge in all present flu types in week 17. Any idea why? Was this due to the viruses becoming more active or something about the way the testing was done?
#2 revere
May 11, 2009

River: This points up the difficulty. It could be there was an upsurge in actual flu, for reasons we don’t know (it seems to include both seasonal and swine flu). But it could just as easily be that the interest in swine flu resulted in many more swabs being sent to the lab. It may be a little of each or yet some other bias. Since we tend not to do as much surveillance outside of flu seasons, it is also possible that there is always a lot of flu around we don’t count.
#3 neil
May 11, 2009

I’d imagine, that with the heightened awareness on ILI and reporting caused by the Novel H1N1 Swine Influenza, that week 17 would show a higher hit rate on the “normal” seasonal virii.

Put a scare in the collection agencies, and they’ll respond by more dilligently collecting samples of any ILI that comes in..,

Many more samples = many more positives.
#4 phytosleuth
May 11, 2009

The last report on the CDC site is a month ago, April 11, Week 14. This seems a bit tardy.
#5 neil
May 11, 2009

Phytosleuth:
You can find results for week 15 by manually entering a URL for the report or for charts in the report. You just can’t get the full report from their normal web site source..,

It looks like Week 15 showed the “normal” continuation of the end of Seasonal Flu data.

However, you can’t find anything on week 16 through either the website or manually creating a URL to find the resource. It seems in all the ruckus they “lost” week 16.

Now we see the result of the NOVEL virus in the universally locatable PDF report:

http://www.cdc.gov/h1n1flu/pubs/pdf/FluView_Week17.pdf
#6 James Unland
May 11, 2009

Toward the top of his post DemFromCt quoted Dr. Anne Schuchat partially. That quote is from my question during the news conference and the full exchange is as follows, on the issue of possibly placing/enhancing ‘testing labs’ in various places like Mexico, to hopefully shorten the ‘outbreak-to-confirm’ time:

>> Operator: NEXT QUESTION, JAMES UNLAND WITH EDITOR OF “HEALTH BUSINESS AND POLICY” YOUR LINE IS OPEN.
>> THANK YOU VERY MUCH FOR TAKING MY QUESTION. MY QUESTION GOES TO SOMETHING THAT CAME OUT IN AN INTERVIEW WITH DR. CAROLYN BRIDGES THE OTHER DAY. SHE REALLY INDICATED THAT THE TIME LAPSE BETWEEN AN OUTBREAK OF SOME KIND AND CONFIRMATION IS CRUCIAL, TRYING TO NARROW THAT TIME LAPSE. IS CDC OR THE U.S. GOVERNMENT CONSIDERING GIVING COUNTRIES LIKE MEXICO MORE FULL SCALE LABS IS SO WE CAN SHORTEN AND NARROW THAT TIME LAG, SO TO SPEAK? BECAUSE IT SEEMS LIKE DURING THAT TIME LAG PEOPLE ARE HIGHLY CONTAGIOUS BUT NONSYMPTOMATIC AND EACH DAY THAT GOES BY IS CRUCIAL. THANK YOU FOR TAKING MY QUESTION.
>> YOU KNOW, OUR INTERNATIONAL STRATEGY HAS BEEN VITAL TO OUR PANDEMIC PREPAREDNESS EFFORT. THE CDC HAS BEEN WORKING WITH THE WORLD HEALTH ORGANIZATION IN A NUMBER OF MINISTRIES OF HEALTH AROUND THE WORLD TO STRENGTHEN THEIR PANDEMIC PREPAREDNESS AND IN SOME OF THE WORLD TO STRENGTHEN THEIR EFFORTS AT BEING READY FOR AVIAN INFLUENZA. THIS PANDEMIC PREPAREDNESS EFFORTS INVOLVED INVESTMENTS IN LABORATORY CAPACITY, INVESTMENTS IN EPIDEMIOLOGIC CAPACITY, RAPID RESPONSE TEAMS TO BE ABLE TO RESPOND TO CLUSTERS OF UNUSUAL ILLNESS, AND WE THINK THOSE ARE REALLY VITALLY IMPORTANT IN SHORTENING THE TIME BETWEEN WHEN A PROBLEM HAPPENS AND WHEN IT’S OF INVESTIGATIVE AND RESPONDED TO, SO THAT THE LABORATORY CAPACITY FOR INFLUENZA IS JUST CENTRAL TO THE RECOGNITION OF A NEW VIRUS, AND IT IS A BIG PART OF OUR INVESTMENT. AS WE’VE SAID ON PREVIOUS CALLS, ONE OF THE ISSUES THAT HAS BEEN PART OF OUR PARTNERSHIP WITH MEXICO HAS BEEN STRENGTHENING THEIR LABORATORY CAPACITY FOR THIS PARTICULAR NEW VIRUS, SO THAT THEY’D BE ABLE TO DETECT THIS STRAIN IN THE BACKGROUND OF ALL THE OTHER RESPIRATORY VIRUSES THAT ARE OUT THERE. WE HAVE DONE IN THE CONTEXT OF THIS RESPONSE, OF COURSE, WE SHIFT THE NEW REAGENT, THE TEST KITS TO DOZENS OF COUNTRIES AROUND THE WORLD, AND WE’VE MADE THIS A PART OF THE W.H.O. PARTNERSHIP REALLY, SO I THINK YOUR QUESTION IS SPOT-ON, BECAUSE WHETHER IT’S THIS H1N1 OR VIRUS OR SOMETHING ELSE, HAVING STRONG LABORATORY CAPACITY AROUND THE WORLD IS VERY IMPORTANT.
#7 DemFromCT
May 11, 2009

Thanks, James. Good question and interesting answer (it struck me as such real time when I heard you ask it.) Several times over the course of the last few days CDC has referred to lab capacity.

The UCLA high throughput lab would be available in theory for testing from all over the world, but from the US as well. Volume has precluded some testing from being as rapid as we would wish, and whereas one way to help is to “send the lab where the illness is”, another way is to improve capacity in the lab where samples are sent.

I know, for example, that the state labs have been running weekends to keep up, and personnel are getting pretty weary. In a bigger outbreak, increased capacity for CDC through something like what i wrote about would be a big help, and Mexican samples could be sent there.

Ideally, there’d be that kind of lab duplicated in several WHO regions.
#8 swampus
May 11, 2009

revere, apologies for posting off topic, but I was wondering if you would be interested in cross-posting your “Call a Canadian” entry on Daily Kos.

I find this idea very appealing and would like to see it discussed in a political arena.
#9 Valerie Brown
May 11, 2009

I don’t know whether this is off-topic or not, but:

Why do Illinois (487) and Wisconsin (384) have the most flu cases, as of 5/11? Because so many college students from that region went to Mexico on spring break and/or engaged in higher-risk behavior than other college students in Mexico? Probably something completely different – better surveillance? too much cheese?
#10 revere
May 11, 2009

Valerie: It could be reporting differences and lag time for reporting lab results (likely). But remember, flu doesn’t happen evenly at all places at once. Some places may be differentially affected, for reasons we don’t always understand, or they may “catch up” at a later time. Best not to place too much emphasis on these differences at this point.
#11 David Riley
May 12, 2009

What’s up in Panama peeps?
#12 Pork Flu
May 12, 2009

“US, Michigan: State officials confirm 131 cases of influenza A (H1N1); only high-risk cases will be tested.” What does high-risk cases mean? So they are not testing everyone with H1N1, only those about to die?
#13 anon
May 12, 2009

the number of tests and thus the number of positives may have gone up due to Mexflu
but the percentage of positives in seasonal flu ???
It went from 8% to ~15% in the graph.
Only ~20% of the positives were Mexflu, so 12% should have been seasonal flu,
well up from 8%.
Or have they changed to testing only really suspectible cases now, other than in week 16 ?
#14 revere
May 12, 2009

Pork: High risk usually means over 65, pregnant or with an underlying medical condition placing them at higher risk for complications of flu.

anon: The percent positive varies throughout the year and depends on what other viruses that cause ILI are circulating. We haven’t been doing much surveillance in the off seasons so the yield on rapid flu tests for ILI isn’t known. These data indicate there is probably more seasonal flu around even when we have assumed it wasn’t and it accounts for a higher percent of the total ILI than we thought.
#15 Jason Dick
May 12, 2009

That week 17 result looks really, really odd to me. The upsurge I might understand as being a result of an upswing from worries related to the pandemic. But why on Earth are all of the various categories represented almost equally? It just seems so…artificial, like there is an error somewhere in the data analysis.
#16 River the writer
May 12, 2009

Revere, you wrote, “These data indicate there is probably more seasonal flu around even when we have assumed it wasn’t and it accounts for a higher percent of the total ILI than we thought.” If that is the case, and the actual flu cases (of all present types) was significantly higher than appeared in the bell curve indicating the “flu season,” how can you be sure of the flu season’s end? I ask because much has been made of the 3 out of 4 pandemic flus striking “out of season.” Have they really, or has the flu season simply continued and was ongoing when the pandemic virus hit?
#17 Racter
May 12, 2009

Jason, I was thinking the same thing. At the end of a season strongly dominated by H1N1, hey presto, suddenly H3N2 and seasonal H1N1 are neck and neck!
#18 revere
May 12, 2009

River: The failure to keep surveillance going during the “off season” is a defect in the surveillance system. I think that is being remedied. But there is no doubt that flu is a strongly seasonal disease. The big question is where it is outside of “flu season.” Some alternatives: in another reservoir (e.g., birds); in the southern hemisphere and returns the next year; still around but at much lower levels. This isn’t settled. I’d vote for the third and I think these data support that. Trying to disentangle a reporting effect here is problematic so how to interpret these figures is up in the air at the moment, IMO.
#19 River the writer
May 12, 2009

Thanks, Revere. I appreciate your response.

Hats off to you and the other excellent Reveres here and elsewhere working on understanding flu viruses. The number of variable you must contend with is simply staggering!

Be well!
#20 CSplat
May 12, 2009

Speaking of laboratory surveillance; does it strike anyone else as odd that the subtype of infection found in TN has not been made public?

Comments like:
“This particular strain is confined to the animals only and is not
spread to people,” says Mr Whitt. “The public is not in any danger.”
(04May2009)

Don’t pass my ‘sniff’ test…

Read more at:
http://promedmail.org/pls/otn/f?p=2400:1001:1758141932873759::NO::F2400_P1001_BACK_PAGE,F2400_P1001_PUB_MAIL_ID:1000,77352

and:
http://promedmail.org/pls/otn/f?p=2400:1001:1758141932873759::NO::F2400_P1001_BACK_PAGE,F2400_P1001_PUB_MAIL_ID:1000,77455
#21 anon
May 17, 2009

maybe they improved their testing and samples
which would have been negative in week 15 are positive now

maybe people and doctors are better aware of the
symptoms now than in week 15

maybe their testing capacity is somehow limited or
they have a limited budget for testing so they are more restrictive now and filter the matching symptoms better